Phase 3 results with dual therapy cabotegravir/rilpivirine long-acting injections: ATLAS and FLAIR studies

Simon Collins, HIV i-Base

Two oral presentations at CROI 2019 showed that dual therapy with long-acting monthly injections resulted in very low levels of virological failure with high participant preference for injections compared to oral combinations.

Although results from cabotegravir/rilpivirine (CAB/RPV) long-acting injectable ART has included more than three years follow-up from phase 2 studies, these are the first phase 3 results.

The international randomised open-label ATLAS and FLAIR studies were therefore highly awaited and were presented in two consecutive oral presentations on the last day of the conference. [1, 2]

Both studies have a similar design, but with different participant characteristics for the control arms. Both are randomised, open-label, non-inferiority studies, based on 6% non-inferiority margin. Both have a primary endpoint of the percentage of participants with detectable viral load (>50 copies/mL) at week 48 by FDA snapshot analysis – which is unusual for reversing the more standard endpoint of the percentage of participants with viral suppression (which is now a key secondary endpoint).

After a 4-week induction with oral CAB plus oral RPV (30 mg/25 mg), the first intramuscular injection uses a 3 mL loading dose of CAB LA 600 mg + RPV LA 900 mg followed by 2 mL injections every four weeks from week 8 using a 400 mg/600 mg dose.

ATLAS study

The ATLAS study, in treatment-experienced participants, randomised 705 HIV positive people on stable PI- NNRTI- or INSTI-based ART plus 2NTRIs to either CAB/RPV long-acting (LA) injections (after 4 week induction) or to remain on their current oral ART. At week 48, participants in the control arm switched to the injections for follow-up to week-96.

The FLAIR study, in treatment-naive individuals, enrolled 629 participants who all started first-line ART with dolutegravir/abacavir/3TC for 20 weeks, before randomising those with undetectable viral load to either the same experimental strategy as ATLAS (4-week oral then switch to injections) or to continue on DTG/ABC/3TC. FLAIR continues for 96 weeks when participants in the control arm are able to switch to injections for extended follow-up.

The results from ATLAS were presented first, by Susan Swindells from University of Nebraska.

Baseline characteristics included median age 42 (range 18–82), with 26% older than 50. Approximately one third of participants were women with 68% white, 23% black and 9% other. Although median CD4 count was 653 cells/mm3, this ranged from 150 to >2500 cells/mm3. Median duration on ART was 4 years, and ranged from 1 to >20 years. BMI at baseline was 26 kg/m2 with a similarly wide range (from 15 to 58 kg/m2). Approximately half the participants were on NNRTI-based ART, with 33% on INSTI and 17% using PIs.

Overall, both groups had high levels of viral suppression with low discontinuations due to viral failure or missing data.

For the primary endpoints, there were very low rates of viral non-responders, with 1.6% vs 1.0% of participants having viral load >50 copies/mL at week 48 in CAB/RPV vs oral ART. This met criteria for non-inferiority with a difference of 0.6% (95% CI:  –1.2 to +2.5).

The secondary endpoint of treatment success (<50 copies/mL) was reported for 92.5% vs 95.5% for the same arms respectively with 5.8% vs 3.6% with missing data. These results numerically favoured the oral therapy arm. Only 3 vs 2 participants in the CAB/RPV vs oral therapy arms discontinued for lack of viral efficacy: and 2/3 in the CAB/RPV arm were later found to have had integrase inhibitor resistance at baseline. Although drug levels were lower than average in the cases of viral failure, median drug levels were consistently well above the protein adjusted IC90 at all time points.

Of the participants without data at week 48, 3.6% vs 1.3% (n=11 vs 4) in the CAB/RPV vs oral therapy arms discontinued due to side effects. The single death (in the oral therapy arm) was an overdose from methamphetamine and judged unrelated to study drugs.

Overall, drug related side effects were reported more frequently for the dual injections (29% vs 3%, nearly all grade 1 or 2): mainly fatigue, fever, headache and nausea (each reported at 4%) but only 3% vs 2% of participants discontinued due to side effects.

Injection site reactions were common following the first treatment (~70%, 98% grade 1 to 2) but reduced to approximately 20% by week 20 and 10% by week 48. Only four participants (1.3%) discontinued due to injection site reactions.

Finally, in the single question survey for patients in the dual therapy arm, 97% (266/273) preferred the injections to their previous oral combination.

FLAIR study

The FLAIR study was then presented by Chloe Orkin from Queen Mary University of London.

Baseline characteristics compared to ATLAS included younger age and lower BMI with a slightly lower percentage of women. Inclusion criteria included being HBV negative and not having NNRTI resistance (other than K103N).

Median age was 34 years (range 18–68), with 11% older than 50. Approximately one-fifth of participants were women with 74% white, 18% black and 8% other. Although median CD4 count was 444 cells/mm3 (IQR: 320 to 604), with 7% <200 cells/mm3. Median viral load was approximately 32,000 copies/mL with 20% >100,000 copies/mL. BMI at baseline was 24 kg/m2 with a similarly wide range (from 13 to 47 kg/m2).

Virological results were similar to ATLAS, with only 2.1% vs 2.5% reporting virology non-response (>50 copies/mL) at week 48 in the CAB/RPV vs oral therapy groups. This again met criteria for non-inferiority with –0.4% in favour of CAB/RPV (95%CI: –2.8 to +2.1). Viral suppression to <50 copies/mL was reported by 93.6% vs 93.3% respectively, with missing data for 4.2% in each group.

Again, there were few discontinuations for virological failure (1.4% vs 1.1%; n=4 vs 3). All 3/3 patients with genotype samples failed with both NNRTI and integrase resistance (two with Q148R and one with G10R – and all with L74I, which was present at baseline).

Drug exposure results were similar to those reported for ATLAS at all time points.

Side effects were very similar to ATLAS in range, frequency and severity with similar incidence of injection site reactions, also reducing over time.

In the single question survey after the study, 91% of participants preferred injections to oral treatment, with only 1% (n=2) preferring oral pills (and the remainder not responding).


Taken together, both studies show that for people who are interested in alternatives to daily tablets there were high rates of viral suppression and tolerable side effects.

There was also a very high level of overall satisfaction with injections, although this was a self-selecting criteria for people to enroll in these studies.


  1. Swindells S et al. Long-acting cabotegravir + rilpivirine as maintenance therapy: atlas week 48 results. CROI 4 – 7 March 2019, Seattle. Oral abstract 139. (abstract) (webcast)
  2. Orkin C et al. LONG-ACTING CABOTEGRAVIR+RILPIVIRINE FOR HIV maintenance: FLAIR week 48 RESULTS. CROI 4 – 7 March 2019, Seattle. Late breaker oral abstract  140LB. (abstract) (webcast)

Links to other websites are current at date of posting but not maintained.