Validation of the REverSe TRanscrIptase Chain Termination assay for measuring tenofovir diphosphate in dried blood spots from a clinical pharmacokinetic trial

Abstract

Background

Tenofovir diphosphate concentration in red blood cells is an objective measure of long-term oral pre-exposure prophylaxis (PrEP) or antiretroviral therapy (ART) adherence. However, current methods for measuring tenofovir diphosphate are equipment and capital intensive, limiting widespread adoption.

Objectives

Low cost, rapid diagnostics for measuring tenofovir diphosphate may drive clinical adoption of routine drug level measurement as a tool for adherence monitoring of tenofovir disoproxil fumarate-based PrEP or ART. We validate a simple and accessible enzymatic assay [REverSe TRanscrIptase Chain Termination (RESTRICT)] for measuring tenofovir diphosphate in dried blood spots (DBS) obtained from a directly observed therapy study of individuals on PrEP.

Methods

We performed RESTRICT measurements on 74 DBS samples from individuals on tenofovir disoproxil fumarate/emtricitabine regimens. We compared RESTRICT measurements with those from a gold-standard method of liquid chromatography tandem mass spectrometry (LC-MS/MS). The ability of RESTRICT to correctly classify DBS tenofovir diphosphate concentrations to established steady-state adherence benchmark concentrations was determined using area under receiver operating characteristic curves (AUCs).

Results

The RESTRICT measurements of DBS samples were highly correlated with LC-MS/MS measurements of tenofovir diphosphate from DBS (r = −0.90; P < 0.0001). The RESTRICT assay correctly classified DBS samples as above or below established steady-state adherence benchmark concentrations corresponding to low (AUC = 0.974), moderate (AUC = 0.936) and high (AUC = 0.955) levels of adherence.

Conclusions

The enzymatic RESTRICT assay can accurately measure tenofovir diphosphate concentrations in DBS specimens using simple procedures and readily available laboratory equipment, offering accessible objective adherence monitoring for persons receiving tenofovir disoproxil fumarate-based PrEP or ART.

Introduction

Oral pre-exposure prophylaxis (PrEP) is highly effective at preventing HIV acquisition, though it is imperative to maintain consistent regimen adherence.^1–3 Self-reported adherence measures can be highly variable and not well-correlated to measured drug levels or patient outcomes, while measured drug levels correspond strongly with PrEP efficacy.^4–8 Recent reports suggest that objective, real-time adherence monitoring may enhance adherence and enable interventions to address adherence lapses.^9–11

Tenofovir disoproxil fumarate-based PrEP is a highly effective first-line recommended regimen with generic versions that are affordable and widely accessible.^12,13 Once ingested, tenofovir disoproxil fumarate is hydrolyzed into tenofovir and intracellularly phosphorylated into its active form of tenofovir diphosphate.^14,15 Tenofovir diphosphate accumulates slowly in red blood cells with a half-life of 17–21 days.^16,17 Due to this comparatively long half-life, tenofovir diphosphate concentrations in red blood cells are indicative of long-term, cumulative regimen adherence over the previous 1–2 months.^17,18 Tenofovir and other nucleotide reverse transcriptase inhibitors (NRTIs), such as emtricitabine, have also been used as objective measures of drug adherence from blood and urine.¹⁹ However, these markers are more reflective of short-term adherence and dosing patterns over the previous 1–7 days as their respective half-lives are significantly shorter (∼15 h).^19,20 These short half-lives also lead these markers to be potentially susceptible to ‘white coat’ effects.²¹

Liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis of dried blood spots (DBS) is currently the gold-standard method to measure tenofovir diphosphate concentrations.^17,22,23 Several directly observed therapy (DOT) studies have established steady-state adherence benchmark tenofovir diphosphate concentrations in DBS that correspond with low, moderate and high levels of adherence.^17,24 DBS are an attractive storage medium as they can be easily collected at the point-of-care, and then stored and shipped to laboratories for later processing.^25,26 Widespread adoption of objective drug-level monitoring is complicated by LC-MS/MS requirements of significant capital investment for equipment and highly trained personnel, limiting its applicability to highly instrumented clinical laboratories.¹⁰

We developed the REverSe TRanscrIptase Chain Termination (RESTRICT) assay as a rapid diagnostic to determine NRTI concentrations from blood samples.^27–29 The RESTRICT assay uses reverse transcriptase (RT) enzymes for complimentary strand synthesis of a DNA template to detect the presence of NRTIs. The template strand can be designed to be selective to specific nucleotide analogues.²⁷ The inhibition of DNA synthesis by NRTIs is measured using an intercalating dye, and samples with higher concentrations of NRTIs result in lower levels of fluorescence due to early chain termination. Previously, we have shown the tuneability of this reaction to specific NRTIs in their respective clinical ranges (e.g. tenofovir diphosphate and emtricitabine triphosphate).^27,28 RESTRICT uses simple sample preparation and readily available laboratory equipment, making it a promising alternative to specialized LC-MS/MS measurement methods.

In this paper, we validate the RESTRICT assay for measuring tenofovir diphosphate in DBS samples collected from individuals receiving tenofovir disoproxil fumarate-based PrEP. We study whether RESTRICT can distinguish drug levels above/below established adherence benchmark concentrations of tenofovir diphosphate corresponding to low (2 doses/week), moderate (4 doses/week) and high adherence (7 doses/week).

Methods

Study population

Clinical dried blood spot samples were sourced from the DOT Tenofovir Adherence to Rapidly Guide and Evaluate PrEP and HIV Therapy (TARGET) study (ClinicalTrials.gov #NCT03012607).²⁴ In TARGET, healthy volunteers (HIV− and HBV−, 18–49 years old) were randomly assigned to three groups (10 participants each, n = 30 total) and received directly-observed doses of tenofovir disoproxil fumarate/emtricitabine (300/200 mg, Truvada) for 6 weeks. The three groups received different dosing regimens: Group 1 received tenofovir disoproxil fumarate/emtricitabine 7 times/week (‘high adherence’), Group 2 received tenofovir disoproxil fumarate/emtricitabine 4 times/week (‘moderate adherence’), and Group 3 received tenofovir disoproxil fumarate/emtricitabine 2 times/week (‘low adherence’). The 6-week dosing schedule was followed by a 4-week wash-out period. Venous blood dried blood spots were collected at Weeks 1, 3, 5, 7, 8, 9 and 10. A total of 74 DBS over the 10-week timespan from 19 participants were analysed via RESTRICT and compared to LC-MS/MS measurements. Samples were selected to provide a wide range of tenofovir diphosphate concentrations and included samples from the initial dosing weeks (as drug was accumulating), steady-state weeks and the wash-out weeks. Consequently, these data are used to determine the correlation between RESTRICT and LC-MS/MS measurements over anticipated ranges and not a measure of PrEP adherence.

Laboratory methods

DBS collected for drug measurement were kept at −70°C to −80°C until analysed. Tenofovir diphosphate concentrations were measured in the TARGET study using a previously validated liquid chromatography mass spectrometry (LC-MS/MS) assay.^17,24,26 Tenofovir diphosphate concentrations are reported in fmol per 3 mm diameter DBS punch.

RESTRICT reaction master mix consisted of 60 mM Tris (77-86-1, Sigma Aldrich, USA), 30 mM KCl (7447-40-7, Sigma Aldrich, USA), 8 mM MgCl₂ (7786-30-3, Sigma Aldrich, USA), 10 mM dithiothreitol (20–265, Sigma Aldrich, USA), 100 nM deoxynucleotide triphosphates (dNTPs) (D7295, Sigma Aldrich, USA), 10 nM primer 16S rRNA forward primer (51-1-19-06, Integrated DNA Technologies, USA) and 1 nM DNA template buffered to pH 8.0 using HCl (H1758, Sigma Aldrich, USA). All values represent final assay concentrations. A custom DNA template was used, which has a 20-nucleotide primer binding site followed by 180-nucleotide TTCA repeating region for chain termination detection of tenofovir diphosphate, as listed in Table S1 (available as Supplementary data at JAC Online) of the Supplementary data.

In the RESTRICT workflow (Figure 1), a 7 mm diameter DBS punch was first eluted into 234 μL of nuclease-free water (AM9906, ThermoFisher Scientific, USA) using a biopsy punch and vigorously vortexed for 1 min. The tube was centrifuged at 10 000 rcf for 5 min before being heated at 95°C for 5 min to coagulate proteinaceous material. The tube was centrifuged again at 10 000 rcf for 5 min. The supernatant was transferred to a 0.45 μm centrifuge filter (8170, Corning Inc., USA) and centrifuged at 10 000 rcf for 5 min. The resulting eluate was agitated briefly before 10 μL of eluate was added to 20 μL of master mix in a flat-bottom polystyrene 96-well plate (3650, Corning Inc., USA). A total of 10 μL of HIV-1 RT (382129-500U, Sigma Aldrich, USA) at a final concentration of 0.0212 units/μL was introduced immediately before incubation at 37°C for 30 min in a microplate reader (SpectraMax iD3, Molecular Devices, USA). After 30 min, 40 μL of PicoGreen dye (P7581, ThermoFisher Scientific, USA) diluted 1:400 in TE buffer was added to stop the reaction, and the fluorescence of the reaction was measured. The underlying mechanisms of the RESTRICT assay are discussed in previous publications.^27,28

Figure 1.

RESTRICT workflow for tenofovir diphosphate measurement from dried blood spots.

A ‘no RT’ negative control was included for each sample, as well as several wells in which various amounts of additional tenofovir diphosphate was spiked into the extracted DBS samples, serially diluted from 900 nM to 0.9 nM (final concentration) to create a dose–response inhibition curve. Experimental conditions for each point of the inhibition curves were repeated in triplicate and averaged, with the exception of a small number of spiked-in conditions where sample volume was limited and duplicates were performed. We also created ‘blank’ DBS cards from a whole blood sample from a single HIV-, PrEP-naïve individual (BioIVT, USA). These DBS were created in quantity from the same sample before analysing clinical samples and acted as a control for day-to-day variation in assay performance. DBS were stored at −80°C until use. A ‘blank’ DBS was included for every three clinical samples and was handled and processed identically. Three clinical samples and one ‘blank’ sample were processed in parallel in each 96-well plate.

Data analysis and statistical methods

Baseline correction of the resulting RESTRICT fluorescence intensities was performed by subtracting the average fluorescence of the ‘no RT’ negative control from the other experimental conditions (i.e. the ‘as-is’ wells and additional tenofovir diphosphate wells). We fit a four-parameter logistic regression curve to the resulting fluorescence intensity as a function of spiked-in tenofovir diphosphate concentrations, with the ‘as-is’ and ‘no RT’ trials artificially pinned at 19 pM and 190 μM, respectively (effectively at zero and infinity, respectively).

The baseline-subtracted fluorescence of the clinical samples was then divided by the maximum of the curve-fit of the included ‘blank’ DBS, and a similar four-parameter logistic regression curve-fit was applied to the resulting normalized ratio. We use the y-axis value of the IC₅₀ point of this normalized ratio curve-fit and compared this against LC-MS/MS tenofovir diphosphate concentrations. An example of this data analysis is shown in the Figure 2.

Figure 2.

RESTRICT data processing protocol. (a) Raw fluorescence RESTRICT measurements for ‘blank’ DBS (as a positive control comparison) and two clinical samples. Note that all of these samples have added tenofovir diphosphate (TFV-DP) and that the third clinical sample curve has been removed for clarity. Data points at TFV-DP concentrations of 19 pM and 190 μM represent ‘as-is’ (i.e. no additional TFV-DP spiked in) and ‘no RT’ (i.e. no RT added to the reaction) measurements, respectively, and are artificially fixed at these concentrations for curve-fitting purposes. (b) Data after baseline subtraction, where all data have been subtracted by the average fluorescence of the ‘no RT’ condition. (c) Data after normalization in which all data are divided by the maximum fluorescence of the ‘blank’ DBS curve fit using a four-parameter logistic regression curve. A second four-parameter logistic regression curve is then fit to the resulting normalized data, and the y-axis coordinates of the IC₅₀ value of each secondary curve-fit of clinical samples (given by the large, filled symbols on the y-axis) are used as the metrics of interest and compared against the LC-MS/MS TFV-DP measurements. Inhibition curve data points are presented as averaged triplicates (with the exception of a small number where only duplicates were possible due to limited sample volume), and error bars represent the standard deviation.

We calculated the Pearson correlation coefficient and the coefficient of determination between the normalized IC₅₀  y-axis values and the log-transformed tenofovir diphosphate concentrations determined via LC-MS/MS. We performed receiver operating characteristic analysis for thresholds of established steady-state adherence benchmarks for tenofovir diphosphate concentrations in DBS. In a previous study, we established adherence benchmarks for this cohort as low adherence (2 doses/week, 466 fmol/3 mm punch), moderate adherence (4 doses/week, 779 fmol/3 mm punch) and high adherence (7 doses/week, 1375 fmol/3 mm punch).³⁰ RESTRICT cut-off values were determined by selecting the point of the receiver operating characteristic curve that minimizes the distance to (0,1) in classifying samples as below the three respective LC-MS/MS adherence benchmark tenofovir diphosphate concentrations.³¹ Data analysis was performed in Prism (GraphPad Software, USA) and MATLAB (MathWorks, USA).

Results

Seventy-four DBS samples from 19 participants (42.1% female) collected over the duration of the 10-week TARGET study were included in the analyses. Tenofovir diphosphate concentrations ranged from 57 to 2580 fmol/3 mm punch. Demographic data of participants and tenofovir diphosphate concentration distributions in clinical DBS samples tested are given in Table 1.

Table 1.

Demographic information for samples from study participants and distribution of tenofovir diphosphate (TFV-DP) concentrations using steady-state adherence benchmark TFV-DP concentrations in DBS^a

n	74
Age (years)	33.5 (28–38)
Female, no. (%)	32 (43.2)

TFV-DP concentrations
x < 466  fmol/punch	466 < x < 779 fmol/punch	779 < x < 1375 fmol/punch	x > 1375 fmol/punch
27	17	19	11

^aData are presented as median (interquartile range) unless otherwise noted.

Figure 3 shows the RESTRICT measurement versus the log of the LC-MS/MS measurements. There is a strong correlation between LC-MS/MS tenofovir diphosphate concentrations and the RESTRICT normalized ratio IC₅₀  y-axis value (n = 74, Pearson r = −0.90; P < 0.0001; 95% confidence interval (CI), −0.94 to −0.85), with a coefficient of determination (r²) of 0.81. The drug level steady-state adherence benchmark concentrations for low, moderate and high levels of drug adherence are represented by dashed vertical lines.

Figure 3.

Correlation of RESTRICT normalized ratio IC₅₀  y-axis values and log₁₀(TFV-DP) LC-MS/MS measurements, with steady-state adherence benchmark concentrations for low, moderate and high levels of drug adherence represented by dashed vertical lines.

Figure 4 shows the sensitivity and specificity analysis to evaluate how well RESTRICT predicts LC-MS/MS concentration categories. In classifying DBS samples below 466 fmol/3 mm punch, the area under the receiver operating characteristic curve (AUC) is 0.974 (95% CI: 0.943–1), with a cut-off normalized ratio IC₅₀  y-axis coordinate value of 0.426 resulting in sensitivity of 96.3% and specificity of 93.6% (1/27 false negatives, 3/47 false positives). In classifying DBS samples below 799 fmol/3 mm punch, the AUC is 0.936 (95% CI: 0.885–0.988), with a cut-off normalized ratio IC₅₀  y-axis coordinate value of 0.327 resulting in sensitivity of 93.2% and specificity of 80% (3/44 false negatives, 6/30 false positives). In classifying DBS samples below 1375 fmol/3 mm punch, the AUC is 0.955 (95% CI: 0.904–1), with a cut-off normalized ratio IC₅₀  y-axis coordinate value of 0.297 resulting in sensitivity of 87.3% and specificity of 90.9% (8/63 false negatives, 1/11 false positives). Plots of receiver operating characteristic curves for low, moderate and high adherence cut-offs can be found in Figure S1 of the Supplementary data. Note, we have selected a ‘positive’ test result to be the RESTRICT result to be below the steady-state adherence benchmark tenofovir diphosphate concentration; however, selecting the opposite, that the sample is above the benchmark concentrations, would have the same results except that the specificity values would be interchanged.

Figure 4.

Population distributions and sensitivity/specificity analyses of RESTRICT measurements using established steady-state adherence benchmarks for TFV-DP concentrations in DBS determined via LC-MS/MS from directly observed therapy cohort. (a) Below threshold of 466 fmol/3 mm punch (concentration consistent with low adherence). (b) Below threshold of 799 fmol/3 mm punch (concentration consistent with moderate adherence). (c) Below threshold of 1375 fmol/3 mm punch (concentration consistent with high adherence).

Discussion

This study used DBS samples collected from DOT participants on tenofovir disoproxil fumarate-based PrEP regimens to validate RESTRICT for measuring tenofovir diphosphate concentrations. RESTRICT measurements were directly compared to LC-MS/MS measurements. A previous pilot study demonstrated that RESTRICT fluorescence values correlate with LC-MS/MS measurements of tenofovir diphosphate in DBS, though this was a smaller study (n = 18, with only six samples having detectable drug levels).²⁹ In this work (n = 74), we show that the normalized ratio IC₅₀  y-axis value is strongly correlated with LC-MS/MS measurements of tenofovir diphosphate in DBS, with higher concentrations corresponding to lower normalized ratio IC₅₀  y-axis values. We demonstrate that the RESTRICT assay is semi-quantitative, accurately distinguishing tenofovir diphosphate concentrations in DBS as below steady-state adherence thresholds corresponding to low (466 fmol/3 mm punch), moderate (779 fmol/3 mm punch) and high (1375 fmol/3 mm punch) levels of drug adherence.

This study represents a significant step towards rapid, objective, long-term regimen adherence monitoring. Oral PrEP efficacy is highly dependent on medication adherence, with significantly diminished HIV-1 risk reduction observed for individuals taking 2 doses/week (low adherence) or less.³² RESTRICT demonstrated 96.3% sensitivity and 93.6% specificity for identifying DBS samples with tenofovir diphosphate concentrations less than 466 fmol/3 mm punch, which corresponds to 2 doses/week of tenofovir disoproxil fumarate-based PrEP. RESTRICT may be a useful near-patient tool for identifying instances of poor PrEP or antiretroviral therapy adherence and enabling timely clinical intervention and adherence counselling.

RESTRICT is performed with commonly available laboratory equipment and features simple sample preparation and a runtime of less than 60 min, which may allow for decentralized testing and removal of barriers to timely linkage with appropriate interventions.¹⁰ Other rapid methods for objective adherence monitoring include urine-based tenofovir point-of-care lateral flow assays, where undetectable tenofovir in urine has been associated with HIV viraemia.^33–35 While these tests benefit from the ease of sample collection compared to blood-based samples, urine-based tenofovir assays are limited to short-term adherence measurement due to the short half-life of tenofovir in urine (∼15 h).^16,36

A limitation of this work is that it validates the RESTRICT assay for tenofovir disoproxil fumarate-based PrEP monitoring but not tenofovir alafenamide-based regimens. Future work is needed to optimize and validate the RESTRICT assay for measuring tenofovir alafenamide-based PrEP. While tenofovir diphosphate is the active form of both tenofovir disoproxil fumarate and tenofovir alafenamide, it is known to accumulate in blood constituents at lower concentrations in tenofovir alafenamide-based regimens compared to tenofovir disoproxil fumarate-based regimens, with ∼1/7th the concentration in DBS, suggesting modifications to the RESTRICT sample preparation protocol may be required.³⁷ As an example, a recent DOT study measuring tenofovir diphosphate concentrations in DBS from tenofovir alafenamide/emtricitabine regimens via LC-MS/MS required the use of two 7 mm punches for each sample, whereas a single 3 mm punch was sufficient for similar tenofovir disoproxil fumarate/emtricitabine DOT DBS studies due to this decrease in tenofovir diphosphate DBS concentrations in tenofovir alafenamide-based regimens.^16,17 Adjustments to the RESTRICT assay itself may also be necessary for tenofovir alafenamide-based regimens, though we have shown that the assay can be tuned to accommodate various targets and target concentrations through modifications of the DNA template and/or reagent concentrations.²⁷

In summary, we evaluated the RESTRICT assay as an objective measure of long-term adherence to tenofovir disoproxil fumarate. We compared the RESTRICT assay measurements against current gold-standard method of LC-MS/MS for measuring tenofovir diphosphate concentrations in DBS from a DOT study. The RESTRICT assay strongly correlates with LC-MS/MS measurements and is able to distinguish tenofovir diphosphate concentrations in DBS as above or below steady-state adherence thresholds corresponding to low, moderate and high adherence. This suggests that RESTRICT can be a useful semi-quantitative test to determine individuals’ adherence levels to tenofovir disoproxil fumarate-based PrEP regimens. This novel assay could be used as a method for objective adherence monitoring to enable more targeted interventions towards minimizing lapses in adherence and increasing overall PrEP regimen efficacy.

Funding

This work was supported by the National Institutes of Health (grant numbers R01AI157756, R01AI136648, R21AI127200). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This work was conducted using equipment in the Biochemical Diagnostics Foundry for Translational Research supported by the M. J. Murdock Charitable Trust. Research reported in this publication was supported by the University of Washington/Fred Hutch Center for AIDS Research, an NIH funded programme under award number AI027757 that is supported by the following NIH institutes and centres: NIAID, NCI, NIMH, NIDA, NICHD, NHLBI, NIA, NIGMS and NIDDK.

Transparency declarations

A.O.O., B.P.S., A.T.B., P.K.D. and J.D.P. are listed as inventors on a patent filed based on this work (PCT/US2020/037609). We declare no other conflicts of interest. B.P.S. wrote the first draft of the manuscript and contributed to formal analysis, software, methodology, visualization and editing and reviewing. C.A.C. contributed to the data collection, methodology, formal analysis, visualization, software and editing and reviewing. A.T.B. contributed to the formal analysis, supervision and editing and reviewing. E.B. contributed to the data collection. O.S. contributed to the data collection. P.-o.S. contributed to the data collection. T.R.C. contributed to the conceptualization and editing and reviewing. P.K.D. contributed to the conceptualization, supervision, project administration, funding acquisition and editing and reviewing. A.O.O. contributed to the conceptualization, methodology, supervision, funding acquisition and editing and reviewing. J.D.P. contributed to the conceptualization, methodology, formal analysis, supervision, project administration, funding acquisition and editing and reviewing.

Supplementary data

Figure S1 and Table S1 are available as Supplementary data at JAC Online.