Urine tenofovir and dried blood spot tenofovir diphosphate concentrations and viraemia in people taking efavirenz and dolutegravir-based antiretroviral therapy

Objective: We aimed to determine whether urine tenofovir (TFV) and dried blood spot (DBS) tenofovir diphosphate (TFV-DP) concentrations are associated with concurrent HIV viraemia. Design: Cross-sectional study among people with HIV (PWH) receiving tenofovir disoproxil fumarate (TDF)-based antiretroviral therapy (ART). Methods: We used dual tandem liquid chromatography and mass spectrometry to measure urine TFV and DBS TFV-DP concentrations, and evaluated their associations with concurrent viraemia at least 1000 copies/ml using logistic regression models. In exploratory analyses, we used receiver operating curves (ROCs) to estimate optimal urine TFV and DBS TFV-DP thresholds to predict concurrent viraemia. Results: Among 124 participants, 68 (54.8%) were women, median age was 39 years [interquartile range (IQR) 34–45] and 74 (59.7%) were receiving efavirenz versus 50 (40.3%) receiving dolutegravir. Higher concentrations of urine TFV [1000 ng/ml increase, odds ratio (OR) 0.97 95% CI 0.94–0.99, P = 0.005] and DBS TFV-DP (100 fmol/punch increase, OR 0.76, 95% CI 0.67–0.86, P < 0.001) were associated with lower odds of viraemia. There was evidence that these associations were stronger among people receiving dolutegravir than among people receiving efavirenz (urine TFV, P = 0.072; DBS TFV-DP, P = 0.003). Nagelkerke pseudo-R2 for the DBS TFV-DP models was higher for the urine TFV models, demonstrating a stronger relationship between DBS TFV-DP and viraemia. Among people receiving dolutegravir, a DBS TFV-DP concentration of 483 fmol/punch had 88% sensitivity and 85% specificity to predict concurrent viraemia ≥1000 copies/ml. Conclusion: Among PWH receiving TDF-based ART, urine TFV concentrations, and in particular DBS TFV-DP concentrations, were strongly associated with concurrent viraemia, especially among people receiving dolutegravir.


Introduction
There is increasing interest in accurately monitoring antiretroviral therapy (ART) adherence for people with HIV (PWH).Tenofovir disoproxil fumarate (TDF) is included in fixed-dose combinations alongside emtricitabine and efavirenz, or lamivudine and dolutegravir, which are used by over 95% of people receiving ART in low-income and middle-income countries (LMICs) [1,2].Therefore, objective tenofovir measurements could identify poor adherence.TDF is converted to tenofovir (TFV), which is metabolized intracellularly to tenofovir diphosphate (TFV-DP).TFV is excreted in urine and correlates with short-term adherence as it has a 12-15 h terminal half-life [3], whereas TFV-DP accumulates in red blood cells and correlates with medium-term adherence as it has a longer half-life of 17 days [4].Studies have shown that qualitative urine TFV [5][6][7][8] and quantitative dried blood spot (DBS) TFV-DP levels [9][10][11] are associated with viral suppression in PWH receiving ART, but none have compared the two measures, or determined thresholds that best predict viral suppression.Furthermore, dolutegravir has a higher genetic barrier to resistance than efavirenz, meaning that measures of adherence should be more closely associated with viral suppression, unlike efavirenz where resistance can cause viraemia despite good adherence.Therefore, we aimed to compare the association between urine TFV, and DBS TFV-DP concentrations, with viraemia among PWH receiving dolutegravir and efavirenz-based ART.In post hoc analyses, we also aimed to estimate optimal urine TFV and DBS TFV-DP thresholds to detect viraemia, and to assess associations between TFV levels and both HIV drug resistance (HIVDR), and self-reported adherence.

Methods
We conducted a cross-sectional analysis at enrolment into a randomised study of point-of-care HIV viral load testing (POwER) [12].We included consecutively enrolled POwER participants receiving TDF as part of dolutegravir or efavirenz-based first-line ART.Eligible PWH had a preenrolment viral load greater than 1000 copies/ml in the past 6 weeks, without having received enhanced adherence counselling.At enrolment, participants self-reported adherence, and had urine, DBS and plasma samples taken and stored at À80 8C, for retrospective testing.
We quantitated urine TFV and DBS TFV-DP concentrations using liquid chromatography and dual tandem mass spectrometry (LC-MS/MS).We tested viral load using the cobas 6800 platform (Roche, Basel, Switzerland), and attempted drug resistance testing for all samples with viral load at least 1000 copies/ml (see testing details supplement, http://links.lww.com/QAD/D80).
We used logistic regression models to assess the relationship between the exposure of either urine TFV concentrations, or DBS TFV-DP concentrations, and the outcome of viraemia.To determine whether associations differed by ARTregimen, we included a variable for ART regimen (dolutegravir versus efavirenz) in the model, with an interaction term between ART regimen and urine TFV, or DBS TFV-DP concentrations.We fitted separate models for the outcomes of viraemia at least 1000 copies/ml, and at least 50 copies/ml, as these thresholds are used in WHO guidelines [1].We compared the Nagelkerke pseudo-R 2 of the urine TFV and DBS TFV-DP models to determine which measure was more strongly associated with viraemia [13].In exploratory, post hoc analyses, we used receiver-operating curves (ROCs) to estimate urine TFV and DBS TFV-DP thresholds that maximize sensitivity and specificity to predict concurrent viraemia.Lastly, we described urine TFV and DBS TFV-DP levels among people with and without HIVDR, and compared self-reported short-term and longer term adherence with urine TFV and DBS TFV-DP levels using logistic regression and linear regression models, respectively.Sample size was determined by the number of participants enrolled into POwER and receiving TDF.
We analysed data using R 4.
Higher concentrations of urine TFV [1000 ng/ml increase, odds ratio (OR) 0.97 95% CI 0.94-0.99,P ¼ 0.005) and DBS TFV-DP (100 fmol/punch increase, OR 0.76, 95% CI 0.67-0.86,P < 0.001) were associated with lower odds of viraemia at least 1000 copies/ml, with similar results at at least 50 copies/ml (Table 1a).There was some evidence that the association between urine TFVand viraemia at 1000 copies/ml (LRT for interaction P ¼ 0.072), and between DBS TFV-DP and viraemia at both 1000 (P ¼ 0.003) and 50 (P ¼ 0.068) copies/ml, was stronger among people receiving dolutegravir than among people receiving efavirenz.There was no evidence of a difference by ART regimen in the association between urine TFV and viraemia at 50 copies/ml (P ¼ 0.797, Table 1a).Overall, at both 1000 and 50 copies/ml thresholds, the Nagelkerke pseudo-R 2 for the DBS TFV-DP models was higher than for the urine TFV models, meaning there was a stronger relationship between DBS TFV-DP and viraemia (Table 1a).

Discussion
In this cross-sectional study, we found that urine TFVand DBS TFV-DP concentrations were negatively associated with concurrent viraemia, and the association was generally stronger with dolutegravir compared with efavirenz.Furthermore, DBS TFV-DP had a better association with viraemia compared with urine TFV.
Our findings are similar to studies which have shown that DBS TFV-DP concentrations are associated with viraemia in PWH receiving TDF-based ART [9][10][11].
A study among 532 PWH in the United States found that higher DBS TFV-DP was associated with viral load less than 20 copies/ml.Thirty-six percent of participants were receiving integrase inhibitors, and 27% were receiving nonnucleoside reverse transcriptase inhibitors.Among people with viral load less than 20 copies/ml, the mean TFV-DP concentration was 1728 (1608-1857) fmol/punch compared with 1469 (1283-1681) fmol/ punch at 20-200 copies/ml, and 633 (542-739) fmol/ punch at greater than 200 copies/ml, but unlike our study, results were not presented by ART class [9].A South African cross-sectional study among 137 people taking efavirenz used ROC curves to demonstrate that DBS TFV-DP was more strongly associated than plasma TFV with viral suppression less than 50 copies/ml [10].Lastly, among 250 virally suppressed PWH receiving efavirenz in South Africa, baseline TFV-DP less than 400 fmol/punch was associated with increased odds of developing viraemia at least 400 copies/ml after 1 month [11].Regarding urine TFV levels, several studies demonstrate qualitative point-of-care urine TFV levels are associated with concurrent viraemia [5][6][7][8], but no studies have assessed the relationship between quantitative TFV concentrations and viraemia.
Although both urine TFV and DBS TFV-DP concentrations were negatively associated with viraemia at least 1000 copies/ml, the association was weaker with efavirenz, likely because the high prevalence of HIVDR means viraemia persists in the presence of measurable adherence.With dolutegravir, HIVDR was rare, meaning the relationship between TFV measures and viraemia was stronger.In exploratory, post hoc analyses using ROC curves, we similarly found that the potential for urine TFV and DBS TFV-DP to predict concurrent viraemia was poor to modest with efavirenz but more acceptable with dolutegravir.Using DBS TFV-DP, a threshold of around 480 fmol/punch would have more than 80% sensitivity and specificity to predict concurrent viraemia at least 1000 copies/ml.Comparing pseudo-R 2 values, models indicated superiority of fit for TFV-DP over urine TFV, suggesting that TFV-DP generally performed better.This is likely because quantitative DBS TFV-DP reflects longer term adherence, which is required to achieve viral suppression, and which is not captured by more transient, shorter term urine TFV concentrations.
Strengths of our study include the comprehensive assessment of adherence using viral load, HIVDR, selfreported adherence, urine TFVand DBS TFV-DP.To our knowledge, this is the first study to directly compare the predictive value of quantitative urine TFV concentrations against DBS TFV-DP in PWH.We used different viraemia thresholds, which reflect WHO guidelines [1].
Our study is limited by the small sample size, with participants enrolled in a clinical trial for people with recent viraemia, meaning results may not be generalizable to other populations.Although we estimate urine TFV and DBS TFV-DP thresholds to optimize sensitivity and specificity to detect viraemia at both at least 50 and at least 1000 copies/ml, these are exploratory analyses with imprecise estimates because of our small sample size.These thresholds should be evaluated in other studies, and should take into account the clinical use scenario, and whether sensitivity or specificity should be maximized [14].Therefore, we provide ranges of thresholds and associated sensitivities and specificities in Table S2, http://links.lww.com/QAD/D84.
Our study suggests that with the global dolutegravir rollout, urine TFV or DBS TFV-DP measures of adherence may be increasingly useful to measure adherence and complement viral load testing.Qualitative point-of-care urine TFVassays have been validated [5][6][7][8], and their clinical effectiveness will need to be confirmed in clinical trials before they can be rolled out in clinical practice [15].Although TFV-DP performed better in our study, it currently requires expensive LC-MS/MS, which limits widespread clinical use in LMICs.Development of point-of-care TFV-DP assays should be prioritized [16], alongside studies to establish if there is a reliable TFV-DP threshold for predicting concurrent viraemia among people receiving dolutegravir.People with viraemia who have a TFV-DP level higher than this threshold (suggesting good adherence) could be more likely to have HIVDR, [5,17] and so could be prioritized for HIVDR testing.
receive funding from this award.J.D., Academic Clinical Lecturer (CL-2022-13-005), is funded by the UK National Institute of Health and Social Care Research (NIHR).The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.

Table 1 .
(a) Logistic regression models of the association between urine tenofovir concentrations, and dried blood spot tenofovir diphosphate concentrations, and viraemia; (b) diagnostic accuracy thresholds and areas under the curve for urine tenofovir and tenofovir diphosphate concentrations to detect viraemia.