Table 3.

Landmark trials leading to FDA approvals for ICIs used to treat unresectable/metastatic, previously treated esophagogastric cancers

Trial	Study arm(s)	Study population*	Stratification factors	Outcome measures used for FDA approval†
KEYNOTE-059‡ NCT02335411 (phase II, single-arm, multicohort) Indication withdrawn	Pembrolizumab monotherapy, cohort 1 (n=143)§	Patients with tumors that are: Progressed after ≥2 prior lines of therapy, including prior 5-FU plus platinum doublet GEJ Siewert types II and III and G AC, HER2-negative or HER2-positive and previously treated with trastuzumab PD-L1 CPS ≥1 MSS or unknown MSI/MMR status	Not applicable	ORR: 13.3% (8.2, 20.0) DOR for responders: ranged from 2.8+ to 19.4+ mo; DOR ≥6 mo 58%; DOR ≥12 mo 26%
KEYNOTE-180 NCT02559687 (phase II, single-arm)	Pembrolizumab monotherapy (n=121)	Patients with tumors that are: ≥2 prior lines of therapy E SCC or AC, GEJ Siewert type I AC HER2 status not specified	Not applicable	ORR for ESCC and CPS ≥10: 20% (8, 37) DOR for ESCC and CPS ≥10: range 4.2 to 25.1+ mo; DOR ≥6 mo 71%; DOR ≥12 mo 57%
KEYNOTE-181 NCT02564263 (phase III, open-label)¶	Pembrolizumab monotherapy (n=314) vs ICC (paclitaxel, docetaxel, or irinotecan) (n=314)	Patients with tumors that are: 1 prior line of therapy ESCC or AC, GEJ Siewert type I AC HER2-negative for GEJ	Histology and geographic region (Asia vs rest of world)	Median OS for ESCC and CPS ≥10: 10.3 vs 6.7 mo (HR 0.64 [0.46 to 0.90]) Median OS ESCC: 8.2 vs 7.1 mo (HR 0.78 [0.63 to 0.96]; p=0.0095) Median OS CPS ≥10: 9.3 vs 6.7 mo; (HR 0.69 [0.52 to 0.93]; p=0.0074) Median OS all randomized: 7.1 vs 7.1 mo (HR 0.89 [0.75 to 1.05]; p=0.0560)
ATTRACTION-3 NCT02569242 (phase III, open-label)	Nivolumab monotherapy (n=210) vs ICC (paclitaxel or docetaxel) (n=209)	Patients with tumors that are: 1 prior line of therapy, including prior 5-FU and platinum E, GEJ SCC, adenosquamous	Geographical region (Japan vs rest of the world), number of organs with metastases (≤1 vs ≥2), PD-L1 expression (<1% vs ≥1%)	Median OS: 10.9 vs 8.4 mo (HR 0.77 [0.62 to 0.96]; p=0.0189) ORR (investigator-assessed): 19.3% (nivolumab) vs 21.5% (chemo) Median DOR: 6.9 mo (5.4 to 11.1) vs 3.9 mo (2.8 to 4.2)

CIs are 95% unless stated otherwise.

*No patients in the study populations had prior treatment with ICIs in earlier lines of therapy.

†Outcomes for the experimental arm are always presented first, when applicable.

‡The FDA approval for third-line pembrolizumab monotherapy for PD-L1-positive, unresectable/metastatic G/GEJ AC was withdrawn following an ODAC review of data from the phase II KEYNOTE-061 and KEYNOTE-062 studies along with consideration of the changing landscape of gastric cancer treatment.

§Total study enrollment for KEYNOTE-059 was n=259, however, the FDA approval in 2017 was based on outcomes observed for cohort 1 (n=143).

¶The FDA approval was based on results of both the phase II KEYNOTE-180 and phase III KEYNOTE-181 studies.

AC, adenocarcinoma; CI, confidence interval; CPS, combined positive score; DOR, duration of response; E, esophageal; ESCC, esophageal squamous cell carcinoma; FDA, Food and Drug Administration; 5-FU, 5-fluorouracil; G, gastric; GEJ, gastroesophageal junction; HR, hazard ratio; ICC, investigator’s choice of chemotherapy; MMR, mismatch repair deficiency; MSS, microsatellite stable; ODAC, Oncologic Drugs Advisory Committee; ORR, objective response rate; PD-L1, programmed death-ligand 1; SCC, squamous cell carcinoma.