CSF cell-free DNA EGFR testing using DdPCR holds promise over conventional modalities for diagnosing leptomeningeal involvement in patients with non-small cell lung cancer. (October 2020)

• Record Type: Journal Article
• Authors: Suryavanshi, Moushumi; Jaipuria, Jiten; Panigrahi, Manoj Kumar; Goyal, Neha; Singal, Rishu; Mehta, Anurag; Batra, Ullas; Doval, D.C.; Talwar, Vineet
• Is Part Of: Lung cancer. Volume 148(2020)
• Page Start: 33
• Page End: 39
• Publication Date: 2020-10
• Subjects: CSF -- Cell-FRee DNA -- Leptomeningeal -- EGFR -- Nonsmall cell lung cancer; Lungs -- Cancer -- Periodicals; Lung Neoplasms -- Abstracts; Lung Neoplasms -- Periodicals; Poumons -- Cancer -- Périodiques; Lungs -- Cancer; Periodicals; Electronic journals; Electronic journals; 616.99424
• DOI: 10.1016/j.lungcan.2020.07.034
• Languages: English
• ISSNs: 0169-5002

Abstract

Highlights: Diagnosing Leptomeningeal involvement in lung cancer is a challenge. cfDNA EGFR testing in CSF using DdPCR has potential against MRI and CSF cytology. cfDNA has higher sensitivity, accuracy and net comparative benefit. Mutation burden is significantly higher in supernantant than pellet of CSF. Abstract: Background: EGFR mutant NSCLC patients have leptomeningeal (LM) involvement in more than 9% cases. Material & methods: We conducted a study evaluating the diagnostic utility of cfDNA EGFR testing in CSF using DdPCR while comparing it against MRI and CSF cytology. We also looked for known EGFR mutations in the CSF sample. These mutations were also tested in paired plasma samples. We further compared which constituent of CSF (pellet/supernatant) had better yield. Results: 21 patients comprised the study. Of these 17 patients were diagnosed to have LM involvement based on conventional criteria. All modalities had 100 % specificity and positive predictive value. However, MRI and CSF cytology had a poor negative predictive value. cfDNA had the highest sensitivity (92.3 %), negative predictive value (75 %), accuracy (94.1 %), and net comparative benefit. Paired plasma samples were available for 19 patients. Primary EGFR mutation was detectable in the CSF sample in 16/19 patients; however, the plasma sample was positive only in 7/19 patients. 3 samples were negative for primary EGFR mutation in both CSF and plasma. None of the CSF samples showed positivity for T790M mutation which could however be observed in two patients in plasma samples. Both supernatant and pellet were analysed for cfDNA mutation analysis in 18/21 patients. The intraclass correlation coefficient regarding the percentage fraction tumor-derived DNA of cfDNA observed was 0.83(95 % CI 0.29 to 0.95) between both samples. Conclusion: EGFR detection in CSF has a potential role in diagnosing LM involvement. T790 M resistance mutations are uncommon in CSF post first and second-generation TKIs. Both supernatant and pellet samples can be used for the extraction of cell-free DNA in CSF.