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

Abstract

Up to 50% of patients with non-small cell lung cancer (NSCLC) develop brain metastasis (BM), yet the study of BM genomics has been limited by tissue access, incomplete clinical data, and a lack of comparison with paired extracranial specimens. Here we report a cohort of 233 patients with resected and sequenced (MSK-IMPACT) NSCLC BM and comprehensive clinical data. With matched samples (47 primary tumor, 42 extracranial metastatic), we show CDKN2A/B deletions and cell cycle pathway alterations to be enriched in the BM samples. Meaningful clinico-genomic correlations are noted, namely EGFR alterations in leptomeningeal disease (LMD) and MYC amplifications in multifocal regional brain progression. Patients who developed early LMD frequently have had uncommon, multiple, and persistently detectable EGFR driver mutations. The distinct mutational patterns identified in BM specimens compared to other tissue sites suggest specific biologic underpinnings of intracranial progression.

Fig. 1. Study design and genomic differences between BM NSCLC and primary tissue (PT) or extracranial metastatic (EM) sites.

A Overview of study design. B Comparison of broad genomic features between brain metastases (BM) samples (n = 233), extracranial metastases (EM) samples (n = 42), and primary tumor (PT) samples (n = 47) (TMB comparison: BM vs. extracranial median: 5.8; p = 0.00766; FGA comparison: BM vs. extracranial metastases: p = 2.765e-06, BM vs. primary: p = 2.273e−07). A two-sided Mann–Whitney U-test was used to assess statistical significance. The center line of the box plots indicates the median. The bounds of the box indicate the interquartile range. The whiskers indicate the highest and lowest values not considered outliers. Asterisks indicate significance between groups being compared. C Oncoprint depicting the most frequent oncogenic alterations in BM, EM, and PT samples. D Comparison of oncogenic signaling pathway alterations across BM, EM, and PT samples. The cell cycle pathway was significantly enriched in BM vs PT tumors (p = 0.004, q = 0.041). A two-sided Fisher’s exact test was used to assess statistical significance. Multiple hypotheses testing was performed using a Benjamini-Hochberg correction. Asterisks indicate significance between groups being compared. E Genome-wide copy number profiles for BM, PT, and EM samples. Source data are provided as a Source Data file for Fig. 1.

Fig. 2. Paired analysis.

A Overview of mutations that were either shared or unique when comparing BM to PT/EM samples when BM samples were obtained before PT/EM samples; the bar plot at the bottom represents the most frequently mutated genes that were private to the BM samples. B Overview of mutations that were either shared or unique when comparing BM to PT/EM samples when BM samples were obtained after PT/EM samples; the bar plot at the bottom represents the most frequently mutated genes that were private to the BM samples. C Overview of mutations that were either shared or unique when comparing BM to CSF samples when BM samples were obtained before CSF samples; the asterisk indicates one patient in which CSF was obtained before BM sample. The bar plot at the bottom represents the most frequently mutated genes that were private to the BM samples. D Shared and unique mutations between patients with synchronous BM and PT/EM tumors. Oncoprint depicts the types of mutations across the samples per patient. E Oncoprint of BM tumor pairs from patients with multiple BM samples showing shared and unique alterations. F Patient vignettes for two patients with multiple samples per patient. Tumor locations are shown in the body maps and the intervals of time between samplings are depicted at the bottom. Oncogenic alterations identified for each tumor are written out, colored by whether they were shared or unique. Source data are provided as a Source Data file for Fig. 2.

Fig. 3. Clinical and genomic correlates including disease progression in BM LUAD cohort.

A Scatterplots comparing driver alteration frequencies between (left to right): BM samples found at diagnosis versus BM samples found as progression of disease, BM samples from patients with one BM at diagnosis versus BM samples from patients with multiple BMs at diagnosis, treatment naïve BM samples versus BM samples from patients with prior treatment, and BM samples from patients with no prior tyrosine kinase inhibitor (TKI) treatment versus BM samples from patients with prior TKI treatment. Genes altered in at least 25% of one of the groups being compared are shown and red coloring of a point indicates significance. B Overall survival (OS) in BM LUAD group from the time of BM diagnosis. C Progression-free survival (PFS) in BM LUAD group from the time of BM diagnosis. D Comparison of oncogenic alterations in BM samples from patients with different types of intracranial disease progression. Comparisons with significant p-value results are shown with the presence of an asterisk by their alteration frequency. The color of the asterisk indicates which groups were being compared. E Pathway-level alterations between BM samples from patients with different types of intracranial disease progression. The MYC pathway was significantly enriched in the patients with LMD (p = 0.013, q = 0.14) and regional progression (both single: p = 0.023, q = 0.255, and multifocal: p = 0.023, q = 0.255) compared to patients with local progression. A two-sided Fisher’s exact test was used to assess statistical significance. Asterisks indicate significance between groups being compared. Source data are provided as a Source Data file for Fig. 3.

Fig. 4. EGFR alteration distributions and individual patient cases.

A Lollipop plot (on the left) of EGFR depicting the most common sites of mutations in the BM samples. The kinase domain is blown out to show the types of mutations by the type of intracranial progression. The stacked bar plot (on the right) depicts the most common types of mutations stratified by the type of intracranial progression. B Vignette of patient B with three sequenced samples. The disease timeline depicting the treatment the patient received and tumor samplings is shown beneath, along with what oncogenic alterations were shared or unique to each of the samples. C Vignette of patient C with multiple sequenced samples. The disease depicting the treatment the patient received and tumor samplings is shown beneath, along with what oncogenic alterations were shared or unique to each of the samples and the circulating tumor cells (CTC) count at each sampling. Source data are provided as a Source Data file for Fig. 4.