Oncology development
When Should a Biomarker Change Cancer Treatment? Lessons from 12 Reports
Compare ALTAIR, KRYSTAL-10, EVOKE-03, IMvigor011 and eight other reports on cancer biomarkers, trial endpoints, treatment decisions, and patient access.
By Lampgen Clinical Research

6 min read · estimated
What the source reported
A biomarker should change a cancer treatment decision when the evidence supports acting on that result for a defined patient population. A blood test can flag molecular recurrence before a scan, and a drug can shrink more tumors, while the strategy still falls short on the trial’s primary outcome. The 12 reports below show what different kinds of evidence can establish, and what sponsors and sites can do to make the next trial answer the question patients are asking.
When earlier detection or higher response fell short
ALTAIR evaluated six months of trifluridine/tipiracil versus placebo in 243 people with resected colorectal cancer, detectable circulating tumor DNA (ctDNA) after standard care, and no radiographic evidence of disease. Median disease-free survival was 9.30 versus 5.55 months, but the prespecified primary test missed statistical significance (hazard ratio 0.79, 95% CI 0.60–1.05). Severe hematologic adverse events were more frequent with treatment. This result applies to the regimen and timing tested.
KRYSTAL-10 tested second-line adagrasib plus cetuximab against chemotherapy in previously treated KRAS G12C–mutated metastatic colorectal cancer. Response rates were 47% versus 16%, yet the Phase 3 study missed both primary endpoints. Median progression-free survival was 7.5 versus 8.1 months and median overall survival was 21.6 versus 21.7 months. The prior U.S. accelerated approval for the colorectal indication was withdrawn in 2026.
EVOKE-03 randomized 620 patients with previously untreated metastatic non-small cell lung cancer and a PD-L1 tumor proportion score of at least 50%. Adding sacituzumab govitecan raised response from 43.7% to 55.6% versus pembrolizumab alone. The progression-free survival analysis missed its prespecified significance threshold, and the interim overall survival analysis also missed significance. Grade 3 or higher treatment-related adverse events occurred in 55.7% versus 16.5%.
COMPANION-002 found that tovecimig plus paclitaxel improved response (18.0% versus 5.3%) and median progression-free survival (4.7 versus 2.6 months) in 168 people receiving second-line treatment for advanced biliary tract cancer. Overall survival did not differ significantly. Fifty-four percent of the control arm crossed over after progression, which complicates interpretation of the survival comparison.
When a treatment strategy improved a clinical outcome
IMvigor011 randomized 250 ctDNA-positive patients who had undergone cystectomy for muscle-invasive bladder cancer to atezolizumab or placebo. Median disease-free survival was 9.9 versus 4.8 months, and median overall survival was 32.8 versus 21.1 months (hazard ratio for death 0.59, 95% CI 0.39–0.90). The main readout was published in 2025. In May 2026, the FDA approved adjuvant atezolizumab for adults in this setting with ctDNA molecular residual disease identified by an FDA-authorized test.
SERENA-6 used serial blood testing to identify an emerging ESR1 mutation during first-line treatment of HR-positive, HER2-negative advanced breast cancer. In 315 randomized patients, switching from an aromatase inhibitor to camizestrant before radiographic progression, while continuing CDK4/6 inhibition, improved median progression-free survival to 16.0 months versus 9.2 months in the FDA analysis. The FDA granted accelerated approval in September 2026. Extended follow-up also favored the early switch on time to second progression or death. Final overall survival remains a confirmatory requirement.
SAFFRON enrolled patients with EGFR-mutated advanced non-small cell lung cancer and MET overexpression, amplification, or both after progression on osimertinib. In 338 randomized patients, osimertinib plus savolitinib improved median progression-free survival (7.6 versus 5.4 months) and overall survival (17.8 versus 12.9 months) compared with platinum plus pemetrexed. Grade 3 or higher adverse events occurred in 64% and 63%, respectively.
INTerpath-001 evaluated investigational intismeran autogene, a personalized mRNA neoantigen therapy, with pembrolizumab after resection of high-risk melanoma. Among 1,137 randomized patients, two-year recurrence-free survival was 82.2% with the combination versus 70.0% with placebo plus pembrolizumab, a 12.2 percentage-point difference. At the October 2026 ESMO readout, overall survival follow-up was still maturing.
What makes a biomarker useful in practice
FIND used ctDNA methylation testing to trigger earlier imaging after surgery for nonmetastatic colorectal cancer. In the 584-person analysis, clinical recurrence was detected a median 3.9 months earlier. Among patients whose cancer recurred, 48.1% received treatment with curative intent versus 23.6% under standard surveillance. Longer survival remains an open question. Moving the date of diagnosis earlier can lengthen measured survival from recurrence even when the date of death stays the same.
CEPHEUS illustrates why the assay threshold and timing matter. In 395 patients with newly diagnosed multiple myeloma who were ineligible for or deferred transplant, adding daratumumab increased measurable residual disease negativity from 39.4% to 60.9% at the 10⁻⁵ threshold. The expanded analysis also favored the combination at the more sensitive 10⁻⁶ threshold and for negativity sustained for at least a year. The threshold describes how deeply the assay can look. Repeated results show whether that depth of response was sustained.
CATALINA evaluated AI-derived scoring of tumor-infiltrating lymphocytes in early triple-negative breast cancer. It improved prognostic discrimination over a model using clinical and pathology factors alone. Once pathologists’ lymphocyte scores were included, adding the AI scores did not improve discrimination significantly. That sets a practical benchmark for future validation against existing workflows.
A JNCI Cancer Spectrum commentary examines the distance from a molecular match to actual treatment. Testing turnaround, access to a trial site, eligibility beyond the biomarker, and drug coverage can each determine whether a match leads to therapy. A screening funnel that tracks only who tests positive leaves that attrition invisible.
What this means for trial teams
Decide what evidence would change practice. Choose the patient population, comparator, treatment trigger, and endpoint together. Specify the primary analysis and how to handle treatment discontinuation, crossover, and subsequent therapy. Collect symptoms and function when they help characterize benefit alongside scans, recurrence, and survival.
Trace a result to a decision. Use the SIGNAL™ framework to investigate a concrete observation, such as reports arriving on time but sitting unreviewed. Compare the time from collection to lab receipt, report release, investigator review, and protocol action. Assign an owner for repeat delays and verify the next eligible result moved through the pathway within the study’s window.
Make access measurable. Track who had a usable sample, who received a result in time, who could complete the next visit, and who received the planned intervention. Review gaps by site or geography and, when lawful and feasible, by relevant demographic groups. Use documented reasons to see where transportation, language support, reimbursement, or local testing could change participation.
Protect follow-up and staff for the protocol. Continue endpoint follow-up after treatment discontinuation where the protocol and participant consent permit. An oncology CRA can verify the specimen and assessment trail, escalate missed windows, and confirm follow-through. A CTM can investigate recurring patterns across sites and vendors. Investigators retain the treatment decision.
Sources
- ALTAIR • Nature Medicine, June 2026
- KRYSTAL-10 • ESMO Gastrointestinal Cancers Congress, July 2026
- Adagrasib • FDA withdrawn cancer accelerated approvals
- EVOKE-03 • IASLC World Conference on Lung Cancer, September 2026
- COMPANION-002 • Journal of Clinical Oncology, October 2026
- IMvigor011 • New England Journal of Medicine, 2025
- IMvigor011 • FDA approval, May 2026
- SERENA-6 • FDA accelerated approval, September 2026
- SERENA-6 • The Lancet Oncology extended analysis, 2026
- Camizestrant • FDA approval letter and confirmatory requirement, September 2026
- SAFFRON • New England Journal of Medicine, October 2026
- INTerpath-001 • ESMO Congress, October 2026
- FIND • Journal of Clinical Oncology, 2026
- CEPHEUS • Blood Advances expanded MRD analysis, 2026
- CATALINA • The Lancet Oncology independent validation study, September 2026
- Treatment access after genomic testing • JNCI Cancer Spectrum, October 2026
- Oncology trial endpoints • FDA guidance
- Lampgen Clinical Insights • Planning biomarker screening and specimen workflows
- Lampgen Clinical Insights • SERENA-6 and the camizestrant approval
- Lampgen Clinical Insights • INTerpath-001 and personalized melanoma vaccine logistics