No effect of tenofovir on infant bone mineral content in African study

CROI 2016Polly Clayden, HIV i-Base

A report from the PROMISE study showed no adverse bone mineral effect, linked to maternal tenofovir disoproxil fumarate (TDF), in exposed infants. [1] But starting a lopinavir/ritonavir (LPV/r)-containing regimen during pregnancy might lead to lower newborn bone mineralisation.

George Siberry presented data from a sub-study of PROMISE (Promoting Maternal-Infant Survival Everywhere/IMPAACT P1084) at CROI 2016.

Results from the main study were presented at CROI 2015. [2,3] The study showed that taking a three-drug ART regimen in pregnancy was more effective in preventing mother-to-child transmission than taking one antiretroviral drug during pregnancy, another in labour and two after delivery.

PROMISE was a multinational study to which asymptomatic women with >350 cells/mm3 (or above the local threshold for starting ART at the time) were enrolled. At 14 weeks of pregnancy women were randomised to: Arm A – AZT plus single-dose nevirapine at delivery plus TDF/ FTC tail vs Arm B – AZT/3TC plus LPV/r vs Arm C – TDF/FTC plus LPV/r.

At the study’s interim review the DSMB reported that the pre-specified efficacy boundary for this part of the study was crossed and there were safety differences between arms. The rate of vertical transmission at 14 days postpartum was significantly lower in the pooled triple ART arms (Arms B and C) compared with Arm A.

The P1084s infant DXA substudy compared newborn bone mineral content (BMC) by exposure to maternal antiretroviral regimens at >14 weeks gestational age (GA). The primary comparison was between the triple ART arms but the investigators also made secondary comparisons between each triple ART arm and Arm 1. P1084s participants were enrolled at eight African sites (Malawi, South Africa, Uganda and Zimbabwe) with DXA capacity within 21 days of PROMISE enrolment and before the onset of labour.

The infants had whole-body (WB) and lumbar spine (LS) DXA BMC measurements 28 days of age. Training was standardised. Analysis was standardised and performed centrally at UCSF.

The accrual target of 150 infants per arm was based on 80% power to detect a pair-wise difference of 4-5% in mean WB-BMC (121 evaluable per arm) and 6-7% in mean LS-BMC (140 evaluable per arm). The investigators compared mean BMC differences using Student’s t-test.

As mothers enrolled in the substudy after PROMISE randomisation, the investigators used multivariable linear regression to adjust for baseline maternal factors and infant factors at time of DXA scan.

Of 452 eligible mothers, data from 426 infants were available for analysis after accounting for twins (6 pairs), foetal (8) or neonatal death (10) and those that dropped out of the study (15). About 15% of infants did not have an evaluable DXA within the time frame. Arms 1, 2 and 3 had 118 (117 LS-BMC and 99 WB-BMC), 129 (127 LS-BMC, and 104 WB-BMC) and 115 (113 LS-BMC and 96 WB-BMC) infants with evaluable data respectively.

Overall mothers of infants with a DXA scan were median age of 26 years (range18-43), and a median CD4 count of 543 cells/mm3 (range 350-1493), entered the study at a median of 28 weeks (range13.7-39.3) of pregnancy and 19% reported alcohol use. The infants were 51% female, with a median GA of 40 weeks (32-43), a median length at birth of 49 cm (IQR 47-50), and weight of 2990 gm (IQR 2700-3210).

Mothers in the triple ART arms were slightly older 25.5 vs 27 years (p=0.03). Infants in the ART arms had slightly lower birth weight 2990 vs 2920 gm (p<0.001) respectively.

In Arms A, B and C, mean LS-BMC were 1.73 vs 1.64 vs 1.72g; and WB-BMC were 73.1 vs 65.1 vs 63.3g respectively.

LS-BMC pairwise comparisons revealed some borderline differences. Mean difference adjusted for maternal factors at baseline and infant factors at time of DXA: Arm B vs Arm C, -0.08 (95% CI -0.16 to 0.0), p=0.04; Arm A vs Arm C, +0.01 (95% CI -0.07 to 0.09), p=0.79; and Arm A vs Arm B, +0.07 (95% CI -0.01 to 0.15), p=0.09.

But significant differences in WB-BMC between Arm A and each triple ART arm persisted after adjustment. Mean difference adjusted for maternal factors at baseline and infant factors at time of DXA: Arm B vs Arm C, +1.22 (95% CI -2.31 to 4.75), p=0.5; Arm A vs Arm C, +8.69 (95% CI 4.78 to 12.60), p<0.001; and Arm A vs Arm B, +5.82 (95% CI 2.10 to 9.54), p=0.002.

Dr Siberry explained that there were some limitations to the study: higher rates of neonatal death in Arm C (TDF/FTC backbone) in the parent study; both ART arms used LPV/r, limiting extrapolation to most women in African countries who receive efavirenz (EFV)-based regimens; and women in PROMISE had high CD4 counts. He noted that the clinical significance and persitance of these findings are unknown.

But P1084s also had a number of strengths: maternal regimens were randomly assigned; the ART regimens only differed between two backbones; women started antiretrovirals in pregnancy; the sample size was large; and DXAs were standardised.

The study did not show an impact of maternal TDF on infant BMC. But, compared to AZT (plus single dose nevirapine plus TDF/FTC tail), LPV/r-based ART in pregnancy was associated with significantly lower WB-BMC, even after adjustment.


These data are reassuring as TDF is widely used in pregnancy and they also reinforce current WHO and national first-line recommendations.

One of the reasons for this substudy was that previous US observational data from PHACS found lower newborn BMC after maternal TDF use. [4] So much of the discussion after the presentation was about this contradiction. It is possible that different TDF-containing regimens or US versus African settings might make a difference. Also PHACS women started ART before conception or early in pregnancy whereas PROMISE women started from the second trimester onwards.

Other mediating factors are also being analysed including the effect of HIV. Dr Siberry pointed out that we have the advantage of other studies that are ongoing or planned in which it will be possible to look at HIV negative women receiving TDF as PrEP in pregnancy, as well as one in Thailand looking at TDF for the prevention of hepatitis B transmission.

A further important finding from the PROMISE substudy is that infant DXAs are feasible in African clinical research and 6-month LS results are also in analysis.


  1. Siberry JK et al. Impact of maternal tenofovir use on HIV-exposed newborn bone mineral content. Conference on Retroviruses and Opportunistic Infections. 22-26 February 2016. Boston. Oral abstract 36. (Abstract) (Webcast)
  2. Fowler MG et al. PROMISE: Efficacy and safety of 2 strategies to prevent perinatal HIV transmission. CROI 2015. 2015 Conference on Retroviruses and Opportunistic Infections (CROI 2015), 23-26 February 2015, Seattle. Oral abstract 31LB.
  3. Clayden P. Three drug ART best for preventing vertical transmission to infants: results from the PROMISE study. HTB. 24 March 2015.
  4. Siberry GK et al for the Pediatric HIV/AIDS Cohort Study. Lower newborn bone mineral content associated with maternal use of tenofovir disoproxil fumarate during pregnancy. Clin Infect Dis 2015; 61(6):996-1003.

Links to other websites are current at date of posting but not maintained.