KEYNOTE-585 FAILS WHILE MATTERHORN SUCCEEDS IN GASTRIC CANCER: WHAT LESSONS CAN WE LEARN?

Takeaway summary:

In the article that accompanies this editorial, the KEYNOTE-585 trial demonstrated that adding pembrolizumab to perioperative chemotherapy with 5-FU and cisplatin in resectable esophagogastric cancer improved pathologic complete response rate but did not significantly improve event free survival over chemotherapy alone, despite trends toward improvement in both event free and overall survival. Although a negative trial, results from KEYNOTE-585 support and reinforce the positive results recently reported from the MATTERHORN Trial, which establishes a new care standard adding durvalumab to perioperative FLOT chemotherapy in resectable esophagogastric cancer.

Two decades of clinical trials have achieved global validation for adding systemic therapy to the surgical management of esophagogastric adenocarcinoma. Perioperative chemotherapy is increasingly being embraced as a care standard. Moreover, multiple contemporary trials have also indicted no clear contribution for the addition of radiation therapy to perioperative chemotherapy.^1-3

In the article that accompanies this editorial, Shitara and colleagues report final results from the phase 3 KEYNOTE-585 trial in resectable esophagogastric cancer.⁴ KEYNOTE-585 compared perioperative chemotherapy with 5-FU and cisplatin with or without the addition of pembrolizumab. The trial failed to meet the primary endpoint of improving event free survival (EFS) for the addition of pembrolizumab, despite a significant improvement in the co-primary endpoint of pathologic complete response rate. A numeric but non statistically significant improvement in EFS with the addition of pembrolizumab was hampered by multiple interim analyses. Given the failure of the primary endpoint of improved EFS in KEYNOTE-585, ultimately the linked co-primary endpoint of overall survival could not undergo formal statistical testing.

From the outset, a criticism of KEYNOTE-585 was the reluctance of the investigators to adopt the FLOT regimen combining 5-FU, oxaliplatin and docetaxel as the perioperative chemotherapy regimen. The investigators instead opted to use the two-drug regimen of perioperative 5-FU and cisplatin as the chemotherapy backbone, despite the clear inferiority of the similar ECF regimen (epirubicin, cisplatin, and 5-FU) compared to FLOT reported in the German FLOT 4 perioperative chemotherapy trial.⁵ FLOT 4 established the FLOT regimen as the superior and preferred perioperative chemotherapy regimen in resectable esophagogastric cancer. As KEYNOTE-585 was designed as a global trial, there were concerns about regional variations in treatment practice and the potential resistance to use of FLOT as the standard therapy arm. In Asia, for example, the doublet of a fluorinated pyrimidine plus platinum represents standard post operative adjuvant chemotherapy, without use of a taxane.^6,7

For KEYNOTE-585, in the 804-patient main trial cohort, an almost 20-month improvement in EFS (from 25.7 to 44.4 months) with the addition of pembrolizumab to perioperative chemotherapy did not achieve statistical significance (HR 0.81). The now reported overall survival (OS) in these patients also showed a numerically superior median OS for pembrolizumab plus chemotherapy, 71.8 months, versus chemotherapy alone, 55.7 months (HR 0.86), with a 5-year OS of 54% and 48% respectively. The trial was amended toward the end of accrual to incorporate an additional 203-patient FLOT cohort to be randomized to receive pembrolizumab or placebo. The pooled analyses of the main and FLOT patient cohorts did not alter the trial outcome for EFS superiority for pembrolizumab (HR 0.80), and there was no change in the outcome for the OS analysis (HR 0.86).

How do we place the negative results for perioperative pembrolizumab added to chemotherapy in the context of the MATTERHORN trial, recently reported as a positive trial?⁸ MATTERHORN treated all 948 patients with perioperative FLOT with or without the addition of monthly durvalumab. The trial achieved superiority for the stand-alone primary endpoint of improved EFS, with a 2-year EFS of 67.5% for durvalumab compared to 58.5% for chemotherapy alone (HR 0.71). Preliminary evaluation of the secondary endpoint of OS at 2 years trended superior for the durvalumab plus chemotherapy arm (75.7%) over chemotherapy alone (70.4%). Similar to KEYNOTE-585, a significant increase was seen in pathologic complete response rate with a gain of 12% with the addition of durvalumab. The positive results for MATTERHORN for the addition of durvalumab to perioperative FLOT chemotherapy have established a new standard of care for operable gastric cancer. The global reach of this trial argues for widespread adoption of this therapy.

Although cross-trial comparisons have limited value, FLOT has behaved consistently as a perioperative therapy across multiple contemporary clinical trials. As reported in the initial FLOT 4 trial establishing FLOT as superior to ECF, the median disease-free survival was 30 months with a 2-year OS 59%.⁵ The FLOT control arm behaved similarly in MATTERHORN with a median EFS of 32.8 months and a 2-year OS 70.4%.⁸ In a prior preliminary report of the outcome for the FLOT cohort from KEYNOTE-585, the median EFS was also similar to MATTERHORN at 30.9 months with a 2-year OS of 66%.⁹ Meanwhile, the experimental immunotherapy plus FLOT containing arms from both KEYNOTE-585 and MATTERHORN appear remarkably similar, with comparable 2-year rates of EFS (66-67.4%), 2-year OS (72%-75.7%) and rates of pathologic complete response (17%-19.2%).

Nonetheless, as written, KEYNOTE-585 is a negative trial for the addition of pembrolizumab to perioperative chemotherapy in gastroesophageal cancer. The flawed trial design and the failure to use the preferred, now global standard regimen of FLOT in all patients represent key deficiencies of the trial. However, the arguably comparable outcomes for FLOT plus pembrolizumab from KEYNOTE-585 and FLOT plus durvalumab from MATTERHORN further support the adoption of immunotherapy as part of perioperative chemotherapy. Although the final OS analyses of MATTERHORN are pending, the significant improvement in EFS and the substantial improvement in pathologic complete response, coupled with no adverse impacts on surgical outcome or therapy related toxicities, support adoption of this strategy.

These results reinforce that the use of a taxane-containing triplet perioperative regimen is critical. Also now mandated is the preoperative use of durvalumab. Interestingly, however, the addition of immunotherapy to chemotherapy following up-front surgery, without preoperative treatment, should not be pursued as a treatment strategy, as adjuvant nivolumab plus chemotherapy after up front gastrectomy did not improve relapse-free survival rates compared to placebo plus chemotherapy in ATTRACTION-5, a 755-patient trial.¹⁰

While Asian practice in the past has emphasized surgery first followed by adjuvant chemotherapy in patients with gastric cancer, recent trials from Asia argue for global adoption of pre and postoperative chemotherapy. The potential superiority of pre and postoperative chemotherapy over adjuvant chemotherapy alone was recently evidenced from updated results from the RESOLVE trial from China and the PRODIGY trial from Japan and Korea.^11,12 Therefore, perioperative FLOT, now with the addition of durvalumab, should and will be embraced globally as the standard of care in the management of operable esophagogastric cancer.

What are the next steps to build upon the new care standard of durvalumab and FLOT chemotherapy? The survival benefits for this new standard therapy over chemotherapy alone are likely to be modest and there is clear room for improvement. Further augmentation of the immune response with novel strategies requires investigation. Identifying biomarkers to identify which patients are most likely to benefit from therapy is critical. Patients at higher risk of recurrence, including those with poorer pathologic response to preoperative therapy and those having persistence of circulating tumor DNA after preoperative therapy and surgery, should lead to exploration of innovative treatment strategies. Promising emerging targets which have resulted in new drug development, including zolbetuximab in Claudin 18.2 positive patients and bemarituzumab in FGFR positive patients, should be further evaluated in the perioperative treatment setting.

Although KEYNOTE-585 is a negative trial as conducted, it supports the adoption of immunotherapy as part of perioperative chemotherapy in esophagogastric cancer. Moreover, results from KEYNOTE-585 will aid in the direction and design for future clinical trials.