Leukocyte Telomere Length at Birth and During the Early Life of Children Exposed to but Uninfected With HIV After In Utero Exposure to Antiretrovirals

Abstract

Background

Maternal combination antiretroviral therapy (cART) during pregnancy could impact the health of human immunodeficiency virus (HIV)–exposed, HIV-uninfected (HEU) children, because some antiretrovirals cross the placenta and can inhibit telomerase. Our objective was to compare leukocyte telomere length (LTL) in HEU children and HIV-unexposed, HIV-uninfected (HUU) children at birth and in early life and to investigate any relationship with cART exposure.

Methods

HEU and HUU children’s blood LTL was compared cross-sectionally at birth, and during the first three years of life. Longitudinal HEU LTL dynamics was evaluated over that same period.

Results

At birth, the LTL in HEU children (n = 114) was not shorter than that in HUU children (n = 86), but female infants had longer LTL than male infants. Maternal cART (duration or type) showed no association with shorter infant LTL. Among 214 HEU children age- and sex-matched at a 1:1 ratio to HUU children, LTL declined similarly in both groups. In a longitudinal analysis, LTL attrition in HEU children was rapid from birth to 1 year of age and gradual thereafter. Zidovudine prophylaxis did not significantly alter LTL.

Conclusions

Our results indicate that from birth to 3 years of age, the LTL in HEU children is not negatively affected by exposure to maternal HIV infection and cART, at least not to the regimens used within this Canadian cohort, a reassuring finding.

Leukocyte telomere length, HIV-exposed uninfected, combination antiretroviral therapy, maternal smoking during pregnancy, longitudinal LTL attrition

The introduction of combination antiretroviral therapy (cART) during pregnancy, together with postnatal infant antiretroviral prophylaxis, has drastically reduced the risk of mother-to-child transmission of human immunodeficiency virus (HIV), with current rates in North America being as low as approximately 1% [1, 2]. Guidelines recommend that women living with HIV initiate cART as soon as HIV infection is diagnosed, and consequently, most will now conceive while receiving cART [3]. With increased cART use during pregnancy, the number of new perinatal HIV infections has considerably declined, and the number of HIV-exposed, HIV-uninfected (HEU) children with in utero cART exposure is rapidly rising. Despite the success of cART in pregnancy, neonates born to women living with HIV are at higher risks of adverse outcomes, including stillbirth, preterm birth, low birth weight, and intrauterine growth restriction [4–6]. It remains unclear whether in utero and postnatal antiretroviral exposure will have long-term consequences on the health of HEU children. Some antiretrovirals are known to cross the placenta [7] and, by doing so, could affect cellular processes in the developing fetus. Neurodevelopmental, cognitive, and language delays [8–10], immune system abnormalities [11, 12], as well as increased risks of morbidity and mortality [12, 13] have been reported in HEU children.

The telomerase complex contains a reverse transcriptase that replicates telomeric DNA (telomeres) protecting the ends of eukaryotic chromosomes. In somatic cells, telomerase is not expressed, and telomeres shorten with each cellular division. However, germ cells, embryonic stem cells, and placental cells [14] can express telomerase, resulting in telomere length (TL) maintenance or slower TL attrition. TL thus provides an account of a cell’s replication history and/or its remaining replicative potential and is a biomarker of aging. Shorter leukocyte TL (LTL) is associated with mortality [15] and has been linked to several age-associated diseases [16–18]. Nucleoside reverse transcriptase inhibitors (NRTIs) used to treat HIV infection, including in pregnancy, are known to inhibit telomerase in vitro [19–21] and shorten telomeres in immortalized telomerase-dependent cell lines [22, 23].

A few cross-sectional studies have reported similar LTL among HEU children and HIV-unexposed, HIV-uninfected (HUU) children at birth [24, 25] or later in life [26, 27], while one study observed a trend toward shorter cord blood TL in HEU children as compared to HUU children [25]. Several of these studies were limited in their sample size. Our objective was to compare LTL at birth in a larger cohort of HEU and HUU children and to investigate the relationship between LTL and in utero exposure to maternal cART. Furthermore, we report for the first time the longitudinal dynamics of LTL over the first 3 years of life among HEU children.

METHODS

Study Population

Study participants were all HEU and HUU children ≤3 years of age. All HEU and half of the HUU children were enrolled in 3 Canadian cohort studies: (1) a pediatric cohort (born between 2003 and 2006 in Vancouver or Toronto) [28], which collected whole-blood specimens longitudinally from HEU children between birth and 9 months of age and single whole-blood specimens from HUU children aged 0–9 months; (2) a pregnancy cohort (born between 2005 and 2009 in Vancouver) [25], which collected single whole-blood specimens from HEU children between birth and 15 months and from HUU children at birth only; and (3) the Children and Women: Antiretrovirals and Markers of Aging (CARMA) cohort (born between 2006 and 2012 in Vancouver, Toronto, Ottawa, or Montreal) [26], which collects whole-blood specimens longitudinally from HEU children between birth and 19 years of age and single specimens from HUU children aged 0–19 years (Supplementary Figure 1). The remaining HUU children were anonymous controls (all from Vancouver), for whom only sex and date of birth were known. Their whole-blood samples were leftovers from blood work during routine hospital visits or for reasons such as elective surgery. Inclusion/exclusion criteria for the 3 cohorts are described in Supplementary Table 1. Of note, all participants with a birth specimen were from Vancouver.

This study was reviewed and approved by the institutional review board of all participating institutions, and written informed consent was obtained from the parents/guardians of all study participants. The latter was not required for anonymous controls.

Maternal demographic and clinical data, including exposures during pregnancy were extracted from the cohort databases. Maternal ethnicity was self-reported. Data on tobacco smoking during pregnancy was self-reported but collected with different levels of detail in each cohort. We therefore categorized smoking within this analysis as self-reported smoking at any time during pregnancy (yes vs no), irrespective of the intensity, frequency, and duration of cigarette smoking. Because many children were enrolled in early childhood during visits to their pediatrician, information relating to maternal clinical characteristics during pregnancy (eg, plasma HIV load and coinfections) was not always available. A detectable plasma HIV load was defined as >50 copies/mL. Preterm delivery was defined as births occurring at <37 weeks of gestation. Small-for-gestational age (SGA) was defined as a birth weight below the tenth percentile relative to Canadian neonates of the same gestational age [29].

LTL Measurement

Whole-blood relative LTL was measured via monochrome multiplex quantitative polymerase chain reaction as previously described [30]. Relative LTL was expressed as a ratio of the quantity of telomeric DNA normalized to the copy number of a single-copy nuclear gene [30]. Ratios were calibrated to approximate the absolute TL (in kilobases) by applying a coefficient, as previously described [30]. All study specimens were randomized, and those collected from a given study participant were assayed simultaneously to avoid interrun variability. Specimens that did not meet quality-control criteria [30], namely a <15% difference between replicates over 2 assay attempts, were excluded. The intraassay and interassay variabilities were 4.2% and 4.3%, respectively.

Statistical Analyses

The χ², Student’s t, and Mann-Whitney U tests were used to compare clinical and demographic characteristics of HEU and HUU study participants. These tests, as well as Pearson and Spearman correlations, were used to investigate univariate associations between LTL at birth and the following explanatory variables: HIV exposure status (HEU vs HUU), infant sex, gestational age at birth, birth weight, SGA, maternal ethnicity, maternal age at delivery, smoking ever during pregnancy, and duration and type of cART (zidovudine [AZT], lamivudine [3TC], and ritonavir-boosted lopinavir [LPV/r]; AZT, 3TC, and nelfinavir [NFV]; AZT, 3TC, and nevirapine [NVP]; abacavir [ABC], 3TC, and ritonavir-boosted protease inhibitor [PI/r]; and tenofovir disoproxil fumarate [TDF], emtricitabine [FTC] or 3TC, and PI/r) during pregnancy. Multiple-group comparisons were done using the Kruskal-Wallis test, followed by the Dunn’s multiple pairwise comparison test if indicated. Apart from cART duration and type, which were forced in as per an a priori decision, factors found to be important univariately (ie, those with P value of < .15) were all considered while developing multivariable analyses of covariance, using backward stepwise selection. Equality of variances for ≥2 groups was verified using the Levene’s test. Interactions between HEU or HUU status and other variables of interest were also examined and, if present, were included in the model. For cross-sectional comparison of LTL during the first 3 years of life, a subset of HEU and HUU children were matched by sex and age (±2 days within the first 2 weeks of life, ±8 days from 2 weeks to 1 year, and ±15 days from 1–3 years) in a 1:1 ratio. The relationship between LTL and age was then evaluated by comparing the linear regression’s slopes of the 2 groups, using GraphPad Prism v7. Analysis of LTL dynamics of HEU children between birth and the closest subsequent visit during the prophylaxis period was performed using the Wilcoxon signed rank paired test. Finally, to assess the relationship between age and LTL, longitudinal LTL dynamics in HEU children was analyzed using a generalized mixed-effects additive model with the “mgcv” package in R.

RESULTS

Characteristics of Study Participants

All Participants

Whole-blood specimens were available for 324 HEU children and 306 HUU children. Of these, 214 HEU children (66%) had ≥2 blood specimens collected between birth and three years of age whereas all HUU children had only a single blood specimen each (Supplementary Figure 1). Characteristics of all the children (and their mothers) are presented in Table 1A. Demographic information was unavailable for 154 anonymous HUU children. For the remainder, HEU and HUU children were very similar, apart from their ethnicity, whereby approximately 50% of HEU children but <1% of HUU children were black/African Canadian. Rates of maternal smoking during pregnancy, although high, were comparable between groups, with 33% of HIV-positive mothers and 30% of HIV-negative mothers self-reporting to have ever smoked during their pregnancy. With respect to in utero cART exposure, the most common regimen backbone was AZT and 3TC. A total of 112 HEU children (35%) were exposed to maternal AZT, 3TC, and LPV/r; 83 (26%), to AZT, 3TC, and NFV; 26 (8%), to AZT, 3TC, and NVP; 32 (10%), to ABC, 3TC, and PI/r; and 27 (8%), to TDF, FTC or 3TC, and PI/r. The remaining 36 HEU children (11%) were exposed to other nonstandard cART regimen (Supplementary Table 2). Seven HEU children (2%) were born to ART-naive HIV-positive mothers.

Table 1.

Demographic and Clinical Characteristics of Human Immunodeficiency Virus (HIV)–Exposed, HIV-Uninfected (HEU) Children and HIV-Unexposed, HIV-Uninfected (HUU) Children Overall and Among Those With a Blood Specimen Collected at Birth

Characteristic	All Children			Children With Birth Specimen
	HEU Group (n = 324)	HUU Group (n = 306)	P	HEU Group (n = 114)	HUU Group (n = 88)a	P
Infants
Male sex	161 (50)	169 (55)	.16	64 (56)	47 (53)	.70
Gestational age, wk	38.4 (27.1–41.7) (n = 323)	39.0 (28.3–42.1) (n = 149)	.07	38.3 (31.3–41.6)	39.4 (28.9–42.1)	<.001
Preterm delivery (<37 wk)	61 (19)	28 (19)	.98	23 (20)	9 (10)	.055
<37 and ≥34 wk	42 (13)	17 (11)		19 (17)	8 (9)
<34 wk	19 (6)	11 (7)		4 (4)	1 (1)
Birth weight, kg	3.1 (1.6–4.2) (n = 184)	3.2 (1.2–5.2) (n = 150)	.32	3.1 (1.6–4.1)	3.4 (1.4–5.2)	<.001
SGA	30 (16) (n = 184)	27 (18) (n = 149)	.66	19 (17)	12 (14)	.55
Mothers
Age at delivery, y	31.5 (16.7–45.3)	32.2 (16.7–44.0) (n = 135)	.15	31.0 (17.4–42.4)	32.5 (21.3–43.0)	.063
Ethnicity			<.001			<.001
Indigenous	42 (13)	9 (3)		33 (29)	7 (8)
Black/African Canadian	158 (49)	2 (0.7)		28 (25)	2 (2)
White	84 (26)	93 (30)		39 (34)	55 (63)
Asian	21 (6)	29 (9)		10 (9)	16 (18)
Other/unknown	12/7 (6)	19/154 (57)		4/0 (4)	1/7 (9)
Smoking ever in pregnancy	105 (33) (n = 322)	45 (30) (n = 152)	.51	63 (55)	38 (43)	.09
Detectable plasma HIV load (>50 copies/ml) close to delivery	18 (12) (n = 145)	NA		12 (12) (n = 99)	NA
cART
Duration of in utero exposure, wk	24.4 (0.0–41.7) (n = 322)	NA		20.6 (0.0–41.1)	NA
Time of initiation
Before conception	124 (38)	NA		40 (35)	NA
1st trimester	39 (12)	NA		15 (13)	NA
2nd trimester	124 (38)	NA		46 (40)	NA
3rd trimester	29 (9)	NA		13 (11)	NA
Naive/unknown	7/1 (2)	NA		0 (0)	NA
Regimen
AZT, 3TC, NVP	26 (8)	NA		7 (6)	NA
AZT, 3TC, NFV	83 (26)	NA		34 (30)	NA
AZT, 3TC, LPV/r	112 (35)	NA		45 (39)	NA
ABC, 3TC, PI/r	32 (10)	NA		10 (9)	NA
TDF, FTC (or 3TC), PI/r	27 (8)	NA		10 (9)	NA
Otherb	36 (11)	NA		8 (7)	NA
Naïve/unknown	7/1 (2)	NA		0 (0)	NA
AZT prophylaxis duration in infants, wk	6 (0–8) (n = 323)	NA		5 (2–6)	NA

Data are no. (%) of individuals or median (range). Twenty-nine sets of siblings (25 HEU children and 4 HUU children) were identified among all study participants, of whom 5 were HEU twin pairs.

Abbreviations: ABC, abacavir; AZT, zidovudine; cART, combination antiretroviral therapy; FTC, emtricitabine; LPV/r, ritonavir-boosted lopinavir; NA, not applicable; NFV, nelfinavir; NVP, nevirapine; PI/r, ritonavir-boosted protease inhibitor; SGA, small for gestational age; TDF, tenofovir disoproxil fumarate; 3TC, lamivudine.

^aThree anonymous HUU controls (with no information apart from age and sex) are not listed in this table because they could not be included in multivariable models.

^bDetailed in Supplementary Table 2.

Participants With a Birth Specimen

A specimen obtained shortly after birth (age, 0–3 days) and demographic information were available for 114 HEU children and 88 HUU children. Their characteristics are shown in Table 1B. The majority of these HEU infants were of Indigenous (29%) and black/African Canadian (25%) ethnicity, while 63% of HUU infants were white. There was no difference in infant sex or SGA between the groups. However, HEU infants had lower gestational age and weight at birth (P < .001 for both comparisons) and tended toward a higher risk of preterm birth (P = .055). Among these 202 participants, rates of maternal smoking during pregnancy were higher (P < .001) than those for the overall group, with 55% of HIV-positive mothers and 43% of HIV-negative mothers self-reporting having ever smoked during their pregnancy. Detailed smoking intensity data were available for a subset of mothers (44 were HIV positive and 11 were HIV negative), and were similar between the 2 groups (median number of cigarettes/day, 10.0 [interquartile range, 5.0–10.5] among HIV-positive mothers vs 5.0 [interquartile range, 3.5–11] among HIV-negative mothers; P = .48). All pregnant women received cART during their pregnancy, and once again, AZT, 3TC, and LPV/r and AZT, 3TC, and NFV were the most common regimens. At the visit closest to delivery, 88% of women had an undetectable plasma HIV load.

Infant LTL at Birth

LTL data were obtained for all infants, but for 2 HUU children measurements did not pass assay quality control (Supplementary Figure 1). In univariate analyses, LTL at birth was similar between HEU and HUU infants (Figure 1A). Longer LTL was associated with female sex in the combined group, as well as in the HEU and HUU subgroups (Figure 1B and Supplementary Table 3). Lower infant birth weight was significantly associated with longer LTL at birth in the combined group and in the HEU subgroup (Supplementary Table 3). Being SGA was associated with longer LTL only in the HEU subgroup (Supplementary Table 3). Additionally, there was an association between LTL and ethnicity (P = .045, by the Kruskal-Wallis test), whereby infants born to black/African Canadian mothers had significantly longer LTL at birth than those born to Indigenous mothers but not those born to white mothers (Supplementary Figure 2). Of note, a significant interaction (P = .009) between HEU/HUU status and maternal smoking ever during pregnancy was observed, with the latter being associated with longer LTL in HUU children but shorter LTL in HEU children (Figure 1C).

Figure 1.

Unadjusted comparisons of leukocyte telomere length (LTL) at birth. A, Comparisons between human immunodeficiency virus (HIV)–exposed, HIV-uninfected (HEU) infants and HIV-unexposed, HIV-uninfected (HUU) infants. The P value was determined by the Mann-Whitney U test. B, Comparisons between males and females. The P values were determined by the Mann-Whitney U test. C, Comparisons between HEU children and HUU children with or without mothers who smoked during pregnancy. The P values were determined by the Mann-Whitney U test. D, Comparisons between HEU children exposed to different combination antiretroviral therapy regimens in utero and HUU children. The P values were determined by the Student t test (solid lines; P = .02, by analysis of variance, for the 3-group comparison involving HEU children exposed to the zidovudine [AZT] and lamivudine [3TC] backbone regimen) or the Kruskal-Wallis test (dashed line). For panels B–D, whiskers of the box plots represent the 5th–95th percentiles. ABC, abacavir; FTC, emtricitabine; LPV/r, ritonavir-boosted lopinavir; NFV, nelfinavir; NVP, nevirapine; NS, not significant; PI/r, ritonavir-boosted protease inhibitor; TDF, tenofovir disoproxil fumarate.

Among HEU infants, neither duration of cART exposure in utero nor maternal plasma HIV load status (detectable vs undetectable) close to delivery were related to LTL at birth among HEU children. However, HEU children exposed to AZT, 3TC, and NVP during pregnancy had significantly longer LTL at birth than HEU children exposed to AZT, 3TC, and LPV/r (P = .01) or HUU controls (P = .034; Figure 1D).

In a multivariable model of 185 children with a birth specimen (Figure 2A) that included HEU/HUU status, infant sex, birth weight, ethnicity, maternal smoking ever during pregnancy, and the HEU/HUU status*maternal smoking interaction term, female sex remained independently associated with longer infant LTL at birth. With respect to the other variables, although the model suggests that HEU status and being born to mothers who smoked during pregnancy were both independently associated with longer LTL at birth, this must be interpreted with caution because of the behavior of the interaction term. The latter clearly indicates that HEU infants born to mothers who smoked during pregnancy had significantly shorter LTL at birth (P < .001) as compared to HUU infants born to mothers who did not smoke during pregnancy, indicating that the direction of the effect of maternal smoking is opposite in the 2 groups (Figure 2C and 2D). In a similar model that separated HEU children according to in utero cART regimen exposure (Figure 2B), HEU children exposed to AZT, 3TC, and NVP and those exposed to AZT, 3TC, and NFV had significantly longer LTL at birth than HUU children. Additionally, although HIV-positive mothers tended to be younger than HIV-negative mothers (P = .063), the results did not change when maternal age at delivery was forced into the models (data not shown).

In the multivariable model restricted to the HEU group (106; Figure 2C), lower birth weight, being born to mothers who never smoked during pregnancy, and being exposed to AZT, 3TC, and NVP (as compared to AZT, 3TC, and LPV/r) remained independently associated with a longer LTL at birth, while duration of cART exposure continued to show no association.

In the final multivariable model of HUU children, restricted to the 79 children who had known ethnicity (Figure 2D), female sex and being born to mothers who smoked during pregnancy but not lower birth weight were independently associated with a longer LTL at birth.

Figure 2.

Multivariable regression analyses of the association between possible predictors of leukocyte telomere length (LTL) at birth. A, Findings for all 185 participants (R² = 0.17). B, Findings for all 185 participants, with human immunodeficiency virus (HIV)–exposed, HIV-uninfected (HEU) children separated by type of combination antiretroviral therapy (cART) exposure in utero (R² = 0.20). C, Findings for 106 HEU children (R² = 0.19). D, Findings for 79 HIV-unexposed, HIV-uninfected (HUU) children (R² = 0.18). Effect sizes are expressed as nonstandardized β values. Strong collinearity between gestational age and birth weight was detected and precluded inclusion of both variables in the same model. Separate models that included either gestational age or birth weight were constructed, but the latter explained a greater portion of the variance (ie, it had a higher R²) and was chosen as the best model. ABC, abacavir; AZT, zidovudine; CI, confidence interval; FTC, emtricitabine; LPV/r, ritonavir-boosted lopinavir; NFV, nelfinavir; NVP, nevirapine; PI/r, ritonavir-boosted protease inhibitor; TDF, tenofovir disoproxil fumarate; 3TC, lamivudine. ^aEight HEU children who were exposed to nonstandard cART regimen in utero were excluded from these analyses.

Given an imbalance in ethnicity between the groups, whereby HEU infants were more likely to be of Indigenous or black/African Canadian ethnicity as compared to HUU infants (53% vs 10%), a sensitivity analysis that included only children born to white mothers (n = 93; 39 HEU children and 54 HUU children) was performed. The results were essentially unchanged from that of the full sample (Supplementary Figure 3), suggesting that the associations observed with LTL at birth were not confounded by ethnicity per se.

Cross-sectional Comparison of Group LTL During the First 3 Years of Life

For a total of 214 distinct HEU children aged 0–3 years, a single specimen was matched at a 1:1 ratio with those from HUU children of the same age and sex. Their characteristics (both infant and maternal) are detailed in Supplementary Table 4. The linear regression’s slope of LTL vs age was similar in both groups (P = .49; Figure 3A). Comparison of LTL ranks (by the Mann-Whitney U test) between the groups also showed no difference. In agreement with differences seen at birth, females had longer LTL than males, especially during the first 18 months of life. However, males maintained their LTL over time, while females showed significant LTL attrition (P < .01; Figure 3B–E).

Figure 3.

Relationship between leukocyte telomere length (LTL) and age during the first 3 years of life among age- and sex-matched children. A, Findings for all human immunodeficiency virus (HIV)–exposed, HIV-uninfected (HEU) and all HIV-unexposed, HIV-uninfected (HUU) infants. B, Findings for HEU males and HUU males. C, Findings for HEU females and HUU females. D, Findings for HEU males and HEU females. E, Findings for HUU males and HUU females.

Longitudinal LTL Dynamics Among HEU Children

Longitudinal changes in LTL were assessed in the 214 HEU children from whom ≥2 blood specimens were collected between birth and 3 years of age. A significant nonlinear relationship between LTL and age was observed, showing a rapid decline in LTL during the first 40 weeks (Figure 4A). This was followed by a leveling out, perhaps even an upward trend, but the confidence intervals around the line were consistent with a flat relationship.

Figure 4.

A, Nonlinear regression model (ie, generalized mixed effects additive model) of leukocyte telomere length (LTL) and age (in weeks) for human immunodeficiency virus (HIV)–exposed, HIV-uninfected (HEU) infants. For clarity, one outlier (LTL value of 19.8) is omitted from the figure but not from the analysis. B, Linear regression model of LTL and age (in weeks) for HIV-unexposed, HIV-uninfected (HUU) children. The solid and dashed lines indicate the regression line and the 95% confidence interval of the estimated line, respectively. C, Comparisons of LTL at birth and the subsequent visit (at age 18–47 days) among HEU children. The P value was calculated by the paired Wilcoxon signed rank test.

For the HUU infants from whom we had 1 specimen per participant, a linear regression model showed an inverse association between LTL and age (P = .004; Figure 4B).

LTL in HEU Children During Prophylaxis

Finally, to evaluate the possible effect of infant AZT prophylaxis on LTL, the change in LTL among HEU children between birth and the closest subsequent visit during prophylaxis was assessed using paired HEU whole-blood specimens from 58 children (Figure 4C). There was no difference in LTL between birth and the closest subsequent visit after birth (median age, 31 days; range, 18–47 days; P = .09).

DISCUSSION

In this cohort, we found no evidence that LTLs are shorter in HEU children as compared to HUU children, whether at birth or over the first 3 years of life. These results are in agreement with those from smaller previous studies [24–27]. The in utero cART exposure among HEU children varied with respect to duration and type, in accordance with the treatment guidelines of the time [31, 32]. HEU children exposed to maternal AZT, 3TC, and NFV or NVP had longer LTLs at birth as compared to HUU children. This may be related to the selective elimination of cell subsets with shorter telomeres and/or proliferation of cells with longer telomeres; our study was not designed to address this. Furthermore, the LTL among HEU children did not change significantly during the AZT prophylaxis period. Taken together, these results are reassuring and suggest that HIV and cART exposure appears to have little or no effect on telomere length early in life.

In contrast, maternal smoking during pregnancy exerts an effect on LTL at birth, albeit in opposite directions in the 2 groups studied. Among HEU infants, being exposed to any maternal smoking in utero was associated with a shorter LTL. This may be related to increased oxidative stress, in addition to stresses related to HIV and cART exposure. Although a similar effect was expected in HUU infants, our data instead suggest that HUU children exposed to maternal smoking have a longer LTL at birth than their unexposed peers. While this counterintuitive observation may be related to differences in smoking intensity, our exploration of this in a subset of women does not support this explanation. However, this is consistent with another study [33], which reported longer CD4⁺ T-cell TL in infants born to smoking mothers. The mechanism behind this effect is unclear, and a larger prospective study with more details about smoking exposure and other confounders would be required to confirm this observation. Early breastfeeding has been associated with a longer LTL in children [34], and reduced breastfeeding rates and duration were reported among mothers who smoked [35]. We lacked breastfeeding information for our controls (approximately 90% of HIV-uninfected women breastfeed in Canada), and none of the HIV-positive mothers breastfed. Neither breastfeeding nor the effects of smoking on breastfeeding are likely confounders in our study, given that the effects observed are in the opposite direction of those expected.

Within both HEU and HUU groups, female infants had longer LTLs at birth as compared to males but also experienced faster attrition, which attenuated any difference before 3 years of age. Our finding of longer TL among female infants is consistent with other studies that reported sex differences in LTL at birth and early childhood [36, 37]. While it is well documented that women have longer telomeres than men [38–40], an effect attributed to the ability of estrogen to stimulate telomerase [41] and act as an antioxidant against reactive oxygen species that damage telomeres [42], it is unclear how this is conferred as early as birth. Nevertheless, LTL early in life has been suggested to predict health outcomes later in life [36], and women do, on average, live longer than men.

Finally, we were uniquely positioned to investigate longitudinal changes in LTL among HEU children early in life. We observed that telomere attrition was rapid from birth to 1 year of age but slowed thereafter, mirroring those in a baboon study reporting a similar rapid decline in granulocyte and lymphocyte TL in the first year of life, followed by a stabilization of TL [43]. This may reflect a period of significant growth, as well as differentiation and maturation of cells forming the innate and adaptive immune systems.

This is the first study to investigate longitudinal changes in children’s LTL early in life. It is also the first detailed investigation of the potential effects of in utero exposure to maternal cART and smoking on infant LTL. However, our study has some limitations. First, we lacked longitudinal samples from HUU infants that would have enabled detailed comparisons of LTL attrition dynamics between the HEU and HUU groups. Our cohort also had imbalanced ethnicity, with nearly half of HEU children but <1% of HUU children being black/African Canadian. It is well known that black individuals have longer telomeres than white individuals [44–48]. Although our sensitivity analysis restricted to white children showed similar results, ethnicity could still be a confounding factor. Another limitation is the heterogeneity in the collection of smoking data across the 3 cohorts, which we addressed by defining smoking categorically, as yes/no ever during pregnancy. We acknowledge that this does not account for smoking frequency or intensity or for smoking duration, as some women may have quit during pregnancy. However, based on the subset of women for whom more-extensive smoking information was available, the number of cigarettes per day reported was similar for the 2 groups. Furthermore, our qualitative data indicate that few women quit smoking and that the majority of women who smoked at any time in pregnancy continued smoking throughout their pregnancy. It is also possible that the observed association between smoking and LTL could be a proxy for the effect of an unmeasured confounder, such as socioeconomic status or parental anxiety. Cytomegalovirus infection is known to be associated with shorter telomeres [49], and we did not have access to information on cytomegalovirus infection for our study participants. Paternal age is known to influence progeny TL, and this information was unavailable. Finally, given the small amount of blood obtained, only LTL was measured, and our results cannot be generalized to TL in specific cell subsets because these may evolve substantially over the first years of life.

In conclusion, we found that LTL was similar between HEU and HUU children at birth and over the first 3 years of life. Further, HIV and cART exposure in utero do not appear to alter telomere dynamics during early life, a reassuring finding. Instead, maternal smoking during pregnancy appears to exert a greater effect on LTL at birth, emphasizing the need for smoking cessation strategies, especially during pregnancy.

STUDY GROUP MEMBERS

Other investigators of the CIHR Team in Cellular Aging and HIV Comorbidities in Women and Children include Neora Pick, MD; Melanie Murray, MD; Patricia Janssen, PhD; Joel Singer, PhD; Normand Lapointe, MD; Jerilynn Prior, MD; Michael Silverman, MD; and Mary Lou Smith, PhD.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank all the study participants and the research staff at our CARMA sites and in the Côté laboratory.

Financial support. This work was supported by the Canadian Foundation for AIDS Research (grants 20-004 and 16-012), the Canadian Institutes of Health Research (grants HET-85515 [in HIV therapy and aging 2007–2012] and TCO-125269 [cellular aging and HIV comorbidities in women and children 2013–2018] to H. C. F. C. and D. M. M.), and the UBC Centre for Blood Research (Internal Collaborative Training Award to the first author A. A.).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Presented in part: 22nd Annual Conference on Retroviruses and Opportunistic Infections, Seattle, Washington, 23–26 February 2015; 24th Annual Canadian Conference on HIV/AIDS Research, Toronto, Canada, 30 April–3 May 2015.

Author notes

Additional study group members are listed at the end of the text.