Effect of Antiretroviral Therapy on Bone and Renal Health in Young Adults Infected With HIV in Early Life

Abstract

Context:

HIV antiretroviral (ARV) therapy is associated with renal and bone toxicity, but little is known about the potential cumulative effects in adults exposed to ARVs from birth.

Objective:

To prospectively evaluate renal and bone health in young adults with lifelong HIV and extensive ARV exposure.

Design:

Cross-sectional comparison of bone mineral density (BMD) by dual-energy X-ray absorptiometry, bone turnover, and renal function in young adults infected with HIV in early life (n = 65) to matched healthy controls (n = 23) and longitudinal evaluation (mean follow-up = 4.4 years) within a subset of the HIV cohort (n = 33).

Setting:

Government outpatient research clinic.

Results:

Albumin/creatinine ratio, protein/creatinine ratio, anion gap, N-terminal telopeptides, and osteocalcin were significantly increased in persons with HIV compared with controls, whereas whole-body BMD and BMD z scores were lower. Within the HIV group, duration of tenofovir disoproxil fumarate (TDF) correlated with higher anion gap but did not correlate with bone parameters. Longer duration of didanosine and stavudine use correlated with lower BMD and BMD z scores. Longitudinal analyses revealed that BMD and bone metabolism significantly improved over time. No subject had an estimated glomerular filtration rate (eGFR) <60, but decline in eGFR correlated with increasing years of TDF exposure.

Conclusions:

Subclinical markers of renal dysfunction were increased in HIV-infected young adults and associated with TDF exposure, whereas lower bone density was associated with didanosine and stavudine exposure. The tendency for improvement in markers of bone health over time and the availability of less toxic ARV alternatives may herald improvements in renal and bone health for perinatally infected patients in adulthood.

Renal dysfunction was increased in young adults with lifelong HIV, whereas bone density was lower compared with controls. Less toxic HIV therapy may herald future improvements in renal and bone health.

With the advent of antiretroviral therapy (ART), individuals with HIV are now experiencing the effects of long-term—and, in some cases, lifelong—exposure to HIV treatment. Individuals with HIV are at increased risk of end stage organ disease relative to their peers and may experience accelerated age-associated comorbidities (1). In particular, long-term HIV infection and its treatment have been implicated in renal disease, abnormalities in bone mineral density (BMD), and osteoporosis (2).

Early in the HIV epidemic, chronic kidney disease (CKD) was seen in children and adults with HIV, most commonly due to HIV-associated nephropathy. Although rates of CKD, proteinuria, and impairment in glomerular filtration rate (GFR) have decreased since the pre–highly active ART era (3), ART use, especially tenofovir disoproxil fumarate (TDF), is linked to renal toxicities (2, 4–6). Further, TDF, along with nucleoside analog reverse-transcriptase inhibitors (NRTIs), has been associated with lower BMD and increased bone turnover compared with age-matched controls (4).

The effect of TDF on kidney and bone health was linked to proximal renal tubular dysfunction (PRTD), which results in decreased reabsorption of renal phosphate, uric acid, and bicarbonate losses. Through this loss of renal phosphate, PRTD was believed to be one factor contributing to TDF-induced BMD loss (4). However, recent observations in preexposure prophylaxis studies suggest BMD loss following TDF/emtricitibine exposure is related to hormonal changes in bone turnover independent of renal phosphate handling (7). Further, other NRTI medications are thought to contribute to decreased BMD because of induced mitochondrial dysfunction and osteoclast stimulation. HIV infection may additionally stimulate osteoclasts and impede osteoblastic differentiation to shift the balance of bone remodeling in favor of bone loss (8, 9).

Despite strong findings of renal toxicity and BMD loss in cross section studies of HIV-infected patients, the full implications for perinatally infected individuals who are now young adults are still unclear (10). There are few studies examining the potential effects of long-term ART exposure in a cohort infected early in life who are now adults. The purpose of this study was to prospectively evaluate the impact of HIV infection and ART use on renal function and bone health in young adults infected with HIV in early life as they approach the age of peak bone mass.

Methods

Subjects

Between March 2007 and April 2016, adolescents and young adults who acquired HIV perinatally or in early childhood were enrolled and followed on prospective, longitudinal, cohort studies at the National Institutes of Health (ClinicalTrials.gov NCT01656564 or NCT00924365). The National Institute of Allergy and Infectious Diseases Institutional Review Board approved these protocols. Written informed consent was obtained from participants, parents, and/or legal guardians when appropriate, and assent was obtained from minor participants.

HIV-infected participants completed annual evaluations, but healthy controls were prospectively recruited for a one-time assessment to serve as a matched comparator group. Targeted recruitment was performed to obtain approximately one control per three HIV-infected subjects seeking similar distributions of age, sex, and race/ethnicity. We conducted a cross-sectional analysis comparing the healthy controls with the HIV-infected subjects at their most recent visit. We excluded HIV-infected subjects who were under 18 at their last follow-up as controls were required to be 18 years old or older. Additionally, we excluded one subject receiving hemodialysis for end-stage renal disease. We also performed a longitudinal analysis on a subset of HIV-infected subjects who had baseline evaluations and at least one follow-up visit that included measures of renal function and bone density.

Assessments

Each participant completed the following at each visit: medical history, physical examination, body composition measurements (height, weight, body mass index, waist/hip circumference), blood pressure, fasting blood work, and urine laboratory tests. Lifetime antiretroviral (ARV) exposure and nadir CD4 cell counts were collected on HIV-infected subjects at baseline from detailed chart review and medical record extraction. HIV-related parameters (HIV viral load and CD4 and CD8 lymphocyte counts) and any changes in ARV treatment were collected at each visit. Total duration (in years) of ARV exposure was measured for each participant. Further delineation of duration by class and agent were also calculated.

Bone specific assessments

BMD was assessed by dual-energy X-ray absorptiometry (Hologic, Bedford, MA) at the lumbar spine from L1 to L4 and for whole-body composition including head. Data were collected, analyzed, and interpreted by a single radiologist (J.R.) at the National Institutes of Health Clinical Center. For lumbar spine, the Hologic database reference standard was used. For whole-body BMD and body composition, a National Health and Nutrition Examination Survey reference database provided by Hologic was used (11) with race categorized as white, black, or Hispanic. For mixed and other races, race assignments were based on patient preference. Dual-energy X-ray absorptiometry was performed at baseline and repeated for HIV-infected subjects at follow-up visits every 1 to 5 years. BMD z scores were calculated using a reference population database from Hologic for sex-, race-, and age-matched youth and adults with the standard formula:

$$\mathrm{BMD}\mathit{z}\text{score}=\frac{\mathrm{BMD}\text{patient}-\mathrm{BMD}\text{mean}\text{of}\text{reference}\text{population}\left(\text{same}\text{age,}\text{race,}\text{and}\text{sex}\text{as}\text{patient}\right)}{\text{Standard}\text{deviation}\text{of}\text{reference}\text{population}\left(\text{same}\text{age,}\text{race,}\text{and}\text{sex}\text{as}\text{patient}\right)}$$

The following markers of bone turnover and metabolism were measured: serum calcium, phosphate, alkaline phosphatase, osteocalcin, parathyroid hormone (PTH), thyroid-stimulating hormone, 25-hydroxyvitamin D, and urine N-terminal telopeptide (NTx) adjusted for creatinine. Osteocalcin and PTH were quantified using electrochemiluminescence immunoassay or Roche Cobas e601 analyzer (Roche Diagnostics, Indianapolis, IN). Thyroid-stimulating hormone, serum phosphate, calcium, and alkaline phosphatase were measured by Roche Cobas 6000 Analyzer. NTx was quantified by Vitrox NTx assay and 25-hydroxyvitamin D by chemiluminescent immunoassay performed on the DiaSorin Liason XL (DiaSorin Inc., Stillwater, MN).

Renal specific assessments

Urine protein, creatinine, albumin, calcium, and phosphorous were measured. Urinary protein/creatinine ratio (PCR) and albumin/creatinine ratio (ACR) were calculated. Microalbuminuria was defined as an ACR of 25 to 355 mg/g in women and 17 to 250 mg/g in men. Proteinuria was defined as a PCR ≥0.5 g/g. We used the CKD Epidemiology Collaboration creatinine equation to estimate GFR in both adolescents and adults. Serum anion gap was calculated using the following equation: sodium (mmol/L) – [chloride (mmol/L) + bicarbonate (mmol/L)]. Proximal tubular dysfunction (4) and tubular maximum for phosphate corrected for GFR (TmP/GFR) (12) were calculated using methods described previously.

Statistical analysis

Descriptive statistics were calculated for all variables, and continuous variables are expressed as median [interquartile range (IQR)] or mean (standard error). For the cross-sectional analysis, nonparametric Wilcoxon rank-sum test and χ² (Pearson) were performed for comparisons between groups. Paired t tests were used for comparing values at baseline with subjects’ last follow-up visit for the longitudinal analysis. Univariate linear regression analyses were performed to evaluate associations between clinical variables, ARV exposure, and markers of bone and renal health within the HIV-infected group. Statistical significance was determined using a P value < 0.05, using SAS JMP statistical software (Version 12.0.1, SAS Institute, Cary, NC).

Results

Cross-sectional analysis of HIV-infected vs controls

We studied 65 young adults (median age = 24 years, IQR = 21 to 28) who acquired HIV within the first 10 years of life and 23 HIV-uninfected controls (median age = 25 years, IQR = 22 to 28; Table 1). The two groups were similar in age, sex, and race distribution. Fifty-four HIV-infected subjects (83%) acquired HIV perinatally, and 11 acquired HIV postnatally through transfusion or feeding in early life. Of the HIV-infected subjects, 85% were on ART at last follow-up with a median duration of 15.8 years, 55% were currently on a TDF-containing regimen, and 57% had an HIV viral load <40 copies/mL.

Table 1.

Demographic and Clinical Characteristics of Participants

	HIV+ (n = 65)	Control (n = 23)	P Value
Age, y	24 (21, 28)	25 (22, 28)	0.25
Sex, male, n (%)	25 (39)	11 (48)	0.43
Race
Caucasian	24 (37)	11 (48)	0.63
African American	33 (51)	11 (48)
Mixed race	6 (9)	1 (4)
Native American	2 (3)	0
Ethnicity, Hispanic, n (%)	9 (14)	4 (17)	0.68
Body mass index, kg/m2	24 (22, 28)	25 (23, 29)	0.40
Height, cm	165 (159, 169)	165 (159, 173)	0.19
Waist/hip ratio	0.91 (0.86, 0.97)	0.84 (0.82, 0.88)	0.004
Smoker ever, n (%)	18 (28)	4 (19)	0.41
Smoking pack-years, y	1.1 (0, 4.5)	0.32 (0.04, 0.6)	0.49
Total cholesterol, mg/dL	164 (136, 178)	171 (154, 191)	0.08
Triglycerides, mg/dL	87 (63, 130)	64 (54, 83)	0.02
High-density lipoprotein, mg/dL	47 (38, 57)	65 (51, 71)	0.0003
Low-density lipoprotein, mg/dL	88 (72, 103)	99 (67, 120)	0.33
Systolic blood pressure, mmHg	119 (109, 128)	118 (108, 121)	0.31
Diastolic blood pressure, mmHg	72 (66, 81)	67 (63, 71)	0.01
Hypertension,a n (%)	4(6)	0	0.24
Diabetes, n (%)	0	0	1
Vitamin D deficiency,b n (%)	21(38)	9 (39)	0.66
CD4 count, cells/μL	542 (281, 805)	787 (631, 969)	0.003
CD8 count, cells/μL	747 (521, 1051)	452 (405, 621)	0.0003
Currently on ARVs, n (%)	55 (85)
Duration of ARVs,c y	15.8 (12.4, 19.5)
Duration of protease inhibitors, y	9.8 (5.9, 13.8)
Duration of NRTIs, y	15.4 (10.4, 18.2)
Duration of non-NRTIs, y	2.7 (0, 7.7)
Duration of TDF, y	3.1 (0, 6.6)
Duration of stavudine, y	3.5 (0, 6.8)
HIV RNA <40 copies/mL, n (%)	37 (57)
HIV viral load, copies/mL	5038 (1332, 60452)
Nadir CD4 <200 cells/μL, n (%)	31 (52)

Values are expressed as median (IQR) unless otherwise indicated. Boldface indicates significant P values (<0.05).

^aHypertension: systolic blood pressure >140, diastolic >90, or on antihypertensive medication.

^bVitamin D deficiency defined as <20 ng/mL (50 nmol/L). Data available on 55 of 65 HIV-infected subjects.

^cThree HIV-infected subjects lacked verifiable documentation of ARV history.

Although body mass index and lean body mass were not significantly different between the two groups, the waist/hip ratio was significantly higher in the HIV group. In addition, triglyceride levels were higher and high-density lipoprotein cholesterol levels lower in HIV compared with controls. There was no significant difference in smoking exposure between the two groups. There was no evidence of diabetes in any study subject. One HIV-infected subject had chronic hepatitis B and one HIV-infected subject was treated for hypertension. Vitamin D deficiency, defined as 25-hydroxyvitamin D level <20 ng/mL, was common and did not differ between groups (38% HIV and 39% controls).

Bone parameters

There were significant differences between groups in bone density and bone remodeling markers (Table 2). Whole-body BMD in the HIV group was 1.15 g/cm² (IQR = 1.08 to 1.22) compared with 1.21 g/cm² (IQR = 1.16 to 1.28) in controls (P = 0.005). Also, this difference remained significant when calculated using subtotal BMD minus head (pediatric convention, data not shown). Further, whole-body BMD z scores were significantly lower in individuals infected with HIV (P = 0.02). One of the HIV subjects had a z score <–2.5. There were no significant differences in lumbar spine BMD or lumbar spine BMD z scores between groups.

Table 2.

Cross-Sectional Markers of Bone and Renal Health

	HIV+ (n = 65)	Control (n = 23)	P Value
Bone density
Whole-body BMD, g/cm2	1.15 (1.08, 1.22)	1.21 (1.16, 1.28)	0.005
Whole-body BMD z score	−0.2 (-1.1, 0.6)	0.4 (-0.2, 1.1)	0.02
Whole-body lean mass, kg	45.9 (39.3, 52.8)	48.6 (40.8, 61.7)	0.2
Lumbar spine BMD, g/cm2	1.03 (0.94, 1.14)	1.08 (0.99, 1.12)	0.31
Lumbar spine z score	−0.5 (-1.3, 0.4)	−0.3 (-0.8, 0.3)	0.42
Bone parameters
25-Hydroxyvitamin D, ng/mL	23 (16, 32)	21 (17, 33)	0.96
NTx, nmol/mmol	52 (34, 70)	32 (18, 54)	0.01
Alkaline phosphatase, U/L	76 (62, 93)	70 (53, 83)	0.05
Osteocalcin, ng/mL	30.1 (23.4, 39.6)	22.1 (15.2, 25.5)	0.007
PTH, pg/mL	40 (34.3, 53.4)	35.4 (29.7, 38.9)	0.03
Phosphorus, inorganic, mg/dL	3.3 (3, 3.7)	3.6 (3.2, 3.9)	0.1
Calcium, mmol/L	2.27 (2.18, 2.35)	2.31 (2.28, 2.38)	0.03
Thyroid-stimulating hormone, mcIU/mL	1.37 (0.88, 2.08)	1.81 (1.17, 2.67)	0.19
Renal parameters
eGFR, mL/min/1.73m2	129 (116, 142)	119 (113, 124)	0.02
Proteinuria, n (%)	2 (3)	0 (0)	0.39
Microalbuminuria, n (%)	5 (9)	2 (9)	0.95
Urinary ACR, mg/g	6.9 (4.6, 15.3)	4.4 (3, 6.1)	0.002
Urinary PCR, mg/mg	0.13 (0.11, 0.17)	0.09 (0.07, 0.13)	0.004
Creatinine, mg/dL	0.71 (0.64, 0.84)	0.81 (0.72, 0.98)	0.02
Anion gap, mEq/L	10 (9, 13)	8 (7, 11)	0.02
Bicarbonate, mmol/L	25 (23, 26)	26 (24, 29)	0.02
Glucosuria, n (%)	2 (3)	1 (4)	0.84

Values are expressed as median (IQR). Boldface indicates significant P values (<0.05). Lumbar spine available on n = 56 HIV, n = 22 control.

NTx, a urinary marker of bone resorption, was significantly elevated in HIV-infected subjects compared with controls (P = 0.01) and remained significantly elevated when adjusted for lean body mass (P = 0.02). Osteocalcin and PTH levels were also significantly elevated in the HIV-infected group (P = 0.007 and P = 0.03, respectively). Additionally, serum calcium was lower in HIV subjects compared with controls (P = 0.03). There were no significant differences in thyroid stimulating hormone or 25-hydroxyvitamin D levels between groups.

Within the HIV-infected group, univariate correlations showed that longer duration of exposure to didanosine and stavudine correlated with lower whole-body BMD (r = –0.28, P = 0.03 and r = –0.28, P = 0.03, respectively) and with lower BMD z scores (didanosine: r = –0.31, P = 0.02). Duration of didanosine exposure also correlated with lower lumbar spine BMD and z scores (r = –0.29, P = 0.04 and r = –0.35, P = 0.01, respectively). These were the only ARV agents associated with lower BMD. Of note, there was no association between TDF exposure and BMD or BMD z score. HIV viral load, presence of viremia, nadir CD4 <200, and CD8+ counts were not associated with any BMD measurement.

Kidney parameters

The prevalence of proteinuria and albuminuria were similar in both groups, and age-adjusted CKD Epidemiology Collaboration estimated glomerular filtration rate (eGFR) did not differ between groups. No subject had an eGFR <60 mL/min/1.73m². As a group, HIV-infected subjects had higher urine PCR (P = 0.004), ACR (P = 0.002), and anion gap (P = 0.02) compared with controls. Further, urinary protein and urinary albumin remained significantly increased when corrected for lean body mass instead of creatinine (P = 0.002 and P = 0.0005, respectively). Only two subjects in the HIV cohort had an anion gap >16 mEq/L; these subjects also had microalbuminuria but normal eGFR. Though no subject had a serum bicarbonate level <20mmol/L, mean bicarbonate was significantly lower in the HIV-infected group (P = 0.02). There were no significant differences in TmP/GFR or glucose between HIV-infected subjects and controls (data not shown).

Univariate correlations within the HIV-infected cohort showed a significant relationship between eGFR and CD8+ T-cell counts (r = –0.26, P = 0.04); higher CD8+ T-cell count was associated with lower eGFR. Higher anion gap correlated with a longer duration of TDF exposure (r = 0.34, P = 0.008). TmP/GFR did not differ based on current TDF use. Higher urinary PCR correlated with increased duration of exposure to lopinavir/ritonavir (LPV/r; r = 0.30, P = 0.03). No other HIV-related parameters were significantly correlated with markers of kidney health.

Longitudinal analysis in HIV

Longitudinal data were available from 33 subjects in the HIV-infected cohort with a mean follow-up period of 4.4 years (mean age = 19 ± 4 years at baseline and 23 ± 4 years at last follow-up). Each subject’s most recent evaluation was compared with his or her baseline values to calculate mean change from initial evaluation. At baseline, two subjects were Tanner stages 3 to 4, nine subjects were Tanner stage 4, and the rest were Tanner stage 5; all participants were Tanner stage 5 at last follow-up. CD4 T-cell count did not change significantly over time for this group. However, at baseline, 56% of this subset was viremic, whereas 39% had a detectable viral load at their most recent follow-up visit.

Bone parameters

Despite the differences in bone density noted relative to controls, BMD increased for patients over the course of follow-up. Increases in both lumber spine BMD (P = 0.0001) and whole-body BMD (P < 0.0001) were observed. Whole-body BMD z scores were less negative over time, indicating a normalization of BMD as subjects enter adulthood (P < 0.0001). Despite significant increases in lumbar spine BMD, there was no significant change in lumbar spine z score during the study period.

Markers of bone turnover tended to normalize over time as well. Alkaline phosphatase, osteocalcin, and NTx levels all significantly decreased in the HIV-infected group during follow-up (P = 0.02, P = 0.03, P = 0.02; Table 3). Fourteen (42%) of participants reported use of a vitamin D supplement at some point during follow-up, but the median duration of vitamin D use was 0.55 years (IQR = 0.2 to 4.1).

Table 3.

Longitudinal Markers of Bone and Renal Health in HIV-Infected Subjects (n = 33)

	Baseline	Follow-Up	Mean Difference	P Value
Bone density
Whole-body BMD, g/cm2	1.07 ± 0.02	1.16 ± 0.02	0.08 ± 0.01	<0.0001
Whole-body BMD z score	−0.69 ± 0.2	−0.25 ± 0.2	0.4 ± 0.1	<0.0001
Lumbar spine BMD, g/cm2	0.97 ± 0.02	1.04 ± 0.02	0.06 ± 0.01	0.0001
Lumbar spine z score	−0.41 ± 0.18	−0.38 ± 0.19	0 ± 0.1	1
Bone parameters
25-Hydroxyvitamin D, ng/mL	28.8 ± 2.8	28.8 ± 2.5	0.07 ± 2.7	0.98
NTx, nmol/mmol	135.9 ± 34.4	57.7 ± 5.6	−78.2 ± 32.1	0.02
Alkaline phosphatase, U/L	122.6 ± 18.5	81.2 ± 3.6	−41.4 ± 17.4	0.02
Osteocalcin, ng/mL	55.8 ± 10.5	34 ± 2.9	−21.8 ± 9.5	0.03
Phosphorus, inorganic, mg/dL	4.1 ± 0.1	3.5 ± 0.1	−0.7 ± 0.1	<0.0001
Calcium, mmol/L	2.3 ± 0.01	2.2 ± 0.02	−0.1 ± 0.02	<0.0001
Renal parameters
eGFR, mL/min/1.73m2	129.2 ± 4.4	128.2 ± 3.5	−1.03 ± 2.8	0.71
Urinary ACR, mg/g	19.3 ± 6.3	34.6 ± 21.7	15.2 ± 18.3	0.41
Urinary PCR, mg/mg	0.14 ± 0.02	0.16 ± 0.02	0.02 ± 0.03	0.45
Creatinine, mg/dL	0.78 ± 0.03	0.76 ± 0.03	−0.02 ± 0.02	0.35
Anion gap, mEq/L	6.8 ± 0.4	9.7 ± 0.4	2.9 ± 0.6	<0.0001
Bicarbonate, mmol/L	26.7 ± 0.35	25.1 ± 0.45	−1.6 ± 0.4	0.0005

Values are expressed as mean (standard error) unless indicated otherwise. Boldface indicates significant P values (<0.05).

Kidney parameters

Measures of proteinuria (PCR) and albuminuria (ACR) did not change significantly from baseline to last follow-up. One subject had PCR ≥0.5 g/g at initial visit but not at follow-up and another had proteinuria only at last follow-up. Six subjects (18%) had microalbuminuria at initial evaluation, five had microalbuminuria at last follow-up, and three had microalbuminuria at both baseline and last follow-up. In the subset of subjects with longitudinal data, no one had an anion gap >16 mEq/L or a bicarbonate level <20mmol/L; however, anion gap increased significantly from baseline to last follow-up (P < 0.0001) and serum bicarbonate decreased significantly during follow-up (P = 0.0005). Further, change in anion gap was positively correlated with change in both PCR (r = 0.48, P = 0.01) and change in years of TDF exposure (r = 0.37, P = 0.04) during the study follow-up. There were significant inverse correlations between change in eGFR and change in duration of exposure to TDF (r = –0.43, P = 0.02) and change in CD8+ T-cells (r = –0.41, P = 0.02); a decrease in eGFR was associated with increasing duration of TDF exposure and increasing CD8+ T-cells.

Discussion

This study is among the first examining bone and renal function in nearly lifelong HIV-infected individuals as they enter adulthood and reach peak bone mass. We show that HIV-infected subjects have lower whole-body BMD and BMD z scores compared with controls as well as significantly elevated markers of bone turnover. Despite the evidence of lower bone density at their most recent evaluation, we observed that osteocalcin and NTx decreased over follow-up, possibly indicating an improvement in bone metabolism in the patient population. Of note, within the HIV-infected group, duration of exposure to stavudine and didanosine was associated with lower bone density. Finally, microalbuminuria, proteinuria, and low eGFR were uncommon in this cohort and did not differ from controls; however, HIV-infected subjects showed significant differences in subclinical markers of renal dysfunction with increased urine PCR and ACR and serum anion gap values.

It is important to recognize the continuum of renal and bone health. Hypovitaminosis D is common in young adult populations (13) and especially in HIV-infected adolescents (14, 15) and can lead to secondary hyperparathyroidism, resulting in increased bone resorption and eventual loss of BMD (16). In our study, there was no difference in vitamin D levels between HIV-infected patients and controls, or in the prevalence of vitamin D deficiency. PTH was significantly increased in the HIV-infected group, and calcium levels were lower in HIV. Sudjaritruk et al. (16) demonstrated lower vitamin D levels and increased activation of the calcium–PTH–vitamin D axis in perinatally infected adolescents with lower CD4+ T-cell counts. HIV associated immunosuppression in our HIV-infected cohort may partially explain the observed differences relative to healthy controls as well as renal dysfunction contributing to secondary hyperparathyroidism.

HIV-mediated effects, including both direct infection of renal tubular epithelium as well as inflammatory responses to HIV in the kidney, may also cause derangement of the bone-kidney axis (3). Advanced HIV disease and immune suppression have been associated with subclinical renal laboratory abnormalities in adults (17) and adolescents with HIV (18). Elevations in CD8 T-cells, a marker of inflammatory non-AIDS-related clinical events (19), were associated with lower eGFR cross-sectionally and decreasing eGFR over time in our cohort. In previous studies, HIV infection has been independently associated with microalbuminuria; risk factors being low CD4+ T-cell count, higher HIV-RNA levels, and NRTI use (17). However, these results came from a patient population of median age 40 years. Although we did not see high rates of microalbuminuria in our cohort, this may be due to the relative younger age of the subjects infected early in life.

Certain ARV agents have been historically associated with nephrotoxicity. In particular, TDF has been associated with proximal tubular dysfunction, likely via deleterious effects in mitochondria (5). Abundant mitochondrial-generated ATP are needed in the proximal tubule to facilitate transcellular and paracellular transport of ions, glucose, and proteins (5), thus patients with proximal tubular dysfunction due to TDF are likely to present with glycosuria, phosphaturia, and low molecular weight proteinuria (3–5, 20, 21). Although the mechanism of mild anion gap elevations is unknown, it might represent subtle systemic mitochondrial toxicity. Similar to previous observations, in our cohort increased exposure to TDF was associated with higher anion gap (20) and decline in eGFR over time (6). Further, increasing anion gap correlated with increasing PCR over the duration of the study period, identifying anion gap as a potential biomarker of ongoing subclinical kidney injury. Although microalbuminuria, proteinuria, and low eGFR were uncommon in this cohort, persistent subclinical renal dysfunction in HIV-infected youth has been linked with increased risk of future renal abnormalities (18, 21). Renal function in these patients, especially for those with other comorbidities or predispositions to renal dysfunction (21), should be monitored as patients enter adulthood. Participants in this cohort did not receive TDF alafenamide during the study period, but this alternative to TDF with less renal toxicity may favorably alter the trajectory of renal impairment in HIV care in the future.

Interestingly, duration of LPV/r exposure was associated with higher PCR. Although protease inhibitors have been implicated in decreases in renal function over time, nephrotoxicity from LPV/r has been less clear (3, 6, 22, 23). LPV/r has been shown to increase TDF levels by more than 50% when coadministered (24); however, we did not observe an effect of TDF exposure on urinary PCR. Nonetheless, LPV/r remains a preferred agent in regimens for initial treatment in HIV-infected children (25) and is commonly used in areas worldwide where perinatal transmission still occurs at high rates. Assessing nephrotoxicity from long-term LPV/r may be important as children reach and enter adulthood.

Mounting evidence has demonstrated lower BMDs in both adults (4, 26, 27) and children (26, 28, 29) with HIV compared with uninfected controls. This has also been shown in patients with perinatal infection (8, 10, 16, 29), similar to results seen in our cohort. The mechanism of bone loss in patients infected early in life is likely multifactorial and influenced by both traditional factors—including nutrition, drug and toxin exposure, physical inactivity, and lower body mass index—and HIV-specific factors—including altered proinflammatory cytokines and levels of hormones, uncontrolled viremia in the presence or absence of AIDS, failure to thrive, and effects from ARVs (8, 10, 29–31). In our cohort, we did not see any statistically significant relationships between past or current severity of HIV disease and/or immunosuppression on BMD. Whole-body BMD increased over time and whole-body BMD z scores became less negative, indicating a trend toward improved bone health. Similarly, markers of bone metabolism decreased over time suggesting a normalization of excess bone resorption with age. Gains in BMD and reduction in markers of bone turnover may also have coincided with completion of pubertal development in the subset of HIV infected participants who were not Tanner stage 5 at baseline. The decrease in pathophysiologic bone metabolism may also represent improvements in general health and less toxic ARV options.

ARV-induced bone loss has long been implicated (32) in HIV, but there has been conflicting evidence about the specific causal mechanisms. Protease inhibitors and NRTIs specifically have been implicated in loss of whole-body BMD (33). In the current study, duration on stavudine was associated with lower BMD and didanosine was independently associated with both lower BMD and BMD z scores in HIV. Mitochondrial dysfunction as a result of ART, particularly with zidovudine and stavudine, is associated with a wide range of pathology in HIV (34). For example, HIV-infected adults randomized to reduced dose stavudine had significant increases in mitochondrial DNA in fat cells and less decreases in BMD after 48 weeks compared with those randomized to standard dose stavudine (35). Further, mouse models have shown osteoclastogenic effects and osteopenia during treatment with didanosine treatment (36). It is likely that a combination of both mitochondrial toxicity and direct induction of osteoclastogenesis contributes to bone loss. Although didanosine and stavudine are no longer commonly used, there remains a substantial cohort of adolescents and young adults who were exposed to these agents during childhood worldwide.

Surprisingly, exposure to TDF, an ARV agent commonly implicated in renal dysfunction, phosphaturia, and bone loss (4, 7), showed no significant effect on BMD in our cohort despite widespread exposure. There were no significant differences in tubular reabsorption of phosphate, TmP/GFR, glucose handling, phosphate wasting, and PRTD between patients and controls in our cohort. As noted in the preexposure prophylaxis trial in young HIV-uninfected males (7), proximal tubular damage and phosphate wasting were not contributing factors to bone density loss in our young adult HIV-infected population.

There are several potential limitations to this study, including relatively small sample size and duration of prospective follow-up less than 5 years. Further, information on history of bone fracture was not collected; therefore, the clinical relevance of lower BMD in this specific cohort is yet to be fully appreciated. The considerable percentage of patients with detectable HIV is assumed to be due to nonadherence to ART and this may limit our ability to fully appreciate potential ARV-associated toxicity.

However, this is one of the first studies to analyze bone and kidney health in young adults infected with HIV early in life with extensive exposure to ART. Exposure to stavudine and didanosine, but not TDF, was associated with lower bone density, whereas renal dysfunction was related to TDF and LPV/r use in this HIV population with nearly lifelong infection. The observed subclinical but significant differences in bone and renal health in this HIV cohort may emerge clinically as the cohort ages and acquires other comorbidities. Longitudinal studies of perinatally infected HIV cohorts as they progress into adulthood, especially in larger resource-poor settings, are needed to verify these findings.