Study design
An adjusted indirect comparison of CAB + RPV LA Q8W with daily oral SoC was conducted according to the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) guidelines on good research practices for indirect treatment comparisons (18). To date, the only trial of CAB + RPV LA Q8W is ATLAS-2M which compared this regimen with CAB + RPV LA Q4W; the available trials that compare CAB + RPV LA with daily oral SoC maintenance treatments have used a Q4W CAB + RPV LA regimen (ATLAS and FLAIR). Thus, the evidence base for this analysis of CAB + RPV LA Q8W versus daily oral SoC was necessarily limited to the ATLAS, FLAIR, and ATLAS-2M trials (14, 15, 17). An anchored Bucher’s frequentist adjusted indirect treatment comparison was used for this analysis of CAB + RPV LA Q8W and oral daily SoC, with CAB + RPV LA Q4W as the common comparator (19). Since combining studies should only be considered if they are clinically and methodologically similar, we employed the PICOS approach to compare trials in terms of population (P), intervention (I), comparator (C), outcome (O), and study design (S) (Table 1). The ATLAS, FLAIR, and ATLAS-2M trials were all 1:1 randomized, open-label studies, evaluating non-inferiority in efficacy after switching to the intervention of interest in a virologically suppressed population at Week 48 with primary and secondary endpoints being identical between the trials included in the analysis (Table 1).
Table 1
Summary of study characteristics of the ATLAS, FLAIR, and ATLAS-2M trials
Trial acronym (Author year) | Study design | Countries/region | Participant eligibility criteria | Intervention/ comparator | Endpoints |
ATLAS (Swindells 2020) (15) | Randomized, multicenter, parallel-group, open-label, Phase 3 | Argentina Australia Canada Europe Mexico Republic of Korea Russian Federation South Africa USA | • On initial or second ARV regimen ≥ 6 months prior to screening • Prior switch only for tolerability/safety, access to medications, or convenience/ simplification, and NOT due to treatment failure • HIV-1 RNA < 50 copies/mL at screening and ≥ 2 HIV-1 RNA measurements < 50 copies/mL in the 12 months prior to screening | • CAB + RPV Q4W • Continued cART SoC therapy | • Primary: % participants with HIV-1 RNA ≥ 50 copies/mL at Week 48 • Key secondary: participants with HIV-1 RNA < 50 copies/mL at Week 48 |
FLAIR (Orkin 2020) (14) | Randomized, multicenter, open-label, noninferiority, Phase 3 | Canada Europe Japan Russian Federation South Africa USA | • Treatment naïve at screening • 20 weeks induction period with DTG/ABC/3TC • HIV-1 RNA < 50 copies/mL prior to randomization | • CAB + RPV Q4W • DTG + ABC + 3TC | • Primary: % participants with HIV-1 RNA ≥ 50 copies/mL at Week 48 • Key secondary: participants with HIV-1 RNA < 50 copies/mL at Week 48 |
ATLAS-2M (Overton 2020) (17, 35) | Randomized, multicenter, parallel-group, open-label, noninferiority, Phase 3 | Argentina Australia Canada Europe Mexico North America Republic of Korea Russian Federation South Africa USA | • Uninterrupted SoC regimen ≥ 6 months • 2 documented HIV-1 RNA viral load measurements < 50 copies/mL in each past 2, 6-month periods and at study entry | • CAB + RPV Q4Wa • CAB + RPV Q8Wa | • Primary: % participants with HIV-1 RNA ≥ 50 copies/mL at Week 48 • Key secondary: participants with HIV-1 RNA < 50 copies/mL at Week 48 |
aParticipants with prior CAB + RPV LA exposure were excluded from the current analysis |
3TC, lamivudine; ABC, abacavir; ATLAS, Antiretroviral Therapy as Long-Acting Suppression; ATLAS-2M, ATLAS every 2 months; CAB, cabotegravir; cART, combination antiretroviral therapy; DTG, dolutegravir; FLAIR, First Long-Acting Injectable Regimen; LA, long-acting; HIV-1, human immunodeficiency virus type 1; PI, protease inhibitor; Q4W, every 4 weeks; Q8W, every 8 weeks; RNA, ribonucleic acid; RPV, rilpivirine; SoC, standard of care |
Participants and treatments
Eligible participants from the ATLAS, FLAIR, and ATLAS-2M trials were all treatment-experienced individuals with virologically suppressed HIV-1 infection, and a viral load of < 50 HIV-1 ribonucleic acid (RNA) copies/mL. Participants from FLAIR, ATLAS and ATLAS-2M received oral lead-in therapy with 30 mg CAB + 25 mg RPV once daily for four weeks before long-acting injectable was started to mitigate the risk of any serious side-effects associated with CAB or RPV before administering long acting injectable regimen. Participants in the ATLAS and ATLAS-2M trials had initiated treatment with any standard daily oral SoC prior to the switch to CAB + RPV LA (14, 15, 17). Some participants in ATLAS-2M had prior CAB + RPV LA Q4W exposure; these participants were excluded from the current analysis, aiming for similarity in baseline participant characteristics as per ISPOR guidelines.
A pooled analysis of the ATLAS and FLAIR trials was previously conducted (16), and no significant heterogeneity between the two trials in terms of trial or participant characteristics for the treatment effect was found. Pooled data from ATLAS and FLAIR studies (n = 591 per treatment group), and data from participants with no prior CAB + RPV exposure (n = 327 per treatment group) in the ATLAS-2M trial, were used to inform this analysis.
In addition, aiming to further satisfy the similarity assumption as per ISPOR guidelines, we explored comparability of participants’ demographics and clinical characteristics at study entry between trials. Participants in the FLAIR trial had DTG-based induction therapy prior to the switch to CAB + RPV LA (14). Pooled data from the ATLAS and FLAIR trials, created overall data for CAB + RPV LA Q4W similar to the CAB + RPV LA Q4W arm of the ATLAS-2M trial (Fig. 1).
A summary of study population demographics and characteristics at baseline for the pooled ATLAS/FLAIR, and ATLAS-2M trials are shown in Table 2. Baseline characteristics between the Q4W arms of the pooled ATLAS/FLAIR and the ATLAS-2M trials showed no significant differences for sex, race, ethnicity, and body mass index ([BMI]; Table 2). A significant difference between treatment groups was found for both median age (ATLAS/FLAIR: 38 years; ATLAS-2M: 41 years; P = 0.0002) and mean CD4 counts (ATLAS/FLAIR: 670.2, ATLAS-2M: 741.0; P = 0.0004); however, as per ISPOR guidelines on assessing similarity for indirect treatment comparisons, although researchers may be able to use statistical information they must rely primarily on clinical judgment of whether differences among studies may affect the comparisons of treatments or make some comparisons inappropriate (18). A mean difference (MD) in age of 3 years between treatment groups, although statistically significant, is very unlikely to be clinically relevant. Additionally, both ATLAS/FLAIR and ATLAS-2M participants had high CD4-cell count at baseline, both are within the range of a normal CD4-cell count, and therefore the differences are not considered clinically relevant. In addition, it was observed in the individual studies (ATLAS/FLAIR/ATLAS-2M) that efficacy outcomes per age group and CD4-cell count were consistent with the overall population, suggesting these factors were not effect modifiers in the clinical trials used in this comparison. The combined data for the ATLAS/FLAIR trials showed 65% of participants received INSTI treatment and 26% received NNRTIs at baseline; in contrast, data for the ATLAS-2M trial showed 42% of participants received INSTI treatment and 47% received NNRTIs at baseline (Table 2). Due to the marked difference between studies in baseline third agent, a subgroup analyses of virologic suppression and lack of virologic suppression stratified by baseline third active drug class were conducted to validate the findings of the main analysis.
Table 2
Study population demographic characteristics at baseline for the ATLAS, FLAIR, pooled ATLAS/FLAIR and ATLAS-2M trials
| ATLAS/FLAIR (14–16) | ATLAS-2M (17) | Statistical comparison |
| SoC (n = 591) | CAB + RPV Q4W (n = 591) | CAB + RPV Q4Wa (n = 327) | CAB + RPV Q8Wa (n = 327) | ATLAS/FLAIR vs ATLAS-2M CAB + RPV Q4W |
Age (years), median (range) | 38 (18, 82) | 38 (19, 74) | 42 (19, 67) | 41 (20, 83) | P = 0.0002b |
Male, n (%) | 423 (72) | 429 (73) | 252 (77) | 254 (78) | P = 0.1563c |
Race, n (%) White Black/African American Asian Other | 408 (69) 133 (23) 28 (5) 20 (3) | 430 (73) 109 (18) 34 (6) 18 (3) | 256 (78) 45 (14) 12 (4) 14 (4) | 238 (73) 57 (17) 17 (5) 15 (5) | P = 0.0955d |
Ethnicity Hispanic/Latino, n (%) | 74 (13) | 63 (11) | 42 (13) | 54 (17) | P = 0.3309c |
BMI (kg/m2), mean (SD) | 25.9 (5.4) | 25.7 (4.8) | 26.2 (5.2) | 26.2 (5.1) | P = 0.1451b |
CD4-cell count (cells/mm2), mean (SD) | 670.2 (273.2) | 672.7 (264.3) | 741.0 (288.8) | 688.6 (266.0) | P = 0.0004b |
CD4-cell count (cells/mm2), n (%) < 350 ≥ 350 to < 500 ≥ 500 | 54 (9) 117 (20) 420 (71) | 42 (7) 120 (20) 429 (73) | 17(5) 49(15) 261(80) | 25(8) 60(18) 242(74) | P = 0.0526d |
Baseline third active class, n (%) INSTI NNRTI PI | 382 (65) 155 (26) 54 (9) | 385 (65) 155 (26) 51 (9) | 141 (43) 156 (48) 30 (9) | 136 (42) 151 (46) 40 (12) | P < 0.0001d - |
aParticipants with prior CAB + RPV LA exposure were excluded |
bWilcoxon Rank Sum test |
cFisher's exact test |
dChi-square test |
All statistical tests compare Q4W groups in ATLAS/FLAIR and ATLAS-2M. |
ATLAS, Antiretroviral Therapy as Long-Acting Suppression; ATLAS-2M, ATLAS every 2 months; BMI, body mass index; CAB, cabotegravir; cART, combination antiretroviral therapy; FLAIR, First Long-Acting Injectable Regimen; HIV-1, human immunodeficiency virus type 1; INSTI, integrase strand inhibitor; IQR, interquartile range; m, months; NA, not applicable; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; Q4W, every 4 weeks; Q8W, every 8 weeks; RNA, ribonucleic acid; RPV, rilpivirine; SD, standard deviation; SoC, standard of care |
This study is a statistical analysis of data from the ATLAS, FLAIR and ATLAS-2M trials and therefore does not require approval from an institutional review board or ethics committee. All trials included in this study were conducted in accordance with the Declaration of Helsinki (20). All participants provided written informed consent and the protocols were approved by an institutional review board or ethics committee from each study site.
Study measures
The most frequently reported safety and efficacy outcomes were selected for quantitative comparison as follows: treatment efficacy was assessed for the primary study endpoint, the estimated proportion of participants with HIV-1 RNA ≥ 50 copies/mL according to the FDA snapshot algorithm at Week 48. Other measures of treatment efficacy were assessed as key secondary endpoints: Week 48 virologic suppression (HIV-1 RNA < 50 copies per mL) as per FDA snapshot algorithm, CD4-cell count changes from baseline, and patients with no virologic data at Week 48. Safety was assessed in terms of discontinuations due to adverse events (AEs), and overall AEs, serious AEs, and Grade 3–5 AEs excluding injection-site reactions (ISRs) at Week 48. AEs were analyzed excluding ISRs to compare the events more likely related to systemic exposure to the respective drugs, rather than primarily to the route of administration. In the ATLAS/FLAIR and ATLAS-2M studies, ISRs for both Q4W and Q8W dosing were mostly mild (Grade 1 or 2) and the frequency progressively reduced to Week 48 among study participants, resulting in very few (< 2%) discontinuations, which are included in the discontinuations due to AEs analysis. Comparative efficacy of virologic suppression and lack of virologic suppression were stratified by baseline third active drug class (INSTI, NNRTI, and protease inhibitor [PI]) in a subgroup analysis to assess the relative effects of each treatment class and whether any statistically significant differences that can impact overall outcomes exist.
Statistical analyses
Assessment of the similarity of population demographic characteristics at baseline and effect of the baseline third active drug class on efficacy, determined by the percentage of < 50 HIV-1 copies/mL in the Q4W arms of the pooled ATLAS/FLAIR trials and ATLAS-2M was carried out using the Wilcoxon Rank Sum Test and the Fisher's Exact Test. The robustness of comparative efficacy conclusions was assessed using alternate summary statistics. Between treatment comparison statistics for ATLAS/FLAIR (Q4W vs SoC) and for ATLAS-2M (Q8W vs Q4W) are shown in Supplementary Table S1.
The statistical methodology published by Bucher et al (19) was then used to calculate relative risk (RR), odds ratio (OR), and risk difference (RD) for the comparison of CAB + RPV LA Q8W versus daily oral SoC, using CAB + RPV Q4W as the common comparator. In this method, a summary statistic of the comparison is expressed on a continuous scale and assumed to have a normal distribution. The MD and RD scales are assumed to be and were normally distributed. The OR and RR scales were transformed to a normally distributed scale using the natural log transformation (lnOR or lnRR). Thus, we have:
![](https://myfiles.space/user_files/58891_537ccc6497188def/58891_custom_files/img1599877291.png)