DREAMseq: A phase III trial of treatment sequences in BRAFV600-mutant (m) metastatic melanoma (MM)—Final clinical results.

9506 Background: The DREAMseq trial compared efficacy and toxicity of the sequence of nivolumab/ipilimumab (N/I) followed by dabrafenib/trametinib (D/T) to the reverse sequence in patients (pts) with BRAFV600m MM. In 9/2021, with 59% of pts 2+ years (yr) from enrollment, the DSMC and NCI CTEP recommended halting the trial and releasing data that showed a 20% difference in 2-yr OS (72% vs 52%) favoring the N/I first sequence. Here we update data to median ~5 yr from entry and report secondary analyses including time to CNS relapse and percent unconfirmed responses (ucOR). Methods: Eligible pts with untreated BRAFV600m MM were stratified by ECOG Performance Status 0 or 1 and LDH, and randomized 1:1 to Step 1 treatment with either N/I (Arm A) or D/T (Arm B) and at disease progression (PD) were eligible for Step 2 alternate therapy, D/T (Arm C) or N/I (Arm D). Imaging was done at baseline and q12 weeks (wks). The primary endpoint was 2-yr OS. Secondary endpoints included: 3-yr OS, efficacy (PFS, ORR and DOR) and toxicity. Results: 267 out of 300 proposed pts were enrolled (135 Arm A; 132 Arm B). As of 7/23/24, median follow-up of 58 months (mo) (range:0-101), 30 pts had switched to Arm C and 52 to Arm D. 2-yr OS for those assigned to Arm A was 68.3% (95% CI: 60.8-76.9) and for Arm B 54.1% (95% CI: 46.1-63.7%) (log-rank p < 0.01). 3 and 5-yr OS by sequence and 2, 3 and 5-yr PFS for initial arms, and median PFS, ORR and DOR for all arms are shown in Table. There were 125 deaths (Arm A-C:47; Arm B-D:78). 76% of responders in Arm A and 24% in Arm B remain in response. At 12 wks, 59 pts on Arm A and 85 on Arm B had RECIST PR of which 10 (16.9%) and 35 (41.2%), respectively were ucOR by wk 24. CNS was the first site of PD in 24 pts on Arm A and 44 pts on Arm B. Median time to CNS PD: Arm A 12.2 mo (0.7-46.5); Arm B 8.4 mo (1.3-78.1) (p < 0.01). Conclusions: At nearly 5 yr median f/up, the N/I first treatment sequence continues to show superior efficacy over the D/T first sequence for treatment-naïve BRAFV600m MM with a near doubling (30% absolute difference) in 5-yr OS and a tripling of 5-yr PFS. While confirmed ORR were similar between Arms A and B, shorter DOR and more ucOR and more and earlier CNS PD were seen with initial D/T contributing to its worse efficacy. Clinical trial information: NCT02224781 . Secondary Endpoint (95% CI) Arm A to C (n=135) Arm B to D (n=132) Log-rank p 3 yr OS rate 65.6% (57.3, 73.9) 44.8% (36.7, 54.6) p<0.01 5 yr OS rate 63.3% (55.4, 72.3) 33.9% (25.9, 44.3) 2 yr PFS rate Arm A50.8% (42.8, 60.3) Arm B22.9% (16.5, 31.7) p<0.01 3 yr PFS rate 45.0% (37.0, 54.8) 15.9% (10.5, 24.0) 5 yr PFS rate 39.4% (31.3, 49.5) 12.8% (7.9, 20.7) Median PFS (mo) Arm A26.7 (11.2-47.3) Arm B8.5 (8.1-12.6) Arm C (n=30)11.2 (9.5, 22.3) Arm D (n=52)5.9 (2.9, 22.4) ORR Arm A (n= 132)51.5% (42.7, 60.3) Arm B (n=131)51.1% (42.3, 60.0) Arm C (n=30)70% (37.4, 74.5) Arm D (n= 52)46.2% (32.2, 60.5) Median DOR (mo) Arm A (n=68)Not reached Arm B (n=67)15.5 (11.2, 23.5) <0.01 Arm C (n=17)14.7 (8.2, NR) Arm D (n=20)45.2 (19.5, NR) p=0.03