Measuring ART Adherence Among Young Adults with Perinatally Acquired HIV: Comparison between Self-report, Telephone-based Pill Count, and Objective Pharmacologic Measures

Abstract

Tenofovir diphosphate (TVF-DP) can be quantified in red blood cells (RBCs) and dried blood spots (DBS) and can objectively measure ART adherence and predict viral suppression. Data on the association of TFV-DP with viral load are very limited in adolescents and young adults (AYA) living with perinatally-acquired HIV (PHIV), as are data comparing TFV-DP to other measures of ART adherence, such as self-report and unannounced telephone pill count. Viral load and ART adherence (self-report, TFV-DP and unannounced telephone pill count) were assessed and compared among 61 AYAPHIV recruited from an ongoing longitudinal study (CASAH) in New York City.

Introduction

Optimal adherence to antiretroviral therapy (ART) among people living with HIV is critical for maintaining viral suppression [1] and preventing onward HIV transmission [2]. Maintaining optimal ART adherence is often difficult, especially for adolescents and young adults (AYA) with perinatally acquired HIV (AYAPHIV) who require lifetime adherence beginning in childhood. Globally, there are 5.1 million people with HIV ≤25 years [3], many millions with PHIV, who assessing ART adherence is critical for adherence research and for clinical care to identify those at greatest risk for poor HIV health outcomes, such as virologic failure.

Assessing ART adherence to identify those at risk for virologic failure has significant challenges. Self-reported adherence, while low-cost, practical [1] and inversely correlated with viral load (VL), is subject to bias (via forgetfulness, social desirability, ceiling effects) which can result in low sensitivity (but high specificity) to detect poor ART adherence [4]. Unannounced telephone pill counts, in which individuals count their pills from home while speaking with a pill count assessor over the phone, provide a more objective and quantifiable measure of adherence compared to self-report and have also been inversely correlated with VL in adults and youth living with HIV, including AYAPHIV [5]. While this procedure can be administered by lay staff, does not require travel, technology, or medical equipment, it may not be feasible in all contexts as it requires considerable staff resources, a closed pharmacy system, and patient privacy (where little or none may exist) [5].

Tenofovir diphosphate (TVF-DP), the phosphorylated anabolite of tenofovir (TFV), can be quantified in red blood cells (RBCs) and dried blood spots (DBS) in PWH who are on a treatment regimen of either tenofovir disoproxil fumarate (TDF) or tenofovir alafenamide (TAF) [6]. TFV-DP has an intracellular half-life of 17 days in RBCs and DBS, making it effective at quantifying cumulative adherence to ART over four to eight weeks preceding blood draw [6]. Concentrations of TFV-DP in DBS can objectively measure ART adherence and predict viral suppression [7-8]. Despite these advantages over self-report and unannounced pill counts, data on the association of TFV-DP with VL are very limited in AYAPHIV, as are data comparing it to self-report and unannounced telephone pill count adherence measures. To address this, we used data from a longitudinal cohort study of AYAPHIV in New York City were we assessed the association between self-reported adherence, unannounced telephone pill counts, and TFV-DP in DBS with VL.

Methods

Participants

Data were collected from AYAPHIV enrolled in the Child and Adolescent Self-Awareness and Health Study (CASAH), an ongoing longitudinal study investigating the mental health and risk behaviors of youth with PHIV and youth perinatally exposed to HIV but uninfected (PHEU). The study began in 2003 and currently has ten waves of data collection. Methods have been described in detail elsewhere [9]. The study enrolled 340 youth (206 PHIV and 134 PHEU) at baseline when youth were 9-16 years (mean=12 years). Since the initial study, nine post-baseline follow-up (FU) visits (FU1 – FU9) have been completed. During FUs 6 and 7, two adherence specific sub-studies were completed. In one sub-study, AYAPHIV provided blood samples for viral load (VL) and DBS for quantification of TFV-DP (N = 61) and completed a self-report measure of adherence. Among those 61 participants, 42 also completed another sub-study that used unannounced telephone pill counts to measure ART adherence [5]. Unannounced pill counts were completed within two weeks (14 days) prior to the blood draw for the first sub-study. Inclusion criteria for both sub-studies mirrored those of the parent study (described above), except only AYAPHIV participants were eligible.

Procedures

Demographic data come from the CASAH parent study; adherence data come from the two sub-studies on adherence (DBS and VL, and unannounced pill count; see Measures). We used data for CASAH participants who completed both sub-studies (N=42) at approximately the same time to compare predictive ability of each adherence measure to predict detectable VL. Participants who completed the unannounced pill count sub-study and DBS sub-study did so within 2-weeks of each other – pill count first, then DBS. Self-reported adherence was also assessed at the same visit when the DBS sample was obtained. Before completing any procedures, all participants provided their written informed consent

Measures

Demographics

Demographics included age, sex at birth, and race/ethnicity.

Self-Reported ART Adherence

Self-reported ART adherence was measured using a 3-item adherence scale previously developed and validated in 2020 [10]. Responses for the three adherence items were linearly transformed to a 0–100% scale, with 0% being the worst adherence and 100% the best. Summary scores were calculated as the mean of the three individual items.

Unannounced Telephone Pill Count

Participants who enrolled in the sub-study were trained to a) describe their ART regimens, b) provide information from their pharmacy labels, and c) count out all pills for each ART medication they were taking. Trained research assistants (RAs) completed four monthly unannounced pill count calls with participants: one baseline call and three consecutive calls approximately 30 days apart to determine adherence during the prior month. During calls, participants counted aloud the number of pills remaining for each of their HIV medications. The study team used this information to calculate the percentage of prescribed doses taken since the last call, resulting in up to three past-month adherence scores from 0-100%. A more detailed account of the protocol has been published elsewhere [5]. For the current study, we used the pill count-derived adherence score that preceded their blood draw.

Tenofovir Diphosphate in Dried Blood Spots

Participants who were on a TDF or TAF-containing ART regimen provided blood samples for VL and DBS for TFV-DP quantification. Blood samples were collected approximately one week after completing the second unannounced telephone pill count for those participants who completed both sub-studies. To prepare the DBS, 25 microliters of whole blood were spotted five times onto a Whatman 903 ProteinSaver cards and dried for a minimum of 3 hours (up to overnight) before being individually packaged in a plastic bag with desiccant and a humidity indicator. The cards were stored frozen at −80°C until shipping to the Colorado Antiviral Pharmacology Laboratory (CAVP) to be assayed using a validated assay for TFV-DP [6].

Viral Load Outcome

Blood for HIV VL was collected (5mL sample) in a citrated tube at the same time as the DBS. VL specimens were assayed by polymerase chain reaction (PCR) at the New York Presbyterian Hospital Clinical Pathology Laboratory utilizing the COBAS AmpliPrep/COBAS TaqMan HIV-1 Test Kit, version 2.0. VL was considered detectable if >200 copies/mL based on the treatment as prevention (TasP) campaign (Undetectable=Untransmittable [U=U]) [2].

Statistical Analysis

We examined the relationship between concentrations of TFV-DP in DBS, self-reported adherence, and unannounced telephone-based pill counts (≤7 days) across VL status groups (≤200 copies/mL vs. >200 copies/mL) and each measure’s ability to detect VL >200 copies/mL among AYAPHIV. Using each adherence measure, we used independent samples T-test to compare mean adherence scores across all adherence measures by VL status, and generated ROCs based on these probabilities. SPSS version 25 was used for all analyses [11].

Results

Sixty-one participants were included in this analysis, with a mean age of 24.17 years (SD = 2.65; Min = 18, Max = 28); 69% were female, 59% (n = 36) identified as Blank/African American and 53% (n = 32) identified as Hispanic/Latino/a. Forty-four percent (n = 27) were on TAF-based regimens, and 56% (n = 34) on TDF-based regimens. Twenty-four (39%) participants had HIV VL >200 copies/mL with a median VL of 74, mean of 10,149 (SD = 38,090). Mean self-report adherence was 73.77% (SD = 23.02); mean pill count adherence was 77.36 (SD = 26.74). The 27 participants on a TAF-based regimen had a mean of TFV-DP concentration of 1,847 fmol/punch; and the 34 participants on TDF-based regimens had a mean TFV-DP concentration of 825 fmol/punch (SD = 826). The mean days between pill count adherence and blood draw was 5.71 (SD = 5.43; Max = 26). Table 1 presents the adherence measures by VL (≤200 vs >200 copies/mL).

Table 1.

Independent Samples T-Test

		Viral Load ≤ 200 copies/mL			Viral Load > 200 copies/mL						95% CI of Difference
Measure	N	n	Mean	SD	n	Mean	SD	F	t	p	Lower	Upper
Self-Report*	61	37	80.15	17.97	24	63.94	26.65	5.82	2.62	.013	4	29
Pill Count*	42	29	84.95	20.60	13	60.42	31.67	8.79	2.56	.005	4	44
TFV-DP (TAF)*	27	17	2751.59	2288.69	10	309.11	358.88	8.10	4.31	.000	1249	3636
TFV-DP (TDF)	34	20	1091.15	789.02	14	445.95	746.87	.003	2.40	.022	97	1193

Note: *Equal variances not assumed

Sensitivity and Specificity to Detect Concurrent Detectable Viral Load

An ROC curve was computed to examine how well each measure predicted concurrent detectable VL. ROC indicated TFV-DP from TAF had the largest area under the curve (0.97 [95% CI: 0.92, 1.00]) followed by TFV-DP from TDF (0.78 [95% CI: 0.60, 0.95]), pill count (0.72 [95% CI: 0.53, 0.90]), and self-report (0.69 [95% CI: 0.48, 0.85]). Maximized sensitivity and specificity were: TFV-DP from TAF 90% and 94% (cut-off = 497 fmol/punches), TFV-DP from TDF 79% and 85% (cut-off = 514 fmol/punch), pill count 54% (cut-off = 48%) and 93%, self-report 58% and 84% (cut-off = 71.7%).

Discussion

In this study, lower mean adherence was associated with HIV VL>200 copies/mL on all adherence measures. Concentrations of TFV-DP (from TAF and TDF) demonstrated the highest sensitivity and specificity to identify an HIV VL >200 copies/mL compared to unannounced pill count and self-reported adherence. Unannounced pill count had slightly higher sensitivity and specificity compared to self-report adherence. Although the AUCs for the DBS assays were higher than unannounced pill count and self-report adherence measures, all three measures predicted HIV VL> 200 copies/mL. This is one of the first studies to examine the predicative validity of TFV-DP in DBS to predict viral load, as well as compare its predictive ability with self-reported adherence and unannounced pill count.

It is important to note that each measure has strengths and weaknesses related to its use in research and clinical practice. With low sensitivity and high specificity, self-report does well at correctly identifying true poor adherence among those (i.e., low false positivity rate), but misses many people with adherence problems that could lead to viral breakthrough. In clinical use, its quick, free, and ease of use is appealing, which must be weighed against how many patients with adherence problems it misses (i.e., its low sensitivity). As a research tool, an inexpensive and quick measure of adherence may help provide population level estimates of adherence problems that can inform policy, but may have limited ability to be used in clinical trials and mechanistic studies. Unannounced telephone pill counts may have better sensitivity and specificity than self-report. However, they require substantial patient and staff training and patient and staff time when the counts occur. While it is not within the scope of this report to estimate the time and costs associated with its use, its resource intensiveness may make it too cumbersome for clinical and research practice.

TFV-DP DBS, which performed the best, requires a phlebotomist, specialized cards for the DBS, proper storage and transportation, and a laboratory that can conduct the assays. Many of these requirements (i.e., phlebotomist, storage, transportation) are available in most clinical settings whether for research or care. However, currently, there are only a few labs in the world that currently assay DBS for TFV-DP and not everyone is taking regimens that include TFV-DP. While it is not in the scope of this report to present costing information for TFV-DP tests (it does cost considerably more than standard HIV-RNA viral load tests [8]), its high cost and limited availability of where assays can be processed likely make it impractical for use in clinical care at the moment. Point-of-care tests, if available, could greatly change this. In clinical studies and trials, DBS for TFV-DP may be ideal because it has been shown to be highly predictive of viral breakthrough several months before it occurs [8].

There are several limitations to this study. First, the sample size is small, which may have limited our ability to detect small effect sizes. Second, these data only provide information on concurrent adherence and HIV VL measurement and is lacking a predictive assessment for future viral breakthrough. Moreover, the population of NYC-based AYAPHIV may not be generalizable to those in other state and countries, nor other populations of AYA affected by HIV. Future research is needed to examine long term predictive value of these adherence measures before viremia is present to inform preventive interventions and inform clinical practice. Despite these limitations, this is one of the first studies to examine TFV-DP concentrations in AYAPHIV and to compare the predictive ability of self-reported adherence, unannounced telephone pill counts, and TFV-DP concentrations in DBS to detect VL >200 copies/mL.