MET alterations are enriched in lung adenocarcinoma brain metastases, defining a distinct biologic subtype

Abstract

Non-small cell lung cancer exhibits the highest rates of brain metastases (BMs) among all solid tumors, presenting a major clinical challenge. The development of novel therapeutic strategies targeting BMs is clearly needed. We identified a significant enrichment of MET amplification in lung adenocarcinoma (LUAD) BMs compared with primary LUAD and extracranial metastases in oncogene driver-negative patients. Of note, MET-amplified BMs were responsive to MET inhibitors in vivo, including models with acquired MET amplification at the time of metastasis. MET alterations (amplifications and/or mutations) were also more frequently detected in circulating tumor DNA from patients with LUAD BMs than in those without BMs. MET-altered BMs also demonstrated unique genomic features compared with non-MET-altered BMs. Transcriptomic analyses revealed that in contrast to MET WT BMs, MET-amplified BMs exhibited a more inflamed tumor microenvironment and displayed evidence of metabolic adaptation, particularly a reliance on glycolysis in contrast to OXPHOS in MET WT BMs. Furthermore, MET-amplified BMs demonstrated evidence of epithelial-mesenchymal transition signaling, including increased expression of TWIST1. Patients with MET-amplified BMs had significantly shorter overall survival. These findings highlight MET amplification as a critical driver of LUAD BMs, emphasizing its potential as a therapeutic target.

Keywords: Biomarkers; Genetics; Lung cancer; Oncogenes; Oncology.

Figure 1. Acquired MET amplification in a LUAD BM that was responsive to capmatinib.

(A) Timeline summarizing the treatment course, tumor biopsies, and MET amplification status. MET FISH images (×40 original magnification) are shown for the primary tumor biopsy at the time of diagnosis and 10 months later at the time of metastatic brain tumor biopsy. Red signals, MET; green signals, centromere 7 (CEP7). (B) A PDX model was established from patient 16-16 BM resection specimen. Mice were randomized to receive vehicle (0.25% w/v methyl cellulose) or capmatinib (5 mg/kg) by oral gavage 5 times per week for 4 weeks. Results are presented as mean tumor volume ± SEM of 6 tumors/group. Data were assessed by 2-tailed Student’s t test; **P = 0.01. (C) Luciferase-labeled H1993 LUAD cells were injected intracardially into SCID mice and monitored for metastatic spread. Mice were randomized to receive either vehicle (0.25% w/v methyl cellulose) or capmatinib (5 mg/kg) via oral gavage, administered 5 times per week for 3 weeks. Bioluminescent signal intensity in the head region was quantified relative to baseline and is presented as the mean ± SD. Statistical analysis was performed using the Mann-Whitney test; *P < 0.05. (D) Longitudinal bioluminescent imaging of individual mice over the course of treatment. All images were acquired and analyzed using Living Image Software (Perkin Elmer) and set to the same intensity scale for comparison. “X” represents mice that died prior to the end of the 3-week treatment period.

Figure 2. MET amplification is more frequently observed in LUAD BMs compared with extracranial metastases and primary LUAD.

(A) Pie charts showing the frequency of MET amplification by FISH (MET/CEP7 ratio ≥ 2.0) in primary LUAD, liver metastases, and BMs in the UPMC cohort. Fisher’s exact test, 2-sided; **P < 0.01 (P = 0.002 exact), ****P < 0.00001. (B) MET protein expression by frequency of IHC staining intensity (0, +1, +2, +3; χ² test; ****P < 0.0001) and MET H-score (Student’s t test, 2-tailed; ****P < 0.0001) in non–MET-amplified and MET-amplified BMs. Horizontal lines represent mean values. (C) Frequency of MET amplification by NGS copy number alteration (cutoff ≥ 6) in primary NSCLC, non-BMs, and BMs in the Caris cohort. χ² test; ****P < 0.0001. (D) Representative MET FISH images (captured at ×40 original magnification and enlarged) from a matched primary LUAD and BM from the same patient. White boxes represent areas of focal MET amplification. Red signals, MET; green signals, CEP7.

Figure 3. MET alterations detected in ctDNA are found more often in patients with BMs.

(A) Percentage of ctDNA-positive MET alterations (amplifications and mutations combined) in patients with (N = 90) and without (N = 187) BMs, as detected with the Guardant360 CDx assay. (B) Percentage of ctDNA-positive MET amplifications in patients with (N = 90) and without (N = 187) BMs, as detected with the Guardant360 CDx assay. (C) Percentage of MET mutations in patients with (N = 90) and without (N =187) BMs. Fisher’s exact test, 1-sided; P values are shown for each comparison.

Figure 4. LUAD BMs have a distinct mutational profile compared with primary LUAD tumors.

OncoPlot of the distribution of mutations for patients with LUAD BMs (N = 74) compared with those with primary LUAD tumors (N = 180 total; N = 147 with variants detected). Frequency of mutations is listed for each gene in order of the highest to lowest frequency in LUAD BMs. The mutation types are color-coded and annotated in the key. Variants annotated as “Multi-hit” are genes that are mutated more than once in the same sample. Fisher’s exact test, 2-sided; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Red asterisks indicate significance after FDR adjustment. A q value < 0.05 was considered significant.

Figure 5. MET-altered BMs are genomically distinct from non–MET-altered BMs.

OncoPlot of the distribution of mutations for patients with MET altered LUAD BMs (N = 31) compared with those with non–MET-altered LUAD BMs (N = 43). Frequency of mutations is listed for each gene in order of highest to lowest. The mutation types are color-coded and annotated in the key. Variants annotated as “Multi-Hit” are genes that are mutated more than once in the same sample. Fisher’s exact test, 1-sided; *P ≤ 0.05.

Figure 6. LUAD BMs have a distinct transcriptional profile compared with matched primary LUAD tumors, and MET-amplified BMs are distinct from non–MET-amplified BMs.

(A) Heat map of 174 differentially expressed genes in 5 matched primary LUAD (yellow) and BMs (blue) (FDR < 0.05, fold change [FC] ≥ 2.0 or ≤ –2.0). (B) GSEA of the Hallmark gene sets from the Molecular Signatures Database (MSigDB) showing increased (orange) and decreased (blue) pathways in BMs compared with primary LUAD. The top 20 pathways are shown sorted by median rank of higher to lower (representing confidence higher to lower). (C) Heat map of 243 differentially expressed genes in MET-amplified (red) and MET WT (black) BMs (FDR < 0.05, fold change ≥ 2.0 or ≤ –2.0). (D) GSEA of the Hallmark gene sets from the MSigDB showing increased (orange) and decreased (blue) pathways in MET-amplified compared with non–MET-amplified BMs. The top 20 pathways are shown sorted by median rank of higher to lower (representing confidence higher to lower). The red dashed lines in B and D represent the threshold of what was considered significant.

Figure 7. Patients with NSCLC MET-amplified BMs have worse OS.

Kaplan-Meier survival analysis showing OS in months from initial diagnosis in patients with NSCLC BMs stratified by MET amplification (red line) versus no amplification (blue line). Median OS in months, HR, and CI were calculated. (A) All patients; (B) all patients excluding those with EGFR mutations. 1-, 3-, and 5-year survival rates are indicated.