Nucleoside reverse transcriptase inhibitor‐reducing strategies in HIV treatment: assessing the evidence

Antiretroviral (ARV) therapy, comprising a backbone of two nucleos(t)ide reverse transcriptase inhibitors (NRTIs) plus another ARV, is the recognized standard of care (SOC), which has helped extend life expectancy in people living with HIV. In a quest to reduce lifelong drug exposure and minimize or avoid the toxicity of NRTIs, “NRTI‐reducing” regimens have been investigated. This descriptive review assessing the results of NRTI‐reducing strategies from the largest randomized trials focuses on virological efficacy, resistance, regimen safety (in terms of bone mineral density, renal function, lipids and central nervous system function) and simplicity. The review considers efficacy across various NRTI‐sparing strategies, for example an integrase strand transfer inhibitor (INSTI) plus a ritonavir‐boosted protease inhibitor (PI/r) or PI/r + lamivudine (3TC), in both naïve and switch regimes. Of 10 key studies in treatment‐naïve adults assessing five NRTI‐reducing strategies, only four studies demonstrated noninferiority vs. SOC [GARDEL, NEAT 001, AIDS Clinical Trials Group 5142 and PROGRESS]. In switch settings, 17 studies (10 randomized) were reviewed that used four strategies, including three studies assessing an INSTI plus a nonnucleoside reverse transcriptase inhibitor . Noninferiority of the NRTI‐reducing arm was shown in six of 10 studies (ATLAS‐M, SALT, DUAL, OLE, LATTE‐2 and SWORD). In general, NRTI‐reducing therapy did not always result in an improvement in short‐ or long‐term adverse events; however, in many cases, these endpoints were not reported. Some of these studies reported higher virological failure rates with more frequent emergence of resistance mutations. None of these NRTI‐reducing strategies has been compared against a single‐pill regimen, including those containing tenofovir alafenamide. Only strategies demonstrating noninferior efficacy, a benefit in safety/tolerability, and a favourable cost‐efficacy ratio, preferably in a single pill, will eventually match the current SOC of triple ARV therapy.


Introduction
Three-drug combination antiretroviral therapy (ART)two nucleos(t)ide reverse transcriptase inhibitors (NRTIs) plus an anchor drug from another antiretroviral class -has resulted in the life expectancy of adults with HIV infection approaching that of the general population [1][2][3][4]. Rates of virological suppression at 48 weeks with current simple (one-pill, once-daily) regimens can exceed 90%, accompanied by improved long-and short-term safety profiles [5][6][7][8].
Two-drug, "NRTI-reducing" regimens remain under investigation; some are included in guidelines as alternatives, such as when abacavir (ABC), tenofovir disoproxil fumarate (TDF) and tenofovir alafenamide (TAF) are not appropriate, but have a more limited evidence base. A recent systematic review comparing all NRTI-reducing regimens with all three-drug ART regimens used as firstline therapy suggested that NRTI-reducing therapy is not a good option in patients with a high baseline viral load (VL) (> 100 000 HIV-1 RNA copies/mL plasma) or low CD4 lymphocyte counts, in whom virological failure and resistance are more likely to occur [10]. However, this pooled analysis did not report the evidence for each specific NRTI-reducing regimen nor whether NRTI-reducing regimens offer any safety or simplicity advantage over standard preferred ART (and not only over the comparator three-drug ART chosen in every particular trial). Here, we review the efficacy, resistance, safety and simplicity of individual NRTI-reducing regimens relative to three-drug ART.

Methods
This descriptive review considers data reported from trials assessing NRTI-reducing regimens in ART-na€ ıve and virologically suppressed adults. The review focuses on the largest, better powered randomized trials, which provide the strongest evidence for both efficacy and safety. Smaller, exploratory studies of NRTI-reducing strategies and monotherapy studies (either PI/r or integrase inhibitor) are listed and described in brief only in tabular format (see Tables 1 and 2).
Individual studies were assessed with reference to virological efficacy, with particular focus on noninferiority and resistance, safety [in terms of serious non-AIDSrelated events, lipids, renal function, central nervous system (CNS) function and bone mineral density (BMD)], and treatment simplicity and dosing.

Studies selected
The key NRTI-reducing studies reviewed in ART-na€ ıve and virologically suppressed HIV-1-infected adults are listed in Tables 1 and 2, respectively.
In treatment-na€ ıve adults, 10 key randomized studies have reported on a variety of NRTI-reducing strategies ( Table 1). In most studies, over 80% of participants were men. Of these studies, five evaluated a PI/r plus an integrase strand transfer inhibitor (INSTI), one a PI/r plus a nonnucleoside reverse transcriptase inhibitor (NNRTI), three a PI/r plus a C-C chemokine receptor type 5 (CCR5) antagonist [maraviroc (MVC)], and one a PI/r plus lamivudine (3TC).
In simplification/switch settings, 17 studies (10 of them randomized) were reviewed (Table 2), including: four of one PI/r plus 3TC; one of one PI/r plus MVC; one of one PI/ r plus one INSTI; and three of one INSTI plus one NNRTI.
Studies in treatment-na€ ıve patients PI/r plus NNRTI: ACTG5142 randomized 753 patients to open-label lopinavir/r (LPV/r) plus efavirenz (EFV) or to standard triple regimens with two NRTIs [3TC plus stavudine (d4T) or TDF or zidovudine (ZDV)] plus either EFV or LPV/r [28]. There were no significant differences between the arms in overall efficacy through 96 weeks, but, among patients with baseline VL > 100 000 copies/ mL, those patients randomized to LPV/r plus EFV had a shorter time to virological failure vs. those randomized to EFV plus two NRTIs. PI/r plus INSTI: Five studies evaluated one PI/r plus one INSTI, although one study was uncontrolled and two had fewer than 100 participants [29][30][31][32][33][34][35][36][37]. Of the three controlled, well-powered studies, noninferiority at week 48 was found in two studies.
In NEAT-001 [31,32,34], darunavir/r (DRV/r) + raltegravir (RAL) was noninferior to DRV/r plus TDF/emtricitabine (FTC) at 48 weeks overall, but was significantly less effective in patients with baseline CD4 cell counts < 200 cells/lL, and noninferiority was not demonstrated in those with baseline VL > 100 000 copies/mL. PI/r plus 3TC: In the open-label GARDEL study, LPV/r plus 3TC was virologically noninferior to LPV/r plus two NRTIs at both 48 and 96 weeks. Similar virological efficacy responses were observed regardless of baseline VL [38]. However, the relevance to contemporary ART is limited somewhat, as 54% of controls received ZDV and both arms received LPV/r, neither of which are recommended any more as part of triple therapy. PI/r plus maraviroc: Three studies evaluated one PI/r plus MVC [39][40][41]. The largest of these -MODERNcompared DRV/r plus MVC 150 mg with DRV/RTV plus TDF/FTC. This study was terminated early because of statistically lower rates of virological suppression with DRV/r plus MVC [39][40][41]. INSTI plus 3TC: In a pilot, proof-of-concept, 48-week study -PADDLEthe efficacy of dolutegravir (DTG) plus 3TC was 90% in 20 treatment-na€ ıve adults with baseline plasma VL < 100 000 copies/mL and CD4 counts > 200 cells/lL [42][43][44]. There was one virological failure and one suicide. The efficacy and safety of DTG plus 3TC is being explored in well-powered, phase 3 studies (GEMINI I and 2; ClinicalTrials.gov identifier NCT02831673).
Switch studies in virologically suppressed patients NRTI-reducing approaches have also been explored in virologically suppressed patients who switched their triple-drug regimens for simplification, or because of drugrelated toxicities, adherence issues, or cost ( Table 2). PI/r plus 3TC: Five studies evaluated the switch from two NRTIs plus one PI/r; the two largest, SALT and ATLAS-M, both evaluated atazanavir/r (ATV/r) [45][46][47][48][49][50]. SALT reported that the switch to ATV/r plus 3TC was noninferior (at both 48 and 96 weeks) to ATV/r plus two NRTIs in virologically suppressed patients (with 40% already receiving ATV/r at enrolment). In the control arm, the anchor drug was also switched to ATV + RTV [47].
ATLAS-M demonstrated that switching to ATV/r plus 3TC was noninferior (superior on post hoc analysis) to continuing ATV/r plus two NRTIs, with efficacy of 89.5% and 79.7% at 48 weeks, respectively [45,46,51,52]. These differences were driven by fewer virological failures and withdrawals in the ATV/r plus 3TC arm and were maintained through to week 96.
Two other studies have demonstrated noninferiority of the combination of PI/r plus 3TC. In DUAL, a switch to once-daily DRV/r plus 3TC was noninferior to DRV/r plus two NRTIs at week 48 (89% and 93% had VL < 50 copies/mL, respectively) [50]. Similarly, OLE found that a switch to LPV/r plus 3TC (61% on TDF/FTC) resulted in noninferior efficacy compared with continued LPV/ r + two NRTIs at week 48 (87.8% vs. 86.6% with VL < 50 copies/mL, respectively), but with no significant differences in discontinuations or adverse events (AEs) [49].
Taken together, the above PI/r studies demonstrated that a PI/r (DRV/r, ATV/r or LPV/r) with 3TC is a potential simplification strategy in subjects already receiving triple regimens comprising that specific PI/r plus two NRTIs. PI/r plus MVC: One study -MARCHhas assessed a switch to DRV/r plus MVC from a PI/r plus two NRTIs in patients with CCR5-tropic HIV. The study reported   LATTE is an induction-maintenance, phase 2b, randomized, partially blinded study in treatment-na€ ıve adults. Oral cabotegravir (CAB; 10, 30 or 60 mg once daily) plus two NRTIs was administered for 24 weeks. Subjects with HIV RNA < 50 copies/mL at week 20 were then simplified to CAB plus RPV 25 mg daily. At 48 weeks, CAB plus RPV had similar efficacy (93%) to the control of EFV plus two NRTIs (94%) in those subjects undergoing the simplification [55].
LATTE-2 evaluated CAB plus RPV administered by long-acting intramuscular injections. A total of 309 ART-na€ ıve subjects received CAB plus ABC/3TC for 20 weeks [56]. Oral RPV was then added for 4 weeks; subjects who were able to tolerate oral RPV were then switched to intramuscular CAB plus RPV every 4 or 8 weeks or maintained on oral CAB plus ABC/3TC. At week 48, both intramuscular CAB plus RPV arms were noninferior to the control arm, with rates of VL < 50 copies/mL of 91%, 92% and 89% for intramuscular CAB plus RPV every 4 and 8 weeks and oral CAB plus ABC/3TC, respectively. Because of variable pharmacokinetics with 8-weekly treatment, maintenance with long-acting CAB plus RPV is only being evaluated 4-weekly in two phase 3 studies [ATLAS (ClinTrials.gov Identifier NCT02951052) and First Long-Acting Injectable Regimen (FLAIR) (ClinTrials.gov Identifier NCT02938520)].
CAB, which may not be submitted for licensed oral administration, has paved the way for other all-oral switch studies combining DTG with RPV [57]. In the phase 3 SWORD 1 and 2 studies, an open-label switch to DTG (50 mg) plus RPV (25 mg) once daily was noninferior at week 48 vs. continued three-or four-drug ART, with efficacy rates of 95% in both arms [57]. As might be expected with a switch regimen including one or two new drugs administered to all participants, slightly more AEs leading to discontinuation occurred with DTG plus RPV than with the continued three-or four-drug ART (3% vs. 1%, respectively), with 2% on DTG plus RPV discontinuing because of CNS-related AEs. It is the first time that an all-oral, two-drug, PI/rand NRTI-free, single-pill regimen has demonstrated noninferiority. A regulatory filing for approval of the DTG 50 mg/RPV 25 mg co-formulation has been submitted.

Resistance
Treatment-na€ ıve and virologically suppressed subjects who experienced virological failure on NRTI-reducing regimens had an overall higher risk of developing antiretroviral resistance mutations at treatment failure. Resistance tended to be more common in ART-na€ ıve groups treated with NRTI-reducing regimens, but was also observed in virologically suppressed patients who were switched to NRTI-reducing regimens (Tables 1 and 2). In the best designed study in treatment-na€ ıve patients, the NEAT-001 study, resistance-associated mutations developed in 29.5% of those who experienced virological failure in the DRV/r plus RAL group (including some major integrase resistance mutations) compared with 0% in the DRV/r plus TDF/FTC arm [31,32]. Resistance was reported to occur more commonly in patients with high VL in studies in treatment-na€ ıve patients. However, this was less common with PI/r-based regimens (e.g. 1% of ACTG5142 participants had virological failure on LPV/r plus EFV, and none of the participants in GARDEL, all of whom were on LPV/r, had virological failure) [28].
In switch/simplification, dual-drug strategies based on a PI/r plus 3TC, or CAB (or DTG) plus 3TC have not shown increased rates of resistance at failure. Only one subject in LATTE, receiving the lowest CAB dose (10 mg, subsequently dismissed), developed an integrase mutation (Q148R) [56]. In SWORD 1 and 2, one subject on DTG plus RPV meeting virological withdrawal criteria had an identified NNRTI resistance-associated mutation (K101K/ E), although the patient subsequently re-suppressed while continuing the same regimen [57].

Safety
Bone mineral density Only four of 10 studies in treatment-na€ ıve patients and three of 17 switch studies listed reported changes in BMD. All studies reporting an increase in BMD with NRTI-reducing ART were those in which the comparator arm included TDF, particularly when combined with a PI/ r. The BMD differences ranged from 2% to 6% in studies in treatment-na€ ıve patients and were only reported in one study in treatment-experienced patients, where the difference was around 3% (Tables 1 and 2).
Of less certain clinical significance, some studies have reported differences in markers of bone turnover; in general, avoidance of TDF resulted in lower levels of these markers, suggesting better bone health.
A switch to ATV/r plus 3TC resulted in a significant improvement in BMD in the lumbar spine (but not the femoral neck or hip) at 96 weeks. However, no differences in BMD were seen in the SALT study [47]. In MARCH, DRV/r plus MVC vs. DRV/r plus two NRTIs was associated with a significant improvement in BMD [53].

Renal function
Despite renal improvement being a consideration in NRTI-reducing strategies, the evidence supporting improvements in renal function has been best provided by studies where TDF is removed from the regimen. The mean difference in estimated glomerular filtration rate (eGFR) in studies in treatment-na€ ıve patients ranged from 5.4 to 8.6 mL/min (two of 10 studies available data) at 48 weeks, and in switch studies was approximately 9 mL/min (two of 17 studies available data).

Cardiovascular disease and lipids
No studies have reported significantly lower cardiovascular event rates with NRTI-reducing strategies. For lipids, there were no or only modest differences, particularly for total: high-density lipoprotein (HDL) cholesterol ratio, the strongest lipid predictor of cardiovascular risk.
Central nervous system CNS endpoints were reported in only four studies (NEAT, SALT, ATLAS-M and SWORD 1 and 2). None of these studies found differences in rates of cognitive function with NRTI-reducing vs. three-drug ART.

Adverse events
No clinically relevant improvements in AEs have been reported with NRTI-reducing ART, although it is important to note that some comparator backbones included ABC, TDF or even ZDV (GARDEL). Rates of grade 3 or 4 AEs ranged from 4.5% to 48.3% with NRTI-reducing regimens vs. 2.9-32% with triple therapy.

Simplicity
The pill burden in studies in ART-na€ ıve patients ranged from 3 to 10 pills per day with three-drug ART and from 2 to 9 pills per day with an NRTI-reducing regimen. The mean difference in pills per day in randomized trials in ART-na€ ıve patients was 1 pill per day.
Similarly, in switch studies, the pill burden ranged from 3 to 6 pills per day with control ART and from 1 to 4 pills per day with an NRTI-reducing regimen. The mean difference in pills per day in randomized switch trials was 2 pills per day. To date, no study has investigated the switch from a triple therapy-based single tablet regimen to an NRTI-reducing regimen.

Discussion
The findings of this descriptive review are consistent with those of a previous review that found that NRTI-reducing therapy may not be as effective as conventional threedrug ART, especially in treatment-na€ ıve patients with high baseline VL and low CD4 counts. In general, NRTIreducing therapy resulted in more resistance. Our analysis has further identified that in these studies NRTI-reducing regimens did not offer advantages in terms of simplicity. Furthermore, improvements in AEs were infrequent, and reductions in the rates of ART-related serious non-AIDSrelated events have not been demonstrated.
One of the key aims of NRTI-reducing therapies is to avoid NRTI-related toxicity such as bone, renal or cardiovascular toxicity. Most studies were too small and/or poorly designed to adequately assess toxicity endpoints such as myocardial infarction, stroke, renal failure or fracture, the very endpoints that have driven the evolution of the NRTI-reducing strategy [27]. Improvements in surrogate biomarkers of bone or kidney tubular toxicity or BMD scans have only been seen specifically with regimens avoiding TDF. Thus, the dual regimen per se might not be as important as the discontinuation or avoidance of TDF, particularly when combined with a PI/r. Also, most studies have not collected all AE/toxicity-related biomarkers or events, thus preventing a full understanding of the potential risks and benefits of most NRTI-reducing regimens.
Cardiovascular outcomes are hardly ever reported. ABC was not commonly used in the triple comparator regimens, so the effect of its removal on potential cardiovascular events is not known. With limited evidence, CNS and/or cognitive function issues have not been seen associated with NRTI-reducing regimens.
Collectively, the available data appear to support the concept that avoiding TDF within combinations may be important in certain patient subgroups at risk of renal or bone toxicity, particularly when associated with a ritonavir-boosted PI or cobicistat-boosted PI. Benefits for bone and renal safety outcomes have also been demonstrated in na€ ıve and switch studies comparing TAF-based triple therapy co-formulations against TDF, with noninferior and superior efficacy in highand low-VL populations, respectively, in initial ART [58][59][60].
Another aim of any improved ART regimen is simplicity, which is strongly associated with maximum adherence. Unfortunately, pill burden was not usually lower in those receiving NRTI-reducing therapies, particularly when compared with the currently available, preferred single-pill regimens. The strong patient preference for single-pill regimens means that any NRTI-reducing regimen will not only need to be effective and safer than current SOC ART, but also co-formulated and once-daily.
We have observed that not all NRTI-reducing strategies are equally successful in switch. Two NRTI-reducing strategies have been successful in the switch scenario and demonstrated noninferiority: PI/r + 3TC and CAB or DTG + RPV. The ATLAS-M study is so far the only NRTI-reducing strategy (ATV/r plus 3TC) showing greater virological efficacy than the triple-drug comparator regimen containing ATV/r plus two NRTIs. Although noninferior, NRTI-reducing strategies have not been demonstrated to offer clear clinical benefits to the patient beyond simply preventing exposure to a third drug and potentially reducing costs.
Given the available data, PI/r plus 3TC strategies only apply to subjects already receiving triple regimens based on that particular PI/r (LPV/r, ATV/r or DRV/r) who would be maintained on that specific PI/r. Clearly, LPV/r is currently considered suboptimal and is not recommended by guidelines, and a triple regimen with elvitegravir/cobicistat plus either FTC/TDF or FTC/TAF has shown superiority in switch to the maintenance of PI/rbased triple strategies [49,58].
Favourable evidence has begun to accumulate from switch studies of INSTI (DTG or CAB) plus NNRTI (RPV) which yielded similar noninferior efficacy and tolerability to the triple therapy arm. DTG plus RPV is the only oral strategy that has demonstrated noninferior efficacy in a situation that approximates clinical practice, where a real control arm of subjects maintain their triple regimen with two NRTIs plus an INSTI, a PI/r or an NNRTI [57]. However, significant safety advantages have not been reported so far despite demonstration of reduced NRTI exposures in lifelong ART.
No completed study has compared a TAF-based tripledrug combination against an NRTI-reducing regimen. The currently ongoing GEMINI 1, GEMINI 2, LATTE, LATTE-2 and SWORD trials do not compare NRTI-reducing strategies against triple regimens containing TAF so cannot fully evaluate the question of whether these are better or equal alternatives to TAF. The concerns raised about increased pill burden with other NRTI-reducing strategies will be allayed by using once-daily co-formulated DTG/RPV in a single-pill regimen. Dual-drug strategies (CAB + RPV) given once-monthly, parenterally, in the ATLAS and FLAIR phase 3 studies are also being evaluated.
NRTI-reducing regimens containing cobicistat-boosted PI fixed-dose combinations are also required to evaluate modern PI co-formulations, particularly given the existence of differences in the trough concentration (C trough ) between DRV/r and DRV/c [61].
In an era where persistent virological suppression can be achieved in most patients with one-pill, once-daily regimens, with improved safety profiles, it is not easy to further simplify regimens by reducing pill burden, treatment frequency, and/or toxicity without jeopardizing efficacy or increasing the costs of HIV therapy.
The number of phase 3 NRTI-reducing studies is increasing. Long-term evidence from well-planned trials comparing NRTI-or drug-reducing regimens with preferred, contemporary, three-drug regimens in diverse patient populations is needed to define any potential benefits of NRTI-reducing strategies.
NRTI-reducing regimens will only be of major value when they offer all the potency of three-drug regimens in both na€ ıve and treatment-experienced patients, and also yield a clinically relevant reduction in toxicity relative to current three-drug ART, and in a single-pill format. No such NRTI-reducing regimen is currently recommended in guidelines, although DTG plus RPV is likely to be added to the list of potential switch/simplification regimens provided that long-term virological, clinical and patientreported outcomes, and financial considerations are favourable.