Abstract 660P
Background
Leptomeningeal metastases (LM) is becoming frequent in EGFR-mutated NSCLC and causes dismal prognosis due to limited treatments. A deeper understanding of the molecular phenotypes of LM is lacking, hindering the exploration of potential therapeutic targets.
Methods
Using Data-independent Acquisition and ctDNA next generation sequencing (panel), we characterize 135 CSF from 120 patients with advanced EGFR-mutated NSCLC and LM (n=69), or brain metastases (BM, n=43), or non LM/B (n=8). Proteomic data was available for 135 samples and genomic data available for 86 smaples.
Results
A higher rated of CSF ctDNA detection was shown in LM than in BM. CSF in LM was distinct from those in BM or non LM/BM group in proteomic profiles: a significantly higher number of proteins were detected in LM; GSVA using the Hallmark gene sets revealed upregulation of NOTCH signaling, glycolysis, and Interferon gamma response pathways in LM. Signatures of extracellular matrix and CD4+ T cells estimation was upregulated and CD8+ T cells estimation was downregulated in LM. NMF subtyping showed that our cohort could be grouped to 3 subtypes, which were correlated with clinical features. Differential expression of proteins were selected as LM-specific proteomics to stratify LM into two subtypes, which was related to survival outcomes. Subtype I, characterized by disrupted extracellular matrix and increased endocytosis, was associated with shorter OS. Subtype II showed upregulation of immune response related pathway. Patients of subtype I had higher rate of positive CSF cytology and concurrent TP53 alteration. The proteomic profiles of BM revealed its heterogenous nature and a subtype which might be related to poor prognosis and development of LM. Finally, selected signature proteins that can be targeted by known FDA-approved drugs or candidate drugs are shown to be enriched in LM as well as in BM subtype that was prone to LM development.
Conclusions
We connected genomic aberration to proteomics which revealed a more aggressive CNS metastasis phenotype. We also revealed the proteomic stratification of LM which provided insight into the underlying biology of this deadly complication and suggested the opportunity for therapeutic targets.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.