OS Increases With Checkpoint Inhibitor Doublet in Advanced Colorectal Cancer

Eric X. Chen, MD

When durvalumab (Imfinzi) and tremelimumab were combined with best supportive care, the combination reduced the risk of progression or death by 28% compared with best supportive care alone, according to findings from a phase II clinical trial. 

Results from the randomized Canadian Cancer Clinical Trials Group (CCCTG) CO.26 trial presented at the 2019 Gastrointestinal Cancers Symposium demonstrated that the addition of the immune checkpoint doublet to best supportive care demonstrated a median overall survival (OS) of 6.6 months compared with 4.1 months with supportive care alone (hazard ratio [HR], 0.72; 90% CI, 0.54-0.97; P= .07). However, median progression-free survival (PFS) did not significantly differ between arms (1.8 months vs. 1.9 months; HR, 1.01; 90% CI, 0.76-1.34; P= .97).

“CCCTG study is the first study demonstrating the effectiveness of immune checkpoint blockade in [patients with] colorectal cancer unselected for mismatch repair deficiency,” said Eric X. Chen, MD, PhD, affiliate scientist, Princess Margaret Cancer Centre, Toronto.

Previously, treatment with immune checkpoint blockade therapy had not proven to be effective in refractory colorectal cancer except in patients with mismatch repair deficiency, Chen explained. However, the researchers hoped that targeting both PD-L1 with durvalumab and CTLA-4 with tremelimumab would have additive and/or synergistic activity as their mechanisms are nonredundant. In addition, a previous phase Ib study showed that the 2 agents could be combined safely.

A total of 180 patients with advanced refractory colorectal cancer and an ECOG performance status of 0 or 1 were randomized 2:1 to durvalumab plus tremelimumab with best supportive care or best supportive care alone. Eligibility criteria included patients with measurable disease for which all available standard regimens, such as bevacizumab (Avastin) and regorafenib (Stivarga), had failed, meaning that an EGFR inhibitor must have been used if the patient had RASwild-type disease. No prior therapy with PD-1/PD-L1 or CTLA-4 inhibitors was permitted.

Both arms received at least 1 prior systemic therapy, including 155 patients who had prior VEGF-targeting therapy, and 65 who had prior radiotherapy. All patients withRASwild-type colorectal cancer in the active treatment arm and 83% in the control arm were previously treated with cetuximab (Erbitux) or panitumumab (Vectibix).

The 119 patients randomized to the doublet regimen received 1,500 mg of durvalumab intravenously on day 1 of every 28-day cycle and 75 mg of tremelimimab intravenously on day 1 for the first 4 cycles.

OS served as the primary endpoint, while secondary endpoints included PFS, safety and toxicity, and objective response rate (ORR).

After a median follow-up of 15.2 months, the unstratified HR for OS was 0.72 in favor of durvalumab and tremelimumab (90% CI, 0.54-0.97) with aPvalue of .07 (a 2-sided&nbsp;Pvalue < .10 was considered statistically significant for this analysis). The unadjusted HR for OS was .70 (P= .03) also favoring active treatment.

All subgroups evaluated in the study also demonstrated superior OS with the doublet regimen, including those by primary tumor location andBRAFandRAS(KRAS, NRAS) status. In the 166 patients with microsatelite stable disease, the HR for OS was 0.66 (90% CI, .49-.89;P= .024) in favor of active treatment.

At baseline in the active treatment arm and the best supportive care arm, median patient age was 65 years and 64 years, respectively. In addition, 62% and 77%, respectively, were male, and liver metastases were present in 67% and 77% of patients. ECOG performance status was 1 in 72% of patients in both arms. The primary site of disease was left colon plus rectum in the doublet arm (73%) and the control arm (67%). The majority of patients in the active treatment arm and the control group hadKRASmutations (78% vs 49%, respectively),NRASmutations (2% vs 16%), andBRAFV600E mutations (7% vs 11%).

Overall, 169 patients had baseline blood samples available for molecular analysis, and 168 had successful circulating tumor DNA assessment based on baseline blood. Sequencing was performed using a 500-gene panel with 93.7% sensitivity and 99.2% specificity for detecting microsatellite instability. One patient in each arm had microsatellite instability-high/mismatch repair-deficient tumors.

Of the 103 patients evaluable for response in the active treatment arm, 1 patient (1%) experienced a partial response and 26 (22%) had stable disease. In the 46 evaluable patients in the best supportive care arm, 4 (7%) had stable disease as a best response. In the intent to treat population, the disease control rate was superior for the active treatment arm (odds ratio [OR], 4.16; 90% CI, 14.-12.3;P= .006). This superiority was consistent when the researchers restricted the analysis to patients with evaluable disease.

Tertiary endpoints of the study were quality of life (QOL) and the evaluation of correlative studies. Compliance with prespecified QOL survey evaluations decreased in both arms over time; however, compliance was lower in the control arm over the study period. By 12 weeks, compliance with evaluations was 56% in the doublet arm compared with just 13% in the best supportive care arm. The proportion of patients reporting deterioration in physical function or deterioration in global health status was not significantly different between the 2 arms at 8 or 16 weeks.

There were no new safety signals with the doublet. Grade 3/4 adverse events in the doublet arm were higher than the control arm, which included abdominal pain (7% vs 0%), fatigue (13% vs 3%), and lymphocytosis (23% vs 11%).

Discussant Michael J. Overman, MD, professor of gastrointestinal medical oncology, The University of Texas MD Anderson Cancer Center, Houston, acknowledged the sifnigicane of the investigators using tumor DNA to evaluate for microsatellite status. &ldquo;I think that&rsquo;s an excellent approach,&rdquo; he said. &ldquo;As a clinical investigator and in the gastrointestinal cancer community as a whole, we really have to actually obtain microsatellite status in all of our clinical trials. It&rsquo;s extremely hard to interpret low level activity without having knowledge in regards to microsatellite status on patients in immune-oncology gastrointestinal oncology trials.&rdquo;

The discordance between OS and PFS observed in CCCTG CO.26 has occurred in other trials as well, Overman noted. To be confident in results from trials with discrepant endpoints, trials should be large, double-blind, and placebo-controlled, he said, and none of these criteria were present in CCCTG CO.26, he added.

&ldquo;We believe that a confirmatory phase III study is warranted,&rdquo; Chen concluded.

Reference:

Chen EX, Jonker DJ, Kennecke HF, et al. CCTG CO.26 trial: A phase II randomized study of durvalumab (D) plus tremelimumab (T) and best supportive care (BSC) versus BSC alone in patients (pts) with advanced refractory colorectal carcinoma (rCRC). Presented at: 2019 Gastrointestinal Cancer Symposium; January 17-19, 2019; San Francisco. Abstract 481.