To the Editor—We appreciate the insightful and important questions raised by Koo and colleagues about our report of an association between maternal tenofovir disoproxil fumarate use during pregnancy and newborn whole-body (WB) bone mineral content (BMC) [1, 2].
The authors lay out a convincing rationale for using WB-BMC with head, rather than less head, as the most appropriate dual-energy x-ray absorptiometry (DXA) outcome measurement for infants. Per protocol, WB-BMC with head was the primary outcome measure for our analysis; we included results for WB-BMC less head because this measure is commonly preferred in older children. We appreciate the additional justification for making WB-BMC (including head) the primary outcome in our infant study.
Dr Koo and colleagues also raise important concerns that we did not adequately account for the impact of DXA measurements being obtained throughout the first month of life, a period of rapid infant growth and bone mass accrual. As pointed out, the range of ages when DXA scanning was performed was further complicated by the observation that tenofovir-exposed infants, on average, had their DXA scans performed at a younger age. To address this important source of variability, we adjusted our analyses for infant length (as a measure of body size) and age at time of DXA. We chose length instead of weight (or mass) as a more stable measure of infant body size because infants normally lose substantial weight in the first days of life (postnatal diuresis) before regaining and surpassing birthweight by 2–3 weeks of age. Based on the important reasons raised by Koo and colleagues for using body mass (or weight) to standardize BMC for age at DXA, we have repeated the model with adjustment for infant body mass at DXA (Table 1). This approach results in a somewhat lower effect size that remains significant, which does not alter our conclusions.
Table 1.
Reanalysis of Adjusted Mean Whole-Body Bone Mineral Content Differences by Various Methods of Adjustment for Infant Body Size
| Method of Adjustment for Infant Body Size | Adjusted Mean Difference (g) in Whole-Body BMC (With Head) Between Tenofovir Exposure vs no Tenofovir Exposure | 95% CI | P Value |
|---|---|---|---|
| Infant body length at DXA (as reported in [1]) | −5.3 | −9.5 to −1.2 | .013 |
| Infant body mass at DXA | −3.0 | −5.5 to −.5 | .021 |
| Infant body length and infant body mass at DXA | −3.2 | −5.7 to −.7 | .014 |
Abbreviations: BMC, bone mineral content; CI, confidence interval; DXA. dual-energy x-ray absorptiometry.
As we noted in our report, potential residual confounding is a limitation of our study design, and we reiterate the importance of corroborating our findings in future studies in which maternal tenofovir exposure is randomly assigned.
Note
Potential conflict of interest. Both authors: No reported conflicts. Both 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.
References
- 1.Siberry GK, Jacobson DL, Kalkwarf HJ et al. Lower newborn bone mineral content associated with maternal use of tenofovir disoproxil fumarate during pregnancy. Clin Infect Dis 2015; 61:996–1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Koo W, Caldito G, Bocchini J Jr. Bone mineral content in neonates. Clin Infect Dis 2016; 62:1323–4. [DOI] [PubMed] [Google Scholar]