Treatment-Naive

DRIVE-FORWARD (P-018, N = 769)^5,6 and DRIVE-AHEAD (P-021, N = 728)^7,8 were similarly designed phase III, randomized (1:1), multi-centre, double-blind (DB), double-dummy, parallel-group, active-controlled, noninferiority trials; and included ART-naive patients with plasma HIV-1 RNA ≥ 1,000 copies/mL at screening. In DRIVE-FORWARD, the efficacy and safety of DOR (100 mg) once daily was compared with DRV/r (800 mg/100 mg) once daily, each given in combination with investigator-selected FTC/TDF 200 mg/300 mg, supplied as Truvada once daily, or ABC/3TC 600 mg/300 mg, supplied as Epzicom or Kivexa once daily. DRIVE-AHEAD was designed to evaluate the comparative safety and efficacy of DOR/3TC/TDF 100 mg/300 mg/TDF 300 mg once daily compared with efavirenz/FTC/TDF 600 mg/200 mg/300 mg (EFV/FTC/TDF, supplied as Atripla) once daily. The primary efficacy end point in both trials was the proportions of patients with HIV-1 RNA < 50 copies/mL assessed at week 48. The following secondary efficacy outcomes were also measured through week 96: the proportions of patients with HIV-1 RNA < and ≥ 50 copies/mL, and change from baseline in CD4 cell count. Notable safety end points included lipid profile and neuropsychiatric adverse events (AEs) including but not limited to dizziness, sleep disorders, and altered sensorium (e.g., depressed level of consciousness, lethargy, somnolence, or syncope).

In both trials, patients who met the entry criteria underwent randomization, which was conducted centrally using an interactive voice/Web response system (IVRS/IVWS) in a 1:1 ratio. Patients in DRIVE-FORWARD were stratified by HIV-1 RNA level at screening (≤ or > 100,000 copies/mL) and NRTI background therapy (FTC/TDF or ABC/3TC, as selected by the investigator); and patients in DRIVE-AHEAD were stratified by screening HIV-1 RNA (≤ or > 100,000 copies/mL) and hepatitis B and/or C co-infection status.

Both DRIVE-FORWARD and DRIVE-AHEAD had a total DB duration of 96 weeks (the base study), in which the first 48 weeks were used for the primary analyses. All eligible patients who provided consent to continue their treatment entered the open-label (OL) study extension, receiving DOR/3TC/TDF once daily or DOR once daily in combination with NRTI background therapy for an additional 96 weeks. Due to the minimal data available for the extension period, this review will be limited to the duration of the base study. Patients who met the criteria for protocol-defined virologic failure (PDVF) at any point during the study discontinued from the trial, regardless of the reason. The criteria for PDVF included having a confirmed (i.e., two consecutive measures at least one week apart) HIV-1 RNA ≥ 200 copies/mL at week 24 or week 36, confirmed HIV-1 RNA ≥ 50 copies/mL at week 48 (termed non-responders), or confirmed HIV-1 RNA ≥ 50 copies/mL after an initial response of HIV-1 RNA < 50 copies/mL at any time during the study (termed rebounders).

Treatment-Switch

DRIVE-SHIFT (P-024, N = 673)⁹ was a phase III, randomized (1:1), multi-centre, OL, parallel-group, active-controlled, noninferiority trial that included virologically suppressed patients (defined as HIV-1 RNA < 40 copies/mL) on a stable ARV regimen. This study evaluated a switch from a stable ARV regimen of a ritonavir- or cobicistat-boosted PI, or cobicistat-boosted integrase strand transfer inhibitor (InSTI), or NNRTI, each administered with two NRTIs, to DOR/3TC/TDF. Following a screening visit to determine the entry criteria, eligible patients were randomized in a 2:1 ratio to the immediate switch group (ISG) to receive DOR/3TC/TDF on study day 1 or to the delayed switch group (DSG) to start DOR/3TC/TDF at week 24. Patients in the DSG arm received their baseline regimens (described above) until week 24. Randomization was conducted centrally using an IVRS/IVWS; an allocation schedule was computer-generated.

The efficacy and safety outcomes were similar to the trials described above, and included the proportions of patients with HIV RNA < 50 copies/mL at week 48 (primary outcome), HIV RNA ≥ 50 copies/mL, CD4 cell count, lipid profile, and neuropsychiatric AEs at week 24 and 48. Due to the differences in DOR/3TC/TDF exposure between the treatment arms, the outcomes were compared between the ISG arm at week 48 and the baseline regimen at week 24 for the DSG arm (primary time point) as well as between the ISG arm at week 24 and baseline regimen at week 24 for the DSG arm (secondary time point).

The trial consisted of a 48-week base study, used for the primary analyses (described in Table 4). Patients who completed the base study and were eligible could enter into the OL study extension, receiving DOR/3TC/TDF once daily for up to 192 weeks total. Due to the minimal data available for the extension period, this review is limited to the duration of the base study. Patients who met the criteria for PDVF at any point during the study discontinued from the trial, regardless of the reason.

This CDR review will be limited to the base study period for each trial.

Inclusion and Exclusion Criteria

Baseline Characteristics

Baseline characteristics were generally similar between groups within trials (Table 5). The majority of the patients were male (> 80%). Patients who were treatment-naive were younger, with a mean age between 32 and 36 years, whereas patients who were treatment-experienced had a mean age of 43 years. Baseline CD4 cell counts were higher among treatment-experienced patients as they were on viral suppressive therapies at baseline. Approximately 20% of treatment-naive patients had a baseline plasma HIV-1 RNA level of > 100,000 copies/mL. The frequency of patients with AIDS ranged from 10% to 15% among treatment-naive patients, and approximately 15% to 18% among treatment-experienced patients.

Table 5

Summary of Baseline Characteristics.

Medical histories between-treatment groups within trials were largely similar; with the exception of gastrointestinal disorders occurring at a greater frequency in the DRV/r arm of DRIVE-FORWARD and psychiatric and immune system disorders occurring at a greater frequency in the ISG arm of DRIVE-SHIFT. The baseline regimens between-treatment groups in DRIVE-SHIFT were similar, and the majority of the patients received a ritonavir-boosted PI (~ 70%) and NNRTIs (~ 24%). A small but similar proportion of patients in each arm within the trials received lipid-lowering therapy.

Treatment-Naive

Both trials used a double-dummy design to maintain blinding, with matching placebo for active treatments. Patients in DRIVE-FORWARD were randomized (1:1) to receive DOR 100 mg or DRV/r (800 mg/100 mg), each given in combination with fixed-dose combination (FDC) FTC/TDF (200 mg/300 mg) or ABC/3TC (600 mg/300 mg), administered orally and once daily. DOR and ABC/3TC were taken without regard to food; whereas DRV/r and FTC/TDF were taken with food. Patients with hepatitis B were selectively given FTC/TDF.

Patients in DRIVE-AHEAD received oral once-daily dosages of either DOR/3TC/TDF (100 mg/300 mg/300 mg, plus placebo for EFV/FTC/TDF) or EFV/FTC/TDF (600 mg/200 mg/300 mg, plus placebo for DOR/3TC/TDF) FDC, as determined by patients’ 1:1 random assignment. DOR/3TC/TDF was taken at approximately the same time of the day without regard to food, whereas EFV/FTC/TDF was taken at bedtime on an empty stomach.

Treatment-Switch

Patients in DRIVE-SHIFT were randomized (2:1) to receive oral once-daily dosages of DOR/3TC/TDF immediately on day 1 in the ISG arm or continued their baseline regimen (ritonavir- or cobicistat-boosted PI, e.g., atazanavir, darunavir [DRV], or lopinavir, or cobicistat-boosted InSTI (e.g., EVG, or an NNRTI, e.g., EFV, NVP, or rilpivirine; each administered with two NRTIs) until they switched to DOR/3TC/TDF at week 24 in the DSG arm, administered at approximately the same time of the day without regard to food.

No rescue or supportive medications or dose modification of the study medication was allowed during the treatment period of any trial; with the exception of a recommended dosing interval adjustment of FTC/TDF in DRIVE-FORWARD to one tablet every 48 hours in patients with creatinine clearance of 30 mL/min to 49 mL/min. Unless specifically prohibited, all trials allowed patients to receive concomitant medications for their clinical conditions barring potential drug-drug interaction, e.g., oral or other hormonal contraception and new HCV treatments. Patients in DRIVE-SHIFT were allowed to initiate or modify lipid-lowering therapy. Notable prohibited medications included immune-therapy agents or other immunosuppressive therapy (except for short courses of corticosteroids, specialized treatment for Kaposi’s sarcoma, malignancy, and hepatitis), moderate or potent inducers of CYP3A, and medications that may interact with any of the study drugs.

Treatment-Naive

The primary efficacy outcome in DRIVE-FORWARD and DRIVE-AHEAD was the proportions of patients with HIV-1 RNA < 50 copies/mL at week 48, as determined by the FDA-defined snapshot algorithm. Under this approach, all missing data were treated as failures regardless of the reason.

Secondary outcomes of interest were identified in the review protocol (Table 3): virologic failure, CD4 cell count, drug resistance, adherence, and health-related quality of life (HRQoL). Virologic failure was defined as the proportions of patients with HIV-1 RNA ≥ 50 copies/mL at week 48, as determined by the FDA snapshot algorithm. Changes in CD4 cell count from baseline were estimated at week 48 (or most-recent screening visit if baseline values were missing). The magnitude and direction of the CD4 cell count was compared with the baseline value rather than a pre-established cutoff.

Genotypic and phenotypic resistance testing to the study drugs and backbone therapies were performed by a central laboratory. Data were summarized for patients who met the criteria for PDVF and those who discontinued the trial for any reason. Patients were classified as PDVF if they experienced a rebound in HIV-1 RNA ≥ 50 copies/mL at any visit after achieving HIV-1 RNA < 50 copies/mL, confirmed HIV-1 RNA ≥ 200 copies/mL at week 24 or 36, or (3) confirmed HIV-1 RNA ≥ 50 copies/mL at week 48.

Drug adherence was calculated based on a subjective rating method — the Study Medication Diary Cards — which was further validated by the study coordinator for completeness and accuracy. Per cent adherence was calculated by dividing the number of days a patient was “on-therapy” by the number of days the patient should be on therapy. Two definitions of an on-therapy day were used. Partial adherence was assigned to a study day if the patient took at least one tablet from any supplied bottle/container, whereas full adherence was assigned to a study day if the patent took the required number of tablets from each bottle/container. It should be noted that adherence was not regarded as an efficacy outcome, and data were therefore presented descriptively.

One relevant subgroup analysis was pre-planned in both study protocols: primary efficacy outcome by baseline viral load (< 100,000 copies/mL and ≥ 100,000 copies/mL). Neither trial assessed HRQoL.

Harms outcomes included the changes from baseline in fasting lipids (LDL, non-HDL, total cholesterol, HDL, and triglycerides) monitoring of all adverse events, and clinical and laboratory tests. An AE was defined as any unfavourable and unintended sign (including an abnormal laboratory finding), symptom, or disease as well as worsening of a pre-existing condition that is temporally associated with the study medication or procedure. A serious adverse event (SAE) was any AE that: resulted in death, was life-threatening, resulted in a persistent or significant disability/incapacity, resulted in or prolonged an existing in-patient hospitalization, was a congenital anomaly/birth defect, or was another important medical event. Any incidence of cancer or overdose-associated AE was also considered an SAE. The severity of laboratory AEs was based on the Division of Acquired Immunodeficiency Syndrome (DAIDS) criteria. In DRIVE-AHEAD, three categories of neuropsychiatric AEs (dizziness, sleep disorders, and altered sensorium, e.g., depressed level of consciousness, lethargy, somnolence, syncope) by week 48 constituted the primary safety end point, and an additional two categories (depression and suicide/self-injury and psychosis/psychotic disorders) were secondary safety end points.

The efficacy and safety assessments described above were carried out at various time points through week 96, and between-treatment comparisons were made at week 96 for the following outcomes: changes from baseline in CD4 cell count, and virologic suppression (HIV-1 RNA < 50 copies/mL) (tested for noninferiority and superiority).

Statistical Analysis for Efficacy End Points

All statistical tests were conducted at an alpha level of 0.05 (two-sided) unless otherwise indicated. In all three trials, the primary outcome (difference in proportions between-treatment groups and the associated 95% confidence interval [CI]) was calculated using the stratum-adjusted Mantel–Haenszel method, with the difference weighted by the harmonic mean of the sample size per arm for each stratum (screening HIV RNA ≤ 100,000 or > 100,000 copies/mL for DRIVE-FORWARD and DRIVE-AHEAD and PI use in baseline regimen for DRIVE-SHIFT). The remaining stratification factors, chronic hepatitis co-infection for DRIVE-AHEAD and the use of lipid-lowering therapy for DRIVE-SHIFT, were not expected to be associated with virologic response. Stratification by these factors was therefore not included in the analyses of virologic response.

The choice of noninferiority margin (NIM) in DRIVE-FORWARD and DRIVE-AHEAD was 10 percentage points, whereas the NIM was eight percentage points in DRIVE-SHIFT. The DOR arm in each trial was considered to be noninferior to the comparator arm if the lower bound of the two-sided 95% CI for the primary outcome (difference in the proportion of patients with HIV-1 RNA < 50 copies/mL) was greater than the chosen NIM in each trial. Provided noninferiority was established (DRIVE-FORWARD) or multiplicity criteria were satisfied (DRIVE-AHEAD and DRIVE-SHIFT), the treatment arm was tested for superiority to the respective comparators if the lower bound of the two-sided 95% CI for the difference in response rates was greater than zero.

The treatment difference in changes from baseline in CD4 cell count at time points of interest between the two treatment groups in all trials was estimated using the two-sample t-test. Genotypic and phenotypic resistance data from patients with PDVF and those who discontinued for any reason were summarized descriptively, provided they had blood samples available with HIV-1 RNA > 400 copies/mL.

Statistical Analysis for Safety End Points

The analyses of safety end points followed a tiered approach that varied by trial and with respect to the analyses that were performed. The list of safety parameters and analyses strategy for all trials is summarized in Table 6. This review focuses on the notable safety end points described in Table 3.

Table 6

Analysis Strategy for Safety Parameters.

Tier 1 safety events included the change from baseline in fasting LDL and non-HDL (all trials) and the proportion of patients with neuropsychiatric AEs in three pre-specified categories — dizziness, sleep disorders and disturbances, and altered sensorium (DRIVE-AHEAD only) — by week 48. The change from baseline in fasting lipids at week 48 (week 24 for DRIVE-SHIFT) was analyzed using analysis of covariance models adjusted by baseline lipids level (all trials) and the use of lipid-lowering therapy at study day 1 (DRIVE-SHIFT). In DRIVE-FORWARD, the superiority of DOR over DRV/r in LDL was demonstrated if the mean change from baseline was statistically significantly lower for the former, with a between-group P value < 0.04998 (P value adjusted for multiple statistical tests). The superiority for non-HDL was tested sequentially at the same alpha level. In DRIVE-SHIFT, after establishing the primary hypothesis of noninferior efficacy, the superiority of switching immediately to DOR over continuing baseline regimens was demonstrated if the mean change from baseline in LDL was statistically significantly lower for the former, with a one-sided between-group P value < 0.025. The superiority for non-HDL was tested sequentially at the same alpha level. In all trials, statistical testing was stopped with the first of these tests failing to reach statistical significance and no subsequent tests were considered for statistical significance.

The remaining fasting lipids (all trials) and neuropsychiatric AEs in the depression, suicide/self-injury, psychosis, and psychotic disorders categories (DRIVE-AHEAD) by week 48 were considered tier 2 events. All tier 1 and 2 neuropsychiatric AEs in DRIVE-AHEAD were analyzed using Miettinen and Nurminen’s method to generate the treatment difference and the associated 95% CI, and P values were provided for the tier 1 events only.

Missing Data

Missing values were of three types:

• Intermittent missing values due to a missed or skipped visit or due to an inadequate sample
• Non-intermittent missing values due to premature discontinuations because of treatment-related reasons such as clinical adverse experience, laboratory abnormalities, and withdrawal due to lack of efficacy (based on HIV-1 RNA results)
• Non-intermittent missing values due to premature discontinuations because of other reasons not related to treatment such as loss to follow-up, protocol violation, consent withdrawal, etc.

The two approaches used to handle missing efficacy values are summarized in Table 7. The primary approach was consistent with the FDA “snapshot” approach, in which all missing data were considered treatment failures regardless of reasons. All non-completers as well as those with an HIV-1 RNA measurement of ≥ 50 copies/mL were therefore considered virologic failures. Only patients with an HIV-1 RNA level of < 50 copies/mL within the pre-specified time window of the DB phase were classified as virologic successes. In DRIVE-FORWARD, patients were allowed to switch the study backbone NRTI regimens to manage toxicity; those who switched after week 2 and had HIV-1 RNA > 40 copies/mL at the time of switching were also regarded as a failure at all time points post-switching. The second approach was observed failure (OF), in which non-intermittent missing data for patients who prematurely discontinued treatment due to lack of efficacy were considered failures at time points thereafter. Patients with other reasons for missing data were excluded from the analyses.

Table 7

Approaches to Handling Missing Data.

For patients who had missing lipid data, the last observation following randomization (or before starting lipid-lowering therapy for those who started lipid-lowering therapy) was carried forward for later time points.

Sensitivity Analyses

A sensitivity analysis for the primary efficacy outcome was performed using the OF approach under which non-intermittent missing data for patients who prematurely discontinued their assigned treatment due to lack of efficacy were regarded as failures thereafter.

Sample Size Calculation

All three trials used an asymptotic method proposed by Farrington and Manning for power calculations. The sample size in DRIVE-FORWARD and DRIVE-AHEAD was chosen to provide 90% power to demonstrate noninferiority of DOR compared with the respective comparator at an overall one-sided 2.5% alpha level for the primary end point — the proportion of patients achieving HIV-1 RNA < 50 copies/mL at Week 48 — assuming a true response rate of 80% for both treatment arms using the FDA snapshot approach. This resulted in 340 patients in each group.

Power estimates for safety end points related to lipid profiles in DRIVE-FORWARD and DRIVE-AHEAD were based on data from a previous study (MK-1439 007, which studied four doses of DOR versus efavirenz, each in combination therapy with Truvada). It was estimated that with a sample size of 340 patients per treatment arm, the studies had > 99% power to detect a between-treatment difference of 0.43 mmol/L (7.7 mg/dL) and 1.11 mmol/L (20 mg/dL) for LDL and non-HDL, respectively.

In DRIVE-AHEAD, data from a previous study (MK-1439 007) was used to estimate the power to detect between-treatment differences in three pre-specified neuropsychiatric AEs: dizziness, sleep disorders and disturbances, and altered sensorium. The expected frequency for these AEs in the DOR and EFV arms, respectively, were 5% versus 24% for dizziness, 16% versus 29% for sleep disorders and disturbances, and 3% versus 10% for altered sensorium. With 340 patients in each treatment arm, the study had > 99%, 97%, and 95% power, respectively, to detect between-treatment differences in these tier 1 AEs if the proportions of patients with neuropsychiatric AEs were similar to those observed in study MK-1439 007.

The sample size in DRIVE-SHIFT was chosen to provide 80% power to demonstrate noninferiority of switching immediately to DOR compared with continuing baseline regimen at an overall one-sided 2.5% alpha level for the primary end point, the proportion of patients achieving HIV-1 RNA < 50 copies/mL, assuming a true response rate of 85% for both arms. This resulted in randomizing 660 patients in a 2:1 ratio between the immediate and delayed switch arm. The assumed response rate for the delayed switch arm was based on the result from a similar switch study (the SPIRIT study).

Power estimates for safety end points in DRIVE-SHIFT were based on data from the SPIRIT study, in which the estimated between-treatment differences in mean changes in fasting LDL and non-HDL were 0.89 mmol/L (16 mg/dL) and 1.16 mmol/L (21 mg/dL), respectively. With 440 patients in the ISG arm and 220 patients in the DSG arm, the study had an estimated > 99% power to detect a between-treatment difference of 0.89 mmol/L (16 mg/dL) and 1.16 mmol/L (21 mg/dL) for LDL and non-HDL, respectively.

Multiplicity

Multiple statistical testing was carried out in a hierarchical manner, as shown in Table 8. The efficacy and safety end points were tested in a sequential manner. Testing was stopped with the first of these tests failing to reach statistical significance and no subsequent tests were considered statistically significant. The overall one-sided type I error rate in testing these hypotheses was controlled at 2.5%.

Table 8

Statistical Testing Hierarchy for Multiplicity.

In DRIVE-FORWARD, no adjustment for multiplicity was made for the superiority test as noninferiority was confirmed if the data supported superiority due to the principles of closed testing. Two interim analyses were conducted, but these were unlikely to affect the type I error rate for the testing of the primary efficacy hypothesis or secondary safety hypotheses. The first interim analysis of neuropsychiatric AEs was done for making a program decision and was not related to any of the efficacy or safety end points. The second interim analysis was an efficacy analysis, but for the sole purpose of stopping the study in the event of a lack of efficacy. Notably, a small alpha level of 0.0001 was allocated to each interim analysis. Both the primary efficacy hypothesis and secondary safety hypotheses were tested at the two-sided alpha level of 0.049998.

In DRIVE-AHEAD, three interim safety analyses were carried out, and an alpha level of 0.0001 was allocated to each analysis. All safety hypotheses were tested at a one-sided alpha level of 0.02497. However, the primary efficacy hypothesis for noninferiority was tested at an unadjusted one-sided alpha level of 0.025 because no interim efficacy analysis was scheduled.

Subgroup Analyses

In DRIVE-FORWARD and DRIVE-AHEAD, results of the efficacy end points were reported by the subgroup relevant for this review — baseline HIV-1 RNA categories (HIV-1 RNA ≤ 100,000 copies/mL, HIV-1 RNA > 100,000 copies/mL). The estimates of between-group treatment difference were reported as a nominal 95% CI unadjusted for stratification factors. No interaction P values were reported. The OF approach was used to handle missing values in these subgroup analyses. Subgroup analyses by baseline viral load were not relevant in DRIVE-SHIFT as all enrolled patients were virologically suppressed.

Analysis Populations

In all included trials, the full-analysis set (FAS) was used as the primary population for the analyses of efficacy end points. The FAS population consisted of all randomized patients who received at least one dose of the study medication and had at least one measurement of the outcome (baseline or post-baseline). Patients in the FAS population were analyzed based on the treatment group to which they were randomized.

Safety analyses in all three trials were done in the “as-treated” population, consisting of all randomized patients who received at least one dose of study medication. Patients were included in the treatment group that corresponded to the medication they actually received.

The per-protocol (PP) population was used in DRIVE-FORWARD and DRIVE-AHEAD to analyze noninferiority of the key efficacy end points/virologic success by week 48. The PP population consisted of a subset of the FAS population that excluded patients with important deviations from the protocol by week 48, including non-compliance with study medication, discontinuation for reasons not related to treatment, and major protocol violations with the potential to affect efficacy.

Treatment-Naive

Overall, the treatment arms in each trial had comparable efficacy responses at week 48 (Table 11). Approximately 80% of patients achieved the FDA-defined snapshot algorithm of HIV-1 RNA < 50 copies/mL. Treatment differences in DRIVE-FORWARD and DRIVE-AHEAD were 3.9% (95% CI, −1.6 to 9.4) and 3.5% (95% CI, −2.0 to 9.0), respectively. In both cases, the pre-specified NIM of 10% was met, as the lower bound of the 95% CI for treatment differences was above the NIM of −10%. The secondary analyses (PP and sensitivity analysis using the OF approach) supported the primary analyses (Table 27). The proportions of patients with HIV-1 RNA ≥ 50 copies/mL using the FDA-defined snapshot approach were 11.2% versus 13.1% between DOR and DRV/r, respectively, in DRIVE-FORWARD, and 10.7% versus 10.2% between DOR/3TC/TDF and EFV/FTC/TDF, respectively, in DRIVE-AHEAD. Approximately 5% to 10% of the patients in the two studies had no virologic data at week 48. The proportion of patients with no virologic data in the EFV arm of DRIVE-AHEAD was noticeably higher compared with the DOR arm (9.1% versus 4.9%), mostly resulting from a disproportionately higher number of patients in the EFV arm discontinuing the study due to AEs or death.

Table 11

Virologic Response – Treatment-Naive Patients.

A similar pattern was seen at week 96 (Table 11), where a greater proportion of patients in the DOR arm achieved the FDA-defined snapshot algorithm of HIV-1 RNA < 50 copies/mL. The between-treatment differences in the two trials were 7.1% (95% CI, 0.5 to 13.7) and 3.8% (95%CI, −2.4 to 10.0); and sensitivity analyses using the OF approach supported these findings (Table 29). The proportions of patients with HIV-1 RNA ≥ 50 copies/mL using the FDA-defined snapshot approach were 17.2% versus 20.2% between DOR and DRV/r, respectively, in DRIVE-FORWARD, and 15.1% versus 12,1% between DOR/3TC/TDF and EFV/FTC/TDF, respectively, in DRIVE-AHEAD. The proportions of patients with no virologic data at week 96 ranged between 7% and 15%. More patients in the comparator arms of both trials had no virologic data compared with the DOR arm, 13.8% versus 9.8% and 14.3% versus 7.4% in DRIVE-FORWARD and DRIVE-AHEAD, respectively, mostly due to a disproportionately higher percentage of patients in the comparator arms discontinuing the studies.

Treatment-Switch

In DRIVE-SHIFT, more than 90% of patients across the treatment arms achieved the FDA-defined snapshot algorithm of HIV-1 RNA < 50 copies/mL (Table 12). The treatment difference between the ISG arm at week 48 and baseline regimen at study week 24 for the DSG arm was 3.8% (95% CI, −7.9 to 0.3), meeting the pre-specified NIM of 8%, as the lower bound of the 95% CI for the between-treatment difference was above the NIM. Of note, noninferiority was not met when data from the initial database lock (dated March 27, 2018) were used for the primary analysis. In this analysis, five patients in the ISG arm had missing HIV-1 RNA data by week 48 (week 24, n = 3; week 48, n = 2) and were analyzed as treatment failures according to the FDA snapshot approach. Based on this, 90.4% and 94.6% of patients achieved HIV-1 RNA < 50 copies/mL with a difference between groups of −4.2% (95% CI, −8.4% to −0.1%) (Appendix 4 Table 28). The NIM was only met following the addition of these missing data into the dataset. The secondary analysis (sensitivity analysis using the OF approach) supported the findings from the primary analyses (Appendix 4 Table 28). The treatment difference between the ISG and baseline regimen (both at week 24) was −0.9% (95%CI, −4.7 to 3.0).

Table 12

Virologic Response – Treatment-Experienced Patients.

The proportion of patients with HIV 1 RNA ≥ 50 copies/mL using the FDA-defined snapshot approach was < 3% across treatment arms at both time points. Patients with no virologic data at week 24 ranged between 3% and 5% in the ISG and DSG arm, and 7.6% in the ISG arm at week 48.

Treatment-Naive

The mean CD4 cell count at baseline ranged between 411 and 435 cells/mm³ across the trials (Table 13). In both trials, patients had an increase in CD4 cell count at weeks 48 and 96, regardless of treatment. The mean differences between the treatment arms at week 48 were 7.1 (95% CI, −20.8 to 35.0) and 10.1 (95% CI, −16.1 to 36.3) in DRIVE-FORWARD and DRIVE-AHEAD, respectively. At week 96, the respective treatment differences in the two trials were 17.4 (95% CI, −14.5 to 49.3) and 14.7 (95% CI, −18.7 to 48.2).

Table 13

CD4 Cell Count – Treatment-Naive Patients.

Treatment-Switch

The mean CD4 cell count at baseline ranged from 649 to 665 cells/mm³ in the two treatment arms (Table 14). The mean differences between the arms at both time points were comparable, albeit numerically greater at the primary time point. The treatment differences at the primary and secondary time points were −4.0 (95% CI, −31.6 to 23.5) and −12.8 (95% CI, −41.1 to 15.4), respectively.

Table 14

CD4 Cell Count – Treatment-Experienced Patients.

Treatment-Naive

Among the patients who met viral resistance-testing criteria, a smaller percentage underwent successful genotypic and phenotypic resistance-testing. Reasons for not performing resistance-testing include limited sample availability (< 400 copies/mL), and other site error (samples not collected [on time or at all] or sent).

Overall, the number of patients with PDVF and those who discontinued without PDVF increased with follow-up duration; the number of patients with a successful resistance test also increased over this period. Likewise, more patients in either arm within the trials developed resistance at week 96 compared with week 48. The incidences of resistance-associated mutations (RAMs), genotypic or phenotypic, were low across treatment arms (< 15 cases in any treatment group) in both trials, and the clinical expert consulted for this review shared this conclusion. There were more cases of genotypic RAMs than phenotypic RAMs, consistent with the notion that not all genotypic RAMs confer phenotypic resistance. Results are summarized in Table 15.

Table 15

Drug Resistance – Treatment-Naive Patients.

Treatment-Switch

One incidence of genotypic and phenotypic RAM against the background NRTI was found in the DSG arm prior to switching treatment at week 24. No other RAMs were found by week 48 in either arm (Table 16).

Table 16

Drug Resistance – Treatment-Experienced Patients.

Treatment-Naive

Data for per cent adherence are presented in Table 17 using the “full-compliance” definition, under which an “on-therapy” study day was reported if the patient took the required number of tablets from each supplied bottle/container. In each trial, while in the study, per cent adherence between the treatment arms was generally similar; more than 85% patients had an adherence of 90% or higher at both time points. No formal statistical test was completed for these end points.

Table 17

Treatment Adherence – Treatment-Naive Patients.

Treatment-Switch

Similar to the treatment-naive patients, while in the study, per cent adherence in DRIVE-SHIFT was high and generally similar between-treatment arms by week 48 (Table 18); adherence with the study medication regimen was ≥ 90% for most participants in the ISG and for the DSG before and after switching to DOR/3TC/TDF. No formal statistical test was completed for these end points.

Table 18

Treatment Adherence – Treatment-Experienced Patients.

Treatment-Naive

The overall frequency of AEs ranged between 82% and 94% across trials. The majority of these events were mild to moderate in intensity. The most common AEs across treatment groups included diarrhea, headache, upper respiratory tract infection (URTI), nausea, viral URTI, nasopharyngitis, pharyngitis, fatigue, back pain, bronchitis, cough, syphilis, upper abdominal pain, insomnia, dizziness, somnolence, abnormal dreams, and rash-related events. AEs that occurred at noticeably different frequencies between-treatment groups (≥ 5%) included diarrhea (17.0% versus 23.8%), URTI (13.3% versus 7.8%), and back pain (7.3% versus 2.9%) in DRIVE-FORWARD and dizziness (10.2% versus 38.2%), abnormal dreams (4.9% versus 12.1%), and rash-related events (7.1% versus 18.1%) in DRIVE-AHEAD.

Treatment-Switch

In the switch study, patients in the ISG arm experienced more AEs compared with the baseline regimen at week 24 for the DSG arm (68.9% versus 52.5%, respectively). The most common AEs between the treatment groups across time points included diarrhea, nasopharyngitis, back pain, and headache.

Lipid Profile

All three trials measured lipid profiles, of which changes from baseline in fasting LDL and fasting non-HDL at week 48 (DRIVE-FORWARD and DRIVE-AHEAD) or week 24 (DRIVE-SHIFT) were part of the statistical testing hierarchy. Overall, baseline mean levels for each type of lipid were balanced between the treatment groups across trials, with the exception of fasting triglycerides in DRIVE-AHEAD, which were higher in the DOR/3TC/TDF arm than in the EFV/FTC/TDF arm (199.5 mg/dL versus 123.0 mg/dL).

Among treatment-naive patients (Table 24), fasting LDL and non-HDL levels were decreased in the DOR arms and increased in their respective comparators arms at week 48 in DRIVE-FORWARD and DRIVE-AHEAD; the mean differences for change from baseline in fasting LDL between the treatment arms (95% CI) were −14.6 mg/dL (−18.1 to −11.1) and −10.0 mg/dL (−13.5 to −6.5), respectively; P < 0.0001 in both cases. For change from baseline in non-HDL, the mean differences between the treatment arms were −19.3 mg/dL (95% CI, −23.3 to −15.3) and −17.02 mg/dL (95% CI, −20.9 to −13.2), respectively; P value < 0.0001 in both cases. Between-treatment differences for fasting total cholesterol and fasting triglycerides followed a similar pattern, with decreases in DOR arms and increases in comparator arms. Changes from baseline in fasting HDL increased in both treatment groups, although a numerically greater increase occurred in the EFV/FTC/TDF arm in DRIVE-AHEAD. Data at week 96 reflected a similar pattern, where patients receiving DOR showed greater improvement in lipid profiles compared with the respective comparator arm (Table 25).

Table 24

Harms – Lipid Profile in All Patients (≤ 48 Weeks).

Table 25

Harms – Lipid Profile in Treatment-Naive Patients (96 Weeks).

Among treatment-switch patients, those in the ISG arm had a numerically greater decrease from baseline in fasting LDL and non-HDL at week 24 compared with the DSG arm; mean difference −15.3 mg/dL (95% CI, −18.9 to −1.6) and −23.9 mg/dL (95%CI, −28.1 to −19.6), respectively; no P value was reported in either case. All statistical comparisons related to lipid profile were done in a subset of patients on a ritonavir-boosted PI regimen (termed ritonavir-boosted PI strata), as the majority (approximately 70%) of the participants in DRIVE-SHIFT were on this regimen at baseline. At week 24, patients in the ISG arm had a statistically significantly greater change from baseline in fasting LDL and non-HDL compared with the DSG arm; between-treatment differences (95% CI) were −14.65 mg/dL (−18.92 to −10.38) and −23.03 mg/dL (−28.00 to −18.05), respectively; P value < 0.0001 in both cases. Data for the full available sample set are provided in Table 24. A similar pattern was found for fasting cholesterol, triglycerides, and HDL; patients in the ISG arm had a numerically greater improvements in lipid levels from baseline compared with the DSG arm.

Approximately 2% to 7% of patients in the three studies received (started, stopped, or modified dosage) lipid-lowering therapy. For all trials and time points, the between-treatment differences were small and not meaningful clinically or statistically (data not presented).

Neuropsychiatric Adverse Events

Neuropsychiatric AEs at week 48 were considered a primary safety outcome in DRIVE-AHEAD, and three categories were specifically analyzed in a pre-specified order of statistical hierarchy: dizziness, sleep disorders and disturbances, and altered sensorium. A statistically significantly lower proportion of patients in the DOR arm reported all three AEs compared with those in the EFV arm; between-treatment differences were −28.3 (95% CI, −34.0 to −22.5), −13.5 (95% CI, −19.1 to −7.9), and −3.8 (95% CI, −7.6 to −0.3) for dizziness, sleep disorders and disturbances, and altered sensorium, respectively. Two additional categories of neuropsychiatric AEs, depression and suicide/self-injury and psychosis and psychotic disorders, occurred at a numerically lower proportion in the DOR arm than in the EFV arm (Table 26).

Table 26

Neuropsychiatric Adverse Events in Treatment-Naive Patients (48 Weeks).

Data from 96 weeks showed a similar pattern for DRIVE-AHEAD patients, all five categories of neuropsychiatric AEs occurred at a lower frequency in the DOR arm compared with the EFV arm. The differences were most prominent for dizziness (10.2% versus 38.2%), sleep disorders and disturbances (14.0% versus 27.5%), and altered sensorium (5.2% versus 7.4%). Psychiatric disorders also occurred at a lower frequency in the DOR arm compared with the EFV arm (21.7% versus 37.4%) In DRIVE-FORWARD, the proportions of patients experiencing neuropsychiatric AEs and psychiatric disorders were largely similar between the groups.

Among treatment-switch patients, the overall rate of neuropsychiatric events was low. However, a greater proportion of patients in the ISG arm reported dizziness and sleep disorders compared with the DSG arm, both at week 24 prior to the switch to DOR and week 48 after the switch. Psychiatric disorders were also seen at a greater frequency in the ISG arm at both time points.

Altered Hepatic Enzymes

Measurements of hepatic enzymes showed an irregular pattern, depending on the grade of severity, treatment group, and trial. No one enzyme or severity grade was found to be consistently increased or decreased between-treatment arms across trials. Notably, the incidence of drug-induced liver injury, defined as having alanine transaminase or aspartate transaminase ≥ 3 × upper limit of normal range (ULN) plus bilirubin ≥ 2 × ULN and alkaline phosphatase < 2 × ULN, was low or lacking in all trials. Results are summarized in Table 22 and Table 23.

Table 22

Harms in Treatment-Naive Patients (96 Weeks).

Cardiovascular Disease or Events

The overall incidence of cardiovascular disease was low in all trials, at < 3% in each treatment group (Table 22 and Table 23).

Renal and Bone-Related Toxicity

The overall incidence of renal and urinary disorders was low in all trials, at < 5% in each treatment group (Table 22 and Table 23).

Skin Disorders

Data for skin and subcutaneous tissue disorders as an organ class are reported here, and summarized in Table 22 and Table 23. Additionally, all rash-related events were reported as a composite outcome, including rash, erythematous rash, follicular rash, genital rash, generalized rash, macular rash, maculopapular rash, papular rash, pruritic rash, pustular rash, vesicular rash, and viral rash.

Skin disorders and rash-related events occurred at a similar frequency between-treatment arms in DRIVE-FORWARD. In DRIVE-AHEAD, the frequency of skin disorders and rash-related events was noticeably lower among patients receiving DOR compared with those receiving EFV. In DRIVE-SHIFT, the overall incidence of skin disorders and rash-related events was low, with generally similar frequency between-treatment arms.