Cerebrospinal fluid circulating tumour DNA genotyping and survival analysis in lung adenocarcinoma with leptomeningeal metastases

Purpose The prognosis of patients with leptomeningeal metastasis (LM) remains poor. Circulating tumour DNA (ctDNA) has been proven to be abundantly present in cerebrospinal fluid (CSF); hence, its clinical implication as a biomarker needs to be further verified. Methods We conducted a retrospective study of 35 lung adenocarcinoma (LUAD) patients with LM, and matched CSF and plasma samples were collected from all patients. All paired samples underwent next-generation sequencing (NGS) of 139 lung cancer-associated genes. The clinical characteristics and genetic profiling of LM were analysed in association with survival prognosis. Results LM showed genetic heterogeneity, in which CSF had a higher detection rate of ctDNA ( P = 0.003), a higher median mutation count ( P < 0.0001), a higher frequency of driver mutations ( P < 0.01), and more copy number variation (CNV) alterations ( P < 0.001) than plasma. The mutation frequencies of the EGFR, TP53, CDKN2A, MYC and CDKN2B genes were easier to detect in CSF than in LUAD tissue ( P < 0.05), possibly reflecting the underlying mechanism of LM metastasis. CSF ctDNA is helpful for analysing the mechanism of EGFR-TKI resistance. In cohort 1, which comprised patients who received 1/2 EGFR-TKIs before the diagnosis of LM, TP53 and CDKN2A were the most common EGFR -independent resistant mutations. In cohort 2, comprising those who progressed after osimertinib and developed LM, 7 patients (43.75%) had EGFR CNV detected in CSF but not plasma. Furthermore, patient characteristics and various genes were included for interactive survival analysis. Patients with EGFR -mutated LUAD ( P = 0.042) had a higher median OS, and CSF ctDNA mutation with TERT ( P = 0.013) indicated a lower median OS. Last, we reported an LM case in which CSF ctDNA dynamic changes were well correlated with clinical treatment. Conclusions CSF ctDNA could provide a more comprehensive genetic landscape of LM, indicating the potential metastasis-related and EGFR-TKI resistance mechanisms of LM patients. In addition, genotyping of CSF combined with clinical outcomes can predict the prognosis of LUAD patients with LM.