Main

The pattern of immune ontogeny in humans has evolved to meet the changing and distinct challenges to the immune system that arise in utero and in early life, through childhood and into adolescence and adulthood. The differences that are observed in terms of disease outcome for a range of infectious diseases, according to the age at which the particular infections arise, illustrate the impact of the distinct phases of immune system development on disease susceptibility. HIV infection is a case in point (Fig. 1). In the absence of antiretroviral therapy (ART), >50% of HIV-infected children die by 2 years of age1, whereas the median survival time is 11 years for ART-naive HIV-positive adults2. In the small subset of ART-naive individuals infected with HIV who maintain normal CD4+ T cell counts and immune function and, therefore, do not progress to disease, the mechanisms operating in children and adults are markedly different. Non-progressing adult infection, in so-called elite controllers, is typically characterized by a strong HIV-specific CD8+ T cell response restricted by 'protective' MHC class I molecules (for example, HLA-B27 and HLA-B57)3,4,5, such that high CD4+ T cell counts are maintained as a consequence of successful suppression of viraemia beneath the levels of detection (normally <50 HIV RNA copies per ml). By contrast, non-progressing HIV-infected children (who are defined here as healthy, ART-naive children aged more than 5 years who maintain absolute CD4+ T cell counts in the normal range for age-matched uninfected children (above 750 cells per mm3 of blood))6 maintain normal CD4+ T cell counts despite median viral loads of 30,000 copies per ml of plasma7, and the HLA-B alleles that strongly influence disease outcome in adult infection do not have a significant role in these children8. It is important to clarify the distinction between non-progressing adult and paediatric infection, which is defined by the maintenance of normal CD4+ T cell counts in the absence of ART, and long-term survivors, who typically have low CD4+ T cell counts and high levels of immune activation7, but may be asymptomatic (presenting late) and, in the case of children, are usually stunted9,10. Non-progressing adult and paediatric infection are both characterized by low levels of immune activation but through distinct mechanisms. In non-progressing adult infection, this is achieved through HLA-mediated suppression of viral replication. In non-progressing paediatric infection, the mechanism by which immune activation remains low in some patients despite persistent high levels of viraemia7 is unknown. The phenotype of normal CD4+ T cell counts and low levels of immune activation, despite high viral loads, is reminiscent of the natural hosts of simian immunodeficiency virus (SIV), such as the sooty mangabey and African green monkey, which live normal lifespans with normal CD4+ T cell counts and low levels of immune activation but have median viral loads of 105 SIV RNA copies per ml (Ref. 11).

Figure 1: Changes in viral load following HIV infection in paediatric and adult cases.
figure 1

Typical, progressing adult and paediatric infection (upper panels), and non-progressing adult and paediatric infection (lower panels)2,7,8,161. Typically, in HIV infection of antiretroviral therapy (ART)-naive adults, a viral set point is reached after 6 weeks, and the median time to AIDS is 10 years. In typical, ART-naive paediatric infection with HIV, a viral set point is reached after 5 years, and the median time to AIDS is 1 year. Non-progressing adult infection is characterized by low or undetectable viral loads and low levels of immune activation. The prevalence of non-progressing paediatric infection is 5–10%7,162,163 and of adult elite controllers is 0.3%146. Non-progressing paediatric infection also features low levels of immune activation but in the setting of high viral loads (30,000 copies per ml).

PowerPoint slide

The differences that exist between adult and paediatric HIV infection, not only in terms of disease outcome and immune response to HIV but also in terms of mode of transmission, have provided opportunities to improve our understanding of the mechanisms of immune control of HIV. The possibility that these differences might also provide opportunities to learn about mechanisms of HIV eradication or 'cure' was initially prompted by the report of the 'Mississippi child' (Refs 12,13) (Box 1). Although virus levels ultimately rebounded in the Mississippi child 28 months after ART discontinuation, the case was sufficiently unusual to prompt the hypothesis that the early initiation of ART during paediatric infection might provide the setting in which HIV eradication, or remission, could most readily be achieved.

A further anecdotal example of post-treatment control of viraemia was recently described in a French HIV-infected child (Box 1). Although anecdotal cases can be misleading (Box 2), this paediatric example follows the description of post-treatment control in the VISCONTI cohort of adults who received ART during acute infection (within months, not days, of infection) and who discontinued ART after a median of 3 years14. It is estimated that 5–15% of adults who initiate ART in early infection and continue for this duration become post-treatment controllers, but the equivalent figure in paediatric infection following early initiation of ART is unknown. Several reports have described viraemic rebound within days of ART discontinuation in children in whom ART was initiated in the first week of life and continued for 2–4 years15,16,17,18. In the much larger CHER study, in which 250 children were treated for either 40 or 96 weeks from a median age of 7 weeks19, and in the PEHSS study20, in which 19 infants received 12 months of ART from a median age of 4 weeks, all patients rebounded soon after ART discontinuation (M. F. Cotton, personal communication; P.J.G., unpublished observations). In summary, it is not currently known whether early initiation of ART in paediatric infection is more or less likely to facilitate post-treatment control than early initiation of ART in adult infection.

This Review considers the probable impact of early initiation of ART on the potential for HIV eradication in children versus adults and discusses the influence of immune ontogeny and maternal transmission in paediatric infection. Finally, we consider the particular opportunities that these features of paediatric infection present in terms of strategies to promote HIV remission or eradication.

Early ART in paediatric versus adult infection

Feasibility. In adult infection, outside of expensive trial settings21,22, the identification of early HIV infection is problematic. By contrast, in utero paediatric infection, which mainly occurs late in the third trimester23,24, can be diagnosed using point-of-care testing25 within hours of birth in the children of chronically HIV-infected mothers, and ART can be initiated in these children within minutes of birth, pending the results of diagnostic tests. In most cases, therefore, treatment can be started earlier in paediatric infection than in adult infection. However, in some cases — for example, when the mother develops acute infection during pregnancy16in utero transmission can occur several weeks before birth (Fig. 2). In addition, paediatric infection can arise from intra-partum or post-partum mother-to-child transmission (MTCT), and in neither case are these infections detectable at birth. Furthermore, there are added challenges to detecting intra-partum or post-partum infection for the practical reason that the mother and child have long departed the hospital. Therefore, initiating ART soon after infection following intra-partum or post-partum transmission of HIV may be substantially more difficult than after in utero transmission that mostly arises late in the third trimester and is detectable at birth. However, it is not known whether the timing of early initiation of ART during infection is crucial to the outcome in terms of post-treatment control. On the one hand, SIV–macaque studies indicate that there is a very rapid seeding of the viral reservoir, within 3 days of infection26. On the other hand, in the VISCONTI cohort subjects, ART was initiated within the first 2–3 months of infection — which is categorized on the basis of Fiebig stages I–VI27 — similarly as in the VISCONTI teenager28, which indicates that ART initiation after the first few days of infection does not preclude subsequent post-treatment control in HIV infection.

Figure 2: Risk of MTCT in utero and in early life.
figure 2

Mother-to-child transmission (MTCT) risk of HIV in the absence of prophylactic measures is approximately 7% from in utero infection, 18% from intra-partum infection and 15% from post-partum transmission (through breast milk from breastfeeding for 18–24 months)164,165,166,167. In utero HIV-infected infants are viraemic and HIV DNA-positive at birth, whereas transmission arising at the time of delivery (intra-partum) results in viraemia approximately 4 weeks later. Treatment with antiretroviral therapy (ART) of all mothers testing HIV-positive in pregnancy has reduced combined in utero and intra-partum MTCT from 25% to 1–2%102, and prevention of breast-milk transmission through ART of the mother has reduced post-partum MTCT rates to 1–2%168. The remaining 1–2% MTCT rate results either from ART non-adherence or from acute infection during pregnancy or after birth (in utero MTCT risk of 20%167,169,170 or post-partum MTCT risk of 20–25%171) as tests to detect HIV antibody are negative during acute maternal infection when peak viraemia is 107 copies per ml and transmission risk is high.

PowerPoint slide

Reduction of the viral reservoir. It is evident that early initiation of ART reduces the frequency of latently infected T cells in both adult and paediatric infection. If ART is initiated during chronic infection in adults, the latently infected resting CD4+ T cells persist with a half-life of 44 months29,30,31. However, treatment of adults during acute infection results in an approximately 10–100-fold lower frequency of latently infected resting CD4+ T cells compared with chronic infection, when measured in terms of either HIV DNA or replication-competent virus32,33,34. Similar results have been demonstrated in SIV-infected macaques when treatment is initiated within 3 days compared with 7 days after infection26. In addition, the amount of cell-associated unspliced HIV RNA, which reflects basal levels of HIV transcription, is significantly lower in individuals who are treated in acute infection compared with chronic infection35. Much of the decrease in HIV DNA and unspliced HIV RNA levels occurs in the first few years of therapy36,37. Latently infected CD4+ T cell subsets have different decay rates, and following long-term ART (more than 10 years), long-lived central memory T cells, memory stem cells and naive T cells are the main T cell subsets enriched with HIV36,38,39.

In infants who received ART at a median age of 8 weeks, the half-life of latently infected resting CD4+ T cells was 11 months during the first 2 years on ART40. Levels of both cell-associated HIV DNA and cell-associated HIV RNA are lower in children in whom ART was initiated earlier after infection17,41,42,43,44. In the CHER study, for example, in which levels of both HIV DNA and cell-associated HIV RNA were measured, initiation of ART in children before 2 months of age (at a median age of 53 days) resulted in a lower frequency of HIV-infected CD4+ T cells 7–8 years later, compared with children in whom ART was started at more than 2 months of age (at a median age of 170 days)43. Furthermore, whereas the levels of cell-associated HIV DNA seem to plateau after 4 years in adult infection45, in children these levels continue to decline for more than a decade41. Treatment of HIV infection in children at a median age of 2 months resulted in a significant decrease in HIV DNA, but the persistence of 2-LTR circles (2-long-terminal repeat circles) and cell-associated unspliced RNA after 96 weeks, and the frequency of HIV DNA was more than 148-fold higher than the frequency of infectious genomes46, as previously described following treatment of HIV in adults47. Thus, ART can generally be initiated earlier in paediatric in utero infection than in adult infection, the half-life of latently infected resting CD4+ T cells is shorter in children, and initial studies indicate that decline in the viral reservoir seems to continue for longer in children than in adults on ART.

Impact of immune ontogeny on cure potential

Several aspects of normal development will have an important influence on the establishment and size of the viral reservoir, some of which are likely to increase, and others to decrease, the potential for HIV cure in infected children. The overall balance of these effects may depend crucially on the timing of ART initiation following birth.

Tolerogenic environment in early life. The immune system in utero and in early life is exquisitely adapted to meet the particular challenges of that phase of life. In utero, the need to avoid making inflammatory responses during the frequent exposure to non-inherited maternal antigens is facilitated by the high levels of transforming growth factor-β (TGFβ) in the fetus that direct the peripheral differentiation of naive T cells, in the presence of the cognate antigen, into 'adaptive' regulatory T cells (TReg cells)48. These memory TReg cells are long-lived and can be found in children aged 7–17 years48. TReg cells make up 15% of fetal blood T cells, compared with 5% of adult blood T cells49. Moreover, innate immune cells at birth, compared with adulthood, respond to a panel of Toll-like receptor (TLR) agonists by secreting higher levels of interleukin-6 (IL-6), IL-1β and IL-23, which support T helper 17 (TH17) cell differentiation, and lower levels of pro-inflammatory cytokines such as type I interferons (IFNs), and hence lower levels of IFNγ, tumour necrosis factor (TNF) and IL-12, which support TH1 cell differentiation50,51,52. The higher levels in neonates of immunoregulatory and anti-inflammatory cytokines (such as IL-10 and TGFβ) and of TH2 cell-supporting cytokines (IL-4, IL-5, IL-9 and IL-13)53,54 further function to dampen down TH1 cell responses. Finally, plasma levels of adenosine are higher in neonates than in adults and may contribute to the TH17 cell bias and low levels of immune activation in early life through increased IL-6 synthesis and decreased TNF production51. The immune response in utero and in early life therefore establishes a tolerogenic environment that is broadly anti-inflammatory and that promotes TH17 cell-mediated host defence against extracellular bacterial and fungal pathogens, at the expense of a strong TH1 cell-driven response which, by contrast, is a feature of antiviral cell-mediated immunity in adults.

In the setting of HIV infection in early life, one effect of these immune adaptations is to eliminate the possibility of HLA class I-restricted CD8+ T cell-mediated suppression of viraemia that is observed in adult elite controllers. However, the tolerogenic environment in early life favours low levels of immune activation, which could interfere with the vicious cycle that normally occurs during adult infection of immune activation, increased bystander cell death through apoptosis and pyroptosis55,56, increased susceptibility of activated CD4+ T cells to further infection and increased virus production (Fig. 3). For similar reasons to those hypothesized in adult post-treatment controllers who express disease-susceptible HLA alleles such as HLA-B35 (Ref. 11) (discussed further below), the tolerogenic environment of early life may interfere with the establishment of latent infection in the long-lived naive, stem cell-like memory and central memory CD4+ T cells. There is some suggestion that early treatment of children, in contrast to adults, can result in lower levels of these latently infected long-lived T cell subsets36,41. Reduced infection of long-lived T cell subsets has also been observed in adult post-treatment controllers and elite controllers14.

Figure 3: Paediatric factors affecting immune activation and size of the HIV reservoir.
figure 3

The vicious cycle of HIV infection — resulting in intestinal mucosal damage, microbial translocation across the intestinal barrier and increased immune activation, leading to increased CD4+ T cell decline, which predisposes to further co-infections, increased immune activation, HIV replication and CD4+ T cell loss — can be prevented by the initiation of antiretroviral therapy (ART) sufficiently early in adult and paediatric infection. There are factors specific to paediatric infection that tend to interfere with this vicious cycle, and others that contribute to it, the balance of which may vary between individuals and also may depend crucially on the timing of ART initiation. Aspects of immune ontogeny that favour decreased immune activation include the tolerogenic immune environment in utero and in early life and the high proportion of naive CD4+ T cells. Aspects of immune ontogeny that tend to decrease paediatric cure potential include the large number of T helper 17 (TH17) cells in the intestinal mucosa. Maternal factors that favour paediatric cure potential include maternal health, exclusive breastfeeding, lack of co-infections and expression of HLA class I molecules that bind leukocyte immunoglobulin-like receptor subfamily B member 2 (LILRB2) with high avidity. BCG, bacille Calmette–Guérin; CMV, cytomegalovirus; IFN, interferon; IL, interleukin; TB, tuberculosis; TGFβ, transforming growth factor-β; TNF, tumour necrosis factor; TReg cell, regulatory T cell.

PowerPoint slide

Role of TH17 cells. A factor that may also affect the ability of HIV to establish long-lived latency is the high proportion of naive CD4+ T cells in utero and in newborns. Indeed, in cord blood, there are almost no activated CC-chemokine receptor 5 (CCR5)-expressing memory CD4+ T cells57. CCR5 is the co-receptor by which HIV gains entry into CD4+ T cells, and activated memory CD4+ T cells expressing CCR5 are the most susceptible targets for HIV infection58,59,60. By contrast, in the intestinal epithelium in utero and at birth, almost one-third of the CD4+ T cells express CCR5, many of which also express CCR6 (Ref. 54), which is characteristic of TH17 cells that are highly permissive for HIV infection61,62. Even resting CD4+ T cells that express CCR6, in the presence of the CCR6 ligand CCL20, have an increased susceptibility to direct infection and establishment of HIV latency63. Thus, the theoretical paucity of CD4+ T cell targets for HIV infection in utero and early life is at least partly compensated for by the high numbers of TH17 cells and other CD4+ T cell targets available in the intestinal epithelium.

It is of note that the immune system at birth has a strong TH17 cell bias, given the importance of TH17 cells both functionally, in host defence against extracellular pathogens, and structurally, in maintaining the integrity of the intestinal mucosal barrier64. Microbial translocation has a central role in the vicious cycle of immune activation that results in CD4+ T cell decline during HIV infection, and the preferential loss of TH17 cells observed in HIV infection65 is therefore likely to have a disproportionate impact on paediatric infection. It is noteworthy that, despite substantial loss of intestinal CD4+ T cells following acute SIV infection in both natural66,67 and experimental hosts68,69,70, TH17 cell numbers in the gut are maintained at normal levels in non-pathogenic SIV infection71. The presence of higher levels of adenosine in mucosal tissues in non-pathogenic infection72 indicates that the adenosine pathway may contribute to limiting TH17 cell loss by reduced immune activation and the promotion of TH17 cell differentiation. In adult HIV infection, if ART is initiated sufficiently early, TH17 cell numbers are preserved and/or restored, and levels of systemic immune activation are normalized22. Thus, the preferential loss of TH17 cells in the intestinal mucosa during HIV infection in utero and in early life may have a disproportionate impact on that age group, leading to increased susceptibility to infections with extracellular pathogens, microbial translocation and immune activation. Therefore, the timing of early ART in paediatric infection may be particularly important.

Increased immune activation in early life. Working against the tolerogenic environment are factors that predispose to increased immune activation in early life. This is directly relevant to the reservoir size because increased immune activation increases viral replication (Fig. 3). For example, microbial translocation across the intestinal barrier is increased in neonates, whether HIV infected or not73,74. This is particularly true for premature infants, and more than one-third of babies born to HIV-infected mothers are premature (that is, born before 37 weeks of gestation). The changing composition of the intestinal microbiota also contributes to higher levels of immune activation in early life73,75. It is proposed that mixed feeding, as opposed to exclusive breastfeeding, predisposes to increased risk of gastrointestinal infections and mucosal inflammation and, therefore, an increased risk of HIV transmission through breast milk76,77. Microbial translocation, leading to increased immune activation, is similarly increased by breaches in the intestinal mucosa, and thus it is reasonable to assume that the type of feeding received by the child may in this way strongly influence the establishment and size of the viral reservoir before ART initiation in the infant.

Immune activation can also be increased by vaccinations and new infections that are common in infancy. The immunization schedule provides crucial protection against diseases to which HIV-infected children may be particularly susceptible, but certain vaccines may also have unwanted side effects. Mycobacterium bovis bacille Calmette–Guérin (BCG) is a particular case in point78. BCG, which is given at birth to all South African infants on the first day of life, protects against miliary tuberculosis (miliary TB) and TB meningitis79, and it may also have non-specific benefits in terms of reducing morbidity resulting from non-mycobacterial infections80. However, the increased levels of immune activation that result from BCG vaccination increase HIV infection risk in uninfected infants and increase disease progression in infected infants78. Among the large number of infections to which newborns are exposed, cytomegalovirus (CMV) is noteworthy both for its effect in increasing immune activation81 and its prevalence in countries in which HIV is abundant. By 3 months of age, almost two-thirds of children in sub-Saharan Africa have become infected with CMV81. CMV co-infection in HIV-infected children results in more rapid HIV disease progression82. All of these factors that function to increase immune activation in early life may oppose the benefits in terms of HIV infection and disease progression that would be expected to result from the tolerogenic immune environment in early life described above. However, initiation of ART in the first 24–48 hours of life, before many of these factors have maximum effect, would markedly reduce their impact.

Differential localization of viral reservoirs. In adults on ART, HIV-infected T cells are found at higher frequency in the gastrointestinal tract and lymph nodes than in the blood83,84. A detailed comparative analysis of other tissue sites, such as the spleen, liver and genital tract, has not been carried out, but latently infected cells are found in all of these tissues85. Recent findings also suggest that latently infected cells can be detected in adipose tissue86,87. Other long-lived infected cells can also persist on ART, including tissue macrophages — as Kupffer cells in the liver88, as alveolar macrophages in the lung89 and as microglia or perivascular macrophages in the central nervous system (CNS)90. HIV latency has also been demonstrated in astrocytes in the CNS91. It is not clear whether HIV DNA that persists in tissue macrophages is replication competent and contributes to viral rebound following cessation of ART.

In HIV-infected children, the location of latently infected T cells or tissue macrophages has not been explored but may be quite different, given the low frequency of memory T cells and different levels of expression of key chemokine receptors. Although detecting HIV persistence in tissues is challenging, recent novel methods developed in the macaque model, such as total-body imaging92 and RNAscope or DNAscope in situ hybridization, should aid future studies of HIV persistence in children on ART. This is an important feature to understand, as the mechanisms that lead to HIV persistence in resting T cells and macrophages differ93, and, therefore, strategies to activate or eliminate infected cells will need to be different. This has recently been demonstrated in terms of the ex vivo response to histone deacetylase (HDAC) inhibitors of virus derived from brain tissue compared with spleen94.

Finally, many recent papers have shown that the majority of HIV DNA measured in blood CD4+ T cells from HIV-infected individuals on ART is replication defective47,95. Furthermore, the conventional viral outgrowth assay may underestimate the true number of cells that carry replication-competent virus owing to the presence of intact proviruses that are not induced through T cell stimulation96. In a single recent study in children treated with ART, it seems that there is a similar relationship between intact and defective viruses in children as in adults46, but more studies are required in children using newer molecular strategies to define the relationship between defective and replication-competent virus in this setting.

T cell vaccine potential. Children who have received ART in the first few days of life often subsequently lack detectable HIV-specific immune responses. Western blot non-reactivity has been proposed as a marker of early and profound suppression of viral replication41, and lack of HIV-specific cellular immunity may be similarly indicative. However, it is well established that HIV-specific CD8+ T cell responses are evident at birth in the majority of children infected in utero97,98. This suggests that rapid removal of the HIV antigen in infancy or early childhood may result in loss of the memory response even though an immune response was induced. Thus, in early ART-treated HIV-infected children, there may be the potential for a T cell vaccine to skew the CD8+ T cell immunodominance pattern in a more favourable direction — such as towards a broad Gag-specific response — than would be the case if the pre-existing, suboptimal T cell responses that are generated in natural infection are predominant.

Successful skewing of the natural immunodominance pattern of the CD8+ T cell response by an HIV T cell vaccine has been demonstrated in recent analyses of participants in the South African Phambili trial99 who subsequently became infected. In individuals expressing the disease-susceptible HLA-B*5802 allele, recipients of the MRKAd5 HIV-1 Gag/Pol/Nef vaccine made broader and higher magnitude Gag-specific responses than HLA-B*5802-positive recipients of a placebo100. The placebo recipients showed the Env-dominated response pattern that is typical of natural infection101. In association with skewing of the CD8+ T cell response towards targeting Gag, viral set points in the HLA-B*5802-positive vaccine recipients were significantly lower than in the HLA-B*5802-positive placebo recipients100.

In summary, a range of factors associated with the developmental stage of the child at which HIV infection occurs, before ART is initiated, are likely to affect markedly the ultimate size of the HIV reservoir. The overall effect of these mixed influences might depend crucially on the exact timing of in utero HIV transmission and of ART initiation. It is probable that if ART can be initiated within the first 24–48 hours of life, then the factors predisposing to increased microbial translocation and increased immune activation would be largely mitigated, thereby providing children with an inherent advantage compared with adults in terms of the potential for HIV cure.

Impact of maternal factors on cure potential

In utero and intra-partum MTCT are effectively prevented entirely if viraemia is suppressed by ART during pregnancy. In countries such as South Africa, prevention strategies have reduced MTCT rates that arise in utero or intra-partum from 25–30% without any therapy to 1–2% currently102. When transmission does occur, this tends to be associated with challenging circumstances. For example, more than 60% of transmitting mothers in South Africa have a daily income of less than US$1, the source of water in more than one-third of cases is either a shared standpipe or river water, and toilet facilities for almost one-half of mothers are a pit latrine20. In these settings in particular, the outcome of the child may have less to do with immunology than with the physical and mental well-being of the mother and the presence or absence of a social support network. There are therefore practical aspects of successful HIV treatment that differ between adults and children, and the dependence of an infected child on the mother to administer ART, in addition to the other needs of childhood, is an added challenge that should not be underestimated.

Maternal HLA type. The risk of MTCT is strongly related to maternal viral load103,104, and therefore transmitting mothers are more likely to express HLA class I molecules that are known to be associated with high viral set points and rapid progression to disease105,106. In African populations, these are HLA-B*1801, HLA-B*4501 and HLA-B*5802. In Caucasian populations, the best-studied example is HLA-B35. As children inherit one HLA haplotype from their mother, in utero HIV-infected children, as a group, are enriched with the HLA class I molecules that are associated with rapid progression in adult infection106. However, as described above, the HLA-B alleles that have a substantial impact on adult infection do not significantly influence paediatric disease outcome8. Indeed, it has been proposed that HLA class I molecules such as HLA-B35 that are associated with rapid disease progression in adults may actively facilitate post-treatment control14. A possible mechanism is suggested by the fact that HLA-B35 and HLA-B*4501 bind leukocyte immunoglobulin-like receptor subfamily B member 2 (LILRB2) with high avidity, whereas protective HLA alleles such as HLA-B27, HLA-B57 and HLA-B*5801 bind LILRB2 with low avidity107. LILRB2 is expressed on cells of the myeloid lineage, such as conventional dendritic cells, and high-avidity HLA class I binding to LILRB2 inhibits dendritic cell function. The failure of disease-susceptible HLA class I molecules such as HLA-B35 to induce broad HIV-specific CD8+ T cell responses in adult infection, leading to high viral set points, may therefore in part be due to the inhibitory action of LILRB2 on dendritic cell activity. However, in the setting of early initiation of ART, factors such as these that reduce antiviral T cell immunity may tend to reduce immune activation and thereby influence the establishment and cellular localization of the latent HIV reservoir. If this proves to be the case, the apparent maternally inherited HLA 'handicap' in infected children may in fact predispose towards post-treatment control following early ART.

Virus type transmitted through MTCT. One of the benefits of early initiation of ART in paediatric and adult infection is to suppress viral replication before the selection of autologous escape mutants. The early initiation of ART might be particularly important in adults, in whom cytotoxic T lymphocyte (CTL) escape mutants are selected from the first few weeks of infection108 and persist in latently infected T cells109. By contrast, CD8+ T cell responses in in utero HIV-infected infants, even when restricted by protective HLA alleles, such as HLA-B57, HLA-B*5801 and HLA-B*8101, seem to be insufficiently functional in the first few months of infection to select virus escape mutants8. However, in MTCT, the virus that is transmitted is in many cases pre-adapted to T cell responses available in the child, as well as to maternal antibodies crossing the placenta. As the child inherits half of the mother's HLA alleles, in theory the virus transmitted from mother to child may be pre-adapted to all of the CD8+ T cell epitopes that are restricted by the HLA alleles shared with the mother110. However, for two reasons, this may be less of a problem in reality. First, the mothers of children infected with HIV tend to express disease-susceptible HLA alleles106 (as described above), and these typically do not select any virus escape mutants in the crucial Gag epitopes111. Second, as prevention of MTCT becomes increasingly effective, the proportion of in utero transmissions that result from acute infection of the mother during pregnancy becomes greater (Fig. 2). In this scenario, in utero transmission may occur before escape mutants have been selected in maternal virus.

In summary, the mother has a crucial role in determining morbidity and survival of the child. However, if an HIV-infected mother is healthy and sufficient health-care support is available for mother and child, there are reasons to believe that the inheritance by the child of maternal disease-susceptible HLA alleles and the transmission of HIV adapted to maternal immune responses may not be a major disadvantage in relation to the potential for HIV cure in children versus adults.

Opportunities for effective cure strategies

Several factors discussed here provide a rationale for the proposition that unique possibilities exist to facilitate the eradication or remission of HIV in children, namely: the feasibility of initiating ART within 24–48 hours of birth; the ability of ART to make faster reductions in the size of the viral reservoir that continue for longer in infants compared with adults; and the ability of ART to minimize the effect of factors in early life that would otherwise counteract the reduced immune activation resulting from a tolerogenic environment. However, there are also specific obstacles that narrow both the window of opportunity and the range of immunotherapeutic interventions that can be used in children. In particular, the very poor levels of ART adherence in adolescence112, coupled with the normal onset of adolescence as early as 8 and 9 years of age in females and males, respectively113,114, effectively place a time limit on the age by which immunotherapeutic interventions need to be in place (Fig. 4).

Figure 4: Principal opportunities for immunotherapeutic interventions to maximize the potential for HIV cure or remission in paediatric infection.
figure 4

In utero paediatric infection provides particular opportunities for antiretroviral therapy (ART) to be initiated with or without additional interventions (such as broadly neutralizing antibodies) very early in the course of infection. As cell-mediated immunity is likely to have a key role in eradication of the viral reservoir109, and T helper 1 (TH1) cell responses take some years to develop fully, a period of time on ART alone may be useful to minimize the size of the viral reservoir before further interventions that are designed to eradicate infection. In late childhood, approximately between the ages of 3 and 9 years, immune ontogeny favours the generation of effective CD8+ T cell immunity, which provides a window of opportunity for therapeutic intervention before the onset of puberty with associated behavioural changes, including loss of ART adherence and onset of sexual activity.

PowerPoint slide

Early ART alone. The universal failure, so far, of ART initiated in the first few weeks of life to result in post-treatment HIV control in paediatric infection, let alone cure, may indicate that ART should be initiated even earlier and/or for a longer duration and that there may be additional interventions, other than those related to the timing of onset and duration of ART, that are likely to be required to achieve post-treatment control. Host genetic factors may contribute to the likelihood of post-treatment control, as has been proposed in the VISCONTI adult cohort in relation to HLA class I type. Defining these factors will require large studies to initiate early ART in sufficient numbers of in utero HIV-infected infants to eliminate some of the misleading conclusions that can arise from anecdotal cases (Box 2).

Broadly neutralizing antibodies at birth. Broadly neutralizing antibodies do not typically neutralize contemporaneous autologous virus115 and tend to be isolated in HIV-infected individuals who have high viral set points and are progressing to AIDS116. Maternal antibodies crossing the placenta do not neutralize transmitted virus or affect paediatric disease progression117. However, heterologous broadly neutralizing antibodies can both prevent infection118,119 and enhance subsequent B cell responses in infected infants120, and they also substantially reduce viral replication in simian–human immunodeficiency virus (SHIV)-infected macaques (PGT121 antibody)121 and in HIV-infected humans (3BNC117 antibody)122. Expression of broadly neutralizing antibodies by adeno-associated virus vectors123 has the potential to prevent infection for life, although long-term protection against infection using this approach has only been demonstrated in animal models so far124,125.

In paediatric infection, broadly neutralizing antibodies could be used at birth at the same time as ART is initiated (Fig. 3). Human monoclonal antibodies protected against SHIV infection following oral challenge in neonatal macaques126. This approach, in combination with ART, may potentially reduce viral loads by 1–2 log10 over 7–10 days after infusion127 in addition to the effect of ART. ART alone can reduce a typical starting viral load of 105–106 copies per ml to less than 50 copies per ml in 3–6 months20, whereas ART in combination with a broadly neutralizing antibody might achieve aviraemia in days or weeks. The timing of ART initiation and the time to viral suppression are both important independent factors that influence the ultimate size of the latent viral reservoir44. Early suppression of viraemia clearly limits the size of the reservoir but importantly may also limit viral diversity, which thereby might increase the efficacy of any immunotherapeutic approaches that are used later, as described below.

Timing of further immunotherapeutic interventions. Having initiated ART early and achieved suppression of viraemia, sufficient time should then elapse before the introduction of further immune interventions: first, to enable the latent viral reservoir to be optimally reduced to levels beneath the level of detection of the most sensitive assays; and, second, to allow TH1 cell-polarizing innate immune responses to develop more fully, for the reason that cell-mediated immunity is likely to have a key role in eradication of the viral reservoir109. Although IL-12, a TH1 cell-polarizing cytokine, remains at lower levels in infants than in adults (even in 12-year-old children)52, by 2 years of age levels of type I IFNs are similar to those in adults, levels of IL-6 and IL-23 (TH17 cell-polarizing cytokines) and of IL-10 (an immunoregulatory cytokine) are lower than in neonates, and TReg cell numbers are similar to those in adults128. Indeed, there is evidence that the middle years of childhood represent a 'honeymoon period' (Ref. 129), during which the immune response is optimally adapted against a range of infectious agents, including TB, influenza, malaria, Epstein–Barr virus, varicella zoster virus, mumps, measles and perhaps also HIV, such that it is sufficient to control the infection without causing immunopathology. This period — from approximately 3 to 9 years of age — may therefore be the optimal window of opportunity for immunotherapeutic interventions designed to facilitate HIV eradication, before the onset of adolescence (Fig. 3).

T cell vaccines. Early generation T cell vaccines did not reduce viral loads overall in vaccine recipients, and the adenovirus type 5 (Ad5) vector used in the Step trial and Phambili trial was associated with increased risk of infection130. However, current approaches such as those using prime–boost regimens of chimpanzee adenovirus and modified vaccinia virus Ankara result in T cell responses that are 1 log10 higher in magnitude and 5–10-fold greater in breadth compared with the T cell responses achieved by MRKAd5 HIV-1 Gag/Pol/Nef131,132 (B. Mothe, personal communication). The safety and immunogenicity of these prime–boost approaches have been established in Phase I and Phase II trials of malaria vaccination in children ranging in age from infants (10 weeks old and 5–12 months old) to mid-childhood (2–6 years old) (K. Ewer, personal communication). The induction of protective and broad Gag-specific CD8+ T cell responses in children treated early by vaccination and/or the use of broadly neutralizing antibodies could provide the long-term antiviral immunity that is necessary to contain or even to eliminate viral replication from previously latently infected cells, once ART is discontinued.

Latency reversal. Several compounds have now been shown to activate latent HIV infection in vitro133. These include epigenetic modifiers, protein kinase C agonists and TLR agonists. HDAC inhibitors were the first latency activators to move into clinical trials in HIV-infected individuals on ART; these studies all demonstrated an increase in cell-associated HIV RNA and, in some studies, an increase in plasma HIV RNA, but no study reported a change in the frequency of latently infected cells134,135,136,137,138. Although HDAC inhibitors have been widely used for the management of malignancy in adults, there is very little experience in paediatric cases, and the sustained changes in host gene expression that result from their use would be a serious safety concern in children136. More recently, disulfiram, which has been in clinical use for decades and with an excellent safety profile, was shown to also activate HIV latent infection139. This agent could potentially be evaluated in children for latency reversal. It remains unknown whether there are factors that would alter the susceptibility of latently infected cells from HIV-infected children on ART to activation. This needs to be evaluated in ex vivo studies to ensure that there are no inherent differences in the stability of maintenance of HIV latency in children compared with adults, before the initiation of clinical trials of latency reversal in children.

Other immunotherapeutic strategies. Several additional approaches are currently being developed in adults; some of these show great promise, but their efficacy and safety profile in children have yet to be determined. These are summarized in Box 3.

Conclusions

The immune system in utero and in early life favours low levels of immune activation, at least initially, following HIV infection. However, in most cases, it seems that ART would need to be initiated as early as possible to minimize the size of the latent viral reservoir and provide a realistic base from which additional immunotherapeutic interventions can be introduced. The timing of their introduction and of a subsequent treatment interruption remain very much open to discussion. In children, as distinct from adults, there is a relatively narrow window of opportunity between the development of an immune response that is sufficient to eliminate activated HIV-infected cells that previously comprised the latent reservoir and the onset of puberty with its accompanying high levels of ART non-adherence. Recent data suggest that even brief ART interruptions in the absence of HIV-specific immunity can lead to a rapid and irreversible increase in the size of the HIV reservoir43, underlying the observation that reservoir size alone does not necessarily predict the time to viral rebound33,140. Defining the factors that determine post-treatment control and identifying biomarkers that can predict which individuals will have prolonged antiretroviral-free remission141 will require additional work, including clinical trials in children. So far, HIV vaccine trials have been rarely undertaken in children, despite evidence of promising immune responses in children who have been vaccinated142. Evaluation of early treated children who receive immunotherapeutic interventions should be a priority.