Genomic analysis and clinical correlations of non-small cell lung cancer brain metastasis

Authors

Anna Skakodub

Henry Walch

Kathryn R. Tringale

Jordan Eichholz

Brandon S. Imber

Harish N. Vasudevan

Bob T. Li

Nelson S. Moss

Kenny Kwok Hei Yu

Boris A. Mueller

Simon Powell

Pedram Razavi

Helena A. Yu

Jorge S. Reis-Filho

Daniel Gomez

Nikolaus Schultz

Luke R. G. Pike

Doi

10.1038/s41467-023-40793-x

PMID: 37591896 · DOI: 10.1038/s41467-023-40793-x · Journal: Nature Communications (2023)

TL;DR

Skakodub et al. profiled 233 resected NSCLC brain metastases (BM) with MSK-IMPACT plus matched primary tumors (n=47) and extracranial metastases (n=42), with detailed clinical annotation. BM specimens showed higher TMB, FGA, and chromosomal instability than non-BM sites, with CDKN2A/B deletions and cell-cycle pathway alterations enriched in BM. EGFR alterations — often uncommon/non-canonical drivers — were enriched in patients who developed leptomeningeal disease (LMD), and MYC amplifications were associated with multifocal regional intracranial progression PMID:37591896.

Cohort & data

  • 233 patients with resected NSCLC BM sequenced by MSK-IMPACT; 180 (77%) LUAD, 23 (10%) LUSC, 30 (13%) other NSCLC. Median age 67; 57% female; 80% current/former smokers.
  • Matched samples: 47 primary tumors (PT) and 42 extracranial metastases (EM); downstream LUAD-restricted analysis used 179 BM, 37 PT, 34 EM.
  • External comparator LUAD PT cohorts: BM+ (N=32), BM−/EM+ (N=1549), BM−/EM− (N=582).
  • Dataset: bm_nsclc_mskcc_2023.
  • Assay: MSK-IMPACT targeted panel (IMPACT341 / IMPACT410 / IMPACT468 / IMPACT505) PMID:37591896.

Key findings

  • TMB was higher in BM vs extracranial metastases (median 8.8 vs 5.8; p=0.00766) and FGA was higher in BM vs EM (p=2.77e-06) and vs PT (p=2.27e-07) PMID:37591896.
  • CDKN2A/CDKN2B alterations were more frequent in BM (34%) than PT (13%) (p=0.003, q=0.04); cell-cycle pathway alterations 56% BM vs 32% PT (p=0.004, q=0.041), driven by CDKN2A/B.
  • In the LUAD sub-cohort, CDKN2A/B enrichment persisted (31% BM vs 18% PT; p=0.004) as did cell-cycle pathway enrichment (52% vs 27%; p=0.007).
  • Comparing LUAD PT BM+ vs BM−/EM−, TP53, MYC, SMARCA4, RB1, ARID1A, and FOXA1 alterations were enriched in PTs from patients who later developed BM; NKX2-1 alterations were enriched in both BM and EM cohorts.
  • Among 179 LUAD BM patients, 101 (56%) had intracranial progression after craniotomy + RT: regional (30%), local (14%), LMD (11%). Median OS 2.7 years (95% CI 2.3–4.0); median iPFS 1.2 years (95% CI 1.0–1.5).
  • LMD patients had more EGFR alterations than non-progressors (45% vs 21%; p=0.044). Local progressors had more RB1 loss (24% vs 6%; p=0.022) and NKX3-1 alterations (16% vs 3%; p=0.044). Multifocal regional progressors had MYC amplifications in 22% vs 0% in local progressors (p=0.023). MYC pathway alterations were enriched in LMD (p=0.013, q=0.14) and regional progression groups.
  • NF1 alterations were more frequent in LMD patients (15%) than other groups.
  • Paired BM–BM samples from multi-lesion patients showed high genomic concordance, contrasting with lower concordance seen in synchronous BM/PT pairs.

Genes & alterations

  • CDKN2A / CDKN2B: deep deletions enriched in BM; cell-cycle pathway driver in BM development.
  • EGFR: enriched in LMD; uncommon drivers (L861Q, G719A/S, A755G, N771_H773dup) in 45% of LMD patients; persistence of these drivers despite TKI therapy; one vignette showed emergence then clearance of T790M under osimertinib with ongoing L861Q/G719S.
  • MYC: amplifications associated with multifocal regional intracranial progression; MYC pathway alterations enriched in LMD and regional progression.
  • TP53, KRAS: commonly shared between BM and PT/EM; often truncal.
  • RB1: loss associated with local intracranial progression.
  • SMARCA4, ARID1A, FOXA1, NKX2-1: enriched in PTs of BM+ patients vs non-metastatic.
  • STK11: 22% vs 0% in LUAD BM vs SCC BM (p=0.01).
  • NF1: enriched in LMD patients.
  • HLA-B: subset of private BM mutations PMID:37591896.

Clinical implications

  • Non-canonical EGFR mutations in BM may identify NSCLC patients at elevated risk for LMD and partial resistance to osimertinib, with persistence across serial tissue/CSF samples despite maximal EGFR-directed and local therapy PMID:37591896.
  • Cell-cycle / CDKN2A-B loss appears to be a near-obligate event for BM development, suggesting a rationale for CDK4/6-directed strategies in CNS-tropic NSCLC (not tested here).
  • MYC amplification may flag patients at risk for multifocal regional intracranial progression.
  • Prior erlotinib exposure features in the EGFR-mutant LMD trajectories described.

Limitations & open questions

  • Retrospective single-institution cohort, reliant on MSK-IMPACT targeted panel (not WES/WGS).
  • Matched PT and EM sample counts are modest (47 and 42 respectively); synchronous BM/PT analysis N=2.
  • Some q-values were non-significant after multiple-testing correction (e.g., RB1 local-progression association q=0.573), warranting validation.
  • Causal mechanism linking non-canonical EGFR drivers to leptomeningeal tropism vs osimertinib partial resistance remains unresolved.

Citations from this paper used in the wiki

  • “CDKN2A/B alterations were more common in the BM samples (34%) compared to PT (13% p=0.003, q=0.04).”
  • “MYC amplifications were more common in patients who later suffered multifocal regional progression, compared to those with local progression, where no MYC amplifications were detected (22% vs 0%, p=0.023).”
  • “Patients who suffered from LMD frequently exhibited less common EGFR mutations (45%), such as L861Q, G719A/S, A755G, or N771_H773dup.”
  • “The median OS and iPFS from BM diagnosis was 2.7 years (95% CI 2.3–4.0) and 1.2 years (95% CI 1.0–1.5).”

This page was processed by crosslinker on 2026-04-11.