Perinatal HIV Infection and Fetal Exposure to Antiretrovirals

Infection acquired through vertical transmission of the human immunodeficiency virus, now preventable in more than 99% of treated pregnancies, whose consequences in the newborn and in the HIV-exposed uninfected child enter the differential diagnosis of genetic syndromes and drug-induced embryofetopathies.

Literature and regulatory information updated: 29 October 2026.

0. Initial Reference Table

1. Introduction

Perinatal infection with the human immunodeficiency virus is an acquired infection of interest to clinical genetics for three distinct and complementary reasons. The first is that it shares with embryofetopathies and several genetic syndromes a repertoire of signs that bring the child to the clinical genetics clinic: intracranial calcification, microcephaly, failure to thrive, hepatosplenomegaly and developmental delay. The second is that prevention of vertical transmission exposes the fetus, throughout gestation, to drugs with teratogenic potential and with mitochondrial toxicity demonstrated in experimental models, which places the exposed child in the same field of assessment as fetal exposure to valproate or alcohol. The third is that host susceptibility to infection and to its rate of progression depends on germline variants in chemokine co-receptor genes and in the major histocompatibility complex, providing one of the most cited examples of genetic protection against an infectious agent.

The founding landmark of prevention is the Pediatric AIDS Clinical Trials Group protocol 076 trial, published by Connor and colleagues in 1994 [5]. It was a randomised, double-blind trial in pregnant women between 14 and 34 weeks of gestation, with a CD4 T-lymphocyte count above 200 cells per cubic millimetre and no antiretroviral therapy in the current pregnancy. The experimental regimen combined maternal oral zidovudine, 100 mg five times daily until the onset of labour, intrapartum intravenous zidovudine at 2 mg/kg over one hour followed by a continuous infusion of 1 mg/kg per hour until cord clamping, and oral zidovudine in the newborn, 2 mg/kg every six hours for six weeks. Between April 1991 and December 1993, 477 pregnant women were enrolled; infection status was known in 363 newborns. The proportion infected at 18 months, estimated by the Kaplan-Meier method, was 8.3% (95% confidence interval: 3.9 to 12.8) in the zidovudine group and 25.5% (18.4 to 32.5) in the placebo group, corresponding to a relative risk reduction of 67.5% (40.7 to 82.1). Haemoglobin at birth was lower in those exposed to zidovudine, but values equalised by 12 weeks of age.

The effect of this trial on practice and regulation was immediate. The indication for prevention of maternal–fetal transmission was incorporated into the zidovudine label in the United States through an efficacy supplement approved on 8 August 1994, on top of an original authorisation of 19 March 1987 [6, 7]. In Portugal, the same indication appears in the summary of product characteristics of Retrovir oral solution, authorised on 8 April 1993 [8].

The following decade replaced monotherapy with combination therapy and shifted the focus from the drug to maternal viral load. The Women and Infants Transmission Study, a prospective North American cohort of 1542 women and deliveries between 1990 and 2000, showed a clear gradient: transmission was 20.0% (16.1 to 23.9) without any prenatal antiretroviral, 10.4% (8.2 to 12.6) with zidovudine monotherapy, 3.8% (1.1 to 6.5) with dual regimens and 1.2% (0 to 2.5) with highly active combination therapy; transmission rose from 1.0% with a viral load below 400 copies/mL to 23.4% above 30,000 copies/mL, with a 2.4-fold increase in odds for each log10 of viral load at delivery [9]. In the French perinatal cohort, with 8075 mother–child pairs between 2000 and 2011, there was no transmission among 2651 children whose mothers started therapy before conception, maintained it throughout pregnancy and had a viral load below 50 copies/mL at delivery, with an upper confidence limit of 0.1%; transmission increased progressively with later initiation of therapy, from 0.2% before conception to 2.2% in the third trimester [10]. In the United Kingdom and Ireland, the rate fell from 2.1% in 2000–2001 to 0.46% (0.21 to 0.86) in 2010–2011 [11].

Nomenclature has followed this transformation. The historical term «maternal–fetal transmission» was progressively replaced by «vertical transmission» and, in the most recent documents, by «perinatal and postnatal transmission», to explicitly include breastfeeding, which now accounts for more than half of new paediatric infections worldwide [12]. The term «newborn prophylaxis» has likewise been split: the US guidelines distinguish antiretroviral prophylaxis, given to an uninfected newborn to reduce the risk of acquiring infection, from presumptive therapy, a three-drug regimen given to high-risk newborns that functions simultaneously as early treatment of an as yet undocumented infection and as enhanced prophylaxis [4]. European terminology retains the acronym PNP, for postnatal prophylaxis, with two pathways defined by risk [3].

Three recent developments define the current state of the problem. The first is the recognition that sustained maternal virological suppression reduces the risk of transmission through breast milk to less than 1%, but not to zero, which led high-income countries to move from a recommendation of exclusive formula feeding to a shared decision-making model: Switzerland in 2018, Australia in 2021, Canada in 2023, the United Kingdom in 2025 and the United States in the 2023 to 2026 updates [3, 4]. The second is the dolutegravir safety signal: an interim analysis of the Tsepamo study, in Botswana, described in 2018 and 2019 a prevalence of neural tube defects of 0.30% with dolutegravir at conception versus 0.10% with other antiretrovirals [13, 14], a signal that extended surveillance up to 2022, the parallel programme in Eswatini and data from US registries did not confirm, with prevalences of 0.11% and 0.098% and no significant difference [3, 15, 16]. The third is the certification by the World Health Organization, on 18 December 2025, of the elimination of vertical transmission of HIV in Brazil, the first country with more than one hundred million inhabitants to meet the criteria of transmission below 2% and coverage above 95% for antenatal care, testing and treatment [17].

This article therefore covers three linked entities. The first is perinatal infection proper, with its three moments of acquisition and the determinants of risk. The second is the HIV-exposed uninfected child, a rapidly growing population with its own morbidity. The third is fetal and neonatal exposure to antiretrovirals, with its profile of teratogenicity, mitochondrial toxicity and obstetric outcomes. Each has distinct relevance for clinical genetics, and all three share the same set of clinical decisions in the antenatal period, at delivery and in the first year of life.

2. Molecular Genetics and Pathophysiology

2.1. Nature of the agent and mechanism of cell entry

HIV-1 is a lentivirus with an RNA genome which, in order to enter host cells, requires expression of the CD4 receptor and of a fusion co-receptor. Macrophage-tropic strains, designated R5, use CCR5 and, to a lesser extent, CCR2; strains adapted to T-lymphocyte lines, designated X4, use CXCR4 and CX3CR1 [18]. The viral glycoprotein gp120 additionally makes use of intermediate receptors, such as DC-SIGN, encoded by CD209, on dendritic cells, and the syndecans on endothelial cells. The natural ligands of these co-receptors — CCL5, CCL3, CCL3L1 and CCL4 for CCR5, and CXCL12 for CXCR4 — compete with the virus and can block entry, so that variants in those ligands alter both susceptibility to infection and the rate of progression.

The ex vivo model of intraepithelial infection in human vagina developed by Hladik and colleagues in 2007 showed that the virus penetrates intraepithelial Langerhans cells and CD4-positive T lymphocytes simultaneously: in T lymphocytes entry is almost exclusively by direct fusion mediated by CD4 and CCR5, with immediate productive infection, whereas in Langerhans cells it occurs by endocytosis through multiple receptors, with persistence of intact virions in the cytoplasm for several days [18]. This difference explains why mucosal exposure during passage through the birth canal constitutes a moment of particularly high risk, and why approaches to mucosal prevention must act on more than one route of entry.

2.2. The three moments of vertical transmission and their determinants

Transmission occurs in three distinct time windows, with different mechanisms and relative weights. Intrauterine transmission results from transplacental passage of the virus or of infected maternal cells and increases with gestational age. There are reports of identification of the virus in fetal tissue at 8, 15 and 20 weeks, and in a study of second-trimester fetal thymuses the polymerase chain reaction detected HIV-1 in 2 of 100 samples; the US panel adopts 20 weeks as a conservative gestational threshold below which intrauterine transmission is unlikely. A model built from the timing of initiation of antiretrovirals estimated that 3% of transmissions occur before 14 weeks, 17% between 14 and 36 weeks, 50% between 36 weeks and the onset of labour and 30% in the intrapartum period [4].

Intrapartum transmission carries the greatest risk and derives from exposure of the newborn to maternal blood and genital secretions during labour and passage through the birth canal. Postnatal transmission occurs through breast milk and, in the absence of maternal therapy and infant prophylaxis, accumulates a risk of 15 to 20% over two years of breastfeeding [4]. The mechanisms of transmission through breastfeeding are not fully understood: maternal plasma viral load correlates with free viral RNA in milk, but cell-associated viral DNA may persist in milk despite plasma suppression, although it may not be replication-competent [4].

The single most important determinant, across the three windows, is maternal viral load. A meta-analysis cited by the US panel estimates transmission at 0.2% or less when maternal RNA remains below 50 copies/mL; in the Canadian Perinatal Surveillance Program, with 4743 mother–child pairs between 1997 and 2020, 0.5% of newborns of mothers with a viral load between 50 and less than 1000 copies/mL in the four weeks before delivery acquired the infection [4]. The duration of effective therapy before delivery and its time of initiation are independent determinants, as the French cohort and the European cohorts have shown [3, 10].

2.3. Mechanisms of pharmacological prevention

Maternal therapy acts by three distinct routes, and understanding them explains why no single component is sufficient. The first is the reduction of viral load in blood and genital secretions, which decreases the quantity of virus to which the fetus and the newborn are exposed. The second is pre-exposure prophylaxis of the fetus: the drugs given to the mother cross the placenta and produce in the fetus concentrations capable of inhibiting viral replication at the moment of greatest exposure, that of delivery. All the preferred nucleoside reverse transcriptase inhibitors, as well as dolutegravir and raltegravir, have high transplacental passage [4]. The third is the post-exposure prophylaxis conferred by the drugs given to the newborn, which protects against free or cell-associated virus that has entered the fetal circulation during delivery. Indirect demonstration of the importance of the latter two components is the lower efficacy of interventions confined to the intrapartum period or to a single neonatal dose, when compared with regimens that include postnatal antiretrovirals [4].

Data on transplacental passage quantify this rationale. Zidovudine crosses the placenta rapidly, with cord blood to maternal plasma concentration ratios of about 0.80 and an amniotic fluid to maternal plasma ratio of 1.5 [4]. For dolutegravir, an ex vivo perfusion model showed a mean fetal/maternal ratio of 0.6 and two in vivo pharmacokinetic studies gave median cord blood to maternal plasma ratios of 1.21 and 1.25 [4]. A single maternal dose of nevirapine reaches an effective concentration in the neonate within two hours, which is maintained for up to ten days, a property used when the newborn cannot receive medication by the oral route [3].

2.4. Host genetics: the CCR5Δ32 allele

The CCR5 gene is located at 3p21.31, has the identifier HGNC:1606 and encodes a G protein-coupled receptor for the inflammatory chemokines CCL3, CCL4 and CCL5, expressed on T lymphocytes and macrophages, where it directs lymphocyte migration to sites of infection. With CD4, it is the main co-receptor for R5-tropic HIV-1 [19]. The best-known null allele is a 32 base pair deletion that causes a frameshift and gives rise to a non-functional receptor, incapable of supporting membrane fusion or infection by macrophage-tropic and dual-tropic strains.

Standardisation of this variant requires attention to nomenclature. The MANE Select transcript is NM_001394783.1 and, in that transcript, the validated description is c.554_585del, with predicted protein consequence NP_001381712.1:p.(Ser185IlefsTer32); the standardised genomic description in GRCh38 is NC_000003.12:g.46373456_46373487del [20]. The corresponding ClinVar record is VCV000008184.6, which retains the title NM_001394783.1(CCR5):c.554_585del (p.Ser185fs), links the identifier rs333 and the reference OMIM 601373.0001, and classifies the germline variant as benign, with criteria provided by a single submitter and without a specified condition [21]. The coordinates are chr3:46,373,453–46,373,484 in GRCh38 and chr3:46,414,944–46,414,975 in GRCh37. The older literature uses different designations for the same deletion: Liu and colleagues, in 1996, described it as a deletion of nucleotides 794 to 825 of their own complementary DNA numbering, corresponding to codons 175 to 185, with a frameshift after amino acid 174, inclusion of 31 novel amino acids and truncation at codon 206; Ansari-Lari and colleagues, in 1997, placed the frameshift at codon 185, with loss of 168 amino acids and gain of 31 residues at the carboxyl terminus [22]. It should be noted that GeneGuidance and part of the pharmacological literature continue to cite the old transcript NM_000579, and that the abbreviated forms «CCR5Δ32» and «CKR5Δ32» are historical nomenclature, not HGVS descriptions.

The allele frequency of the deletion is about 0.092 in populations of European ancestry and about 0.10 in the European American population of the United States, and it was absent from the samples tested from western and central Africa and from Japan [19, 22]. Approximately 1% of people of western European origin are homozygous and about 20% heterozygous. The protective effect was established in 1996 by two independent groups [22, 23] and quantified by Dean and colleagues: in six AIDS cohorts that together comprised 1955 people, the 17 homozygotes for the deletion were all among the 612 exposed seronegative individuals and none among the 1343 infected; heterozygotes progressed more slowly to AIDS [19, 24]. The protection does not extend to viruses that use CXCR4. From the standpoint of counselling, this is a modifier of the natural history of the infection and not a diagnosis: there is no established Mendelian disease for CCR5, ClinGen has no gene-disease validity or actionability assertion for this gene, and all the associated OMIM entries are susceptibility or resistance phenotypes recorded in braces [19].

2.5. Other host modifiers and the variant with a perinatal effect

OMIM entry 609423 brings together, under a susceptibility and resistance phenotype, a broad set of loci: IL10 at 1q32.1, CXCR1 at 2q35 (slow progression), CX3CR1 at 3p22.2 (rapid progression), CCR2 and CCR5 at 3p21.31, TLR3 at 4q35.1 (resistance), HLA-C at 6p21.33 (viraemia), CXCL12 at 10q11.21, IFNG at 12q15 (rapid progression), CCL3L1 at 17q11.2, CCL2, CCL11 and CCL5 at 17q12 and CD209 at 19p13.2 [18]. The immunogenetic axis is particularly robust: the allele groups HLA-B35 and HLA-B53 are associated with a worse prognosis and higher viral loads, whereas HLA-B57 and HLA-B27 are associated with protection; class I heterozygosity confers an advantage, and Carrington and colleagues attributed in 1999 the prolonged survival of 28 to 40% of Caucasian patients who avoided AIDS for ten years or more to complete heterozygosity at the class I loci, to the absence of the B35 and Cw04 alleles, or to both. A genome-wide association study published by Fellay and colleagues in 2007 identified polymorphisms that explain almost 15% of the interindividual variation in viral load during the asymptomatic steady-state period, one of them associated with HLA-B*5701 and another close to HLA-C. Thomas and colleagues showed in 2009, with 1698 patients of European ancestry, that the −35C allele of the HLA-C promoter is an indicator of high cell surface expression and is associated with better control of viraemia. Ramsuran and colleagues, analysing 9763 individuals from 21 cohorts in 2018, showed that higher levels of HLA-A confer worse control, by providing a signal peptide that determines the expression of HLA-E, the ligand of the inhibitory receptor NKG2A on natural killer cells [18]. The copy number of CCL3L1, which varies by segmental duplication at 17q, modulates susceptibility not through its absolute value but by being below the population-specific mean, and is higher in Africans (Gonzalez and colleagues, 2005). Homozygosity for the APOE4 allele was associated with accelerated progression in a European and African American cohort (Burt and colleagues, 2008) [18].

For the subject of this article, the most relevant variant is not Δ32 but the one listed as OMIM 601373.0007, entitled precisely «increased perinatal transmission of HIV type 1». This is a C to T polymorphism at position 59356 of the CCR5 promoter (rs41469351). Kostrikis and colleagues described in 1999, in African American infants, an association between homozygosity for the 59356T allele and a relative risk of 5.9 of perinatal transmission; the frequency of that allele was significantly higher in African Americans (21%) than in Hispanic (6%) or Caucasian (3%) populations, whereas the variants associated with lower perinatal transmission, the Δ32 deletion and the 59402G allele, were less common in African Americans [22]. This reading is indirect, made through the summary of the OMIM entry, and the primary study was not read in this review; there is no recommendation for host genotyping in any perinatal guideline consulted, and the tropism testing that determines the use of maraviroc is a test of the virus and not of the person's genotype [19].

2.6. Pathophysiology of established infection in the infant and of the reservoir

When transmission occurs, infection in the infant differs from that in the adult by very high viral loads and by more rapid progression, with depletion of CD4 T lymphocytes in an immature immune system. The latent viral reservoir is established early, but its size depends on the timing of initiation of therapy: very early and effective treatment restricts the reservoir, reduces viral genetic variability and modifies the immune response [4]. The practical consequence of this property is twofold. On the one hand, it justifies immediate presumptive therapy in newborns at high risk. On the other, it led to research into treatment-free remission, whose most important result is that of the IMPAACT P1115 study, discussed in section 7, and whose inaugural case is that of the Mississippi child, described in section 10. Restriction of the reservoir by very early treatment may, paradoxically, hinder definitive diagnosis: in a Cape Town study, total viral DNA became undetectable in infants treated very early on the sixth day, at three months and at four months of life [4].

Exposure to antiretrovirals adds a pathophysiological mechanism of its own. Nucleoside reverse transcriptase inhibitors inhibit, to a degree that varies with the affinity of each drug, mitochondrial DNA polymerase gamma, interfering with the replication of mitochondrial DNA and potentially producing depletion and dysfunction of the mitochondrion [4]. This action explains the existence of acquired mitochondrial DNA depletion syndromes, described in adults treated with azidothymidine for HIV-1 infection and with fialuridine for hepatitis B, in which the myopathy with depletion induced by zidovudine reverses on withdrawal of the drug in animal models; there is also documentation of an infant with HIV infection treated with zidovudine, didanosine and a protease inhibitor who had secondary mitochondrial DNA depletion with near-fatal metabolic acidosis and liver failure, with a 79% reduction in muscle mitochondrial DNA that normalised after therapy was stopped [25]. The inaugural description of mitochondrial dysfunction in exposed uninfected children was French: Blanche and colleagues identified, in 1999, eight children without HIV-1 infection with abnormally low activities, absolute or relative, of complexes I, IV or both of the respiratory chain, months or years after the end of therapy — five with late neurological symptoms, two of whom died, and three asymptomatic but with severe biological or neurological abnormalities; four had been exposed to zidovudine with lamivudine and four to zidovudine alone, and none had known mutations causing constitutional disease. The authors considered that the findings supported a link between mitochondrial dysfunction and the perinatal administration of nucleoside analogues, but recommended maintaining the recommendations then in force for zidovudine monotherapy [26]. Later studies in the United States and in Europe did not corroborate this French observation, and in the PACTG 219/219C trial small changes in mitochondrial DNA and in the enzyme activities of oxidative phosphorylation were documented in stored samples, of unknown clinical significance [4]. The pathophysiology of the exposed uninfected child is discussed in detail in section 3 and includes, besides this mitochondrial axis, disturbances of cellular immunity and of monocyte activation attributable to intrauterine exposure to HIV and to an altered maternal immune system [4].

3. Clinical Spectrum and Manifestations by System

3.1. Risk of acquisition by time window and by viral load

The clinical spectrum of this condition begins before any sign: it begins with the probability that the child becomes infected or not. The US guidelines set out that probability in a stratification table defining intrauterine and intrapartum risk from maternal viral load in three periods of gestation [4].

Viraemia may be documented in the laboratory or presumed from clinical features, such as a new diagnosis, discontinuation of therapy before delivery or a report of poor adherence over several weeks. The choice of the 50 copies/mL threshold to define high risk does not follow from a biological effect demonstrated at that exact value, but from the reasoning that low-level viraemia may be a marker of higher undetected viraemia, and that any newborn at increased risk should receive the potential benefit of robust prophylaxis [4]. Viraemia between 50 and 200 copies/mL does not, on available data, confer high intrauterine risk, but is associated with a greater probability of later maternal virological failure.

3.2. Natural history of untreated infection

The natural history of perinatally acquired paediatric infection, now rarely seen in countries with effective antenatal care, is characterised by faster progression than in adults. The timing of the first positive virological result distinguishes two forms: a positive result within 48 hours of life indicates intrauterine infection, with a reservoir already established and usually more rapid progression; a negative result in the first week followed by positivity, in a child who is not breastfed, indicates intrapartum infection [4]. In the pre-therapy era, Pneumocystis jirovecii pneumonia, progressive encephalopathy and recurrent invasive bacterial infections dominated the picture in the first months of life. The clinical justification that persists from this natural history is prophylaxis with trimethoprim-sulfamethoxazole from 4 to 6 weeks of age in any infant whose status remains indeterminate, because of the high mortality and morbidity of P. jirovecii pneumonia in the infected child [3, 4].

Overlap with genetic syndromes arises precisely in this setting. Intracranial calcification is common, although not invariable, in congenital infections, and has been described after cytomegalovirus, rubella, Zika, toxoplasma, parvovirus, varicella, herpes, lymphocytic choriomeningitis virus and HIV; diagnosis is straightforward when there is active neonatal infection identifiable by serology or by polymerase chain reaction, but becomes difficult after the neonatal period, and it is then the radiological phenotype that allows genetic entities to be recognised [27].

3.3. Treated infection: a picture almost without manifestations

Under antiretroviral therapy started early, the clinical spectrum of the infected child approaches normality, with preserved growth and development and undetectable viral load. The case described in section 10 illustrates this transformation in extreme form: the child maintained normal growth and development and a CD4 T-lymphocyte percentage within or above the normal range for age at every assessed time point, even after interruption of therapy [28]. The distinction between natural history and treated disease is, in this article, sharper than in any genetic disease: the same agent produces, depending on when treatment starts, either rapidly progressive disease or a clinically silent picture.

3.4. The HIV-exposed uninfected child

The population of exposed uninfected children grows as prevention becomes effective, and it has its own morbidity. The US panel is explicit about the state of the evidence: available data do not allow definitive conclusions as to whether intrauterine exposure to HIV and to antiretrovirals affects immune function, infectious morbidity, growth, cardiometabolic health, neurodevelopment, mitochondrial function or cancer risk, from childhood into adulthood; ongoing long-term studies include the PACTG Late Outcomes Study and the SMARTT study, from the PHACS cohort [4].

In terms of mortality and overall morbidity, the risk is higher in exposed uninfected children than in unexposed children, and the difference is more pronounced in low- and middle-income countries than in high-income ones. In Botswana, the strongest predictors of mortality at 24 months were exposure status and formula feeding. A meta-analysis of studies comparing exposed uninfected infants and children with unexposed infants and children found higher all-cause mortality in the exposed [29]. In a Canadian longitudinal cohort between 1988 and 2015, exposed children had higher lifetime hospitalisation rates, with added risk in those born small for gestational age and in those whose mothers had a detectable viral load at delivery; by contrast, a Zambian cohort followed in the first six months found no excess hospitalisation or mortality [4].

The infectious axis is the most consistent. A meta-analysis estimated a 50% increase in the risk of diarrhoea and a 70% increase in the risk of pneumonia in the first six months of life, compared with unexposed children. Studies in Malawi and South Africa described more lower respiratory tract infections, although others in South Africa and Botswana did not document excess infectious morbidity. The French perinatal cohort observed an increased risk of severe bacterial infections by encapsulated organisms in the children of women with low CD4 T-lymphocyte counts close to delivery. A retrospective Spanish study of 195,941 infants aged 90 days or less, between 2008 and 2017, found a sevenfold higher risk of group B streptococcal infection and a twenty-nine-fold higher risk of meningitis by that organism in exposed children. In the United States and Canada, early hospitalisation rates were higher, with a possible contribution from respiratory syncytial virus and parainfluenza. Immunological phenotyping suggests that intrauterine exposure is associated more with disturbances of CD4 and CD8 T-lymphocyte cellular responses, with T-cell dysfunction and altered vaccine responses, than with defects of the humoral response, and there are data on increased monocyte activation and pro-inflammatory responses with underexpression of genes of neutrophil-mediated immunity [4]. This evidence underpins the UK recommendation to ensure complete and timely vaccination and to alert primary and secondary care professionals to the possibility of increased infectious risk when assessing an exposed child [3].

In a first period of the Tsepamo study, between 2014 and 2016, with 47,027 births in Botswana, adverse birth outcomes were more frequent in HIV-exposed than in unexposed infants, at 39.6% versus 28.9% and an adjusted relative risk of 1.40 (1.36 to 1.44), and the regimen of tenofovir disoproxil, emtricitabine and efavirenz from conception was associated with fewer adverse outcomes than regimens with nevirapine or with ritonavir-boosted lopinavir [30].

For growth, the pattern diverges by socioeconomic setting. Studies in South Africa, Zambia, Malawi and Uganda, and a systematic review with pooled analysis of 28 studies between 1989 and 2022, 25 of them African and 23 from the combination therapy era, documented persistently lower weight- and length-for-age z scores and more stunting in exposed children; one explanatory hypothesis is disturbance of the growth hormone axis, and raised markers of inflammation and of intestinal microbial translocation are associated with poorer growth and poorer neurodevelopment. An Ethiopian study linked maternal therapy at conception to more stunting, an association that a study in Malawi did not confirm; in Kenya, with universal maternal therapy, growth trajectories were similar to those of unexposed children. In a large Danish study of postnatal growth up to five years there were no differences in weight after two weeks of life or in length after six months, compared with a matched group. The SMARTT study, in the United States, described above-average weight relative to the general paediatric population, high rates of obesity in exposed children and adolescents and, among obese exposed children, a greater risk of systolic and diastolic hypertension than among obese children from the general population; a South African cohort that assessed body composition, glucose metabolism, lipids and blood pressure at 5 to 8 years detected no differences [4].

For neurodevelopment, the results are inconclusive and the heterogeneity of populations and designs complicates interpretation. Several studies found no differences in early development; others described poorer performance. A review of three studies found a higher prevalence of psychiatric disorders, associated with socioeconomic level, stigma and psychosocial stress. In Kenya, exposed children had significantly lower mean z scores for global cognitive ability, short-term and delayed memory, attention and processing speed, even after adjustment for nutritional status, household food security and orphanhood. In a South African cohort, impaired language development was observed across the preschool years together with thicker prefrontal cortices; along the same lines, smaller caudate and total grey matter volumes were described in the first weeks of life, associated with maternal immunosuppression. In the SMARTT study, intrauterine exposure to efavirenz was associated with a greater risk of microcephaly, and exposed children with microcephaly had lower mean scores and a higher prevalence of neurodevelopmental impairment at one and at five years of age than exposed children without microcephaly; there is, however, no definitive evidence linking intrauterine exposure to specific antiretrovirals with poorer neurodevelopment [4].

As for cancer risk, there is transplacental genotoxicity of nucleoside analogues described in macaques and micronucleated erythrocytes identified in exposed infants. The French perinatal cohort described 21 cancers among 15,163 uninfected children exposed to at least one nucleoside inhibitor, at a median age of 9.9 years. A US study based on state health service registries, with 13,617 exposed children followed for a median of 9.3 years and a maximum of 20, found a borderline risk of brain tumour, based on six cases, with no increase in the risk of leukaemia; didanosine, no longer recommended, was the potentially associated drug. In another US study, four cancer diagnoses among 3087 exposed children did not differ significantly from the number expected from national reference rates [4].

3.5. Pregnancy outcomes under antiretroviral therapy

Pregnancy affected by HIV infection carries, independently of the drugs, an increased risk of adverse neonatal outcomes: preterm birth before 37 weeks, low birth weight below 2500 g, small-for-gestational-age newborn with weight below the tenth centile, and stillbirth. The US panel warns of the central methodological difficulty: comparing treated with untreated pregnancies is inadequate, because the latter carry increased risk from the state of immunosuppression, and comparing treated pregnancies with uninfected pregnancies confounds the drug effect with the viral effect; the proposed approach is comparative safety between regimens [4].

In that comparison between regimens, most meta-analyses do not demonstrate a significant association between antiretrovirals and preterm birth. A meta-analysis of 61 observational studies, with 409,781 participants, found no difference in the overall risk of preterm birth between treated and untreated women, but treated women had less spontaneous preterm birth, with a relative risk of 0.46 (0.32 to 0.67). In a British Columbia study that excluded multiple pregnancies and outdated regimens, women with HIV infection had twice the risk of preterm birth compared with the general population, but treated women had less spontaneous preterm birth than untreated women, with a cumulative protective effect for each week of therapy, and neither preconception use nor first-trimester use was associated with preterm birth. The integrase inhibitor class does not increase the risk; in the VESTED trial, preterm birth was more frequent with efavirenz, tenofovir disoproxil and emtricitabine than with dolutegravir, tenofovir alafenamide and emtricitabine (12% versus 6%). Protease inhibitors, especially ritonavir-boosted, are the class most frequently implicated: a meta-analysis of ten studies estimated an adjusted odds ratio of 1.32 (1.04 to 1.6) and a network meta-analysis of seven trials found, for zidovudine, lamivudine and lopinavir/ritonavir compared with zidovudine, lamivudine and abacavir, a relative risk of spontaneous preterm birth of 1.81 (1.21 to 2.71). When a protease inhibitor is necessary, the panel recommends darunavir with ritonavir rather than lopinavir with ritonavir. A meta-analysis of 24 observational studies, with 38,293 pregnancies in 15 countries, described a greater risk of preterm birth when therapy was started before pregnancy compared with initiation during pregnancy, with a relative risk of 1.16 (1.03 to 1.31), an association attenuated in some multivariable analyses; the recommendation is explicit that therapy should not be stopped or delayed to prevent preterm birth [4].

For birth weight, the data are mixed. A meta-analysis of 12 studies with 40,495 participants found no excess of low birth weight in treated women. For the small-for-gestational-age newborn, a meta-analysis of five studies with 6818 participants found a relative risk of 1.38 (1.09 to 1.75) in treated women, and regimens with protease inhibitors were associated with more small-for-gestational-age infants than regimens without inhibitors (relative risk 1.24; 1.08 to 1.43) and than regimens with non-nucleoside inhibitors (1.28; 1.09 to 1.51), although a systematic review of 28 studies between 1997 and 2020 found no significant association. The practical recommendation is to consider enhanced fetal growth surveillance in the third trimester, especially with boosted protease inhibitors [4]. As for stillbirth, a rare outcome in high-resource settings, regimens with tenofovir disoproxil were associated with lower risk (pooled relative risk 0.6; 0.43 to 0.84) and protease inhibitors showed no excess. Data on hypertensive disorders of pregnancy are limited and contradictory, and the guidance is not to avoid or discontinue therapy for this reason [4].

4. Clinical Recognition and Warning Signs

4.1. Identifying exposure: screening is the true diagnostic test

In this condition, early clinical recognition does not come from observing signs in the child, but from identifying maternal serological status before or during delivery. That inversion is the most important feature of the reasoning. The root cause analysis carried out after the rise in transmissions in three referral centres in Maryland in 2022, following a year without any infection, identified missed opportunities for prevention: late entry into perinatal care, difficulties with maternal adherence particularly in the context of substance use, and failure to diagnose the mother in good time [4]. In England, enhanced ISOSS surveillance showed that two thirds of the 156 children with vertically acquired infection born between 2006 and 2021 were born to women whose infection was not diagnosed before or during labour, because of seroconversion during pregnancy or breastfeeding, declined screening or absent antenatal care; among women with a known diagnosis, poor adherence, late entry into care and undisclosed breastfeeding contributed, and in at least seven cases no contributing factor was identified. More than half of the 48 transmissions reported since 2014 occurred in the children of women with concurrent safeguarding, mental health, substance use or insecure housing problems [3].

Screening recommendations reflect this epidemiology. HIV testing is recommended for all sexually active people and should form part of the preconception consultation; in every pregnancy it should be done as early as possible on an opt-out basis, that is, included in the usual panel unless the person declines; and partners of unknown status should be referred for testing. Repeat testing in the third trimester is recommended when the initial test is negative and there is increased risk of acquisition, which includes people cared for in institutions with an incidence of one or more cases per thousand pregnant women per year, residents of jurisdictions with an incidence above 17 per 100,000 women aged 15 to 45 years, and jurisdictions where it is required by law. Testing is also repeated when a sexually transmitted infection arises in pregnancy, when there are signs or symptoms of acute infection or when exposure is ongoing; if the result is negative, pre-exposure prophylaxis is indicated. Some clinicians perform the third-trimester test at around 28 weeks, together with syphilis testing, to limit the number of venepunctures and to allow timely treatment and prophylaxis of congenital syphilis, and some add a third test on admission for delivery [4].

4.2. Warning signs and emergency situations

The situations requiring decisions within hours, rather than days, are well defined. The first is the woman in labour with undocumented serological status, or with a negative test in early pregnancy but increased risk and no repeat test in the third trimester: in these cases a rapid combined antigen/antibody test should be performed, available 24 hours a day in every unit with a maternity or neonatal intensive care unit, with a result in less than one hour. If the result is positive, intravenous zidovudine is started immediately and, in parallel, an HIV-1/HIV-2 antibody differentiation test and an HIV-1 RNA assay are requested. When acute or recent infection is suspected, the RNA assay should be requested at the same time as the combined test, because the phase of high viraemia that precedes seroconversion is precisely the one with the greatest risk of transmission [4].

The second is a positive result obtained during labour or postpartum. In that situation maternal differentiation testing and RNA assay are performed, a nucleic acid amplification test is obtained in the newborn and presumptive therapy appropriate to high risk is started immediately. When maternal results are not available at birth, the newborn is tested with a rapid antibody test to identify exposure; if positive, an amplification test is obtained in the infant and standard maternal testing is offered as rapidly as possible. If there is an intention to breastfeed in this situation, the recommendation is not to breastfeed and to express and store milk until all supplementary tests have been reviewed and are negative [4].

The third emergency is the appearance of maternal viraemia during breastfeeding. A detectable viral load, at or above 50 copies/mL, requires immediate, at least temporary, cessation of breastfeeding and the start of replacement feeding, with repeat viral load, assessment of the cause and reinforced adherence counselling; most experts recommend definitive cessation when the viral load reaches 200 copies/mL or more [4]. UK guidance is equivalent and adds a request for resistance testing on the rebound sample, as it is relevant to the infant's treatment should transmission have occurred [3].

The fourth is the identification of unplanned breastfeeding. In the Brazilian protocol, where breastfeeding is contraindicated, detection of breastfeeding at any point during follow-up requires it to be stopped and a viral load to be requested for the newborn [31].

4.3. When to suspect in a clinical genetics setting

Outside the perinatal pathway, three scenarios should raise the possibility. The first is the child with intracranial calcification, microcephaly or developmental delay under investigation for suspected genetic disease, in whom the maternal history is unknown or incomplete. The second is the child in foster care or adopted whose maternal serological status is unknown: the guidelines explicitly recommend HIV testing in these cases, to identify perinatal exposure and possible infection [4]. The third is the child with recurrent invasive bacterial infections, persistent oral candidiasis or failure to thrive, in whom age-appropriate testing is recommended even without documented or suspected exposure [4].

There are two further laboratory interpretation situations that can cause confusion in genetics and immunology settings. The first is late seroreversion: maternal antibodies transferred across the placenta may persist up to 24 months of age, and in one study 14% of exposed uninfected children seroreverted after 18 months, so a positive antibody test at this age requires a nucleic acid amplification test to exclude or confirm infection [4]. The second is the possibility of falsely positive results on amplification tests after chimeric antigen receptor T-cell therapy and after gene therapy with lentiviral vectors, used for example in severe combined immunodeficiency: gammaretroviral and lentiviral vectors may interfere with the amplified long terminal repeat regions, and laboratories should have alternative platforms available for this population [4].

4.4. Newborn screening

There is no population newborn screening for HIV on a dried blood spot card. The functional equivalent is nucleic acid amplification testing in all newborns with known perinatal exposure, discussed in section 5. In the same context, the US guidelines added in 2025 screening for congenital cytomegalovirus within the first 21 days of life by polymerase chain reaction on urine or saliva in all HIV-exposed infants, because the prevalence of congenital cytomegalovirus is higher in this population than in the general population and early diagnosis allows antiviral intervention with ganciclovir or valganciclovir, with benefit for sensorineural deafness; in some US states universal dried blood spot screening exists and is used to complement urine and saliva testing while awaiting further validation [4].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Basic principle: virological tests below 18 months

In infants and children under 18 months with perinatal exposure, diagnosis requires tests that detect the virus directly: quantitative RNA assays, qualitative HIV-1/HIV-2 RNA assays approved by the US regulatory authority, or qualitative DNA tests. Antibody tests and combined antigen and antibody assays should not be used for diagnosis in this age group, because maternal antibodies transferred across the placenta produce positive results that do not indicate infection of the child. Plasma RNA assays and cell-associated DNA assays are generally equally recommended [4]. The only legitimate use of an antibody or combined test in an infant of this age is to document exposure when the mother is not available for testing, for example in an abandoned child, a use that is explicitly off-label [4].

5.2. Recommended testing schedule

The US schedule, revised on 25 June 2026, specifies amplification tests at birth, at 14 to 21 days, at one to two months and at four to six months of age, with an additional test two to six weeks after antiretrovirals are stopped when presumptive therapy was used. Testing at birth should generally be performed in all infants with perinatal exposure, but is not necessary in low-risk infants who will not be breastfed; it is nevertheless performed in low-risk infants when there is an intention to breastfeed or concern about loss to follow-up, because it provides an initial reference point should a positive result appear later. Samples should be obtained before or immediately after antiretrovirals are started, without delaying prophylaxis or presumptive therapy, and never from cord blood, because of the risk of contamination with maternal blood [4].

For breastfed infants, the schedule retains the same time points and adds a test at two to three months when the interval between the one-to-two-month test and the four-to-six-month test exceeds three months, testing at least every three months throughout breastfeeding, and two tests after cessation: one at four to six weeks and another at three months, whatever the age at which breastfeeding ends. Where there has been prolonged prophylaxis during breastfeeding, exclusion requires two negative tests obtained at least two to six weeks after the drugs are stopped and three months after the end of breastfeeding. The last test should be performed three months after cessation of breastfeeding and at least two weeks after the end of prophylaxis [4].

UK guidance defines separate schedules for the three situations and is more detailed as to the type of assay. In high-risk infants it recommends RNA and DNA within the first 24 hours, before discharge, repeat RNA and DNA at two weeks, and RNA (or DNA) at six and at twelve weeks, with an antibody test at 24 months to document seroreversion. In low-risk infants who are not breastfed, a single type of assay suffices at each time point: within the first 24 hours, at four to six weeks (at least two weeks after the end of prophylaxis) and at ten to twelve weeks, with antibodies at 24 months. In low-risk breastfed infants, it recommends testing within the first 24 hours, monthly RNA testing throughout breastfeeding — an interval that, by shared decision-making and provided maternal viral load is monitored monthly, may be extended to a maximum of two months — and tests at four and at eight weeks after cessation, with antibodies at 24 months or, if later, at least eight weeks after the end of breastfeeding [3].

5.3. Sensitivity, specificity and limits of the tests

The sensitivity of virological assays depends critically on timing and on drug exposure. RNA assays identify between 25% and 58% of infected infants from birth to the end of the first week of life, 89% at one month and 90% to 100% at two to three months, with specificity of 100% at birth and at 1, 3 and 6 months. DNA polymerase chain reaction assays identify 20% to 55% in the first week, more than 90% at two to four weeks and 100% at three and at six months, with specificity of 99.8% at birth and 100% at subsequent time points [4]. This difference between the first days and the following weeks is intrinsic: a test at birth detects infection acquired in utero but not infection acquired at delivery or immediately before.

There are four limitations that should be made explicit in the report and in communication with the family. The first is that a negative result in the first days does not exclude infection, for the reasons given above. The second is that antiretroviral therapy, whether maternal via the placenta or administered to the infant, reduces the sensitivity of the assays: in 47 infants with negative DNA at birth, 68% were identified at four to six weeks, during prophylaxis, and all by three months; and in a Cape Town study, in infants treated very early, DNA became undetectable in some cases at day six, at three months and at four months, which may complicate definitive diagnosis [4]. The third is that DNA assays are frequently developed in-house by laboratories, so their sensitivity and specificity may differ from approved commercial assays; UK guidance requires that, when DNA is used, a maternal sample is always obtained for amplification with, or before, the infant's first sample, to confirm that the primers detect the maternal virus, and that the primer set or assay is changed if they do not [3]. The fourth is volumetric: RNA assays typically require 1 mL of plasma and have a shorter turnaround time, whereas DNA assays can be performed on smaller samples but take longer; dilution of paediatric samples of insufficient volume raises the lower limit of detection and reduces assay sensitivity [3].

The question of subtypes has been resolved. The first assays were developed with group M subtype B, predominant in the United States, but current assays detect all HIV-1 subtypes, generally amplifying more than one conserved region of the genome; the update of 25 June 2026 removed from the guidelines all recommendations and all text on non-B subtype HIV-1, as it is no longer a diagnostic concern. If the picture suggests infection but virological tests do not amplify, HIV-2 testing is indicated and, if that is also negative, a novel non-amplifying variant should be considered, with discussion with a specialist [4].

5.4. Criteria for confirmation and exclusion

Confirmation of infection in children under 18 months is based on positive virological test results in two separate blood samples, because false positives occur with both RNA and DNA assays; the confirmatory sample should be obtained before therapy is started, but therapy does not wait for the result, given the low probability of a false positive and the benefit of immediate treatment [4].

Exclusion has two levels. Presumptive exclusion, in a child who is not breastfed, is accepted with two or more negative virological tests, one obtained at two weeks of age or more and another at four weeks or more; or with a negative virological test at eight weeks or more, obtained at least two weeks after multidrug prophylaxis or presumptive therapy was stopped; or with a negative antibody test at six months or more. Definitive exclusion requires two or more negative virological tests, one obtained at one month of age or more and at least two to six weeks after antiretrovirals were stopped, and another at four months or more; alternatively, two negative antibody tests on separate samples obtained at six months or more. Both levels require the absence of any other laboratory or clinical evidence of infection and the absence of breastfeeding. Once the criteria for definitive exclusion are met, and in the absence of known or suspected exposure after birth, no further routine testing of any kind is needed [4]. Even so, UK guidance retains antibody testing at 24 months, because late postnatal infection rarely occurs — through breastfeeding, through food pre-chewed by an infected carer, or through intrafamilial or unknown exposure — even when all early virological tests were negative [3].

A practical aspect of the schedule is that presumptive exclusion allows prophylaxis against Pneumocystis jirovecii pneumonia to be stopped or avoided. With adequate follow-up, most infants with perinatal exposure do not need trimethoprim-sulfamethoxazole, because infection can be presumptively excluded by the end of the postnatal antiretroviral regimen [4].

5.5. Diagnosis of HIV-2

HIV-2 is endemic in Angola, Mozambique and an extensive list of West African countries — Benin, Burkina Faso, Cape Verde, The Gambia, Ghana, Guinea, Guinea-Bissau, Côte d'Ivoire, Liberia, Mali, Mauritania, Niger, Nigeria, São Tomé, Senegal, Sierra Leone and Togo — and in parts of India, and it is well documented in France and in Portugal, countries with large migrant populations from those regions. In the United States it is rare: between 2010 and 2017, more than 99.9% of identified infections were classified as HIV-1 and diagnoses of HIV-2, alone or as coinfection, remained below 0.03% [4]. The clinical relevance is twofold: HIV-2 is intrinsically resistant to non-nucleoside reverse transcriptase inhibitors and to several other drugs developed for HIV-1, and HIV-2 viral load can only be measured in reference laboratories, in the United States the University of Washington and the New York State department of health [4].

HIV-2 should be suspected when infection is linked to an endemic area or when the laboratory pattern is suggestive, that is, a positive initial HIV-1/2 immunoassay with an HIV-1 viral load at or below the limit of detection, in the absence of treatment. The recommendation is to use an HIV-1/HIV-2 antibody differentiation immunoassay as the supplemental test. In the infant exposed to known or suspected HIV-2, HIV-2-specific DNA tests or qualitative HIV-1/HIV-2 RNA assays should be performed at the same time points used for HIV-1; the HIV-2 DNA test is not commercially available and must be arranged through the state or local health department surveillance programme, or through the centres for disease control. Children with HIV-2 may have falsely negative amplification tests with a persistently positive immunoassay [4]. UK guidance adds that samples should be sent to a laboratory that routinely provides HIV-2 testing [3].

5.6. Practical algorithm in the exposed newborn

The algorithm can be summarised in five steps. First, classify risk by maternal viral load in the three periods defined in section 3 and by clinical features of adherence. Second, obtain blood from the infant — never from the cord — for amplification testing before or immediately after the first antiretroviral dose, except in low-risk infants who are not breastfed, in whom testing at birth is dispensable. Third, start the antiretroviral regimen appropriate to the risk within the first six hours of life, according to US guidance, or within the first four hours, according to UK guidance, without waiting for results. Fourth, use the result of the birth test to decide on de-escalation: a negative result excludes intrauterine infection and allows a triple regimen to be reduced to zidovudine alone in selected infants. Fifth, follow the testing schedule corresponding to the risk and the feeding method until the criteria for definitive exclusion are met, or start therapy if a test is positive [3, 4].

5.7. Report content and genetic counselling issues

The report of an amplification test should identify the type of assay (quantitative RNA, qualitative RNA or DNA), the limit of detection applied, the time of collection in relation to birth and to antiretrovirals, and the nature of the sample, explicitly excluding cord blood. When an in-house DNA assay is used, confirmation that the primers amplify the maternal virus should be stated [3]. In cases where host genotyping is performed — a situation not recommended by any perinatal guideline, but which arises in research or reservoir studies — the report should use HGVS nomenclature with transcript and version, for example NM_001394783.1:c.554_585del p.(Ser185IlefsTer32) for the Δ32 deletion, and classify the variant according to its nature as a resistance allele, not as a pathogenic variant [20, 21].

Invasive prenatal diagnosis has no role in the diagnosis of fetal infection: the invasive sampling needed to document intrauterine infection carries a risk that is not ethically justifiable, and it is this impossibility that explains the imprecision of risk estimates by gestational age [4]. UK guidance includes a section on invasive prenatal testing in the usual obstetric context, that is, for genetic indications, and the general position is that screening and prenatal diagnosis of aneuploidies and genetic diseases follow the same rules as in unaffected pregnancies, on condition of maternal virological suppression [3]. Preimplantation genetic diagnosis does not apply to this condition, which is not hereditary.

6. Differential Diagnosis

The differential diagnosis of this condition runs in two opposite directions, and both are of interest to clinical genetics. The first asks whether a picture attributed to congenital infection may in fact be a genetic disease. The second asks whether findings in a child under genetic investigation may be explained by perinatal HIV infection or by exposure to antiretrovirals. The following table brings together the most relevant entities in both directions.

In prose, three differentials deserve development. The first is intracranial calcification. Distinguishing congenital infection from genetic disease can be difficult, because intracranial calcification is common in congenital infections without being obligatory, and because several genetic diseases produce it. When there is active neonatal infection with identification of the agent by serology or polymerase chain reaction, the diagnosis is immediate; after the neonatal period, diagnosis becomes more difficult and it is careful analysis of the radiological phenotype that allows specific entities to be recognised. The historical designation pseudo-TORCH syndrome, now Aicardi-Goutières syndrome, arose precisely from this resemblance, and cystic leukoencephalopathy from RNASET2 variants mimics congenital cytomegalovirus infection particularly closely [27].

The second is the mitochondrial DNA depletion syndromes. There are genetic forms, caused by variants in nuclear genes involved in the replication and repair of mitochondrial DNA, and forms acquired through pharmacological inhibition of DNA polymerase gamma. The acquired forms have been described in adults treated with azidothymidine for HIV-1 infection and with fialuridine for hepatitis B; in animals, myopathy with zidovudine-induced depletion reverses when the drug is stopped. Zidovudine toxicity more closely resembles the non-hepatic, predominantly encephalomyopathic forms, and fialuridine toxicity the hepatic forms. There is a documented case of an infant with HIV infection treated with zidovudine, didanosine and a protease inhibitor who developed secondary depletion with near-fatal metabolic acidosis and hepatic failure, with a 79% reduction in muscle mitochondrial DNA that normalised after antiviral therapy was stopped [25]. Reversibility on stopping the drug is therefore the most useful discriminating feature, although causality is difficult to establish when other diseases coexist.

The third is neural tube defects in a child exposed to dolutegravir. Here the reasoning has to be epidemiological rather than dysmorphological, because the defect is indistinguishable from one of multifactorial cause. The background prevalence in the general population is 0.5 to 1 case per thousand live births, or 0.05% to 0.1% [15], and in Brazil it is estimated at about 0.06% [31]. The extended surveillance data, discussed in section 7, show no difference between exposure to dolutegravir at conception and other regimens. In an individual child, exposure to dolutegravir is not a sufficient explanation and assessment should follow the usual protocol for a neural tube defect, with attention to prognosis: in the follow-up of children with neural tube defects identified in the Tsepamo study between August 2014 and May 2023, 178 of 261,441 deliveries (0.068%) had a defect, 99 of them (55.6%) were stillborn or died before discharge and, among the 79 alive at discharge with follow-up at a median age of five years, 20 died, corresponding to 42% mortality in that group and 67% overall; thirty developed hydrocephalus, 26 of them with a shunt and 14 with shunt complications, and of the 28 children alive at follow-up 20 (71%) were living with significant physical disability and 10 (36%) with intellectual or developmental disability [32]. Assessment should include the history of folic acid supplementation, which is a known modifier of risk and which in a murine model determined the presence or absence of dolutegravir-induced exencephaly [33].

7. Treatments and Risk Management Strategies

7.1. Principles and architecture of prevention

Prevention of vertical transmission rests on three components that act at different moments and whose sum determines the outcome: maternal antiretroviral therapy started as early as possible and ideally before conception, to suppress viral load and provide pre-exposure prophylaxis to the fetus; intrapartum interventions, including intravenous zidovudine and, in defined situations, planned caesarean section; and prophylaxis or presumptive therapy in the newborn, with duration and number of drugs determined by risk. Therapy should be started in every pregnancy, irrespective of viral load and CD4 T-lymphocyte count, and whatever support is needed should be provided to achieve and maintain undetectable suppression before conception, during pregnancy, postpartum and thereafter [4].

7.2. Antiretroviral therapy in pregnancy

When pregnancy occurs in a person already on a fully suppressive regimen, the rule is to continue that regimen, unless it contains a drug not recommended in non-pregnant adults or there are safety concerns or concerns about lower efficacy in pregnancy. Regimen changes should ideally be made before attempting to conceive, so that suppression is established at conception [4].

In people without previous therapy, the preferred dual nucleoside inhibitor backbones are tenofovir alafenamide with emtricitabine (or with lamivudine) and tenofovir disoproxil with emtricitabine (or with lamivudine); both allow once-daily dosing, exist as fixed-dose combinations, have activity against hepatitis B virus and need no dose adjustment in pregnancy. Combined with dolutegravir, tenofovir alafenamide was associated with fewer adverse birth outcomes and a lower risk of insufficient weight gain than tenofovir disoproxil, but also with more emergent obesity in non-pregnant adult women; tenofovir disoproxil has potential renal toxicity and requires caution in renal impairment [4].

Two options from the integrase strand transfer inhibitor class are preferred anchor drugs. The fixed-dose combination of bictegravir with tenofovir alafenamide and emtricitabine became a preferred regimen in the update of 12 June 2025: bictegravir concentrations are lower in the second and third trimesters than outside pregnancy and fall more than those of dolutegravir late in gestation, but remain above the protein binding-adjusted 95% effective concentration. Dolutegravir with a preferred backbone is the other option, with the advantage of lowering viral load rapidly in late presentations and of having, in non-pregnant adults, less emergent class resistance than raltegravir. Neither of these two options is preferred when there is a history of exposure to long-acting injectable cabotegravir as pre-exposure prophylaxis, because of the risk of class resistance mutations, unless genotyping demonstrates the absence of those mutations; in that situation the preferred regimen is ritonavir-boosted darunavir [4].

Among alternative regimens, the update of 31 March 2026 classified dolutegravir with lamivudine as an alternative in pregnancy, under four cumulative conditions: no previous exposure to long-acting cabotegravir as prophylaxis, viral load at or below 500,000 copies/mL, absence of hepatitis B virus coinfection and confirmed susceptibility to lamivudine on current or historical resistance testing; this combination is not recommended in acute or recent infection, where viral load is frequently above 500,000 copies/mL and rapid initiation is critical. Also alternatives are dolutegravir with abacavir and lamivudine, which requires prior HLA-B5701 testing, raltegravir 400 mg twice daily with a preferred backbone, ritonavir-boosted atazanavir or darunavir, the backbones abacavir with lamivudine and zidovudine with lamivudine, and regimens with efavirenz. Abacavir was downgraded from preferred to alternative in 2025, because of the need for HLA-B5701 testing before initiation and concerns about cardiovascular safety, and it has no activity against hepatitis B virus [4].

The European position in version 13.0 of the European AIDS Clinical Society guidelines, revised in 2025, is close: for pregnant people without previous therapy, the recommended initial regimens now include tenofovir alafenamide with emtricitabine and bictegravir once daily, in addition to tenofovir disoproxil or tenofovir alafenamide with lamivudine or emtricitabine combined with dolutegravir, and the same combinations with ritonavir-boosted darunavir 600/100 mg twice daily, which were already preferred options [34]. In Portugal there is no separate national guideline consulted in this review; practice follows the European recommendations.

The Brazilian protocol defines as the preferred initial regimen in pregnant women the combination of tenofovir disoproxil, lamivudine and dolutegravir, with the particularity that, in the version approved by Ordinance SCTIE/MS no. 55 of 11 November 2020, dolutegravir is only started after 12 weeks of gestation and is listed as contraindicated in the first trimester; alternative regimens are tenofovir with lamivudine combined with efavirenz (requiring demonstrated susceptibility on pre-treatment genotyping, and permissible from the first trimester), with atazanavir plus ritonavir (preferred when initiation in the first trimester is required), with darunavir plus ritonavir, or with raltegravir (preferred in the late third-trimester presenter). Where tenofovir is contraindicated, zidovudine is used and, if both are contraindicated, abacavir with a negative HLA-B*5701 test. For the woman who becomes pregnant while already on dolutegravir, the protocol itself reasons that neural tube closure defects are a rare event, with a prevalence of about 0.06% in the general Brazilian population, that the additional risk attributable to dolutegravir is small and that neural tube closure is complete by the eighth week, so that presentation after that period makes a change of regimen irrelevant [31]. This first-trimester restriction reflected the evidence available in 2020; the extended surveillance published since does not support it, and neither the US nor the UK guidelines retain it. The revised 2022 edition of the Brazilian protocol was not consulted in this review.

7.3. Laboratory monitoring in pregnancy

Viral load should be monitored more frequently than outside pregnancy, because rapid and sustained suppression up to delivery is the central determinant of prevention. The US schedule provides for measurement at the first visit, with review of previous values, two to four weeks after starting or changing therapy, monthly until undetectable and, thereafter, at least every three months; and a measurement at around 36 weeks, or in the four weeks before the expected delivery, which informs the decision on mode of delivery, on the need for intravenous zidovudine and on management of the newborn. With adherence and no resistance mutations, suppression is usually achieved in three to twelve weeks with the preferred regimens, more rapidly with integrase inhibitors; higher baseline viral loads and lower CD4 T-lymphocyte counts require longer. More frequent monitoring is recommended when there are adherence concerns, especially in early pregnancy, and with regimens containing rilpivirine, cobicistat-boosted elvitegravir, atazanavir or darunavir [4]. European version 13.0 specifies the frequency according to pre-pregnancy viral load: every trimester and at 36 weeks if there is durable suppression without adherence problems; every two months and at 36 weeks in the remaining cases; and every two weeks until below 50 copies/mL, and then at 36 weeks, if suppression has not been achieved by the third trimester [34].

The CD4 T-lymphocyte count is measured at the first visit, with review of previous values. In people on therapy for two years or more, with consistent suppression and counts maintained at 300 cells per cubic millimetre or above, adherence and good tolerance, the count need not be repeated in that pregnancy. With less than two years of therapy and counts below 300, inconsistent adherence or detectable viral load, it is repeated every three months; with less than two years of therapy and counts of 300 or above, every six months. There is a physiological fall in the count during pregnancy, which should be taken into account in interpretation [4].

Genotypic resistance testing, and phenotypic testing if indicated, should be reviewed together with the antiretroviral history and performed when the viral load is above the technical threshold, usually more than 500 to 1000 copies/mL, although some laboratories manage with values above 200 copies/mL. It is done before starting therapy in those who have never taken antiretrovirals and have never been tested, before starting in those who have taken antiretrovirals, including pre-exposure prophylaxis, and before modifying the regimen for virological failure or insufficient response. Therapy is started without waiting for the result and modified afterwards, if necessary. Laboratory monitoring of drug toxicity is based on the known profile of each drug in use, and gestational diabetes screening follows the usual protocol [4].

This initial assessment also includes two coinfections with direct implications for the newborn. For hepatitis B, the surface antigen should be tested in every pregnancy, with a triple panel — surface antigen, anti-core antibody and anti-surface antibody — in people without documented triple screening after the age of 18 demonstrating vaccine immunity or infection; it is not acceptable to use lamivudine or emtricitabine as the only drug with activity against hepatitis B, because of the risk of resistance, and the regimen should retain drugs active against hepatitis B before, during and after pregnancy; isolated anti-core antibody justifies considering testing for viral DNA, with adjustment of the regimen to two active drugs if positive, and many experts treat the newborn as potentially exposed, giving vaccine and immunoglobulin within the first 12 hours of life. For hepatitis C, antibody screening is recommended in every pregnancy and the exposed infant should have viral RNA testing between two and six months of age [4].

7.4. Intrapartum care and mode of delivery

The intrapartum decision depends on the viral load measured at around 36 weeks or in the four weeks before delivery. The maternal antiretroviral regimen should be maintained on schedule during labour and before a planned caesarean section; tablets may be taken with small amounts of water in the preoperative period and, where absorption requires food, with liquid dietary supplements, after agreement with the anaesthetist. If the regimen has to be interrupted temporarily, for less than 24 hours, all drugs should be stopped and restarted simultaneously, to minimise the risk of resistance [4].

Intrapartum intravenous zidovudine is recommended when the viral load is above 1000 copies/mL, when it is unknown, when there is known or suspected poor adherence since the last measurement, or when the rapid test at delivery is positive. It is started on admission in labour or at least three hours before a planned caesarean section, with a loading dose of 2 mg/kg over one hour followed by a continuous infusion of 1 mg/kg per hour for at least two hours, for a minimum total of three hours, until cord clamping, with no maximum dose defined in the source. It is not required when three conditions are met: ongoing therapy, viral load below 50 copies/mL in the four weeks before delivery and adherence achieved. It may be considered with viral loads between 50 and 1000 copies/mL, a case-by-case decision taking into account recent adherence and the person's preference, because the data are insufficient to determine whether it adds protection in that range. It remains recommended even with known or suspected zidovudine resistance when the viral load is above 1000 copies/mL, except where there is a documented history of hypersensitivity, because of the proven record of reduced transmission even in that context. Replacement by oral zidovudine is not recommended: levels achieved orally during labour are lower and absorption is erratic [4]. UK guidance indicates the same scenarios but starts the infusion four hours before a planned caesarean section, and considers its use with viral loads between 50 and 1000 copies/mL [3].

The quantitative evidence supporting these thresholds comes mainly from the French perinatal cohort. Between 1997 and 2010, in more than 11,000 pregnant women on therapy, 72% with triple regimens and 95% with intrapartum intravenous zidovudine, in women with a viral load at or above 1000 copies/mL whose newborns received zidovudine alone, the risk of transmission was significantly higher without maternal intravenous zidovudine (10.2%) than with it (2.5%; p<0.01), a difference that disappeared when the newborn received combination prophylaxis with two or more drugs (4.8% versus 4.1%; p=0.8). In a European cohort of high-risk newborns, the absence of intravenous zidovudine was associated with transmission in univariable analysis but not after adjustment for maternal viral load and other factors, with an adjusted odds ratio of 0.79 (0.55 to 1.15). In a cohort of 717 women who gave birth in Miami between 1996 and 2008, non-administration of intravenous zidovudine in 67 cases was not associated with greater risk. It should be noted, however, that 6% of women suppressed during pregnancy had viral load rebound close to delivery, which justifies clinical judgement irrespective of the value from the preceding four weeks [4].

As for mode of delivery, recommendations diverge between jurisdictions on one important point, the viral load threshold.

Planned caesarean section for the sole indication of prevention is not recommended with a viral load at or below 1000 copies/mL on therapy, given the low transmission rate in this group, and labour should not be induced at 38 weeks for that purpose; where there is an obstetric indication, caesarean section or induction is performed at the standard time. The 2025 update added the possibility that, when the viral load is above 1000 copies/mL or unknown, individualised birth plans extending gestation beyond 38 weeks may be considered to avoid caesarean section, given how rapidly viral load can fall with current regimens, subject to specialist consultation and shared decision-making. If a person with a planned caesarean section for high viral load goes into labour or has rupture of membranes, management must be individualised, because the evidence is insufficient to determine whether caesarean section reduces risk in that circumstance [4].

The data on rupture of membranes and mode of delivery with virological suppression are reassuring. In the United Kingdom and Ireland, between 2000 and 2011, in women on therapy with a viral load below 1000 copies/mL, transmission was 0.3% (13 of 3544) with planned caesarean section and 0.3% (6 of 2238) with planned vaginal delivery; the French cohort found 0.3% in both groups in suppressed women. In prospective population surveillance in the United Kingdom and Ireland covering 2116 pregnancies between 2007 and 2012, there was no difference between rupture of membranes of four hours or more (0.64%) and less than four hours (0.34%); in women with a viral load below 50 copies/mL the rates were 0.14% and 0.12%. By contrast, caesarean section has its own maternal morbidity: in a US administrative study, any complication associated with caesarean section occurred in 117 per thousand deliveries in women with HIV infection versus 67 per thousand in women without infection, and a meta-analysis of mainly observational studies estimated an odds ratio of 3.12 for morbidity with elective caesarean section compared with vaginal delivery, without reduction in transmission in women on therapy [4]. The Brazilian protocol reaches an equivalent conclusion, adding that the incidence of postoperative complications and of hospitalisations after caesarean section is higher in women with HIV infection, so elective caesarean section with a viral load below 1000 copies/mL is not recommended for prevention [31].

Three further intrapartum aspects deserve note. Fetal scalp electrodes should be avoided where there is suppression and should not be used when the viral load is at or above 50 copies/mL or unknown. Amniotomy and instrumental vaginal delivery follow obstetric indications, but should be avoided if possible with a viral load at or above 50 copies/mL; in a review of 9072 deliveries in the United Kingdom between 2008 and 2016, with 249 of 3023 vaginal deliveries instrumental (8.2%), there was a single transmission among 222 children with known status at 18 months, with multiple possible causes, which led the authors to conclude that instrumental delivery is safe in suppressed women. Finally, postpartum haemorrhage from uterine atony requires attention to interactions: methylergometrine and other ergot alkaloids should not be co-administered with potent cytochrome P450 3A4 inhibitors, namely protease inhibitors and cobicistat, because of the risk of exaggerated vasoconstriction with lower limb ischaemia, so they should only be used if alternatives such as prostaglandin F2-alpha, misoprostol or oxytocin are unavailable, and then at the lowest effective dose and for the shortest possible time; with inducers of the same cytochrome, such as nevirapine, efavirenz or etravirine, additional uterotonics may be needed because of reduced methylergometrine levels. Ritonavir reduces fentanyl elimination by 67%, but a pharmacokinetic simulation study suggested that, even with maximum clinical doses of epidural fentanyl over 24 hours, plasma concentrations associated with reduced minute ventilation are unlikely to be reached, so epidural analgesia can be used with any regimen [4].

When the diagnosis is made during labour itself, UK guidance proposes an explicit fetal loading regimen: immediate triple therapy with tenofovir disoproxil, emtricitabine and dolutegravir 50 mg, plus an immediate additional oral dose of tenofovir disoproxil 245 mg, plus immediate oral nevirapine 200 mg, and intravenous zidovudine throughout labour. Single-dose nevirapine is given irrespective of the CD4 T-lymphocyte count, hepatitis status or previous resistance, because it crosses the placenta and reaches an effective concentration in the neonate within two hours, maintained for up to ten days, which is particularly important in a newborn unable to take oral medication; nevirapine and dolutegravir should be given even where there is known maternal resistance to those drugs. The additional dose of tenofovir disoproxil is aimed at loading the infant, given that levels are lower and clearance increased in the second and third trimesters [3].

7.5. Prophylaxis and presumptive therapy in the newborn

All newborns with intrauterine or intrapartum exposure should receive one or more antiretrovirals, dosed according to gestational age and weight, started as early as possible, preferably within the first six hours of life according to US guidance and at the latest within the first four hours according to UK guidance. The maximum interval after which prophylaxis ceases to be beneficial is not known, but the studies started it early, animal data show a clear relationship between timing of initiation and efficacy, with no benefit demonstrated beyond 72 hours, and in an early zidovudine monotherapy cohort prophylaxis was only effective when started within the first 72 hours [3, 4].

The US architecture defines three scenarios. At high risk, that is, exposure to a maternal viral load at or above 50 copies/mL in the four weeks before delivery, documented or presumed, presumptive therapy with three drugs is recommended — zidovudine and lamivudine plus dolutegravir or nevirapine at treatment dose — from birth for two to six weeks, with no consensus as to duration; if the triple regimen lasts less than six weeks and the amplification test at birth is negative, zidovudine alone is continued to complete six weeks of prophylaxis. At low risk, defined by a viral load below 50 copies/mL from 20 weeks of gestation until delivery, ideally documented by two consecutive tests four weeks apart, zidovudine alone is recommended for two weeks. In the remaining scenarios, with a viral load at or above 50 copies/mL after 20 weeks but below 50 in the final four weeks, two options are accepted: triple presumptive therapy with de-escalation to zidovudine alone if the test at birth is negative, completing two to six weeks in total, or zidovudine alone for two to six weeks, with a minimum of two weeks in all cases [4]. The panel did not reach consensus in this intermediate scenario, and sets out both arguments: some members consider that the potential benefit of early treatment of an infant infected in utero justifies the presumptive approach; others consider that the marginal benefit and the low expected frequency of intrauterine infection do not outweigh the complexity and potential toxicity in uninfected infants [4].

The preferred presumptive regimen depends on gestational age and weight: dolutegravir with zidovudine and lamivudine (or emtricitabine) in term newborns, at 37 weeks or more and weighing 2 kg or more; nevirapine with zidovudine and lamivudine (or emtricitabine) in all newborns under 2 kg and in those of 32 to less than 37 weeks. Raltegravir has become an alternative, as dolutegravir is considered easier to prepare and administer and to have a higher genetic barrier. Abacavir is not recommended in the presumptive regimen, but may replace zidovudine when the latter is unavailable or causes toxicity, such as anaemia or neutropenia, always after confirmation of a negative test for the HLA-B*5701 allele; it is not approved by the US regulatory authority below 3 months of age, but is recommended by the World Health Organization in term newborns, with a dose derived from the IMPAACT P1106 trial and from two observational cohorts, European and African. In maternal HIV-2 infection or in HIV-1 and HIV-2 coinfection, the presumptive regimen must use dolutegravir, because HIV-2 is not susceptible to nevirapine [4].

The following table brings together the neonatal doses from the US guideline, which are the operational reference.

Source: Table 11.1 of the US perinatal guidelines, updated on 25 June 2026 [4]; doses are per kilogram of body weight or fixed by weight band, with no maximum dose defined in the source. In the triple regimen, zidovudine is continued for six weeks and the other two drugs last two to six weeks. The nevirapine dose for 32 to less than 37 weeks is not approved by the US regulatory authority, and the dose for 32 to less than 34 weeks is based on modelling, which may underestimate potential toxicity because the doses used to build the model were lower than those now recommended [4].

UK guidance uses a simpler architecture, with two pathways. At low risk, zidovudine monotherapy for two weeks, graded 1C, provided four maternal conditions are met: therapy started at least ten weeks before delivery; evidence of good engagement with obstetric, antenatal and HIV services; at least one viral load measurement in the six weeks before delivery; and all measurements in the ten weeks before delivery below 50 copies/mL, with individual risk assessment accepted for values between 50 and 200 copies/mL, particularly if the next measurement is below 50. The previous four-week option was removed as a routine recommendation, remaining only on a case-by-case basis. At high risk, when those conditions are not met, nevirapine for two weeks with zidovudine and lamivudine for four weeks is recommended, graded 1D. The most relevant change from the previous version is the use of the treatment dose of nevirapine, 6 mg/kg, instead of the prophylactic dose of 2 mg/kg, to reduce confusion between regimens and optimise very early treatment should the child already be infected; nevirapine should be stopped two weeks before the other drugs, because of its long half-life, to avoid exposure to monotherapy and the development of resistance to the non-nucleoside inhibitor class. For newborns of 34 to 37 weeks, 4 mg/kg is advised in the first week, and for lower gestational ages there are no standardised doses; the guidance does not define a maximum dose [3].

UK guidance introduces three further useful specifications. The first concerns prematurity: cohort data historically indicated prematurity as a risk factor, but more recent observations suggest that this may not hold in the low-risk scenario, so prematurity is no longer considered a strong risk factor and does not alter stratification, although it may be weighed individually when other factors coexist. The second is the intravenous route: zidovudine is the only antiretroviral licensed for intravenous use in sick or preterm neonates unable to take oral medication, and in the extremely preterm infant, at risk of necrotising enterocolitis, intravenous zidovudine is used until enteral feeding is established — in a large French case-control study, being the child of a woman with HIV infection was associated with an increased risk of necrotising enterocolitis, with an odds ratio of 6.63 (1.26 to 34.8; p=0.025), although the numbers were too small to distinguish the effect of maternal or infant antiretrovirals. In the high-risk preterm infant, combination cover is obtained by adding intravenous zidovudine to the transplacental loading achieved by oral maternal doses of drugs that cross the placenta efficiently and have a long half-life in the neonate: single-dose nevirapine, double-dose tenofovir disoproxil and dolutegravir or raltegravir; with this strategy, exposure of the preterm infant should be equivalent to combination prophylaxis during the first five to seven days of life. The third is maternal resistance: the infant's regimen should be discussed in advance in a multidisciplinary group and, if no opinion is available in good time, the standard regimen of zidovudine, lamivudine and nevirapine is started and modified afterwards [3].

The data supporting the choice between monotherapy and combination regimens are heterogeneous. Randomised trials showed a benefit of combination prophylaxis over monotherapy when the mother received no therapy before delivery, but two trials showed no advantage of dual regimens when the pregnant woman received short-course therapy immediately before delivery. In the HPTN 040/PACTG 1043 trial, with 1684 formula-fed newborns whose mothers received no antenatal therapy, intrapartum transmission was 4.8% (3.2 to 7.1) with zidovudine for six weeks, 2.2% (1.2 to 3.9) with zidovudine plus three doses of nevirapine in the first week and 2.4% (1.4 to 4.3) with zidovudine plus nelfinavir and lamivudine for two weeks (p=0.046 in both comparisons), with neutropenia significantly more frequent in the triple arm; the conclusion was that the two-drug regimen is superior to monotherapy with less toxicity than the three-drug regimen [35]. An adaptive phase 3 trial with a Bayesian design in Thailand, in women who started therapy eight weeks or less before delivery, added single-dose maternal nevirapine at the onset of labour to the usual triple regimen and gave the newborn triple prophylaxis (nevirapine for two weeks plus lamivudine and zidovudine for four weeks), without breastfeeding: there were no transmissions in 88 pairs, with a 94.4% probability of superiority of intensification over usual care [3]. In the opposite direction, European observational data from the EPPICC collaboration, with 5285 high-risk pairs between 1996 and 2010, did not demonstrate an advantage of combination prophylaxis, which the authors attributed to confounding or to the possibility that benefit is limited to a subgroup [3]. The proportion of newborns receiving triple regimens is in any case falling as the proportion of suppressed pregnant women rises: in England, only 54 of 1269 infants (4.3%) received triple prophylaxis, for indications such as transient rises in maternal viral load during breastfeeding, high viral load at delivery, resistance, reduced adherence and late or absent antenatal care [3].

The minimum duration of zidovudine monotherapy has never been determined in a trial. Trial 076 studied six weeks; a randomised trial in Thailand, in the pre-combination era, showed that three days and six weeks of neonatal prophylaxis were equally effective when the mother started treatment at 28 weeks, suggesting that a short duration suffices after adequate maternal therapy. The reduction to two weeks rests on German experience, with more than ten years of use in the lowest-risk situations without any signal of increased transmission and with less haematological toxicity, and on French data showing no transmissions with therapy from conception. There are no studies directly comparing two and four weeks, but two-to-four-week regimens were associated with earlier recovery from anaemia. Some countries use no prophylaxis at all in low-risk situations: in Switzerland, extrapolation from the evidence on treatment as prevention led to a recommendation of the option of giving no prophylaxis to newborns of mothers on therapy with an undetectable viral load on at least two occasions four weeks apart, including after 36 weeks, particularly if therapy was suppressive before conception and throughout pregnancy, with preliminary observational data showing no increase in risk, albeit in small numbers. The UK writing group considers, however, that evidence on treatment as prevention in adults should not be wholly extrapolated: the risk of transmission without intervention in peripartum exposure is 10 to 20%, against 0.1 to 1.5% in sexual or occupational exposure; the nature of the exposure is different, the fetus potentially being exposed at any time from conception; and there is transplacental traffic of maternal cells, including CD4 T lymphocytes, which may persist in the infant's circulation after birth, a phenomenon whose relevance to HIV transmission is unknown but which has been invoked for hepatitis B and for human T-lymphotropic virus type 1, and which underlay the justification for six weeks of prophylaxis in trial 076 [3].

When the maternal viral load from the day of delivery turns out to be above 50 copies/mL after the infant has started two weeks of zidovudine on the basis of low-risk criteria, UK guidance requires urgent discussion with a paediatrician experienced in the prevention of vertical transmission, to consider escalation to three drugs; the probability of additional benefit decreases with time elapsed since delivery and is very unlikely beyond 72 hours, although it may be considered individually. The converse situation, in which an infant started a triple regimen because low-risk criteria were not met and the maternal viral load at delivery subsequently proves to be below 50 copies/mL, allows simplification to zidovudine monotherapy to complete two weeks, a decision likewise taken with an experienced paediatrician [3].

The World Health Organization position, presented at the satellite session of the International AIDS Society conference in Kigali on 14 July 2025 and formalised in the updated recommendations document announced on 7 January 2026, differs from earlier ones in substantive respects: all exposed infants should receive six weeks of postnatal prophylaxis, preferably with nevirapine, with dolutegravir — only in term newborns, over 37 weeks — and lamivudine as alternatives; infants at highest risk should receive enhanced triple prophylaxis, the preferred option being abacavir, lamivudine and dolutegravir in a dispersible paediatric formulation, followed by monotherapy during breastfeeding or until the maternal viral load becomes undetectable; and the paediatric doses of that formulation now cover all term newborns over 3 kg [36, 37]. Version 13.0 of the European guidelines incorporated dolutegravir as the preferred first-line option in term newborns, on the basis of the PETITE-DTG study and in alignment with the World Health Organization's programmatic approach, and adopted the principle of using treatment doses, rather than prophylactic doses, in postnatal prophylaxis, to reduce confusion and simplify the transition to treatment in infants who are diagnosed [34].

The 2020 Brazilian protocol defines a regimen distinct from the foregoing. All children born to mothers with HIV infection receive zidovudine oral solution, preferably still in the delivery room after immediate care, or within the first four hours, continued through the first four weeks of life: 4 mg/kg per dose every 12 hours in newborns of 35 weeks or more; 2 mg/kg every 12 hours for 14 days and 3 mg/kg every 12 hours from day 15 in those of 30 to 35 weeks; and 2 mg/kg every 12 hours in those under 30 weeks. Nevirapine is added, started within 48 hours of life, in five scenarios: no antiretrovirals in pregnancy, irrespective of peripartum zidovudine; use of antiretrovirals in pregnancy but unknown viral load or above 1000 copies/mL in the third trimester; a history of poor adherence, even with a documented viral load below 1000 copies/mL in the third trimester; a mother with a sexually transmitted infection, especially syphilis; and a woman in labour with a reactive result at the time of delivery. Nevirapine is given in three doses — at birth, 48 hours after the first and 96 hours after the second — at a dose of 12 mg (1.2 mL) for a birth weight above 2 kg and 8 mg (0.8 mL) for a weight between 1.5 and 2 kg, and is not used below 1.5 kg. When the oral route is impossible, injectable zidovudine is used at 3 mg/kg per dose every 12 hours in those of 35 weeks or more, at 1.5 mg/kg every 12 hours for the first 14 days and 2.3 mg/kg from day 15 in those of 30 to 35 weeks, and at 1.5 mg/kg every 12 hours in those under 30 weeks, always for four weeks, with no maximum dose defined in the protocol; where swallowing is impossible and nevirapine is indicated, administration by nasoenteric tube may be considered, as there is no injectable formulation. The indication for chemoprophylaxis after 48 hours of life should be discussed case by case, preferably with a specialist [31].

7.6. Infant feeding

Feeding is today the area of greatest international divergence and the one that has changed most. The starting point is quantitative: replacement feeding with correctly prepared formula or with pasteurised donor human milk eliminates the risk of postnatal transmission; maternal antiretroviral therapy with sustained suppression reduces that risk to less than 1%, but not to zero [4].

Current US recommendations distinguish two situations. When therapy is not being taken or suppression was not achieved during pregnancy — at a minimum throughout the third trimester — and at delivery, replacement feeding with formula or pasteurised donor human milk is recommended, to eliminate the risk. When therapy is taken consistently and the viral load remains below 50 copies/mL for at least three months before delivery, counselling on the three options — formula, donor milk or breastfeeding — is recommended, with support for those who choose to breastfeed. Those who choose formula should equally be supported, with identification and resolution of access barriers. The update of 25 June 2026 added two clarifications: there is no evidence that formula supplementation increases the risk of acquisition in the breastfed infant in the context of parental therapy and suppression; and women with mastitis or other unilateral breast disease should stop feeding from the affected breast and feed from the contralateral breast, resuming both when symptoms resolve [4].

Management of breastfeeding requires concrete measures: antenatal coordination between obstetrics, infectious diseases, paediatrics, lactation consultancy and social work; adherence support and assured continuous access to the drugs, recognising that changes in insurance cover may interrupt access; screening and support for postnatal depression, more frequent in women with HIV infection; documentation of sustained suppression before delivery and during breastfeeding, with one possible approach being viral load measurement every one to two months, there being no data to define the ideal frequency; exclusive breastfeeding for the first six months, with later introduction of complementary foods if desired; breast care, with prevention and early treatment of engorgement, stasis, fissures, mastitis and candidiasis; and a joint weaning plan over two to four weeks, because very rapid weaning was associated, in the pre-therapy era, with greater viral shedding in milk and greater risk of transmission. When temporary interruption is necessary, the options include giving milk previously stored from a period of suppression, expressing and flash-heating the milk — a procedure documented to eliminate the virus from milk —, giving formula or donor milk while continuing to express so that feeding can be resumed, or stopping definitively. The guidelines are explicit on one ethical point: involving child protection services is not an appropriate response to feeding choices, there having been reports of threats or actual referrals to those services after a person expressed an intention to breastfeed, a practice harmful to families, which aggravates stigma and is applied disproportionately to families with fewer resources and to Black and Indigenous children [4].

The evidence on transmission through breastfeeding under therapy is the basis of this balance. Without maternal therapy or infant prophylaxis, the risk is 15 to 20% over two years. A systematic review with meta-analysis published in 2025, with 13 studies, estimated the monthly risk of transmission during breastfeeding at 0.1% (0.0 to 0.2) with a recent maternal viral load below 50 copies/mL and at 0.3% (0.0 to 1.8) with a viral load at or above 50; an earlier meta-analysis, from 2017, with six studies of therapy started at some point in pregnancy and continued for at least six months postpartum, estimated pooled postnatal transmission of 1.1% at six months (0.32 to 1.85) [38], with substantial heterogeneity and individual rates between 0.2% and 3.1%. In the PROMISE trial, with more than 2400 women with CD4 T-lymphocyte counts at or above 350 cells per cubic millimetre, estimated transmission was 0.3% at six months and 0.6% at twelve months, in both the maternal therapy arm and the arm of prolonged infant prophylaxis with daily nevirapine, continued until cessation of breastfeeding or to 18 months; maternal viral load and resistance were independently associated with transmission through breastfeeding [4, 39]. In a substudy of the BAN trial, with 31 infected and 232 uninfected infants, there were no transmissions when the mother consistently maintained a viral load below 100 copies/mL. In a post hoc analysis of HPTN 046, with a postnatal transmission risk of 0.5% (0 to 1.4) in mothers on therapy in both arms, adding infant prophylaxis did not further reduce transmission [4].

There are nonetheless documented cases of transmission with a maternal viral load below 50 copies/mL close to the time of transmission: at least six published, two of them from an observational study in Malawi, two from the Mma Bana study in Botswana, in the three-nucleoside-inhibitor arm, with plasma and milk viral load below 50 copies/mL at the first and third month, and one in the efavirenz arm of the DolPHIN-2 trial, detected at 72 weeks postpartum in a context of optimal maternal suppression and without other known risk factors such as mastitis, resistance or milk from another woman. Possible explanations include viraemia between measurements, discordance between plasma and milk viral load, and cell-associated virus reservoirs in milk, and data are lacking on the effect of prolonged therapy and of newer drugs on those reservoirs [3, 4]. It is on this body of observations that the explicit position of the UK guidelines rests, namely that the message «undetectable equals untransmittable» applies only to sexual transmission and not to breastfeeding, nor to vertical transmission in general: the transmission rate in women with a viral load below 50 copies/mL at delivery in England was 0.11% in 2017 and 2018 [3].

Data from high-income countries are sparse but consistent. Ten observational studies, with groups of 7 to 77 breastfed infants and an estimated total of 184 to 214 distinct children in Germany, Italy, Switzerland, the United States and Canada, identified no transmission, although none was designed to assess transmission; the participants were on therapy started before or in the first trimester and documented as suppressed at delivery or at the end of pregnancy, and infant prophylaxis varied between none, zidovudine or nevirapine for two to six weeks, and three-drug regimens until weaning [4]. In the United Kingdom, between 2012 and 2021, 203 women (2.4% of women with HIV infection) were supported to breastfeed, with no reported cases of postnatal transmission at the date of publication, with 34 infants awaiting confirmatory testing and five without complete testing; the number of women choosing to breastfeed quadrupled, from fewer than 10 per year between 2012 and 2014 to 40 to 50 per year between 2019 and 2021. In a UK survey of 94 women, 38% would like to breastfeed, but only 27% considered it safe to do so with a viral load below 40 copies/mL. The Swiss mother and child cohort described 25 pairs, with a median breastfeeding duration of 6.3 months, without transmission at the date of publication in 2023. In the Tshilo Dikotla study, in Botswana, with frequent viral load monitoring and four weeks of prophylaxis with zidovudine or nevirapine, 19 of 247 participants had a viral load at or above 40 copies/mL at some point during breastfeeding — 12 chose to stop and 7 continued with counselling and frequent measurements — and no transmission occurred [3, 4].

The divergences in management in the face of viraemia during breastfeeding are set out below.

Prolonged prophylaxis during breastfeeding, when used, should be continued until four weeks after the last exposure to milk or until four weeks after concerns about maternal suppression have been resolved. If prolonged nevirapine is planned, it should ideally be started from birth, replacing zidovudine as initial prophylaxis; if lamivudine is planned, it is started after zidovudine is completed, because there are no clinical data supporting lamivudine as initial monotherapy from birth. The choice between the two drugs takes into account ease of administration — nevirapine is taken once daily and has been studied up to 18 months, lamivudine twice daily and has been studied up to 12 months —, parental preference, and the need to avoid a drug to which resistance of the maternal virus is anticipated; lamivudine is preferred when there is evidence or suspicion of nevirapine resistance, including HIV-2 infection or coinfection, or when the infant does not tolerate nevirapine [4].

The doses for prolonged prophylaxis are set out in table 12.1 of the US guidelines. Nevirapine, as a 10 mg/mL syrup once daily, in newborns of 36 weeks or more, is given at 10 mg (1 mL) from birth to six weeks if the weight is 2 to less than 3 kg and at 15 mg (1.5 mL) if it is 3 kg or more; then 20 mg (2 mL) from six weeks to six months, 30 mg (3 mL) from six to nine months and 40 mg (4 mL) from nine to 18 months, with doses adapted from the World Health Organization consolidated guidelines of July 2021; for less than 36 weeks, the guidelines refer to specialist consultation. Lamivudine is given at 2 mg/kg per dose twice daily between two and four weeks and, from four weeks to 12 months, by weight band as a 10 mg/mL solution twice daily: 10 mg (1 mL) for 2 to less than 3 kg, 15 mg (1.5 mL) for 3 to less than 4 kg, 25 mg (2.5 mL) for 4 to less than 8 kg and 50 mg (5 mL) for 8 kg or more, a regimen derived from established treatment and from the bands used in the PROMISE-EPI trial, with no maximum dose defined in the source beyond the band of 8 kg or more [4]. When maternal viraemia develops or a new diagnosis is made during breastfeeding, the presumptive regimen is zidovudine with lamivudine plus dolutegravir or nevirapine for four weeks, a duration aligned with the recommendations for non-occupational post-exposure prophylaxis [4]. UK guidance uses, in the same situation, dolutegravir with lamivudine and zidovudine for four weeks, following the CHIVA paediatric association standard for post-exposure prophylaxis in children, with ritonavir-boosted lopinavir or raltegravir as alternatives if dispersible dolutegravir is unavailable [3].

The US position expressly diverges from that of the World Health Organization on one point: the latter recommends dolutegravir as an alternative to nevirapine for postnatal monotherapy prophylaxis during breastfeeding, and the US panel decided not to recommend it, because of limited experience during breastfeeding and the potential implications of breakthrough infections developing resistance to the integrase inhibitor class [4]. The World Health Organization, for its part, recommends exclusive breastfeeding for the first six months, continued to 12 months and possibly to 24 months or beyond, with effective maternal therapy and adequate complementary feeding, and has adopted a rights-based approach supporting the option to breastfeed even where national policy recommends formula [36, 37]. The 2020 Brazilian protocol retains the contraindication to breastfeeding, with lactation inhibition by cabergoline and replacement with formula up to six months, allowing pasteurised human milk from an accredited bank — for example in preterm or low-birth-weight newborns — and strictly contraindicating mixed feeding, cross-nursing by another woman and the use of human milk pasteurised at home [31].

7.7. Safety of the antiretrovirals used in the infant

When used alone for prophylaxis in low-risk scenarios, in the neonatal period and during breastfeeding, antiretrovirals appear very safe. The most relevant toxicity in the neonatal period is anaemia from zidovudine, alone or in combination, which appears between two and four weeks and recovers after discontinuation to values similar to those of controls. Lamivudine has shown safety during breastfeeding up to at least 41 weeks of life and nevirapine from birth to 18 months; three-drug combinations appear safe at the currently recommended durations, of two to six weeks. On the basis of these data, longitudinal laboratory monitoring for adverse events is not required in otherwise healthy infants receiving the currently recommended antiretrovirals in the first six weeks of life, although the American Academy of Pediatrics recommends periodic monitoring of haematological and hepatic toxicity in breastfed infants receiving prophylaxis beyond that period and for prolonged durations [4].

The drug-by-drug data are as follows. For zidovudine, in trial 076 haemoglobin was lower at the end of the six weeks of treatment but equalised between arms by 12 weeks, with no other differing toxicities and no new toxicities up to 18 months of follow-up; a four-week regimen, compared with six, was associated with earlier recovery from anaemia; severe cases may be treated with erythropoietin or lead to replacement by lamivudine or nevirapine. For lamivudine, the Mitra study treated 398 infants with zidovudine and lamivudine from birth to the first week and then with lamivudine alone during breastfeeding, for a maximum of six months plus two weeks after cessation, without serious events attributed to the drug; in the ANRS 12174 trial, 1273 infants were randomised at seven days to lopinavir with ritonavir or lamivudine until the end of breastfeeding, with a median duration of 41 weeks, with no differences in adverse event rates. For nevirapine, the SWEN trial randomised infants to a single dose (n=1047) or six weeks of daily administration (n=977), with adverse events common overall, about 40%, without differences between arms; in the BAN trial, with 2369 dyads, a hypersensitivity reaction with rash, with or without fever, developed in 16 infants (1.9%), ten of them with eosinophilia, within four weeks of initiation, and resolved on replacement by lamivudine, hypersensitivity being the only adverse event significantly different between arms; in HPTN 046, with 1519 breastfed infants randomised to six months of daily nevirapine or placebo after six weeks of nevirapine from birth, there were no differences up to 12 months of follow-up; in the PEPI-Malawi trial, with 3016 infants, the arm of prolonged nevirapine with zidovudine had significantly more serious adverse events considered possibly related, neutropenia being the most common; in the breastfeeding component of PROMISE, with 2431 dyads and a median breastfeeding duration of 16 months, only 20 of 1204 infants (2%) in the nevirapine arm stopped the recommended regimen because of toxicity concerns, with no cases of hypersensitivity; and the French perinatal cohort observed grade 2 or higher anaemia less frequently in 830 infants on daily nevirapine than on zidovudine, in a high-resource setting [4].

In the IMPAACT P1115 trial, 438 newborns of 34 weeks or more gestational age at high risk started nevirapine at treatment dose — 6 mg/kg twice daily in term infants and 4 mg/kg twice daily for one week followed by 6 mg/kg in those of 34 to less than 37 weeks — with two nucleoside inhibitors, zidovudine and lamivudine in 96% of cases, within the first 48 hours of life, continued until intrauterine infection was excluded by DNA testing, with a median duration of 13 days; 389 (89%) were born at term and 36 proved to be infected in utero. Grade 3 or 4 adverse events at least possibly related to the antiretrovirals occurred in 30 infants (7%; 5 to 10), without leading to discontinuation of nevirapine in any case; neutropenia, in 25 (6%), and anaemia, in 6 (1%), were the most frequent toxicities, and zidovudine was permanently discontinued in nine infants (2%) for those reasons [4]. In a prospective cohort in Thailand, grade 2 anaemia at one month of age was more frequent in infants on zidovudine, lamivudine and nevirapine for six weeks (38%) than in those on zidovudine alone (21%; p=0.007), but the risk of grade 3 anaemia was similar (9.2% and 10.2%; p=0.81) and haemoglobin recovered by two months, with no differences in neutropenia or hepatotoxicity. In a retrospective study of 148 high-risk Canadian infants treated with three drugs for six weeks, the combinations were associated with lower haemoglobin over six months than zidovudine alone, although the subgroup on a nevirapine-based regimen did not differ from monotherapy [4].

For abacavir, a meta-analysis and a population pharmacokinetic model that included three trials with a total of 45 infants under three months attributed no grade 3 or higher adverse event to the drug. For dolutegravir, metabolised 90% by uridine diphosphate glucuronosyltransferase 1A1 and 10% by cytochrome P450 3A4, neonatal data are limited but favourable: the PETITE-DTG study, phase 1/2, assessed the safety and pharmacokinetics of multiple doses in exposed newborns aged 0 to 28 days without related adverse events; in IMPAACT 2023, 18 infants received dolutegravir on alternate days to day 14 without grade 3 events; in IMPAACT P1093, in infants of four weeks or more and 3 kg or more, no grade 3 or higher event was attributed to the drug; and in the ODYSSEY trial, in children of 3 to less than 14 kg, there were no related events. The specific concern with raltegravir in the neonatal period is displacement of unconjugated bilirubin from albumin binding sites, compounded by the drug being metabolised by the same uridine diphosphate glucuronosyltransferase 1A1 responsible for bilirubin elimination, whose activity is low at birth; in vitro studies suggest safety at the exposures achieved for treatment, and in the regulatory trial IMPAACT P1110, a single-arm phase 1/2 study that established the dosing in 52 infants of 34 weeks or more gestational age [40], no adverse event was attributed to raltegravir and no bilirubin value exceeded 16 mg/dL [4]. UK guidance nevertheless recommends monitoring total and fractionated bilirubin in the first week of life and restricting raltegravir in preterm infants to exceptional circumstances, after specialist advice and with level monitoring, as they are more vulnerable to hyperbilirubinaemia [3]. As for ritonavir-boosted lopinavir, adrenal suppression has been documented in some neonates, especially preterm, in addition to reports of cardiac, renal and neurological toxicity, including one death during postnatal prophylaxis, without it being clear whether the effect is specific to this drug or common to boosted protease inhibitors; the UK recommendation is to avoid it in routine prophylaxis and reserve it for preterm infants in exceptional circumstances, after specialist advice and where there is maternal resistance to non-nucleoside and integrase inhibitors, with close metabolic monitoring in the first five days of life [3]. Maraviroc has a neonatal dose defined in a small pharmacokinetic study, but without efficacy data as prophylaxis and without consideration of the tropism of the potentially transmitted virus, so it should be avoided in routine prophylaxis until more data are available [3]; it is approved for infants of 2 kg or more and may be an additional option in newborns exposed to multidrug-resistant HIV-1 that remains CCR5-tropic, although drug interactions and the absence of data limit its role [4].

There are four limitations to generalisation that the US panel flags and that should be kept in mind. Most trials were conducted at a time when the mothers of infants on prophylaxis were not receiving therapy, so it is possible that antiretrovirals passed in milk add to toxicity in the infant receiving prophylaxis; current data indicate that only clinically insignificant quantities of the approved drugs enter milk, but this theoretical possibility should be considered for new agents. The data reviewed focus on serious events, grade 3 or 4, and most trials did not report frequencies of lower-grade toxicities. The overall adverse event rate was high in many trials — in the breastfeeding component of PROMISE, about 35% of infants in both arms had grade 3 events or died —, a rate that reflects limited health resources in sub-Saharan African settings and reference ranges that may not be applicable outside the United States, which makes individual risk difficult to extrapolate; comparison between arms of randomised trials nevertheless still allows the added risk attributable to each agent to be assessed. Finally, the review covers only the agents and approaches currently recommended, and safety data for abandoned approaches are in the archived versions of the guidelines [4].

Explicit knowledge gaps have been identified: pharmacokinetic and safety studies are lacking to establish safe doses of dolutegravir in preterm infants, since for infants of less than 37 weeks and less than 2 kg only nevirapine, lamivudine and zidovudine have defined dosing; data are lacking to establish the dose and safety of long-acting antiretrovirals from birth to the end of breastfeeding, which have the potential to resolve difficulties with oral administration; and data are lacking on the risk conferred by mastitis and by mixed feeding where there is virological suppression, as well as implementation research on how to optimise treatment and testing of the mother–child pair [4].

7.8. Teratogenicity and the case of dolutegravir

Assessment of the fetal safety of antiretrovirals rests largely on the Antiretroviral Pregnancy Registry, a prospective registry maintained by manufacturers with an advisory committee that includes teratology, infectious diseases, epidemiology, biostatistics and professionals from obstetrics, maternal-fetal medicine and paediatrics; reporting of all prenatal exposures, to a single drug or to combinations, is strongly recommended and should be done as early as possible in pregnancy. The registry has predefined analytical methods: 200 first-trimester exposures suffice to detect a doubling of the overall risk of birth defects relative to the general population, and a cohort of one thousand exposures allows a 1.5-fold increase to be detected. The reference populations used are the metropolitan Atlanta birth defects surveillance programme, with a prevalence of 2.72 per hundred live births, and the Texas birth defects registry, with 5.07 per hundred [4].

In the analysis up to 31 July 2025, the prevalence of birth defects was 3.0 per hundred live births with exposure to any antiretroviral in the first trimester (393 of 13,093 exposures; 2.7 to 3.3), not significantly different from the 2.9 per hundred observed with initial exposure in the second or third trimester, with a prevalence ratio of 1.05 (0.90 to 1.22) [4]. The drug-by-drug data fall into three categories: for bictegravir, cobicistat, stavudine, didanosine, darunavir, elvitegravir, indinavir, raltegravir and rilpivirine there are sufficient first-trimester exposures to detect at least a doubling of the overall risk, with no increase observed; for lamivudine, abacavir, atazanavir, dolutegravir, efavirenz, emtricitabine, lopinavir with ritonavir, nelfinavir, nevirapine, ritonavir, tenofovir alafenamide, tenofovir disoproxil and zidovudine there are sufficient exposures to detect a 1.5-fold increase in overall risk and a twofold increase in the risk of cardiovascular and genitourinary defects, likewise with no increase detected; and for cabotegravir, doravirine, etravirine, fostemsavir, lenacapavir and enfuvirtide the reported exposures are insufficient to assess risk [4].

There are limitations that should not be omitted in counselling. The registry is not powered to detect rare defects, so these data exclude 1.5-fold or twofold increases in overall risk but do not exclude an increase in rare defects such as those of the neural tube; to exclude a threefold increase in the risk of neural tube defects, with a background prevalence of 0.1%, at least two thousand preconception exposures would be needed [3]. Detection in live births only underestimates the risk, by excluding more severe defects resulting in stillbirth or termination, and joint analysis of all antiretrovirals may mask the signal from a specific drug. Confounders associated with the infection itself must also be considered: exposure to folate antagonists such as co-trimoxazole, nutritional and folate status, and tobacco and alcohol use, in addition to confounding by indication, where the reason for using a particular drug is associated with a greater risk of defects, such as more advanced maternal age or more advanced disease. In a 2018 study of 9546 pregnant women, 97.1% reported taking at least one medicine during pregnancy, and 89.3% of the 290 therapies submitted to the US regulatory authority between 2010 and 2019 lacked human pregnancy data [4].

Experience with efavirenz and with dolutegravir illustrates the importance of sufficient data. Initial data from animal studies and retrospective human reports raised concern about congenital anomalies of the nervous system and neural tube defects with efavirenz taken close to conception and in early pregnancy, a concern that later data did not confirm. For dolutegravir, an initial analysis of the Tsepamo study, an active surveillance study of birth outcomes in Botswana, suggested an association with neural tube defects: in the report published in 2019, between August 2014 and March 2019, 119,477 deliveries were captured, with an evaluable surface examination in 119,033 (99.6%) and 98 neural tube defects identified (0.08% of deliveries); among 1683 deliveries in which the mother was taking dolutegravir at conception, 5 defects were found (0.30%: two myelomeningoceles, one anencephaly, one encephalocele and one iniencephaly), against 15 in 14,792 deliveries with other antiretrovirals at conception (0.10%), 3 in 7959 with efavirenz (0.04%), 1 in 3840 with dolutegravir started during pregnancy (0.03%) and 70 in 89,372 in mothers without infection (0.08%); the difference between dolutegravir and other antiretrovirals at conception was 0.20 percentage points (0.01 to 0.59), and the authors summarised the finding as three per thousand deliveries against one per thousand [14].

Extended surveillance altered this conclusion, in a process that the Tsepamo investigators themselves later analysed as an exercise in tolerating uncertainty in early safety signals [41]. With Tsepamo data between August 2014 and May 2023 — a study that covers about 70% of all births in Botswana and is the largest study of birth defects in children of women on therapy, as well as the only one powered to detect rare anomalies —, the prevalence of neural tube defects with dolutegravir at conception was 0.11% (10 in 9460; 0.06 to 0.19), the same as that observed with any regimen without dolutegravir at conception (0.11%; 25 in 23,664; 0.07 to 0.16), and comparable to 0.08% with efavirenz at conception (11 in 14,432), 0.06% with dolutegravir started during pregnancy (4 in 6551) and 0.07% in women without infection (108 in 170,723), with a difference of 0.00% (−0.07 to 0.10) between dolutegravir and other regimens at conception. The parallel programme in Eswatini found a prevalence of 0.08% in 4832 pregnancies with dolutegravir at conception, against 0.16% in 1248 pregnancies with regimens without dolutegravir and 0.08% in 17,270 pregnancies without infection; combining the two programmes, the prevalence is 0.098% in more than 14,000 women with dolutegravir at conception, with no significant difference [3, 4]. A US cohort study based on health insurance claims data likewise found no increase in risk [4]. This sequence is now reflected in the labels: the European product information for Tivicay states that two large surveillance studies, with more than 14,000 pregnancy outcomes in Botswana and Eswatini, and other sources, do not indicate an increased risk of neural tube defects after exposure to dolutegravir, and places the incidence in the general population at between 0.5 and 1 case per thousand live births [15]; the US label presents the same figures with confidence intervals [16].

Two other, smaller series should be kept in mind so that the matter is not treated as closed. In the US pharmacokinetic study of 29 pregnant women on dolutegravir, anomalies were described in 7 newborns: three normal variants, one with total anomalous pulmonary venous drainage (dolutegravir started at 16 weeks), one with a polycystic right kidney (at 11 weeks), one with an isolated left renal cyst (at 12 weeks) and one with chin tremors and irritability (at 28 weeks); in a parallel study in African women, with dolutegravir started at 28 weeks or later, there were no anomalies in 28 live births. In reviews of clinical experience, defects were noted in 4 of 81 newborns of European women, in 2 of 66 of US women and in none of 116 of Botswanan women with first-trimester exposure, and in 57 French women no increase in defects or in adverse perinatal outcomes was observed compared with matched controls without integrase inhibitors. By contrast, in a series of 257 pregnancies in Denver, Colorado, various congenital anomalies were detected, without any neural tube defect, in 7 of 24 newborns (29.2%) exposed to dolutegravir in the first trimester, against 13 of 109 (11.9%) with exposure to antiretrovirals without integrase inhibitors [4].

The folate axis adds a mechanistic piece. In CD-1 mice fed a normal (3 mg/kg) or low (0.3 mg/kg) folic acid diet and treated with dolutegravir at a dose equivalent to the human therapeutic or a supratherapeutic dose between embryonic days 6.5 and 12.5, fetuses with exencephaly, a neural tube defect, appeared at both exposures but only in females on the low-folate diet; cleft palate occurred under both folate conditions. The authors concluded that adequate dietary levels of folic acid attenuate the developmental defects associated with dolutegravir exposure and hypothesised that folate status might explain, at least in part, the signal observed in Botswana [33]. Integrase inhibitors, including dolutegravir, are associated with weight gain and other metabolic effects, such as hyperglycaemia and altered folate metabolism, both in pregnancy and in non-pregnant adults, and the impact of those effects on obstetric outcomes is unknown: in high-resource countries, where undernutrition is rare, the greater gestational weight gain associated with dolutegravir may lead to more gestational hypertension, gestational diabetes and macrosomia; in settings where undernutrition is frequent, it may reduce the number of women at risk of outcomes linked to low maternal weight and increase the number at risk of those linked to high weight. In the trial that randomised pregnant women from nine countries, 88% of them in sub-Saharan Africa, women on dolutegravir had greater antepartum weight gain, associated with fewer adverse pregnancy outcomes [4].

The practical translation of these data into folic acid supplementation is as follows: UK guidance recommends 400 micrograms per day from before conception until 12 weeks of pregnancy, with the higher dose of 5 mg per day reserved for the usual situations of increased risk — neural tube defect in the person themselves or in their partner, a previously affected child, family history, diabetes, epilepsy, body mass index above 30 kg/m² or previous bariatric surgery — and makes explicit that, given the absence of evidence of significantly increased risk of neural tube defects in women on therapy, people with HIV infection who are trying to conceive should follow exactly the same rule [3]. The Brazilian protocol indicates routine iron supplementation of 40 mg per day of elemental iron combined with folic acid at a dose of 5 mg per day up to 12 weeks of gestational age, and supplementation from three months before becoming pregnant in women intending to conceive [31].

7.9. Comparative regulatory status

The following table summarises the regulatory status of the central drugs of prevention and of neonatal therapy, with the relevant discrepancies between what the guidelines recommend and what the labels authorise.

Sources of the dates: Drugs@FDA database search on 29 October 2026 [6], European summaries of product characteristics [15, 42, 43], US labels [7, 16, 44] and the Portuguese summary of product characteristics from Infomed [8]. It was not possible to consult the ANVISA electronic medicines compendium, so the Brazilian regulatory status of the individual drugs remains a limitation; the Brazilian national protocol was nonetheless consulted in full [31].

From the standpoint of zidovudine, the Portuguese label adds relevant elements of neonatal safety that do not appear in the dose tables of the guidelines: the contraindication of oral formulations in patients with a neutrophil count below 0.75×10⁹/L or haemoglobin below 7.5 g/dL, the specific contraindication in newborns with hyperbilirubinaemia requiring treatment beyond phototherapy or with transaminases five times above the upper limit of normal, and the warnings about excipients — sodium benzoate, whose displacement of bilirubin from albumin may aggravate neonatal jaundice up to kernicterus, propylene glycol, which in concomitant administration with any alcohol dehydrogenase substrate, such as ethanol, may induce serious adverse effects in newborns, and maltitol, to be avoided in hereditary fructose intolerance. The label also recalls that anaemia is not usually seen before six weeks of therapy and requires the use of a syringe graduated in 0.1 mL, given the small volumes involved [8]. The US label adds the thresholds for interruption because of toxicity: significant anaemia, defined by haemoglobin below 7.5 g/dL or a reduction of more than 25% from baseline, and significant neutropenia, defined by granulocytes below 750 per cubic millimetre or a reduction of more than 50% from baseline, may require interruption until there is evidence of marrow recovery, and erythropoietin may be used as an adjunctive measure on reintroduction [7].

7.10. Pivotal trials and developments of the last twelve months

These trials are not comparable with one another: they differ in populations, in comparators, in the viral load thresholds used as outcomes and in the era in which they were conducted, and relative reductions have to be read together with the corresponding absolute rate — a two-thirds reduction on 25.5% in 1994 and a difference of 0.2 percentage points on 1.1% in 2026 are quantities of a different nature. Two of the included trials had negative or non-significant results: PROMISE-EPI did not reach statistical significance, being underpowered, and the breastfeeding component of PROMISE showed no superiority of either strategy. The CREATE trial, of cabotegravir and rilpivirine in pregnancy, was withdrawn without enrolling participants, and the CALENDULA trial, of cabotegravir with lenacapavir, was likewise withdrawn [51].

The developments of the last twelve to eighteen months, classified by level of evidence, are as follows. As published results: IMPAACT P1115, published in 2025, demonstrated that remission without therapy for 48 weeks or more can be achieved in children with intrauterine infection treated very early, although only 6 of 54 met the criteria for interruption and 3 of the 6 had rebound, which makes close monitoring indispensable and does not constitute a basis for interrupting therapy outside a trial [48]; and IMPAACT 2023, published in 2026, established the dosing and safety of dolutegravir in term newborns [49]. As guideline changes: the US update of 25 June 2026 included dolutegravir in the triple presumptive regimen and simplified the diagnostic schedules [4]; version 13.0 of the European guidelines, from 2025, adopted dolutegravir as first line in term newborns and included bictegravir with tenofovir alafenamide and emtricitabine in pregnancy [34]; and the World Health Organization published updated recommendations, announced on 7 January 2026, with nevirapine preferred for six weeks of prophylaxis and abacavir, lamivudine and dolutegravir in a paediatric formulation for high-risk infants [36]. As a regulatory decision: lenacapavir was approved for pre-exposure prophylaxis in the United States on 18 June 2025 [6] and recommended by the World Health Organization on 14 July 2025, with the note that all approved prophylaxis options may be used in pregnancy and that a systematic review based on 193 pregnancies occurring in PURPOSE 1 concluded that lenacapavir probably has no effect on adverse pregnancy and birth outcomes compared with oral emtricitabine and tenofovir disoproxil, with moderate certainty of evidence [37]. As a public health certification: the elimination of vertical transmission in Brazil, certified on 18 December 2025 [17].

In the preclinical and early trial phase, the most promising line is that of broadly neutralising antibodies. Proof-of-concept, pharmacokinetic and safety studies have already been carried out in neonates; they require subcutaneous injection, probably monthly, and could constitute a particularly effective prophylaxis for breastfed infants, but efficacy data are lacking [3]. There is pharmacokinetic and pharmacodynamic modelling of doses of the antibody VRC07-523LS in infants [51], and since 5 January 2026 a phase 1 trial of the antibody PGT121.414.LS alone and in combination with VRC07-523LS has been recruiting in 48 HIV-1-exposed infants (NCT06517693), in addition to two phase 1/2 trials of neutralising antibodies in infected infants with a planned start in January 2027 (NCT07655128 and NCT07655141). Also under way are a phase 4 trial of dolutegravir with lamivudine in pregnant women without previous therapy, recruiting since 10 September 2026 with 210 planned participants (NCT07616739), and the Tshireletso trial of injectable cabotegravir as pre-exposure prophylaxis in a breastfeeding population in Botswana, active without recruiting with 500 participants (NCT05986084); an observational study of injectable cabotegravir and rilpivirine in pregnancy (NCT07637942) was due to start on 15 October 2026 [51].

7.11. Agents to avoid, perioperative care and transition

The agents to avoid in this condition are determined by the maternal regimen and by the infant's age, and not by intrinsic susceptibility. In the mother, methylergometrine and other ergot alkaloids should be avoided with potent cytochrome P450 3A4 inhibitors, because of the risk of exaggerated vasoconstriction, and the need for additional uterotonics should be anticipated with inducers of the same cytochrome [4]. In the infant, nevirapine should not be used where there is HIV-2 infection or coinfection, because of its lack of activity [4]; ritonavir-boosted lopinavir should be avoided in routine prophylaxis and especially in preterm infants [3]; abacavir requires a negative HLA-B*5701 test before use [4]; and maraviroc should be avoided in routine prophylaxis [3]. In the maternal perioperative setting, epidural analgesia is considered safe with any regimen [4]. In maternal therapy, hepatitis B screening requires that lamivudine or emtricitabine not be the only drug with activity against that virus [4].

Transition of care has two dimensions. The first concerns the exposed uninfected child, who after meeting the criteria for definitive exclusion is followed in primary care; in the Brazilian model, exposed uninfected children may be followed in primary care alone, while those with a confirmed diagnosis continue under shared care between the specialist service and primary care [31]. The cross-cutting recommendation, of particular importance for clinical genetics, is that the long-term clinical record of an uninfected child should include information on perinatal exposure to HIV and to antiretrovirals, in case unusual symptoms arise later in life or late adverse effects are identified in these children in the future [4]. The second dimension concerns the adolescent or woman with perinatally acquired infection who becomes pregnant: the US guidelines devote a section to her, because of the particular features of a long history of antiretroviral exposure, accumulated resistance and specific psychosocial needs [4].

8. Reproductive Options and Family Counselling

8.1. Recurrence risk and nature of the condition

Perinatal HIV infection is not hereditary and therefore has no recurrence risk in the Mendelian sense. What recurs between pregnancies is the exposure: a woman with HIV infection who becomes pregnant again is once more in the same situation of risk, and the determinant of that risk is the degree and duration of virological suppression, not a genotype. This is the central message of counselling, and its quantitative translation is direct: with therapy started before conception, maintained throughout pregnancy and with a viral load below 50 copies/mL close to delivery, there was no transmission in 2651 children in the French perinatal cohort, nor in 5482 children in a broader set from the same cohort, with upper confidence limits of 0.1% and 0.07% [3, 10]; with therapy started after conception and an undetectable viral load close to delivery, rare transmissions nonetheless occurred, at a rate of 0.57% (26 of 4596; 0.37 to 0.83), and the risk increased with the gestational age at initiation and with preterm birth [3].

Host genetic modifiers exist, as described in section 2, but have no application in reproductive counselling. There is no recommendation for CCR5 genotyping or for genotyping of any other modifier gene in any perinatal guideline consulted; the only mandatory genetic test in some scenarios is for the HLA-B*5701 allele, for abacavir safety, and that is a maternal or infant pharmacogenetic test, not a test of transmission risk [4]. The CCR5 promoter variant associated with increased perinatal transmission in African American infants, with a relative risk of 5.9 for homozygotes of the 59356T allele, is of pathophysiological and epidemiological interest but has no established clinical utility [22].

8.2. Conception in couples with different serological status

Before attempting to conceive, the person with HIV infection should be on therapy and have achieved sustained suppression, ideally documented by two viral load measurements below the limit of detection obtained at least three months apart; this requirement maximises the person's own health, prevents sexual transmission and minimises the risk of transmission to the infant when conception occurs. For couples in which one member has the infection and is on therapy with sustained suppression, condomless sexual intercourse is a method of conception that does not transmit the infection to the uninfected partner; people on therapy who maintain a viral load below 200 copies/mL do not transmit the virus to sexual partners [4].

The basis for this statement is two large studies in heterosexual couples with different serological status. In the randomised HPTN 052 trial, which assessed whether immediate initiation of therapy in people with CD4 T-lymphocyte counts between 350 and 550 cells per cubic millimetre prevented sexual transmission better than deferral, and in which 54% of participants were from Africa, 30% from Asia and 16% from the Americas, early initiation led to a 93% reduction in the rate of sexual transmission; neither in that trial nor in the PARTNER study did genetically linked transmissions occur while the partner with infection was virologically suppressed, with thresholds of 400 and 200 copies/mL respectively. In a prospective cohort study of 161 couples with different status who planned to conceive, 133 of them with the man infected, in which the person with infection had been on suppressive therapy for at least six months and the couple opted for natural conception, 144 natural pregnancies occurred and 107 children were born, without any sexual transmission to the partner or vertical transmission to the infant [4].

Before attempting to conceive, both partners should be screened for genital tract infections, and repeat screening during attempts may be considered according to individual risk and the duration of the preconception period; under therapy with sustained suppression and/or pre-exposure prophylaxis, sexually transmitted infections do not increase the risk of HIV transmission, but their screening and treatment are important for the health of both and for the prevention of pregnancy complications. If conception does not occur within 12 months, or sooner if indicated by age or by obstetric factors, infertility should be investigated, including semen analysis, because HIV infection and possibly antiretrovirals are associated with a higher prevalence of semen abnormalities, such as low sperm count, reduced motility, a high proportion of abnormal forms and reduced seminal volume; in some cases early assessment is justified by concern about higher rates of infertility [4].

8.3. Pre-exposure prophylaxis and assisted conception options

Pre-exposure prophylaxis should be discussed with all sexually active people without infection, including when they are trying to conceive, and counselling should cover the benefits in preventing acquisition and perinatal transmission, as well as potential adverse effects in the periconception, antenatal, postpartum and breastfeeding periods; it should be offered to those who want it or have an indication. The combination of tenofovir disoproxil with emtricitabine is approved for all populations; the combination of tenofovir alafenamide with emtricitabine is likewise approved but not for prevention of transmission through receptive vaginal sex; long-acting injectable cabotegravir was approved in 2021 for men and women, but safety and pharmacokinetic data in pregnancy are limited and, if pregnancy occurs on this drug, the long half-life and the sparse data should be discussed with shared decision-making about continuing or switching to tenofovir disoproxil with emtricitabine. When the partner with infection has not achieved suppression, has unknown suppression status or may have inconsistent adherence in the periconception period, pre-exposure prophylaxis for the uninfected partner is recommended; if this is declined, additional guidance for safer conception may be needed [4]. The update of 31 March 2026 recognised that there are now several oral and long-acting injectable pre-exposure prophylaxis options approved to prevent transmission through receptive vaginal sex, including in pregnancy and during breastfeeding, and that the choice should be made by shared decision-making [4]; the World Health Organization added that all approved options can be used safely in pregnancy [37].

For conception without vaginal intercourse there are three routes. When the person with infection is the one becoming pregnant and the inseminating partner is uninfected, assisted insemination in the periovulatory period, at home or in the clinic, with the partner's semen, eliminates the risk of transmission to the partner. When the person with infection is the inseminating partner, the use of donor semen from an uninfected donor eliminates the risk. Sperm preparation techniques, such as washing followed by HIV RNA testing of the sample, combined with intrauterine insemination, in vitro fertilisation or fertilisation with intracytoplasmic sperm injection, are no longer routinely recommended: they were developed before the efficacy of therapy and prophylaxis in reducing risk was demonstrated, they are expensive and technically demanding, and their current role is unclear. Gamete donation and surrogacy are options, and when these occur through health services they should include HIV testing and viral load monitoring in the respective protocols [4].

8.4. Counselling on the choice of regimen before and during pregnancy

Counselling about drugs has to balance two risks of a different nature: the risk of malformation, uncertain and small, and the risk of losing virological suppression when changing regimen, concrete and with a direct consequence for transmission. The US guidelines are explicit: therapy should be started as early as possible in every pregnancy and should not be deferred over concerns about teratogenicity in the first trimester; when assessing whether to maintain an effective regimen in early pregnancy, the potential risk of virological rebound on switching should be weighed against the known or unknown risks of birth defects associated with the current regimen and with the timing in gestation [4]. The minimum content of this counselling includes: the information that antiretrovirals, in general, do not increase the risk of birth defects, with the caveat that data for many drugs are limited and evolving; the explanation that pregnancy registry data exclude 1.5-fold or twofold increases in overall risk but do not exclude increases in rare defects, such as those of the neural tube; discussion of future reproductive plans and their timing, of the risks and benefits of conceiving on specific drugs, and of contraceptive options to avoid unplanned pregnancies; and folic acid supplementation according to the usual rules for the general population [3, 4].

Counselling about infant feeding is today an integral part of reproductive counselling and should begin before conception or as early as possible in pregnancy, being revisited throughout gestation and again after delivery. One possible way to open the conversation, proposed in the guidelines, is to ask how the person would like to feed the baby, explain that formula eliminates the risk of transmission through milk, state that fewer than one in a hundred breastfed infants would acquire the infection through breast milk when therapy is taken and suppression is sustained throughout lactation, add that the risk is not zero, and ask what information would help in deciding. The conversation should be private and non-judgemental, seek to understand the motivations for breastfeeding — bonding, health benefits for the dyad, cultural expectation, avoiding disclosure of serological status, inability to buy formula — and the barriers to formula, and address the risk of transmission, the importance of sustained suppression and the common difficulties with adherence postpartum. The shared decision-making process should be documented in a way accessible to all professionals involved; some services require a signed form confirming that counselling took place, a practice that others consider stigmatising and replace with an entry in the clinical record [4].

8.5. Cascade testing, carriers and specific issues

There is no cascade testing in the genetic sense, but there is an epidemiological equivalent of equal importance: screening of sexual partners and of previous children. The guidelines recommend, in every pregnancy, that partners of unknown status be referred for testing; they recommend assessing knowledge of the partner's status and the need for sexually transmitted infection screening, discussing whether serological status has been disclosed to sexual partners and offering pre-exposure prophylaxis where indicated [4]. For children already born, age-appropriate testing is recommended whenever there are compatible signs or symptoms, even without documented or suspected exposure, and in children in foster care or adopted whose maternal serological status is unknown [4].

Three specific issues deserve note. The first concerns people with perinatally acquired infection who reach reproductive age, a group with a long history of antiretroviral exposure and potential accumulated resistance, for whom the US guidelines maintain a dedicated section on antenatal care and management of therapy [4]. The second is gender identity and expression: counselling should ask, without judgement, how the person wishes to be referred to as a parent and how they wish to name infant feeding, and the UK guidelines devote a subsection to the care of transgender and gender-diverse people; as for resuming testosterone after delivery or during lactation, there are no evidence-based guidelines, there being one case report of restarting at 13 months postpartum during lactation in which the calculated milk-to-plasma ratio was below 1.0, the estimated relative infant dose below 1%, without observable effects and with undetectable serum testosterone concentrations in the infant [4]. The third is the social context: more than half of the transmissions reported in England since 2014 occurred in the children of women with concurrent safeguarding, mental health, substance use or insecure housing problems, which highlights social inequalities and makes multidisciplinary and multi-agency support a component of counselling, not an accessory; the message that a negative result reflects only the absence of infection at that moment, and urgent screening of women presenting in labour without antenatal care, are parts of that same strategy [3].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Global picture and what each figure measures

In 2025 there were 41.0 million people living with HIV infection worldwide (35.3 to 47.5 million), of whom 1.3 million (1.0 to 1.7 million) were children aged 0 to 14 years, and 53% of all infected people were women and girls. New infections in children aged 0 to 14 years fell from 300,000 (200,000 to 450,000) in 2010 to 190,000 in 2015, 140,000 in 2020, 100,000 (70,000 to 160,000) in 2024 and 93,000 (63,000 to 140,000) in 2025, a 69% reduction since 2010; it is the group with the greatest progress, but progress has stalled in recent years. In 2025, 87% of pregnant women with infection (73 to more than 98%) had access to antiretrovirals to prevent transmission to their child. Access to therapy was 78% for all infected people, but only 55% (39 to 78%) in children aged 0 to 14 years against 79% in adults, and in children the first pillar of the 95-95-95 targets stood at 65% [1].

It is worth specifying what these figures measure. The value of 93,000 is an estimate of new infections in children aged 0 to 14 years, not a count of notified cases or a transmission rate; it covers intrauterine, intrapartum and breastfeeding transmission and includes children infected in previous years who have only now acquired the infection through breast milk. The difference between 87% coverage of antiretrovirals in pregnant women and 55% coverage of therapy in children is the numerical expression of a structural problem: prevention has advanced faster than paediatric diagnosis and treatment.

9.2. Europe and Portugal

In the World Health Organization European Region, 105,922 people were diagnosed with HIV infection in 2024, reported by 49 of the 53 countries, of whom 24,164 in the countries of the European Union and the European Economic Area, corresponding to a rate of 5.3 per 100,000 inhabitants in the latter area; more than half of the diagnoses (54%) were late, with a CD4 T-lymphocyte count below 350 cells per cubic millimetre. An estimated 3,200,000 people (2,800,000 to 3,400,000) are living with infection in the European Region, of whom about 63% are on therapy [2].

Vertical transmission accounted, in 2024, for 1.1% of diagnoses in the European Union and the European Economic Area with a known mode of transmission, corresponding to 196 people; 75% of them (147) were migrants, and 41.8% of the migrants (82) originated from sub-Saharan Africa. The proportion rose from 0.9% in 2015 to 1.1% in 2024, but the absolute number fell from 227 in 2023 to 196 in 2024 [2]. This is one of the figures that requires careful reading: these are diagnoses attributed to vertical transmission, made in people of any age who may have acquired the infection in another country and many years earlier; it is not a measure of vertical transmission occurring in Europe in 2024, and the high proportion of migrants shows precisely that.

The country data, from 2015 to 2024, illustrate the same ambiguity. In France, diagnoses attributed to vertical transmission numbered 48 in 2015 and 59 in 2024, with a cumulative total of 907; in Germany 29 and 22, with a peak of 61 in 2022 and a cumulative total of 623; in Italy 16 and 3, cumulative 244; in Spain 3 and 5, cumulative 167. In Portugal the annual values were 8, 6, 10, 5, 2, 3, 6, 6, 7 and 5 between 2015 and 2024, with a cumulative total of 601; in the European Union and European Economic Area as a whole, 204 in 2015, 107 in 2020, 313 in 2022, 227 in 2023 and 196 in 2024, with a cumulative total of 5953; in the countries outside that area, 415 in 2024, and in the European Region as a whole, 611 in 2024, with a cumulative total of 16,194 [2]. The World Health Organization Regional Office for Europe, with the European Centre for Disease Prevention and Control, supports Member States in preparing and submitting dossiers for validation and maintenance of the elimination of vertical transmission of HIV and syphilis [2]; among the countries already validated by the World Health Organization for the elimination of vertical transmission of HIV are, outside the Americas, Armenia, Belarus, Malaysia, the Maldives, Oman, Sri Lanka and Thailand [17].

In the United Kingdom, whose data are the most detailed in Europe, the number of children born in the country and diagnosed with vertically acquired infection before the age of 16 fell from 65 in 2006–2007 to fewer than 5 per year since 2014, and the transmission rate in women with a viral load below 50 copies/mL at delivery was 0.11% in England in 2017–2018 [3].

Portugal has an epidemiological particularity relevant to diagnosis: HIV-2, endemic in West Africa, in Angola and in Mozambique, is well documented in France and in Portugal, countries with large migrant populations from those regions [4]. This circumstance has three direct clinical consequences: the need to consider HIV-2 when the laboratory pattern is suggestive, the contraindication of nevirapine in the presumptive regimen of the exposed newborn, and the need to send samples to laboratories that have specific tests available [3, 4].

9.3. United States

In the United States, universal antenatal screening, therapy in every pregnancy, planned caesarean section when the viral load exceeds 1000 copies/mL close to delivery, appropriate antiretroviral management of the newborn and feeding counselling have reduced perinatal transmission to 1% or less, a value likewise achieved in Europe. The goal set by the centres for disease control is elimination, defined as an incidence below one infection per 100,000 live births and a rate below 1% among exposed infants, and the obstacle identified is incomplete implementation of routine antenatal testing and its repetition, as well as of the other recommended interventions [4]. HIV-2 is rare: between 2010 and 2017, more than 99.9% of identified infections were classified as HIV-1 and HIV-2 diagnoses remained below 0.03% [4].

9.4. Brazil and Latin America

Brazil was certified by the World Health Organization on 18 December 2025 for the elimination of vertical transmission of HIV, becoming the first country with more than one hundred million inhabitants to reach that milestone and the largest country in the Americas to do so. The criteria met include reduction of vertical transmission to less than 2% and coverage above 95% for antenatal care, routine testing and timely treatment of pregnant women with infection; beyond the validation targets, the country demonstrated the provision of quality services, robust data and laboratory systems, and a commitment to human rights, gender equality and community participation. The process followed a progressive, subnational approach, first certifying states and municipalities with more than one hundred thousand inhabitants, a practice begun in 2023; the assessment was conducted by independent experts and formally recommended by the World Health Organization Global Validation Advisory Committee, after submission of the final report in June 2025 [17].

Brazil is one of 19 countries and territories validated by the World Health Organization for the elimination of vertical transmission of HIV, twelve of them in the Americas: Cuba was the first in the world, in 2015, followed by Anguilla, Antigua and Barbuda, Bermuda, the Cayman Islands, Montserrat and Saint Kitts and Nevis in 2017, Dominica in 2020, Belize in 2023, and Jamaica and Saint Vincent and the Grenadines in 2024. Over the decade from 2015 to 2024, more than 50,000 paediatric infections were averted in the Region of the Americas through implementation of the elimination initiative, now integrated into the EMTCT Plus initiative, which covers the elimination of vertical transmission of HIV, syphilis, hepatitis B and congenital Chagas disease, in collaboration with the United Nations Children's Fund and the Joint United Nations Programme on HIV/AIDS [17].

Some media reports mentioned an additional criterion of an incidence below 0.5 cases per thousand live births; that value does not appear in the official statement consulted and is not used here.

9.5. Founder populations and population variation of the genetic modifiers

The population variation relevant in this condition is not in the prevalence of the disease but in modifier alleles, and the most studied case is CCR5Δ32. The allele frequency is about 0.092 in populations of European ancestry and about 0.10 in the European American population of the United States, with about 1% homozygotes and 20% heterozygotes in people of western European origin, and the allele was absent in the tested samples from western and central Africa and from Japan [19, 22]. This distribution means that the protection conferred by this allele is practically non-existent in the populations where the paediatric epidemic is most intense.

The converse applies to the promoter variant associated with greater perinatal transmission: the CCR5 59356T allele, homozygosity for which was associated with a relative risk of 5.9 for perinatal transmission, had a frequency of 21% in African Americans against 6% in Hispanic and 3% in Caucasian populations, while the variants associated with lower perinatal transmission were less common in African Americans [22]. The copy number of CCL3L1, another modifier, is higher in Africans, followed by East Asians, Amerindians, Central and South Asians, Middle Eastern populations and Europeans, but susceptibility is associated not with the absolute value but with having a number below the population-specific mean (Gonzalez and colleagues, 2005) [18]. The immunogenetic axis likewise has population structure: the association of HLA-B*35-Px with accelerated progression was observed in both Black and White people, which supports a direct effect of those alleles on the immune response, and in a study of female sex workers in Côte d'Ivoire the exposed seronegative women showed an increased frequency of inhibitory KIR genes in the absence of their HLA ligands [18].

9.6. Special populations

Three groups stand out because of disproportionate risk. The first is migrant women in Europe, who account for 75% of diagnoses attributed to vertical transmission in the European Union and the European Economic Area, with 41.8% of these originating from sub-Saharan Africa [2]; the practical relevance is twofold, because of the need to consider HIV-2 and because of barriers of access and literacy. The second is women with multiple social vulnerabilities: more than half of the 48 transmissions reported in England since 2014 occurred in the children of women with concurrent safeguarding, mental health, substance use or insecure housing problems [3]. The third, documented in the US guidelines in relation to feeding decisions, is Black women, disproportionately affected by the infection in the United States, whose children carry a greater burden of conditions that breastfeeding could alleviate, in a context where structural racism, poverty and segregation contribute to lower initiation and continuation of breastfeeding, even in people without infection; the document adds that, even in the United States, some people have limited access to safe water — an estimated 17% of the population depended, in 2010, on unregulated private wells — or difficulty obtaining formula [4].

9.7. Networks, resources and gaps

There is no European Reference Network dedicated to this condition, as it is an acquired infection rather than a rare disease, and Orphanet assigns it no code. The functional networks exist outside that framework. Penta is an international paediatric infectious diseases research network that maintains, with the European AIDS Clinical Society, the paediatric section of the European guidelines and an international multidisciplinary perinatal virtual clinic, which accepts referrals on the treatment of children and adolescents, adults with perinatally acquired infection and management of the prevention of vertical transmission [34]. In the United Kingdom, the CHIVA association maintains paediatric standards and a network of regional centres, and there is a national paediatric virtual clinic, in addition to extra-regional referral centres [3]. In the United States, the national perinatal hotline (1-888-448-8765) is a federally funded service providing free clinical consultation, 24 hours a day, to professionals caring for pregnant women with infection and for mother–child pairs, and refers to local or regional specialists [4]. The Antiretroviral Pregnancy Registry maintains the prospective registry of exposures and publishes interim reports twice a year, in June and December [4].

Among patient and community organisations, the European AIDS Treatment Group stands out at European level. In Portugal there are GAT — Grupo de Ativistas em Tratamentos, Associação Abraço, which has offered free testing and support since 1992, and Associação Positivo, in addition to the information portal sida.pt. In Brazil there is the Rede Nacional de Pessoas Vivendo com HIV/Aids.

The knowledge gaps declared by the guidelines themselves are relevant for future research: the dose and safety of dolutegravir in preterm infants remain to be established, since below 37 weeks and 2 kg only nevirapine, lamivudine and zidovudine have defined dosing; the dose and safety of long-acting antiretrovirals from birth to the end of breastfeeding remain to be defined; the risk conferred by mastitis and by mixed feeding under virological suppression remains to be understood; implementation research is needed on how to optimise treatment and testing of the mother–child pair throughout breastfeeding; amplification diagnostics with shorter turnaround times are lacking; the sensitivity of RNA and DNA assays during prophylaxis and treatment needs to be assessed, and the optimal interval between cessation of prophylaxis and testing determined when long-acting agents are used; and the prevalence, significance and replication competence of cell-associated DNA in breast milk in the context of plasma suppression need to be assessed [4]. From 2026, the US panel began explicitly flagging these gaps in each section of the guidelines, with the aim of motivating research to resolve them [4].

10. Illustrative Clinical Case

The most instructive case in the literature in this area is that of the child known as the «Mississippi child», described by Persaud and colleagues in 2013 and followed up by Luzuriaga and colleagues in 2015. It is of interest here for three reasons: it documents a complete failure of the prevention chain, it shows what very early treatment can do to the viral reservoir, and it demonstrates, through the later relapse, the limits of that achievement.

The mother had no antenatal care at all. The rapid antibody test was positive during labour, and delivery, vaginal and spontaneous, occurred at 35 weeks of gestation before any antiretroviral prophylaxis was given. Maternal infection was confirmed by Western blot. The maternal viral load at 24 hours was 2423 copies/mL, the CD4 T-lymphocyte count at 14 days was 644 cells per cubic millimetre, and the virus was wild type, subtype B. The mother did not start therapy: at 26 months after delivery she had 513 cells per cubic millimetre and a viral load of 6763 copies/mL, and at 28 months fully replication-competent virus was recovered from her blood, at a frequency of 137 infectious units per million resting CD4 T lymphocytes, with replication kinetics similar to those of the laboratory reference strain. Neither mother nor child had class I HLA alleles associated with spontaneous control, such as HLA-B27 or HLA-B*57, and both were wild type for CCR5 [28].

The child started therapy at 30 hours of life, with a three-drug regimen — zidovudine 2 mg/kg every six hours, lamivudine 4 mg/kg twice daily and nevirapine 2 mg/kg twice daily — chosen to serve as prophylaxis for a high-risk exposure and to minimise the emergence of resistant variants should infection have already occurred in utero. Viral DNA was detected in peripheral blood mononuclear cells at 30 hours of life and viral RNA was 19,812 copies/mL in a separate sample taken at 31 hours, which met the standard diagnostic criteria for infection; therapy was therefore continued. In the first week, nevirapine was replaced by ritonavir-boosted lopinavir to reduce the risk of resistance in case of incomplete adherence — a decision taken before the US regulatory authority's warnings against the use of lopinavir with ritonavir in infants under 14 days of age. The viral load remained detectable at three subsequent time points, with 2617 copies/mL at 6 days, 516 at 11 days and 265 at 19 days, before falling below the limit of detection at 29 days of age, with a biphasic decline similar to that described in infants and adults on therapy. The CD4 T-lymphocyte percentage at 8 days was 69%. The child was not breastfed [28].

Adherence in the first year was considered adequate, based on pharmacy records showing timely prescription refills, undetectable viral loads and erythrocyte macrocytosis during zidovudine treatment. Doubt arose at 18 months, when the mean corpuscular volume, which had been above 101 femtolitres before 15 months, fell to 95 femtolitres, although the viral load remained undetectable, below 20 copies/mL. Between 18 and 23 months the child missed several appointments; when she returned, at 23 months, the mother reported that therapy had been stopped at 18 months, while pharmacy records indicated that the last prescription had been collected at 15 months. At 23 and 24 months the viral load was undetectable, and at 24 months both the polymerase chain reaction for viral DNA and the antibody test were negative [28].

At the time of the first report, the child was 30 months old, had received no antiretroviral since 18 months, the viral load remained undetectable by routine clinical assays and no circulating antibodies were detected; CD4 T-lymphocyte percentages were within or above the normal range for age at every time point tested, and growth and development were normal. Ultrasensitive studies quantified persistence: viral RNA was detected at a level of one copy per millilitre in plasma at 24 months, but not at 26 months (below 2 copies/mL); total proviral DNA was detected close to the limit of detection in mononuclear cells, with fewer than 2.7 copies per million at 24 months and 4.2 copies per million at 26 months, but was not detected in resting CD4 T lymphocytes at either time point; and a culture of 22 million resting CD4 T lymphocytes obtained at 24 months, during a period without therapy, produced no replication-competent virus, with a limit below 0.05 infectious units per million. T-cell activation markers were within normal limits and no specific CD4 and CD8 T-lymphocyte responses were detected at 28 months [28]. Figure 1 of the article, available at https://pmc.ncbi.nlm.nih.gov/articles/PMC3954754/figure/F1/, shows in its three panels the biphasic decline of viral load before and after the start of therapy, sustained control during and after therapy with the time points of the last prescription at 15 months and of the last reported administration at 18 months, and the CD4 T-lymphocyte percentage from birth to 30 months in relation to the normal range.

Follow-up altered the interpretation. After the first report, the viral load remained undetectable by standard clinical assays, with normal CD4 and CD8 T-lymphocyte counts, and ultrasensitive single-copy assays confirmed the absence of viraemia up to 21.9 months without therapy, with values below 0.4 and 0.5 copies per millilitre. At a routine visit at 46.4 months of age, after 27.6 months without therapy, the viral load rose to 16,750 copies/mL, a result confirmed on repeat testing. The child had no symptoms or signs of acute retroviral syndrome and no intercurrent illnesses, including Epstein-Barr virus and cytomegalovirus, and had received no vaccines; the epidemiological risk factors for reinfection — breast milk, pre-chewed food and sexual abuse — were investigated and excluded. Within 72 hours of restarting therapy the viral load fell to 2658 copies/mL, the percentage of circulating CD4 T lymphocytes rose to 43% and the CD4 to CD8 lymphocyte ratio normalised [52].

During the period of remission, viral DNA was detected intermittently in unfractionated mononuclear cells and in fractions enriched for resting and activated CD4 T lymphocytes, with medians of 2.7 (interquartile range 2.2 to 3.4), 3.5 (2.1 to 4.4) and 3.3 (1.8 to 5.5) copies per million cells respectively, before relapse; however, no replication-competent reservoirs were detected at any time point, in a cumulative total of 64 million resting CD4 T lymphocytes cultured. Specific humoral and cellular responses were not detected up to 21.9 months without therapy, but appeared at the time of relapse: a positive enzyme immunoassay was confirmed by Western blot, with reactivity to Env (gp160) and Gag (p24) and, 72 hours later, reactivity to Env (gp160, gp120 and gp41), Gag (p55, p24 and p17) and polymerase (p66); the absence of a response to integrase p31 was compatible with replication of at least one month but less than two. Single-genome amplification and sequencing of the complete env gene directly from plasma virus at relapse revealed 98.6% sequence identity with the maternal plasma viral sequences collected 24 months after delivery, with a diversity of 0.1%, a value compatible with those described in transmitted founder viruses in vertical transmission and with the absence of continuous replication during the period of remission; phenotypic and genotypic analysis showed R5 tropism [52].

The authors' conclusion, which is also the clinical message, is twofold. The reappearance of viraemia after a substantial period of quiescence is compatible with the model of latency in which long-lived resting memory CD4 T lymphocytes are generated in the normal course of immunological memory formation, although other reservoirs are possible; and very early treatment can markedly restrict, but not eradicate, the reservoirs. Until further studies are completed, starting therapy as early as possible in infants and continuing it without interruption appears prudent [52]. Figure 1 of that second article, at https://pmc.ncbi.nlm.nih.gov/articles/PMC4440331/figure/F1/, shows the clinical viral load off therapy, with the limits of detection applied, and figure 2, at https://pmc.ncbi.nlm.nih.gov/articles/PMC4440331/figure/F2/, presents the neighbour-joining phylogenetic tree of the complete gp160 sequences of the child's relapse virus and of the maternal plasma virus, with four previously described clade B mother–child pairs included for comparison of diversity.

Two clarifications are necessary. The management of this case does not represent routine practice in any of its aspects: the initial regimen corresponds to what is now called presumptive therapy in a high-risk newborn, but the replacement of nevirapine by ritonavir-boosted lopinavir in the first week of life is now expressly advised against in infants under 14 days, and the interruption of therapy at 18 months was not a clinical decision but an unplanned interruption; no guideline recommends interrupting therapy in a child with confirmed infection outside the context of a trial with monitored analytical interruption [28, 52]. The framework that now exists for that question is the IMPAACT P1115 trial, described in section 7: of 54 children with intrauterine infection treated very early, only 6 (11%) met the stringent criteria for analytical interruption, at a median age of 5.5 years, and of the 6, four achieved remission of 48 weeks or more (67%; exact 95% confidence interval 22 to 96), one of them with late relapse at 79.3 weeks, and two had early relapse at 3.4 and 9.4 weeks, with all resuppressing the viral load on resumption of therapy and two showing mild symptoms of acute retroviral syndrome during relapse [48].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis of the infected child

The prognosis of the child with perinatally acquired infection depends almost entirely on the timing of diagnosis and of starting therapy. When the test is positive, paediatric guidelines recommend rapid initiation of therapy, defined as immediate or within a few days of diagnosis, without waiting for the confirmatory test, given the low probability of a falsely positive result; therapy is stopped if subsequent negative tests exclude infection [4]. Early and effective treatment restricts the size of the reservoir, reduces viral genetic variability and modifies the immune response, with two prognostic effects: durable virological suppression and, in some cases, the possibility of prolonged remission without therapy, demonstrated in the IMPAACT P1115 trial [4, 48]. Conversely, late diagnosis in an untreated child retains the grim prognosis of the pre-therapy era, with faster progression than in adults. Globally, the difference between 87% coverage of prevention in pregnant women and 55% coverage of therapy in children aged 0 to 14 years quantifies the scale of the problem that remains [1].

11.2. Prognosis of the exposed uninfected child

For the exposed uninfected child, the overall prognosis is good, but not identical to that of an unexposed child, and the magnitude of the difference depends on the setting. The risk of morbidity and mortality is higher than in unexposed children, with more pronounced differences in low- and middle-income countries; in Botswana, the strongest predictors of mortality at 24 months were exposure status and formula feeding, and in a Canadian cohort hospitalisations were more frequent across the lifespan, especially in children born small for gestational age and in those whose mothers had a detectable viral load at delivery. The most consistent consequences are infectious, in the first six months and the first year, with increases of 50% in the risk of diarrhoea and 70% in the risk of pneumonia in meta-analysis, and with the Spanish data showing a sevenfold higher risk of invasive group B streptococcal infection and a twenty-nine-fold higher risk of meningitis by that organism. For growth, the findings diverge by setting: growth faltering in Africa, above-average weight and frequent obesity in the US SMARTT study, and no differences in a large Danish study up to five years. For neurodevelopment, the results are inconclusive, with signs in some cohorts of poorer cognitive and language performance and of early structural brain changes, and without definitive evidence that any specific antiretroviral is responsible [4]. The practical recommendation arising from this uncertain prognosis is twofold: ensure complete and timely vaccination and alert the professionals following the child to the possibility of increased infectious risk [3]; and record, in the long-term clinical record, the perinatal exposure to HIV and to antiretrovirals, because it may be needed if unusual symptoms arise later or if late effects are identified in the future [4].

11.3. Surveillance table

The following table brings together the examinations, ages at onset, intervals and action thresholds recommended, with the source for each row. The intervals coincide with those described in sections 5 and 7.

11.4. Overall prognosis and public health targets

At population level, the prognosis of this condition is defined by the possibility of its elimination. In the United States and in Europe the transmission rate is at or below 1%; the elimination goal set by the US centres for disease control is an incidence below one infection per 100,000 live births and a rate below 1% among the exposed [4]; the World Health Organization validation criteria require transmission below 2% and coverage above 95% for antenatal care, testing and treatment [17]. Nineteen countries and territories have already been validated, and Brazil has shown that scale is not an obstacle [17]. What keeps the condition on the clinical horizon is not the absence of effective tools, but their incomplete implementation: late maternal diagnosis, interruptions of adherence postpartum, accumulated social vulnerabilities and, globally, the lag of paediatric diagnosis and treatment behind the progress of prevention [1, 3, 4].

12. Summary and Final Considerations

Perinatal HIV infection is today a condition in which the determinant of outcome is a number: the maternal viral load at the end of pregnancy. With antiretroviral therapy started before conception, maintained throughout gestation and with a viral load below 50 copies/mL close to delivery, transmission is practically nil, with no cases in series of thousands of mother–child pairs; without any intervention, the risk is 15 to 30% in pregnancy and at delivery and a further 15 to 20% over two years of breastfeeding. The entire clinical architecture of this area — opt-out screening in every pregnancy, repeat testing in the third trimester in at-risk situations, therapy in every pregnancy irrespective of viral load and lymphocyte count, intravenous zidovudine and planned caesarean section at defined thresholds, prophylaxis or presumptive therapy in the newborn, nucleic acid amplification schedules and shared decision-making about infant feeding — serves the aim of reducing and keeping that number below the threshold.

For clinical genetics, this condition is of interest in three distinct registers. It enters the differential diagnosis of syndromes with intracranial calcification, microcephaly and failure to thrive, alongside Aicardi-Goutières syndrome, cystic leukoencephalopathy from RNASET2 variants and the other congenital infections; the difference is not dysmorphological but laboratory-based, and depends on a maternal history that is often absent. It produces a growing population of exposed uninfected children, with increased infectious risk in the first year, context-dependent growth abnormalities, inconclusive neurodevelopmental signals and exposure to drugs with documented mitochondrial and haematological toxicity, which justifies recording perinatal exposure in the clinical record throughout life. And it offers one of the best characterised examples of host genetic modification in the face of an infectious agent, with the 32 base pair deletion in CCR5 — NM_001394783.1:c.554_585del, p.(Ser185IlefsTer32) — conferring recessive protection against infection by R5-tropic virus and a promoter polymorphism of the same gene being associated with greater perinatal transmission in African American infants, without either of these findings having, today, established clinical utility.

The international divergences are substantial and are not resolved by a simple hierarchy of quality. In newborn prophylaxis, the US guideline uses zidovudine for two weeks at low risk and a triple regimen with dolutegravir or nevirapine at high risk; the UK guideline uses zidovudine for two weeks or zidovudine with lamivudine for four weeks plus nevirapine for two weeks at treatment dose; the World Health Organization recommends six weeks of nevirapine for all exposed infants and abacavir with lamivudine and dolutegravir for those at high risk; and the 2020 Brazilian protocol retains four weeks of zidovudine for all, with three doses of nevirapine in at-risk scenarios. For the viral load threshold for caesarean section, the values range from 400 copies/mL in the United Kingdom and Europe to 1000 in the United States and Brazil. For infant feeding, the divergence is greater still, between the absolute contraindication of breastfeeding in the Brazilian protocol and the model of shared decision-making with support for breastfeeding under suppression adopted in North America, in the United Kingdom, in Switzerland and by the World Health Organization. These differences reflect distinct contexts of risk, of resources and of publication dates, and require the clinician to identify the standard applicable to their jurisdiction before transposing doses and thresholds.

Two methodological points deserve retention. The first is dolutegravir: a robust safety signal, with a prevalence of neural tube defects of 0.30% against 0.10%, published in a leading journal and with immediate consequences for practice, was wholly revised by extended surveillance within the same study and by an independent programme, until the prevalence converged at 0.11% and 0.098% without significant difference, now reflected in the European and US labels; the lesson is that the sample size needed to exclude an increase in rare defects is of the order of thousands of preconception exposures, and that reading an early signal requires that scale. The second is breastfeeding: the message that an undetectable viral load equates to the absence of transmission applies to sexual transmission but not to vertical or breastfeeding transmission, with residual rates of 0.11% at delivery in suppressed women in England and a monthly risk of 0.1% during breastfeeding under suppression, and with at least six published cases of transmission through breastfeeding with a maternal viral load below 50 copies/mL.

The immediate outlook includes the demonstration, in the IMPAACT P1115 trial, that very early therapy allows drug-free remission for 48 weeks or more in some children with intrauterine infection, although only 6 of 54 children met the criteria for interruption and 3 of the 6 relapsed; the definition of neonatal dosing of dolutegravir in IMPAACT 2023; and research into broadly neutralising antibodies as monthly prophylaxis for the breastfed infant, still without efficacy data. On the policy front, the landmark is the certification of the elimination of vertical transmission in Brazil, in December 2025, which makes explicit that the 93,000 new paediatric infections of 2025 do not result from a lack of tools, but from incomplete implementation — and that the most evident gap today is the coverage of therapy in children already infected, at 55% against 87% coverage of prevention in pregnant women.

13. References

1. UNAIDS. Global HIV & AIDS statistics — Fact sheet 2026. Geneva: UNAIDS; July 2026 [accessed 29 Oct 2026]. Available at: https://www.unaids.org/sites/default/files/2026-07/2026_UNAIDS_Global_HIV_Factsheet_en.pdf
2. European Centre for Disease Prevention and Control, WHO Regional Office for Europe. HIV/AIDS surveillance in Europe 2025 – 2024 data. Stockholm: ECDC; 2025 [accessed 29 Oct 2026]. Available at: https://www.ecdc.europa.eu/en/publications-data/hivaids-surveillance-europe-2025-2024-data
3. Byrne L, Short CE, Bamford A, Bradshaw D, Cheserem E, Clarke E, et al. BHIVA guidelines on the management of HIV in pregnancy and the postpartum period 2025. HIV Med. 2025;26(11):1757-1880. doi:10.1111/hiv.70091. PMID: 41201115.
4. Panel on Treatment of HIV During Pregnancy and Prevention of Perinatal Transmission. Recommendations for the Use of Antiretroviral Drugs During Pregnancy and Interventions to Reduce Perinatal HIV Transmission in the United States. Department of Health and Human Services; updated 25 June 2026 [accessed 29 Oct 2026]. Available at: https://clinicalinfo.hiv.gov/en/guidelines/perinatal
5. Connor EM, Sperling RS, Gelber R, Kiselev P, Scott G, O'Sullivan MJ, et al. Reduction of maternal-infant transmission of human immunodeficiency virus type 1 with zidovudine treatment. Pediatric AIDS Clinical Trials Group Protocol 076 Study Group. N Engl J Med. 1994;331(18):1173-1180. doi:10.1056/nejm199411033311801. PMID: 7935654.
6. U.S. Food and Drug Administration. Drugs@FDA: FDA-approved drugs (NDA 019655, 019910, 020564, 020636, 020933, 022128, 022145, 204790, 205786, 213983, 220020) [accessed 29 Oct 2026]. Available at: https://www.accessdata.fda.gov/scripts/cder/daf/
7. U.S. Food and Drug Administration. Zidovudine tablets — prescribing information (ANDA 090092; DailyMed set ID 437f55b2-f6d9-0065-e063-6294a90ab929, version 2, 19 January 2026) [accessed 29 Oct 2026]. Available at: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=437f55b2-f6d9-0065-e063-6294a90ab929
8. INFARMED. Retrovir 10 mg/ml solução oral (zidovudine) — Summary of Product Characteristics. ViiV Healthcare BV. Infomed, med_id 7496 [accessed 29 Oct 2026]. Available at: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=7496&tipo_doc=rcm
9. Cooper ER, Charurat M, Mofenson L, Hanson IC, Pitt J, Diaz C, et al. Combination antiretroviral strategies for the treatment of pregnant HIV-1-infected women and prevention of perinatal HIV-1 transmission. J Acquir Immune Defic Syndr. 2002;29(5):484-494. doi:10.1097/00126334-200204150-00009. PMID: 11981365.
10. Mandelbrot L, Tubiana R, Le Chenadec J, Dollfus C, Faye A, Pannier E, et al. No perinatal HIV-1 transmission from women with effective antiretroviral therapy starting before conception. Clin Infect Dis. 2015;61(11):1715-1725. doi:10.1093/cid/civ578. PMID: 26197844.
11. Townsend CL, Byrne L, Cortina-Borja M, Thorne C, de Ruiter A, Lyall H, et al. Earlier initiation of ART and further decline in mother-to-child HIV transmission rates, 2000-2011. AIDS. 2014;28(7):1049-1057. doi:10.1097/qad.0000000000000212. PMID: 24566097.
12. Kankasa C, Mennecier A, Sakana BLD, Molès JP, Mwiya M, Chunda-Liyoka C, et al. Optimised prevention of postnatal HIV transmission in Zambia and Burkina Faso (PROMISE-EPI): a phase 3, open-label, randomised controlled trial. Lancet. 2024;403(10434):1362-1371. doi:10.1016/s0140-6736(23)02464-9. PMID: 38484756.
13. Zash R, Makhema J, Shapiro RL. Neural-Tube Defects with Dolutegravir Treatment from the Time of Conception. N Engl J Med. 2018;379(10):979-981. doi:10.1056/nejmc1807653. PMID: 30037297. PMCID: PMC6550482.
14. Zash R, Holmes L, Diseko M, Jacobson DL, Brummel S, Mayondi G, et al. Neural-Tube Defects and Antiretroviral Treatment Regimens in Botswana. N Engl J Med. 2019;381(9):827-840. doi:10.1056/nejmoa1905230. PMID: 31329379. PMCID: PMC6995896.
15. European Medicines Agency. Tivicay (dolutegravir) — EPAR product information [accessed 29 Oct 2026]. Available at: https://www.ema.europa.eu/en/documents/product-information/tivicay-epar-product-information_en.pdf
16. U.S. Food and Drug Administration. TIVICAY and TIVICAY PD (dolutegravir) — prescribing information (NDA 204790; DailyMed set ID 485bc9db-8665-9f5a-e063-6394a90a7921, 22 May 2026) [accessed 29 Oct 2026]. Available at: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=485bc9db-8665-9f5a-e063-6394a90a7921
17. Pan American Health Organization. OMS certifica Brasil pela eliminação da transmissão do HIV de mãe para filho [news item]. Washington: PAHO; 18 December 2025 [accessed 29 Oct 2026]. Available at: https://www.paho.org/pt/noticias/18-12-2025-oms-certifica-brasil-pela-eliminacao-da-transmissao-do-hiv-mae-para-filho
18. Online Mendelian Inheritance in Man, OMIM. #609423: Human immunodeficiency virus type 1, susceptibility/resistance to. Baltimore: Johns Hopkins University; entry created 16 June 2005, last edited 16 October 2026 [accessed 29 Oct 2026]. Available at: https://omim.org/entry/609423
19. GeneGuidance. CCR5 — C-C motif chemokine receptor 5 [curated database]. Version of 2026-09-26.
20. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166.
21. National Center for Biotechnology Information. ClinVar VCV000008184.6: NM_001394783.1(CCR5):c.554_585del (p.Ser185fs) [accessed 29 Oct 2026]. Available at: https://www.ncbi.nlm.nih.gov/clinvar/variation/8184/
22. Online Mendelian Inheritance in Man, OMIM. *601373: Chemokine, CC motif, receptor 5; CCR5. Baltimore: Johns Hopkins University; entry created 11 August 1996, last edited 3 September 2020 [accessed 29 Oct 2026]. Available at: https://omim.org/entry/601373
23. Samson M, Libert F, Doranz BJ, Rucker J, Liesnard C, Farber CM, et al. Resistance to HIV-1 infection in caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature. 1996;382(6593):722-725. doi:10.1038/382722a0. PMID: 8751444.
24. Dean M, Carrington M, Winkler C, Huttley GA, Smith MW, Allikmets R, et al. Genetic restriction of HIV-1 infection and progression to AIDS by a deletion allele of the CKR5 structural gene. Hemophilia Growth and Development Study, Multicenter AIDS Cohort Study, Multicenter Hemophilia Cohort Study, San Francisco City Cohort, ALIVE Study. Science. 1996;273(5283):1856-1862. doi:10.1126/science.273.5283.1856. PMID: 8791590.
25. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. The mitochondrial DNA depletion syndromes — mitochondrial DNA polymerase deficiency. p. 415–422.
26. Blanche S, Tardieu M, Rustin P, Slama A, Barret B, Firtion G, et al. Persistent mitochondrial dysfunction and perinatal exposure to antiretroviral nucleoside analogues. Lancet. 1999;354(9184):1084-1089. doi:10.1016/s0140-6736(99)07219-0. PMID: 10509500.
27. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Intracranial calcification. p. 200–203.
28. Persaud D, Gay H, Ziemniak C, Chen YH, Piatak M, Chun TW, et al. Absence of detectable HIV-1 viremia after treatment cessation in an infant. N Engl J Med. 2013;369(19):1828-1835. doi:10.1056/nejmoa1302976. PMID: 24152233. PMCID: PMC3954754.
29. Brennan AT, Bonawitz R, Gill CJ, Thea DM, Kleinman M, Useem J, et al. A meta-analysis assessing all-cause mortality in HIV-exposed uninfected compared with HIV-unexposed uninfected infants and children. AIDS. 2016;30(15):2351-2360. doi:10.1097/qad.0000000000001211. PMID: 27456985.
30. Zash R, Jacobson DL, Diseko M, Mayondi G, Mmalane M, Essex M, et al. Comparative Safety of Antiretroviral Treatment Regimens in Pregnancy. JAMA Pediatr. 2017;171(10):e172222. doi:10.1001/jamapediatrics.2017.2222. PMID: 28783807. PMCID: PMC5726309.
31. Brasil. Ministério da Saúde. Protocolo Clínico e Diretrizes Terapêuticas para Prevenção da Transmissão Vertical do HIV, Sífilis e Hepatites Virais (Portaria SCTIE/MS n.º 55, de 11 de novembro de 2020). Brasília: Ministério da Saúde; page updated 21 January 2025 [accessed 29 Oct 2026]. Available at: https://www.gov.br/saude/pt-br/assuntos/pcdt/p/prevencao-da-transmissao-vertical-do-hiv-sifilis-e-hepatites-virais
32. Isaacson A, Diseko M, Mabuta J, Kgannyeng A, Mayondi G, Lockman S, et al. Long-term outcomes of children born with neural-tube defects in Botswana. BMC Pediatr. 2025;25(1):996. doi:10.1186/s12887-025-06313-0. PMID: 41275147. PMCID: PMC12751222.
33. Tukeman GL, Wei H, Lin YL, Wlodarczyk BJ, Finnell RH, Cabrera RM. Dolutegravir-induced neural tube defects in mice are folate responsive. AIDS. 2024;38(4):439-446. doi:10.1097/qad.0000000000003639. PMID: 37382903. PMCID: PMC10755062.
34. Ambrosioni J, Levi LI, Alagaratnam J, Sempere A, Mastrangelo A, Paioni P, et al. Major revision version 13.0 of the European AIDS Clinical Society guidelines 2025. HIV Med. 2026;27(1):18-32. doi:10.1111/hiv.70120. PMID: 41088922. PMCID: PMC12757896.
35. Nielsen-Saines K, Watts DH, Veloso VG, Bryson YJ, Joao EC, Pilotto JH, et al. Three postpartum antiretroviral regimens to prevent intrapartum HIV infection. N Engl J Med. 2012;366(25):2368-2379. doi:10.1056/nejmoa1108275. PMID: 22716975. PMCID: PMC3590113.
36. World Health Organization. WHO releases updated recommendations on HIV clinical management [news item]. Geneva: WHO; 7 January 2026 [accessed 29 Oct 2026]. Available at: https://www.who.int/news/item/07-01-2026-who-releases-updated-recommendations-on-hiv-clinical-management
37. Collins S. IAS 2025: WHO guidelines: major updates on injectables, AHD, breastfeeding, infant prophylaxis and peer support. HIV Treatment Bulletin. London: HIV i-Base; 2 August 2025 [accessed 29 Oct 2026]. Available at: https://i-base.info/htb/52185
38. Bispo S, Chikhungu L, Rollins N, Siegfried N, Newell ML. Postnatal HIV transmission in breastfed infants of HIV-infected women on ART: a systematic review and meta-analysis. J Int AIDS Soc. 2017;20(1):21251. doi:10.7448/ias.20.1.21251. PMID: 28362072. PMCID: PMC5467610.
39. Flynn PM, Taha TE, Cababasay M, Fowler MG, Mofenson LM, Owor M, et al. Prevention of HIV-1 Transmission Through Breastfeeding: Efficacy and Safety of Maternal Antiretroviral Therapy Versus Infant Nevirapine Prophylaxis for Duration of Breastfeeding in HIV-1-Infected Women With High CD4 Cell Count (IMPAACT PROMISE): A Randomized, Open-Label, Clinical Trial. J Acquir Immune Defic Syndr. 2018;77(4):383-392. doi:10.1097/qai.0000000000001612. PMID: 29239901. PMCID: PMC5825265.
40. Clarke DF, Acosta EP, Cababasay M, Wang J, Chain A, Teppler H, et al. Raltegravir (RAL) in Neonates: Dosing, Pharmacokinetics (PK), and Safety in HIV-1-Exposed Neonates at Risk of Infection (IMPAACT P1110). J Acquir Immune Defic Syndr. 2020;84(1):70-77. doi:10.1097/qai.0000000000002294. PMID: 31913995. PMCID: PMC7144886.
41. Brummel SS, Swanson S, Caniglia E, Lockman S, Zash R, Shapiro R. A lesson in embracing uncertainty in early safety signals from the Tsepamo study. AIDS. 2025;39(15):2266-2273. doi:10.1097/qad.0000000000004326. PMID: 40880283. PMCID: PMC13296845.
42. European Medicines Agency. Viramune (nevirapine) — EPAR product information [accessed 29 Oct 2026]. Available at: https://www.ema.europa.eu/en/documents/product-information/viramune-epar-product-information_en.pdf
43. European Medicines Agency. Isentress (raltegravir) — EPAR product information [accessed 29 Oct 2026]. Available at: https://www.ema.europa.eu/en/documents/product-information/isentress-epar-product-information_en.pdf
44. U.S. Food and Drug Administration. ISENTRESS (raltegravir) — prescribing information (NDA 022145; DailyMed set ID ccc6f2f9-696f-4ef9-bb4c-e0198d957b0b, 1 October 2024) [accessed 29 Oct 2026]. Available at: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=ccc6f2f9-696f-4ef9-bb4c-e0198d957b0b
45. Fowler MG, Qin M, Fiscus SA, Currier JS, Flynn PM, Chipato T, et al. Benefits and Risks of Antiretroviral Therapy for Perinatal HIV Prevention. N Engl J Med. 2016;375(18):1726-1737. doi:10.1056/nejmoa1511691. PMID: 27806243. PMCID: PMC5214343.
46. Lockman S, Brummel SS, Ziemba L, Stranix-Chibanda L, McCarthy K, Coletti A, et al. Efficacy and safety of dolutegravir with emtricitabine and tenofovir alafenamide fumarate or tenofovir disoproxil fumarate, and efavirenz, emtricitabine, and tenofovir disoproxil fumarate HIV antiretroviral therapy regimens started in pregnancy (IMPAACT 2010/VESTED): a multicentre, open-label, randomised, controlled, phase 3 trial. Lancet. 2021;397(10281):1276-1292. doi:10.1016/s0140-6736(21)00314-7. PMID: 33812487. PMCID: PMC8132194.
47. Kintu K, Malaba TR, Nakibuka J, Papamichael C, Colbers A, Byrne K, et al. Dolutegravir versus efavirenz in women starting HIV therapy in late pregnancy (DolPHIN-2): an open-label, randomised controlled trial. Lancet HIV. 2020;7(5):e332-e339. doi:10.1016/s2352-3018(20)30050-3. PMID: 32386721. PMCID: PMC10877544.
48. Persaud D, Coletti A, Nelson BS, Jao J, Capparelli EV, Costello D, et al. ART-free HIV-1 remission in children with in-utero HIV-1 after very early ART (IMPAACT P1115): a multicentre, open-label, phase 1/2 proof-of-concept study. Lancet HIV. 2025;12(11):e743-e752. doi:10.1016/s2352-3018(25)00189-4. PMID: 41015049. PMCID: PMC13047515.
49. Momper JD, Wang J, Bradford S, Mustich I, Johnston B, Whitson K, et al. Pharmacokinetics and safety of dolutegravir in term neonates exposed to HIV-1 (IMPAACT 2023): a phase 1, open-label, multicentre study. Lancet HIV. 2026;:S2352-3018(26)00226-2. doi:10.1016/s2352-3018(26)00226-2. PMID: 42906569.
50. Bekker LG, Das M, Abdool Karim Q, Ahmed K, Batting J, Brumskine W, et al. Twice-Yearly Lenacapavir or Daily F/TAF for HIV Prevention in Cisgender Women. N Engl J Med. 2024;391(13):1179-1192. doi:10.1056/nejmoa2407001. PMID: 39046157.
51. ClinicalTrials.gov [database]. Bethesda: National Library of Medicine; searched via API v2 on 29 October 2026. Available at: https://clinicaltrials.gov
52. Luzuriaga K, Gay H, Ziemniak C, Sanborn KB, Somasundaran M, Rainwater-Lovett K, et al. Viremic relapse after HIV-1 remission in a perinatally infected child. N Engl J Med. 2015;372(8):786-788. doi:10.1056/nejmc1413931. PMID: 25693029. PMCID: PMC4440331.