Pembrolizumab Resets TMB-High Threshold to 13 Mut/Mb

The regulatory cutoff has changed

The FDA’s October 16, 2026, traditional approval resets the threshold for tumor mutational burden-high disease under pembrolizumab’s tumor-agnostic indication. TMB-high is now defined as at least 13 mutations per megabase (mut/Mb), rather than the 10-mut/Mb threshold associated with the earlier accelerated approval.

The distinction is clinically consequential. A result of 10, 11 or 12 mut/Mb no longer meets the biomarker definition for this indication solely by virtue of being above the previous cutoff. A result reported as exactly 13 mut/Mb does meet an “at least 13” threshold, subject to the current label’s other eligibility requirements and use of an appropriate test.

This is not simply a terminology update. TMB is a continuous measurement converted into a treatment-selection category by a regulatory cutoff. Moving that boundary changes which reports qualify, particularly those clustered near it. It also creates a reconciliation task for oncology practices, laboratories, payers and clinical pathways that may still encode TMB-high as 10 mut/Mb or greater.

The threshold applies to this specific tumor-agnostic pathway. It does not erase tumor-specific pembrolizumab indications or other biomarker-defined indications, such as those based on microsatellite instability, mismatch-repair deficiency or programmed death ligand 1 expression. A tumor below 13 mut/Mb may therefore remain eligible through a different labeled pathway, but not on TMB status alone under the revised definition.

What the earlier evidence established

The original accelerated approval was supported by a prospectively planned retrospective biomarker analysis of KEYNOTE-158, an open-label, multicohort, nonrandomized phase 2 study. Among 790 patients with evaluable TMB results, 102 had tumors with at least 10 mut/Mb and 688 had lower TMB. There was no randomized control arm for this comparison.

In the TMB-high group, the objective response rate was 29% (95% CI, 21% to 39%). The response rate was 6% (95% CI, 5% to 8%) among patients with < 10 mut/Mb. Median follow-up was approximately 37 months. Among responders in the TMB-high group, 57% had responses lasting at least 12 months and 50% had responses lasting at least 24 months.

Those results supported antitumor activity across selected previously treated solid-tumor cohorts, but they did not prove that selecting patients at a particular TMB threshold caused better survival. The analysis was nonrandomized, tumor types were unevenly represented, and the study was not designed to compare 10 mut/Mb directly with 13 mut/Mb as competing clinical cutoffs.

The 2026 FDA action is therefore the controlling source for the revised regulatory definition. The notification establishes the traditional-approval threshold, while the current prescribing information and FDA review materials should be used for the complete indicated population, test requirements and supporting efficacy analysis. Historical trial reports remain useful context but should not override the updated label.

Verifying a result before treatment

The first check is the number itself. The report should state TMB in mutations per megabase and show a value of at least 13. Descriptions such as “intermediate,” “elevated” or “high” are insufficient if the laboratory’s category was built around a different cutoff. The numerical result and the assay’s interpretive notes matter more than the adjective.

Values near the boundary require particular care. A reported value of 12.8 mut/Mb should not be independently rounded to 13 unless the validated assay and laboratory reporting policy explicitly support that interpretation. If a report displays only a rounded whole number, the treating team may need clarification from the laboratory about the underlying value, precision and classification rule.

The second check is the assay. TMB estimates can vary with panel size, genomic regions covered, sequencing depth, variant inclusion rules, germline filtering and bioinformatic methods. A threshold validated on one platform cannot automatically transfer to every laboratory-developed test. The report should identify the test, specimen type and method clearly enough to determine whether it is appropriate under the FDA label.

The third check is the specimen. Tissue adequacy, tumor content, fixation quality and sample age can affect sequencing performance. Clinicians should distinguish tissue TMB from blood-based TMB; the two are not interchangeable merely because both use mut/Mb. A plasma result should not be treated as satisfying a tissue-based requirement unless the label and an authorized test expressly support that use.

The fourth check is the clinical context. Confirm that the cancer is an advanced solid tumor covered by the tumor-agnostic indication and that the patient meets the label’s disease-setting requirements. Also review whether another biomarker or tumor-specific indication provides a separate basis for pembrolizumab. TMB should complement, not replace, assessment of histology, prior treatment, contraindications and available alternatives.

Operationally, institutions should update electronic order sets, molecular tumor board templates, payer authorization language and clinical decision support. Reports generated before the traditional approval may retain the older “TMB-high” designation at 10 mut/Mb. That legacy classification should not be copied forward without checking the underlying numeric value against the current threshold.

Important uncertainties remain

TMB is an imperfect, assay-dependent proxy for the number of tumor-derived alterations that could generate immunogenic neoantigens. It does not directly measure immune recognition or response to checkpoint blockade. Tumor lineage, mutation type, clonality, antigen presentation and the surrounding immune microenvironment may all influence benefit at the same numerical TMB.

Evidence from a single-arm biomarker analysis also has limitations. KEYNOTE-158 enrolled selected tumor cohorts rather than a population representative of every solid tumor, and some common cancers were not part of the efficacy dataset used for the original approval. Cross-tumor response estimates can obscure substantial variation by histology. Without randomization, differences between TMB groups cannot establish that TMB itself caused the observed response difference.

A sharper regulatory boundary does not eliminate measurement uncertainty around that boundary. Two assays may classify the same specimen differently, and intratumoral heterogeneity can produce different estimates from separate samples. Traditional approval provides a decision threshold, not proof that 12.9 and 13.0 mut/Mb identify biologically distinct populations.

Questions clinicians ask

Does a tumor reported at 10 mut/Mb still qualify as TMB-high?

Not under the revised tumor-agnostic definition of at least 13 mut/Mb. However, pembrolizumab could still be supported by another labeled indication based on tumor type, microsatellite instability, mismatch-repair deficiency, programmed death ligand 1 expression or other qualifying criteria. Check the current label rather than treating TMB as the only route to eligibility.

Can I round 12.8 mut/Mb up to 13?

Not independently. Eligibility should follow the validated assay’s reporting precision and classification rules. If the displayed value is rounded or the report does not disclose sufficient precision, ask the testing laboratory to clarify the underlying result and whether it meets the FDA-defined threshold; do not substitute informal arithmetic for the laboratory’s validated interpretation.

Is a blood TMB result interchangeable with tissue TMB?

No. Blood and tissue assays sample different material and use different analytic pipelines, and their numerical values are not automatically equivalent. Confirm that the specimen type and test match the current label and authorized testing framework before using a blood-based estimate to support treatment under this tumor-agnostic indication.

What should be rechecked on an older molecular report?

Review the exact numeric TMB, units, assay name, specimen type, test date and laboratory interpretation. An older report may call a tumor “TMB-high” at 10 mut/Mb because it used the accelerated-approval cutoff. Do not accept the adjective without confirming that the underlying value reaches at least 13 mut/Mb.

References

1. Oncology (Cancer)/Hematologic Malignancies Approval Notifications — US Food and Drug Administration, 2026 2. FDA Approves Pembrolizumab for Adults and Children With TMB-H Solid Tumors — US Food and Drug Administration, 2020 3. Study of Pembrolizumab in Participants With Advanced Solid Tumors (KEYNOTE-158) — ClinicalTrials.gov, 2026 4. Association of Tumour Mutational Burden With Outcomes in Patients With Advanced Solid Tumours Treated With Pembrolizumab: Prospective Biomarker Analysis of the Multicohort, Open-Label, Phase 2 KEYNOTE-158 Study — The Lancet Oncology, 2020