Phosphaturia in HIV-Exposed Uninfected Neonates Associated with Maternal Use of Tenofovir Disoproxil Fumarate in Late Pregnancy

Abstract

Background

Tenofovir disoproxil fumarate (TDF) is often used in treating pregnant women living with HIV. Third-trimester TDF exposure is associated with a 12% reduction in bone mineral content in HIV-exposed uninfected (HEU) neonates. The potential mechanisms underlying this observation are unknown.

Methods

The TDF study enrolled newborns of gestational age ≥36 weeks from the Surveillance Monitoring for Antiretroviral Therapy and Toxicities study based on in utero TDF exposure (TDF use ≥8 weeks in the third trimester vs none). Blood and urine samples were collected cross-sectionally within 30 days of birth to assess renal function (serum creatinine, serum phosphate, eGFR, percent tubular reabsorption of phosphate [PTRP]), and bone turnover (serum parathyroid hormone, 25-OH vitamin D [25(OH)D], and urinary cross-linked N-telopeptide of type 1 collagen). For each biomarker, a LOESS plot was fit using values at age at specimen collection; regression lines over age were fit among samples collected from 4 to 30 days, to compare slopes by TDF exposure.

Results

Among 141 neonates, 77 were TDF-exposed and 64 TDF-unexposed. Between age 4 and 30 days, PTRP decreased more rapidly in the TDF-exposed compared to the unexposed group with slopes of −0.58 vs −0.08/day (difference −0.50/day [95% CI −0.88, −0.11]). Slopes for 25(OH)D were similar in both groups, but serum levels were lower in TDF-exposed neonates (median [IQR]: 22 [19, 29] vs 26 [22, 37] ng/mL). No differences were observed for other biomarkers.

Conclusions

Third-trimester in utero exposure to TDF is associated with increased urinary loss of phosphate and lower serum concentrations of 25(OH)D in HEU neonates.

TDF use during the third trimester of pregnancy in women living with HIV is associated with lower serum 25-OH vitamin D levels and increased urinary phosphate loss in their exposed but uninfected offspring within the first 30 days of life.

INTRODUCTION

Tenofovir disoproxil fumarate (TDF) is a nucleotide reverse transcriptase inhibitor (NRTI) used worldwide for the treatment of HIV and chronic hepatitis B virus (HBV) infection. It is an option in evidence-based guidelines for the treatment of HIV and chronic HBV in adults and adolescents, in pregnant people living with HIV (WLHIV), and with chronic HBV infection for the prevention of vertical transmission of these viruses. It is also used for HIV pre-exposure prophylaxis [1–6].

In clinical practice, TDF-containing regimens are generally well-tolerated. However, clinically significant reductions in bone mineral density (BMD) can manifest soon after initiating a TDF-containing regimen [7, 8]. Renal toxicity occurs in some children and adults receiving TDF, manifesting as proximal tubular dysfunction, with phosphaturia, glycosuria, uricosuria, and aminoaciduria, and can range from asymptomatic proteinuria to progressively declining glomerular filtration rate [9, 10]. A systematic review and cohort study on TDF use during pregnancy and its outcomes have concluded that TDF use seems safe, but both acknowledge the limited data and need for more prospective studies in neonates and infants, addressing renal and bone health in particular [11, 12].

Subsequent studies examined bone toxicity related to TDF exposure in young children. One showed maternal TDF use from 28 weeks gestation to 2 months postpartum to prevent mother-to-child transmission of HBV had no significant effect on BMD in mother–infant dyads measured a year after delivery [13]. A pilot study of pregnant women co-infected with HIV and HBV showed a trend towards lower whole-body bone mineral content (BMC) at 6 months of age in in utero TDF-exposed infants, compared to TDF-unexposed infants [14]. Another study showed no differences in bone turnover markers according to in utero and breastfeeding TDF exposure in 6-month-old infants, compared to those not exposed [15]. A study of infants born to WLHIV in Brazil and Malawi with exposure to a very short peripartum TDF course who had hand, wrist, and spine radiographs performed at ages 3 days and 12 weeks did not show any association of bony abnormalities with TDF exposure [16]. Our initial study, in which neonates born to WLHIV underwent dual-energy X-ray absorptiometry scans, demonstrated a 12% reduction in BMC in neonates exposed to TDF during the third trimester, compared to neonates on non-TDF regimens [17].

The hypothesized mechanism underlying proximal tubular cell injury caused by tenofovir (TFV), released from the prodrug TDF, is by inhibition of mitochondrial DNA polymerase gamma, demonstrated in murine models, and human kidney histology. There is mitochondrial clumping, loss, reduction in mitochondrial DNA content, and eventual apoptosis. Injured tubular cells fail to function, losing protein, phosphate, and other kidney-related biomarkers, including vitamin D binding protein with resultant vitamin D insufficiency and a compensatory increase in parathyroid (PTH) levels [18, 19]. Bone toxicity is thought to occur by inhibition of ATP release with stimulation of osteoclast differentiation and osteoblast inhibition, causing bone resorption and reduced BMD [20]. Impaired hydroxylation of 25-OH vitamin D (25(OH)D) by renal tubular cells reduces the synthesis of calcitriol, the active form of vitamin D, linking bone with kidney toxicity.

Here we examined bone metabolism and renal function in a cohort of neonates to elucidate mechanisms behind the observed reduction in BMC in neonates exposed to TDF during the third trimester.

METHODS

Study Design

The Surveillance Monitoring for Antiretroviral Therapy and Toxicities (SMARTT) study is a prospective cohort study of infants and children who were HIV-exposed but uninfected (HEU). The study is designed to assess the long-term effects of in utero exposure to ARVs across various domains and organs. It commenced in 2007 at 23 sites within the United States, including Puerto Rico, and is still ongoing, with IRB approval at each site and at the Harvard TH Chan School of Public Health. The dynamic component of SMARTT enrolled participants late in pregnancy or within 72 hours of delivery. Parents or legal guardians provided informed consent for participation by themselves and their child or children.

This TDF study was conducted at 14 of 23 sites between April 2011 and June 2013 [21]. Singleton neonates born to WLHIV who received ≥8 weeks of TDF in the third trimester (TDF-exposed group) and those born to WLHIV who never received TDF during their pregnancy (TDF-unexposed, comparator group) were enrolled if they had a gestational age (GA) ≥36 weeks. Maternal clinical data and infant birth weight, birth length, and GA were abstracted from the SMARTT study database. Z-scores were calculated for weight and length using CDC 2000 growth charts [22]. Serum and urine samples were collected once from each neonate between 0 and 30 days of life and frozen at −80°C.

Assessment of Renal Function

Serum creatinine and phosphate measurements were performed locally in real-time at the clinical laboratory at SMARTT sites. Urine was stored at the SMARTT central repository until samples were sent for batch analyses to Quest Diagnostics (Secaucus, NJ) for urine creatinine and phosphate assays. The estimated glomerular filtration rate (eGFR) in mL/min/m² was calculated as follows:

• For neonates ≥36 to <37 weeks GA, eGFR = 0.33 × infant length (cm)/serum creatinine (mg/dL).

• For neonates ≥37 weeks GA, eGFR = 0.45 × infant length (cm)/serum creatinine (mg/dL).

Percentage tubular reabsorption of phosphate (PTRP) was calculated for each infant as follows:

$$\begin{array}{l}(1-[(\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{s}\mathrm{p}\mathrm{h}\mathrm{a}\mathrm{t}\mathrm{e}\times \mathrm{s}\mathrm{e}\mathrm{r}\mathrm{u}\mathrm{m}\mathrm{c}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{e})/(\mathrm{s}\mathrm{e}\mathrm{r}\mathrm{u}\mathrm{m}\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{s}\mathrm{p}\mathrm{h}\mathrm{a}\mathrm{t}\mathrm{e}\times \mathrm{u}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{c}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{e})])\times 100.\end{array}$$

The reference range for markers of renal function is listed in the footnote to Figure 1 [23].

Figure 1.

LOESS plots for neonatal renal and bone biomarkers by age at specimen collection (0–30 days since birth) for individual TDF-exposed and TDF-unexposed participants. Clockwise from top left: serum creatinine, serum phosphate, estimated eGFR (estimated glomerular filtration rate), urine NTx/cr (cross-linked N-telopeptide of type 1 collagen normalized for creatinine), serum PTH (parathyroid hormone), serum 25-OH vitamin D, and PTRP (percentage tubular reabsorption of phosphate) from 0–30 days of age. The vertical line divides the participants into the 0–3- and 4- to 30-day age groups. The reference range for serum creatinine in the neonate is 0.3–0.9 mg/dL (first 14 days of life) and 0.1–0.4 mg/dL (15 days to 2 years of age). For serum phosphate the range is 5.6–10.5 mg/dL (first 14 days of life) and 4.8–8.4 mg/dL (15 days to 1 year of age). Normal PTRP is >80% across all ages. With eGFR the reference range (± standard deviation) in the first week of life is 41 ± 15, and between 2 and 8 weeks 66 ± 25 mL/min/m². Reference ranges for 25(OH)D is 30–100 ng/mL (deficiency < 20 ng/mL, insufficiency 20 to <30 ng/mL), and for PTH, 10–65 pg/mL. No normative values for NTX/cr are available for neonates and children. Abbreviations: eGFR, estimated glomerular filtration rate; NTx/cr, urine NTx normalized for urine creatinine; PTH, parathyroid hormone; PTRP, percentage tubular reabsorption of phosphate; TDF, tenofovir disoproxil fumarate.

Assessment of Bone Parameters

Serum 25(OH)D and PTH levels were measured. Cross-linked N-telopeptide of type 1 collagen (NTx), a specific marker for bone resorption, was measured in urine [24]. For analysis, urine NTx was normalized for urine creatinine (NTx/cr) using the equation urine NTx (mg/dL)/urine creatinine (mg/dL), and assays performed at Quest Diagnostics (Secaucus, NJ).

Reference ranges established for children and adults are used for bone markers due to the lack of data on neonates and are listed in the footnote to Figure 1 [25].

Statistical Analysis

Neonatal biomarker concentrations in the first few days after birth are strongly influenced by maternal levels of these biomarkers, and thereafter decline gradually achieving concentrations that reflect the newborn’s own metabolism [26, 27]. Therefore, we performed analyses separately, depending on whether a participant sample was collected at 0–3 or at 4–30 days after birth. Because serum and urine were only collected once for each neonate, the analysis was cross-sectional rather than longitudinal, and the sample from each neonate fell into one of the two time periods. For the analysis of PTRP, we only included infants from whom the serum and urine were collected on the same day.

All descriptive analyses and models were fit separately by age group at the time of sample collection (0–3 days, 4–30 days), except for locally weighted scatter smoother (LOESS) plots, which were fit for all samples collected between 0 and 30 days. Specifically, we described the distribution of sociodemographic, clinical, and laboratory values in TDF-exposed and -unexposed neonates. We used LOESS plots stratified by participant TDF exposure to examine trends in the value of each of the seven individual laboratory outcomes over the age at which the sample was drawn. We produced scatter plots for each laboratory variable by TDF exposure across age with a fitted regression line, individually for the two age groups. Linear regression models were fit using generalized estimating equations with robust variance to compare the difference in slopes for each laboratory outcome over age between the TDF-exposed and TDF-unexposed groups.

We fit models for the 4- to 30-day samples, but not the 0- to 3-day samples, because the latter are more reflective of maternal biomarkers and not the infant’s physiologic responses to TDF exposure. The models were adjusted for infant sex (male vs female), race (Black vs non-Black), and yearly household income (<$10 000 vs ≥$10 000) and included an effect modification term for TDF exposure by age at sample collection. When there was no evidence for a slope difference in biomarkers between TDF groups, we used a linear regression model for each biomarker to estimate the adjusted mean difference in biomarker concentrations between TDF-exposed and -unexposed groups, adjusted for age at sample collection, sex, race, and income without the term for TDF exposure by age. We did not have adequate data to adjust for maternal CD4 and HIV viral load at the time of initiation of ARVs in the TDF-exposed or unexposed groups.

RESULTS

Participant Population

Of 195 infants enrolled in the TDF Study, 141 participants (77 TDF-exposed and 64 TDF-unexposed) met the study criteria and were included in the analysis (Supplementary Figure 1). There were 34 and 43 neonates, respectively, exposed to TDF in the (0- to 3-day and 4- to 30-day age groups, with 45 and 19 neonates, respectively, not exposed to TDF. Demographic, socioeconomic, and antiretroviral regimens for both age groups are shown in Table 1. This analysis was performed in the 4- to 30-day group, in which 63% participants were male, and 81% Black. Of their mothers, 77% had at least a high school education and 50% had an annual income <$10 000. TDF-exposed and -unexposed infants in this age group were similar except more exposed infants were born to women with maternal annual income of <$10 000 (55% vs 39%). The median (interquartile range) days in pregnancy when TDF started in the exposed group was 99 (0–163). When the mother was on TDF at the last menstrual period, that day was designated 0 (TDF start date). A boosted protease inhibitor (PI) was the most common anchor drug administered with TDF, in both the 0–3 and 4–30-age groups (approximately 3-quarters of maternal antiretroviral regimens in the third trimester). In the 4–30-age group the percent of TDF-exposed participants whose mothers received a boosted PI compared to TDF-unexposed participants was 79% vs 63% respectively. For atazanavir/ritonavir (ATV/r) these were 58% vs 5%; with lopinavir/ritonavir (LPV/r), 0% vs 47%; and for darunavir/ritonavir (DRV/r) 16% vs 11%. Maternal CD4 count and log₁₀ HIV viral load in the third trimester of pregnancy were similar between TDF-exposed and -unexposed (Table 1).

Table 1.

Socioeconomic and Anthropometric Characteristics by In Utero TDF Exposure Among Those With Renal and Bone Biomarkers Measured Between 0–3 and 4–30 Days of Birth

Characteristics	Specimens Collected 0–3 Days Since Birth			Specimens Collected 4–30 Days Since Birth
	TDF-Exposed (N = 34)	TDF-Unexposed (N = 45)	Total (N = 79)	TDF-Exposed (N = 43)	TDF-Unexposed (N = 19)	Total (N = 62)
Mother—age at delivery (y)
Median (Q1, Q3)	29.2 (23.7, 33.4)	28.1 (25.0, 32.3)	28.7 (23.8, 33.1)	29.5a (25.8, 34.0)	30.4 (21.9, 33.5)	29.9 (25.1, 33.7)
Mother—high school education, number (%)b
Less than high school	12 (36)	16 (36)	28 (36)	10 (23)	4 (21)	14 (23)
At least high school	21 (64)	28 (64)	49 (64)	33 (77)	15 (79)	48 (77)
Income/year <$10 000, number (%)c
No	7 (22)	15 (39)	22 (31)	19 (45)	11 (61)	30 (50)
Yes	25 (78)	23 (61)	48 (69)	23 (55)	7 (39)	30 (50)
Days of pregnancy when TDF started (LMP = 0)
Day (Q1, Q3)	63 (0, 143)	–	–	99 (0, 163)	–	–
Sex assigned at birth, number (%)
Male	19 (56)	24 (53)	43 (54)	29 (67)	10 (53)	39 (63)
Infant—race/ethnicity, number (%)
White non-Hispanic	0 (0)	2 (4)	2 (3)	3 (7)	0 (0)	3 (5)
Black non-Hispanic	14 (41)	21 (47)	35 (44)	33 (77)	15 (79)	48 (77)
Hispanic (regardless of race)	17 (50)	20 (44)	37 (47)	5 (12)	4 (21)	9 (15)
More than one race or other	3 (9)	2 (4)	5 (6)	2 (5)	0 (0)	2 (3)
Infant—Black race, number, (%)d
No	7 (24)	11 (29)	18 (27)	8 (19)	4 (21)	12 (19)
Yes	22 (76)	27 (71)	49 (73)	35 (81)	15 (79)	50 (81)
Gestational age (weeks)
Median (Q1, Q3)	38.4 (37.9, 39.0)	38.0 (37.9, 38.6)	38.1 (37.9, 38.9)	38.6a (37.4, 39.3)	38.4 (37.9, 39.9)	38.6 (37.9, 39.4)
Infant—length (cm), near birth
Median (Q1, Q3)	49.3 (47.2, 50.5)	50.0 (47.5, 50.8)	49.4 (47.4, 50.8)	49.7e (48.0, 50.8)	49.0 (46.8, 51.1)	49.6 (47.5, 51.0)
Infant—length z-score near birth
Median (Q1, Q3)	−0.2 (−0.9, 0.4)	0.1 (−0.8, 0.6)	0.1 (−0.8, 0.6)	−0.1a (−1.1, 0.2)	−0.3 (−1.2, 0.4)	−0.1 (−1.1, 0.2)
Infant—birth weight, z-score
Median (Q1, Q3)	−0.6 (−1.3, −0.2)	−0.6 (−1.0, −0.0)	−0.6 (−1.2, −0.1)	−0.7a (−1.3, −0.4)	−0.5 (−1.0, −0.0)	−0.6 (−1.1, −0.2)
Maternal antiretroviral regimens in third trimester
Boosted PI regimens	30 (88%)	29 (64%)	59 (75%)	34 (79%)	12 (63%)	46 (74%)
Atazanavir/ritonavir	17 (50%)	3 (7%)	20 (25%)	25 (58%)	1 (5%)	26 (42%)
Lopinavir/ritonavir	3 (9%)	19 (42%)	22 (28%)	0 (0%)	9 (47%)	9 (15%)
Darunavir/ritonavir	8 (24%)	7 (16%)	15 (19%)	7 (16%)	2 (11%)	9 (15%)
Fosamprenavir/ritonavir	2 (6%)	0 (0%)	2 (3%)	1 (2%)	0 (0%)	1 (2%)
Saquinavir/ritonavir	0 (0%)	0 (0%)	0 (0%)	1 (2%)	0 (0%)	1 (2%)
Other regimens	4f (12%)	16g (36%)	20 (25%)	9h (21%)	7i (37%)	16 (26%)
Maternal CD4 count (cells/mm3) in the third trimester
Median (Q1, Q3)	435 (344, 554)	540 (340, 809)	452 (340, 663)	483j (347, 651)	455j (392, 614)	481 (348, 647)
Maternal log10 HIV viral load (copies/mL) in the third trimester
Median (Q1, Q3)	1.9 (1.7, 1.9)	1.7 (1.3, 1.9)	1.9 (1.7, 1.9)	1.7k (1.3, 2.0)	1.7k (1.3, 2.0)	1.7 (1.3, 2.0)

Q1, Q3, interquartile range; LMP, last menstrual period; PI, protease inhibitor.

^aData not available in one participant.

^bNot available in one participant in each of the TDF-exposed and -unexposed groups at 0–3 days since birth.

^cNot available in 2 and 7 participants in the TDF-exposed and -unexposed groups respectively, at 0–3 days since birth, and one in each of the TDF-exposed and TDF-unexposed groups at 4–30 days since birth.

^dNot available in 5 and 7 participants in the TDF-exposed and TDF-unexposed groups respectively, at 0–3 days since birth.

^eNot available in 2 participants.

^fTDF/emtricitabine/nelfinavir 1, TDF/emtricitabine/zidovudine/rilpivirine 1, TDF/emtricitabine/rilpivirine 1, TDF/emtricitabine/raltegravir 1.

^gZidovudine/lamivudine/abacavir 9, zidovudine/emtricitabine/nelfinavir 4, zidovudine/lamivudine/raltegravir 2, zidovudine/lamivudine/abacavir/raltegravir 1.

^hTDF/emtricitabine/raltegravir 5, TDF/emtricitabine/abacavir/raltegravir 1, TDF/emtricitabine/nevirapine 1, TDF/emtricitabine/rilpivirine 1, TDF/emtricitabine/zidovudine/rilpivirine 1.

ⁱZidovudine/lamivudine/abacavir 3, zidovudine/lamivudine/nevirapine 1, zidovudine/lamivudine/abacavir/nevirapine 1, zidovudine/lamivudine/maraviroc 1, zidovudine/lamivudine/nelfinavir 1.

^jCD4 not available in 8 TDF-exposed and 10 TDF-unexposed mothers.

^kLog₁₀ HIV viral load unavailable in 8 TDF-exposed and 5 TDF-unexposed mothers. Due to rounding out regimen percentages to 2 significant figures, totals may not be exactly 100%.

The age at which biomarker samples in the first month of life were collected varied slightly for each biomarker, so we describe the number of participants who have a sample in different age categories in the 4- to 30-day age group for TDF-exposed and TDF-unexposed. For example, for serum creatinine and phosphate, among the TDF-exposed (N = 43) these assays were measured in 15 (35%) at 4–7 days, 16 (37%) at 8–14 days, and 12 (28%) at 15–30 days. For the TDF-unexposed (N = 19), 2 (11%) at 4–7 days, 8 (42%) at 8–14 days, and 9 (47%) at 15–30 days. For 25(OH)D this distribution in TDF-exposed (N = 32) was 14 (44%) at 4–7 days, 9 (28%) at 8–14 days, and 9 (28%) at 15–30 days. For TDF-unexposed (N = 16) 25(OH)D distribution was 1 (6%) at 4–7 days, 5 (31%) at 8–14 days, and 10 (63%) at 15–30 days.

Laboratory Outcomes Across Age

Across all values in both the TDF-exposed and -unexposed arms, serum creatinine values ranged from 0.1–1.3 mg/dL. Serum creatinine decreased rapidly over the first few days of life and the rate of decrease slowed by 30 days of life (Figure 1). Values for eGFR ranged from 15–233 mL/min/1.73 m² overall and increased over the first few days of life. The eGFR increase gradually tapered off by age 30 days, consistent with changes in serum creatinine, from which the eGFR is calculated.

Serum phosphate levels ranged from 1.8 to 9.1 mg/dL overall. They increased over the first week of life, and then leveled off, consistent with normal physiological changes in neonates [28]. PTRP values ranged between 82.3% and 99.4% overall and declined more rapidly after the first week of life among the TDF-exposed participants, compared to the unexposed participants. Serum 25(OH)D levels remained relatively flat and ranged from 9 to 51 ng/mL but were lower in the TDF-exposed compared to the unexposed participants. Serum PTH levels ranged from 1 to 245 pg/mL and peaked within the first week of life, then remained stable up to 30 days of life, with no difference between the two groups. Urine NTx/cr (range 5.95–584) showed a rapid increase throughout the first 30 days of life and was not different between the two groups.

Scatter plots of values for each of the seven biomarkers for the 4- to 30-day period are shown with fitted linear regression lines by TDF exposure (Figure 2). For participants aged 4–30 days (clockwise, Table 2, Figure 2), there were no detectable differences in slopes between TDF groups for serum creatinine (difference in slope 0.002 mg/dL/day, 95% CI −0.008, 0.012), serum phosphate (−0.01 mg/dL/day, 95% CI −0.07, 0.05), eGFR (0.31, 95% CI −1.83, 2.46), urine NTx/cr (−4.49, 95% CI −12.75, 3.74), and serum 25(OH)D (0.03 ng/mL/day, 95% CI −0.68, 0.73). In contrast, PTRP decreased more rapidly in the TDF-exposed group (slope = −0.58/day, 95% CI −0.74, −0.04) compared to the -unexposed group (slope = −0.08/day, 95% CI −0.44, 0.27) with a slope difference of −0.50/day (95% CI −0.88, −0.11). For serum PTH, there was not sufficient evidence to support a difference in slope between the two groups (−1.40 pg/mL/day, 95% CI −3.68, 0.89).

Figure 2.

Linear regression lines for neonatal renal and bone biomarkers by age at specimen collection (4–30 days since birth) for individual TDF-exposed and TDF-unexposed participants. Clockwise from top left: serum creatinine, serum phosphate, estimated eGFR (estimated glomerular filtration rate), urine NTx/cr (cross-linked N-telopeptide of type 1 collagen normalized for creatinine), serum PTH (parathyroid hormone), serum 25-OH vitamin D, and PTRP (percentage tubular reabsorption of phosphate) from 4–30 days of age. Abbreviations: eGFR, estimated glomerular filtration rate; NTx/cr, urine NTx normalized for urine creatinine; PTH, parathyroid hormone; PTRP, percentage tubular reabsorption of phosphate; TDF, tenofovir disoproxil fumarate.

Table 2.

Estimated Slope and Difference in Slopes of Each Neonatal Biomarker by TDF Exposure Between 4 and 30 Days of Birth

Outcome	Estimated Slope (95% CI)		Estimated Difference in Slopes (95% CI)
	TDF-Exposeda	TDF-Unexposeda
Creatinine—mg/dL/day	−0.006 (−0.010, −0.002)	−0.008 (−0.015, −0.0002)	0.002 (−0.008, 0.012)
Phosphate—mg/dL/day	−0.06 (−0.11, −0.01)	−0.05, (−0.09, −0.02)	−0.01 (−0.08, 0.06)
eGFR—mL/min/1.73 m2/day	0.96 (0.16, 1.76)	0.65 (−1.27, 2.56)	0.31 (−1.84, 2.46)
PTRPb—per day	−0.58 (−0.74, −0.04)	−0.084 (−0.44, 0.27)	−0.50 (−0.88, −0.11)
25(OH)Db—ng/mL/day	0.048 (−0.35, 0.44)	0.02 (−0.52, 0.56)	0.03 (−0.68, 0.73)
PTHb—pg/mL/day	−1.00, (−2.27, 0.27)	0.40 (−1.55, 2.35)	−1.40 (−3.68, 0.89)
Urine NTxb—per day	12.82 (5.17, 20.47)	17.32 (13.44, 21.19)	−4.49 (−12.73, 3.74)

^aNumber of TDF-exposed and TDF-unexposed in each outcome model: creatinine N = 42, N = 18; phosphate N = 42, N = 18; eGFR N = 39, N = 18; PTRP N = 29, N = 9; 25(OH)D N = 30, N = 13; PTH N = 24, N = 11; NTx N = 30, N = 18.

^b25(OH)D, 25-hydroxy vitamin D; PTH, parathyroid hormone; PTRP, percent tubular reabsorption of phosphate; urine NTx, urine cross-linked N-telopeptide of type 1 collagen normalized for creatinine.

Fitted linear regression lines for all participants (4–30 days of age) for serum phosphate concentrations normalized for creatinine were not different by TDF exposure, similar to serum concentrations. However, for urine phosphate normalized for creatinine, there was an increase in the slope over the 4- to 30-day period for TDF-exposed compared to TDF-unexposed neonates indicating increased phosphate excretion in urine and accounting for the decreased PTRP slope in these participants (Supplementary Figure 2).

The median (interquartile range) of 25(OH)D of subjects sampled across ages 4–30 days was 22 (19, 29) ng/mL in those TDF-exposed and 26 (22, 37) ng/mL in those TDF-unexposed; The percent with 25(OH)D <20 ng/mL, the cut-off for vitamin D deficiency, was 34% (11/32) in those TDF-exposed and 7% (1/15) in those TDF-unexposed. While there was no difference in slope of 25(OH)D between the TDF groups, 25(OH)D concentrations were on average −5.2 ng/mL lower in TDF-exposed compared to TDF-unexposed (−5.22 95% CI −10.83, 0.39) in the 4- to 30-day period, adjusted for age, sex, race, and income. For the other biomarkers with no evident difference in slope between TDF-exposed and -unexposed, difference in adjusted mean concentrations appeared similar across groups for serum creatinine (−0.03, 95% CI −0.29, 0.03 mg/dL), serum phosphate (−0.03 95% CI −0.42, 0.37 mg/dL), eGFR (−1.15 95% CI −18.52, 16.22), PTH (−7.05 95% CI −22.97, 8.86 pg/mL), and urine NTx/cr (19.68 95% CI −28.99, 68.34).

DISCUSSION

This study examined markers of bone turnover and renal function in 141 infants in their first month of life. Among those with specimens collected between 4 and 30 days, neonates born to WLHIV exposed to TDF late in pregnancy had increased urinary phosphate loss, manifested by reduced PTRP, compared to those not exposed. During this period serum 25(OH)D levels were lower in the TDF-exposed compared to the TDF-unexposed groups, and the percent of neonates with vitamin D deficiency (25(OH)D < 20 ng/mL) was higher in those exposed.

The finding of urinary phosphate loss is consistent with the mechanism by which TDF causes renal injury. TDF, a prodrug with a very short half-life in plasma of 0.4 minutes, is rapidly hydrolyzed by intestinal and plasma esterases to TFV which circulates in plasma and subsequently enters peripheral blood mononuclear cells where it is phosphorylated to the active form, TFV-diphosphate (TFV-DP) [29, 30]. Circulating TFV is filtered by the glomerulus and further taken up from blood at the basolateral surface of the proximal convoluted tubular cells by the organic anion transporter-1 and to a lesser extent by organic anion transporter-3. TFV is then secreted into interstitial tubular fluid by multidrug-resistant proteins 2, 4, and possibly 3 [10]. With prolonged use, TFV eventually accumulates in renal proximal tubular cells, where it can be toxic, as previously described. The proximal convoluted tubules are the site at which >90% of filtered phosphate, glucose, uric acid, and amino acids are reabsorbed, and injury at this site leads to renal loss of these solutes, resulting in a partial or full Fanconi syndrome [31]. TFV levels have been measured in maternal, umbilical cord, and neonatal serum, and amniotic fluid, demonstrating in utero transfer to the fetus and consequent fetal exposure [32, 33]. Himes et al. detected tenofovir in meconium collected within 72 hours of birth, and Pintye et al. demonstrated the presence of tenofovir in the hair of neonates from 1 to 14 days of life [34–36]. Thus, the increased urinary phosphate loss described here may be an early manifestation of fetal and neonatal proximal convoluted tubular cell toxicity due to TFV exposure.

PTH plays an important role in the homeostatic control of extracellular calcium and inorganic phosphate concentrations. Serum PTH levels rise in the first few days of a newborn’s life, a physiologic response to the decline in serum calcium at birth, when active placental transfer of calcium abruptly ceases. This early rise was observed in both the TDF-exposed and -unexposed groups. Receiving a TDF-containing regimen is associated with increased serum levels of PTH in people living with and without HIV [37]. However, in the present study, PTH and serum phosphate concentrations were not different between the two groups in their first 30 days of life. Since PTH can also decrease phosphate reabsorption by proximal renal tubular cells and lead to increased urinary phosphate loss, this lack of difference supports TDF exposure with tubular cell toxicity as the explanation for increased urinary phosphate loss.

In contrast, circulating 25(OH)D, also an important predictor of bone health, was lower in the TDF-exposed group compared to the TDF-unexposed group, a finding consistent with a study showing the current use of TDF as a risk factor for vitamin D deficiency in adults living with HIV on combination antiretroviral therapy [38]. The state of lower 25(OH) vitamin D can be associated with decreased intestinal absorption of calcium and phosphorous in the TDF-exposed group, the main source of these minerals in the neonate [28]. It is possible that the combination of urinary phosphate loss and lower 25(OH)D levels observed in the TDF-exposed group contributed to the lower BMC reported in the earlier study, although other factors may contribute [17]. With this state of relative 25(OH)D deficiency, a compensatory increase in serum PTH concentrations could be expected in the TDF-exposed group, as has been reported in a recent study of premature infants with osteopenia [39]. The present study did not demonstrate this physiological response. Havens et al. have suggested that TDF can disrupt the relationship between 25(OH)D and PTH in youth, leading us to speculate that TDF exposure similarly disrupts this relationship in the neonate [40].

This study has limitations. Blood and urine assays were not collected longitudinally on the same participants, but on individual neonates seen at different time points for their study visit in the first month of their life. Whether urinary phosphate loss persisted outside of the neonatal period, and for how long, is unknown. Additionally, creatinine is passively transported across the placenta in a bidirectional fashion and consequently, serum creatinine levels in the newborn within the first 72 hours after birth are highly correlated with maternal levels [26, 41, 42]. This made it necessary for us to stratify the data by age into 0–3 days and 4–30 days of life to evaluate differences in neonatal renal and bone metabolism that may be attributable to TDF exposure. At 4–30 days of age maternal boosted PI regimens containing ATV/r were more common in the TDF-exposed group, with regimens containing LPV/r more common in the TDF-unexposed group. Among contemporary boosted PIs, ATV/r is associated with increased risk for chronic kidney disease, reported with long-term use, and using reduced eGFR < 60 mL/min/1.73m² as the main outcome [43]. It is worth noting that the renal complications with boosted PI use are due to crystalluria and urolithiasis and not injury to proximal tubular cells, the overall percent of boosted PI use was similar in both the TDF-exposed and -unexposed groups (79% vs 63% respectively), and the three main boosted PIs used all increase TFV area under the curve by approximately the same amount (22% for DRV/r, 32% for LPV/r and 24–37% for ATV/r), making it unlikely that this could affect the outcomes we observed for the TDF-exposed group. Further, Siberry et al. adjusted for boosted PI use and specifically ATV/r in their model on the association between TDF use and reduced BMC, with minimal impact on TDF effect [17]. Also, we did not have a large enough sample size to compare lab outcomes within the TDF exposed group comparing those with vs without boosted PI regimens. Finally, the total number of children with samples 4–30 days from birth is small and a larger study is needed to confirm these results.

Tenofovir alafenamide (TAF), a newer generation prodrug, has a longer half-life of 30 minutes in plasma where it is more stable, is mostly metabolized intracellularly by cathepsin A to TFV, and then converted to TFV-DP. Hence plasma levels of TFV are 91% lower, and intracellular concentrations of TFV-DP are 4–7-fold higher with TAF compared to TDF [30, 44, 45]. This translates to reduced exposure to organs such as bone and kidney and a safer toxicity profile, allowing for a smaller dose of TAF to be bioequivalent to TDF. For this reason, the US Department of Health and Human Services currently includes either TDF or TAF as part of the preferred reverse transcriptase backbone for the treatment of people living with HIV, including in pregnancy [5, 6], and TAF is increasingly used preferentially now in the United States. This study was conducted in the United States from 2011 to 2013 when only TDF was available. It is noteworthy that TDF in combination with lamivudine and dolutegravir continues to be recommended as first-line treatment by the World Health Organization (WHO) in their March 2022 update because of concerns about clinical obesity and an adverse effect on lipid profile that can occur with TAF in addition to other programmatic public health issues that affect roll-out of the drug, including cost and access [46]. These findings, although preliminary, can provide additional information to the WHO guidelines committee to consider when deciding on the benefits of TAF in resource-limited countries.

While the results of this study show neonatal renal toxicity and lower 25(OH)D concentration, further studies to determine whether enhanced supplementation of vitamin D is needed for TDF-exposed infants, and if potential renal and bone metabolic effects of TDF exposure produce clinically relevant or lasting consequences. Long-term studies have established that HIV infection in of itself (perinatal and horizontally acquired) is associated with incident clinical fractures [47, 48]. A short-term study in infants showed a significant decrease in lumbosacral BMC associated with TDF exposure through breastmilk but did not report on clinical outcomes such as fracture risk; they reported no differences in creatinine clearance [49]. Of note renal damage can occur without significant reduction in eGFR in the early stages of chronic kidney disease. In contrast, switch studies (from TDF to TAF-containing regimens) have shown that reduction in BMC and renal tubular toxicity are largely reversible [50, 51]. Longitudinal studies need to include sensitive clinical biomarkers to look for evidence of chronic kidney disease and an excess prevalence of osteopenia and incident fragility fractures in TDF-exposed HEU children followed out to early childhood.

In the United States, HIV guidelines at the time of this study did not recommend breastfeeding for babies born with WLHIV. Infant formula preparations are vitamin D-fortified, and since 2008 the American Academy of Pediatrics recommended daily oral supplementation with 400 units of vitamin D for infants under the age of one year until they are able to consume at least one liter of vitamin D-fortified formula a day, reiterated in their 2014 clinical report [52]. Since all infants in this study were formula fed, we did not expect vitamin D fortification through formula to be different between the TDF-exposed and -unexposed groups but did not have the data to examine differences between groups by oral intake of supplemental vitamin D. We contrast this situation to resource-limited countries where babies born to WLHIV are exclusively breastfed. The WHO has not made specific recommendations for vitamin D supplementation of breastfed babies in general, with or without HIV exposure. Further research is critical to establishing the need for such newborn supplementation, particularly with the ongoing use of TDF for pregnant WLHIV.

CONCLUSION

TDF use in pregnancy is generally thought to be safe, but data on renal and bone outcomes in neonates and early childhood are limited and inconsistent. In this study of 141 neonates with 77 TDF-exposed and 64 TDF-unexposed participants, there was increased urinary phosphate loss and lower serum 25(OH)D levels in those who were exposed to TDF. These findings are likely due to in utero exposure to tenofovir in late pregnancy.

Supplementary Data

Supplementary materials are available at the Journal of The Pediatric Infectious Diseases Society online (http://jpids.oxfordjournals.org).

Author Contributions

MP contributed to the acquisition, analysis, and interpretation of data for the work, drafting and critical revision of the manuscript, and approval of the final version, and agrees to be accountable for all aspects of the work. DJ and GS contributed to the design, analysis, and interpretation of data for the work, critical revision of the manuscript, approval of the final version, and agree to be accountable for all aspects of the work. LD, TJY, JK, RVD, and WY contributed to the interpretation of data, critical revision of the manuscript, approval of the final version, and agree to be accountable for all aspects of the work.

Data Availability

Data available upon request at the Pediatric HIV/AIDS Cohort Study (PHACS) website at https://phacsstudy.org/Our-Research/Data-and-Specimen-Sharing.