Table 2 Main efficacy outcomes at week 96

From: Cabotegravir and rilpivirine for treatment of HIV infection in Africa: week 96 results from the phase 3b randomized, open-label, noninferiority CARES trial

Outcome

Long-acting therapy (n = 255)

Oral therapy (n = 257)

Difference (95% CI)*, percentage points

Noninferiority margin (%)

Primary outcome

HIV-1 viral load level, no. (%)

    

<50 copies ml−1†

247 (97%)

250 (97%)

−0.4 (−3.1 to 2.0)

–10%

≥50 copies ml−1‡

4 (2%)

2 (1%)

0.8 (−0.7 to 3.2)

–

No virological data§

4 (2%)

5 (2%)

–

–

Primary outcome, sensitivity analyses

HIV-1 viral load level, no. (%)

    

<50 copies ml−1 (additional adjustment)

97%

97%

−0.4 (−3.4 to 2.5)

–

<50 copies ml−1 (per protocol)‖

220/224 (98%)

214/215 (>99%)

−1.3 (−4.2 to -0.1)

–

<50 copies ml−1 (complete case)

247/251 (98%)

250/252 (99%)

−0.8 (−3.3 to 0.7)

–

Secondary and other efficacy outcomes

    

HIV-1 viral load <200 copies ml−1, no. (%)

249 (98%)

251 (98%)

−0.01 (−2.4 to 2.4)

–

Confirmed virological failure, no. (%)**

4 (2%)

0

1.6 (0.4 to 4.2)

4%

Confirmed virological failure (per protocol), no. (%)‖

4 (2%)

0

1.6 (0.4 to 4.2)

–

Confirmed virological failure with ≥1 major acquired resistance mutation, no. (%)

3 (1%)

0

–

–

CD4+ cell count change from baseline, cells mm−3††

−7 ± 213

18 ± 228

−25 (−64 to 13)

–

  1. Data are n (%) or n/N (%) unless otherwise specified. All analyses of viral load suppression above or below threshold use the FDA snapshot algorithm and are done in the intention-to-treat exposed population, except where stated. Analyses of confirmed virological failure are done in the intention-to-treat exposed population unless otherwise stated. Analysis of change in CD4+ cell count uses complete case analysis. The widths of CIs have not been adjusted for multiplicity and cannot be used to infer treatment effects on secondary and other efficacy outcomes.
  2. *Differences are expressed in percentage points (except for change from baseline in CD4 count, where differences are expressed in cells per mm³). Estimates of difference and 95% CI for primary and secondary virological outcomes used the Cochran–Mantel–Haenszel-weighted Miettinen and Nurminen method, adjusting for third-drug class (INSTI or NNRTI) at screening; except for sensitivity analysis (additional adjustment) that used modified Poisson regression with unbiased sandwich estimate of variance; and except for analysis of confirmed virological failure for which no adjustment for third-drug class was performed. Estimate of difference and 95% CI for CD4 cell count change used Student’s -test. Estimated differences are for the long-acting therapy group minus the oral therapy group. For the primary outcome (viral suppression <50 copies ml−1), noninferiority is met if the lower limit of the 95%CI for the difference (shown in bold) lies above the prespecified noninferiority margin (criterion met). For the key secondary outcome (confirmed virological failure), noninferiority is met if the upper limit of the 95% CI for the difference (shown in bold) lies below the prespecified noninferiority margin (criterion not met).
  3. †In four participants from one site, the week 96 viral load test was delayed deliberately to coincide with a rescheduled injection visit (test done between weeks 103 and 109); to avoid a selective bias in one arm, these four viral load results were considered to be within the week 96 window period (prespecified as weeks 84 to 102).
  4. ‡Viral load ≥50 copies ml−1 is specified as a secondary outcome. Three of the viral load values ≥50 copies ml−1 at week 96 in the long-acting therapy group were values assigned due to earlier treatment switch for virological failure; all others were observed values from a test done during the week 96 window period.
  5. §Of the nine participants with no virological data within the window period, seven withdrew before week 96 for reasons other than adverse events and two died.
  6. Adjusted for third-drug class (INSTI or NNRTI) at screening, site and sex.
  7. ‖The per-protocol population excludes four participants in the long-acting therapy group and five in the oral therapy group who withdrew from the trial or died before week 96; 22 in the long-acting group with injection more than 14 days after the target date without oral bridging and a further five who missed more than 7 days of oral treatment; and 37 in the oral treatment group who missed more than 7 days of treatment at one or more visits.
  8. **Confirmed virological failure is defined as two consecutive viral load values ≥200 copies ml−1. One participant in the long-acting group had a viral load of 44,984 copies ml−1 at week 48 that could not be confirmed (participant died 9 days after the visit, before viral load test could be repeated, following an operation for strangulated umbilical hernia). The level of viremia and accompanying resistance profile were considered to have precluded resuppression on long-acting therapy, if a confirmatory viral load had been performed.
  9. ††CD4+ cell count data were available for 501 participants; missing values were due to withdrawal or death before week 96 (four in the long-acting group, five in the oral therapy group) and missed sample collection (one in each group).
Back to article page