Why did the KEYNOTE‑585 trial fail to meet its primary endpoint in resectable gastric or gastro‑oesophageal junction adenocarcinoma?

Medical Advisory BoardAll articles are reviewed for accuracy by our Medical Advisory Board
Educational purpose only • Exercise caution as content is pending human review
Article Review Status
Submitted
Under Review
Approved

Last updated: August 22, 2026 • View editorial policy

Have a follow-up question?

Our Medical A.I. is used by practicing medical doctors at top research institutions around the world. Ask any follow up question and get world-class guideline-backed answers instantly.

Why KEYNOTE-585 Failed to Meet Its Primary Endpoint

KEYNOTE-585 failed because adding pembrolizumab to perioperative chemotherapy in resectable gastric/gastroesophageal junction (GEJ) adenocarcinoma significantly improved pathological complete response (pCR) but this did not translate into meaningful event-free survival (EFS) or overall survival (OS) benefits. 1

The Trial Results

The KEYNOTE-585 study randomized 804 patients with locally advanced resectable gastric or GEJ adenocarcinoma to receive neoadjuvant and adjuvant pembrolizumab versus placebo, both combined with cisplatin-based chemotherapy 1. After a median follow-up of 47.7 months:

What Worked

  • Pathological complete response was dramatically improved: 12.9% with pembrolizumab versus only 2.0% with placebo (p<0.00001) 1
  • This represented a 10.9% absolute difference in pCR rates 1

What Failed

  • Event-free survival narrowly missed statistical significance: Median EFS was 44.4 months with pembrolizumab versus 25.3 months with placebo (HR 0.81, p=0.0198), but this did not meet the prespecified threshold of p=0.0178 1
  • Overall survival showed no benefit: Median OS was 60.7 months with pembrolizumab versus 58.0 months with placebo (HR 0.90, p=0.174) 1

Why the Disconnect Between pCR and Survival?

1. Biological Heterogeneity of Gastroesophageal Adenocarcinoma

The fundamental problem is that GEJ adenocarcinoma behaves differently from pure gastric adenocarcinoma, and immunotherapy benefits are consistently weaker in GEJ tumors:

  • In CheckMate 649, the HR for OS with nivolumab plus chemotherapy was 0.64 in gastric adenocarcinoma but only 0.82 in GEJ adenocarcinoma 2
  • In KEYNOTE-062, the HR was 0.81 for gastric cancer but 0.96 for GEJ cancer 2
  • In ATTRACTION-4, the HR was 0.87 for gastric cancer but 1.00 (no benefit) for GEJ cancer 2

This pattern suggests GEJ tumors are inherently less responsive to PD-1 blockade, likely due to different tumor microenvironments and immune evasion mechanisms 3.

2. Pathological Response Does Not Equal Cure in This Disease

Unlike some solid tumors where pCR strongly predicts survival, the 6-fold increase in pCR in KEYNOTE-585 did not prevent recurrence:

  • Even with a 12.9% pCR rate, nearly half of patients still experienced disease recurrence 4
  • This suggests that microscopic residual disease or micrometastases not detected at surgery remain the dominant driver of outcomes 3
  • Genomic heterogeneity between primary tumors and metastatic sites may explain why treating the primary tumor (even to pCR) doesn't prevent distant recurrence 3

3. Insufficient PD-L1 Selection

KEYNOTE-585 did not select patients based on PD-L1 expression, which is now known to be critical:

  • In the metastatic setting, pembrolizumab benefits are concentrated in patients with PD-L1 CPS ≥5 or ≥10 5, 6
  • The trial likely diluted efficacy by including many PD-L1-negative patients who derive minimal benefit from immunotherapy 2
  • Current guidelines strongly recommend PD-L1 testing before initiating immunotherapy 2

4. MSI-H Frequency Too Low

MSI-high tumors show dramatic responses to pembrolizumab (HR 0.38 for OS in CheckMate 649) 2, but:

  • MSI-high frequency is only 3-5% in gastric cancer and even lower in GEJ adenocarcinoma 2
  • KEYNOTE-585 likely had too few MSI-high patients to drive a survival signal 1

5. Timing and Duration Issues

The perioperative approach may not be optimal for immunotherapy:

  • In KEYNOTE-577 (esophageal cancer), adjuvant nivolumab after neoadjuvant chemoradiation showed an HR of only 0.87 in GEJ adenocarcinoma versus 0.61 in lower esophageal cancer 2
  • Immunotherapy may need to be given earlier, longer, or in different combinations to overcome the immunosuppressive effects of surgery and the tumor microenvironment 7

Clinical Implications

What This Means for Practice

Do not routinely add pembrolizumab to perioperative chemotherapy for resectable gastric/GEJ cancer based on KEYNOTE-585 alone 1. However:

  • For patients with MSI-high tumors, pembrolizumab should still be strongly considered given the dramatic benefits seen in other trials 2, 5
  • For patients with high PD-L1 expression (CPS ≥10), shared decision-making about adding immunotherapy is reasonable, acknowledging the pCR benefit but uncertain survival impact 6, 8
  • Standard perioperative FLOT chemotherapy remains the backbone for most patients 9, 2

Emerging Alternatives

The MATTERHORN trial (durvalumab plus FLOT) has shown improvements in both pCR and EFS, suggesting that different immunotherapy agents or combinations may succeed where KEYNOTE-585 did not 4. This remains investigational pending full publication.

Common Pitfalls to Avoid

  1. Do not assume pCR equals cure: The KEYNOTE-585 results prove that pathological response alone is insufficient in this disease 1

  2. Do not give immunotherapy without biomarker testing: PD-L1 CPS and MSI status are now essential for treatment selection 2, 8

  3. Do not extrapolate metastatic data to resectable disease: The biology and treatment responses differ significantly between settings 1, 7

  4. Do not ignore tumor location: GEJ adenocarcinoma consistently shows weaker immunotherapy responses than pure gastric cancer 2

Professional Medical Disclaimer

This information is intended for healthcare professionals. Any medical decision-making should rely on clinical judgment and independently verified information. The content provided herein does not replace professional discretion and should be considered supplementary to established clinical guidelines. Healthcare providers should verify all information against primary literature and current practice standards before application in patient care. Dr.Oracle assumes no liability for clinical decisions based on this content.