Diffuse large B-cell lymphoma mutation landscape

What is altered, by cohort

One row per alteration, not per gene: a gene that is amplified and rarely mutated (ERBB2, MYCN, EGFR) gets a row for each. Every cell divides by its own denominator — the patients in that cohort on whom that gene could be called. Copy-number rows are shown only where at least one cohort reaches 2%.

observed — shade scales with frequency, full at 30% assayed, none found not on this cohort's panel cohort not readable

Key findings

KMT2D is mutated in 247 of 1001 patients in Diffuse Large B-Cell Lymphoma (Duke, Cell 2017).
Numerator: 247 · Denominator: 1001 · Frequency: 24.68% · Observed in 2 cohorts · Confidence: moderate · Source: dlbcl_duke_2017 · Retrieved: 2026-09-18

MYD88 is mutated in 172 of 1001 patients in Diffuse Large B-Cell Lymphoma (Duke, Cell 2017).
Numerator: 172 · Denominator: 1001 · Frequency: 17.18% · Observed in 2 cohorts · Confidence: moderate · Source: dlbcl_duke_2017 · Retrieved: 2026-09-18

PIM1 is mutated in 146 of 1001 patients in Diffuse Large B-Cell Lymphoma (Duke, Cell 2017).
Numerator: 146 · Denominator: 1001 · Frequency: 14.59% · Observed in 2 cohorts · Confidence: moderate · Source: dlbcl_duke_2017 · Retrieved: 2026-09-18

Gene table — reference cohort

Headline values are from the reference cohort, dlbcl_duke_2017; the matrix above keeps every cohort separate. "Curated" marks a gene the disease briefing lists as a target; the rest are here because they are among the most frequently mutated genes in the reference cohort. Recurrent changes are the reference cohort's commonest protein changes.

FDA biomarker marks a gene the FDA recognises as a biomarker of response to an approved drug; FDA, tumour-agnostic marks the five that apply whatever the primary site. Hover for the drug, the tumour type and the year. This is level 1 only, United States only, and frozen at November 2022 — a dash means the gene was not FDA-recognised on that date, not that it is undruggable, and not that no trial exists. The frequency beside it is how often the gene is altered in this disease, which is a different question from whether these patients are eligible for the drug. Source: Quantifying the Expanding Landscape of Clinical Actionability for Patients with Cancer. Cancer Discovery 2024;14(1):49-65. doi:10.1158/2159-8290.CD-23-0467, Table 1

Cohorts

Listed in the disease profile, not searched: a name search returns the same patients under several accessions. Patients are unique patient ids in the study's sequenced sample list. Hypermutated: more than ten times the cohort's median non-silent mutations per sample, and at least 100.

Copy-number events

Discrete calls from each study's copy-number profile: 2 is high-level amplification, −2 deep deletion. Gains and shallow losses are not counted. Denominators are the cohort's copy-number sample list, which differs from its sequenced list. Rows at 2% or more.

Cohort-aware frequencies

Each row is calculated from unique patients in that study's sequenced sample list. The range is descriptive; no pooled estimate is shown because cross-study overlap and assay comparability have not been checked.

What this page does not do

Structural variants
Read the structural-variant profiles the studies carry; fusions are the defining event in several of these diseases.

Context
Stage, subtype, age and treatment line are not attached to any count; the cohorts differ on all four.

Interpretation
Activating versus inactivating, actionable versus not, and evidence level are not inferred here; the briefing's target table carries the drug and trial facts.

Limitations

• A cBioPortal public-API snapshot retrieved 2026-09-18; the page does not refresh source data at request time.
• Counts are patients with at least one non-silent call in the study's sequenced sample list; silent, intronic and UTR calls are excluded.
• For targeted-panel cohorts each gene divides by the patients whose panel carried it; a gene absent from the panel is shown as not assayed, not as zero.
• Copy-number rows use discrete calls (2 = high-level amplification, −2 = deep deletion) against the cohort's copy-number sample list, which is a different roster from the sequenced one.
• Promoter mutations are a separate row. A cohort that reports none is marked not assayed, not zero: whole-exome studies such as the TCGA pan-cancer atlas do not cover the TERT promoter, so a low coding-only TERT figure there is not a low TERT rate.
• Cohorts are not pooled. Cross-study patient overlap has not been checked and no disease-wide frequency is reported.
• Structural variants and fusions are not read in this snapshot even where the study carries a profile; germline variants, mutational signatures, TMB and MSI are not reported.
• The gene set is the briefing's curated targets plus the reference cohort's most frequently mutated genes; it is not genome-wide.