Diagnosis and monitoring of Leptomeningeal Disease using Circula5ng free DNA in the cerebrospinal fluid (CSF cfDNA) R. H. Shah 1 , E. I. Pentsova 2 , J. Tang 5 , A. Boire 2 , D. You 5 , S. Briggs 2 , A. Omuro 2 , X. Lin 2 , M. Fleisher 3 , C. Grommes 2 , F. Meng 5 , S. D. Selcuklu 5 , S. Ogilvie 4 , N. Distefano 4 , L.Shagabayeva 2 , M.Rosenblum 2 , L. M. DeAngelis 2 , A. Viale 5 , I. K. Mellinghoff 2, , M. F. Berger 1,5, 1 Department of Pathology, Memorial Sloan KeTering Cancer Center, New York, NY 10065, USA. 2 Department of Neurology, Memorial Sloan KeTering Cancer Center, New York, NY 10065, USA. 3 Department of Laboratory Medicine, Memorial Sloan KeTering Cancer Center, New York, NY 10065, USA. 4 Department of Neurosurgery; Memorial Sloan KeTering Cancer Center, New York, NY 10065, USA. 5 Center for Molecular Oncology, Memorial Sloan KeTering Cancer Center, New York, NY 10065, USA Background Conclusion Leptomeningeal metastases (LM) in solid tumors (ST) represent a devasta]ng complica]on of cancer with a median survival of only 1214 weeks a_er diagnosis [1]; however, establishing the diagnosis of LM can be difficult, par]cularly at early stages before the pa]ent is disabled. The diagnosis is based on CSF cytologic analysis and/or MRI findings.[24] Brain and spine MRIs have been increasingly preferred for the ini]al evalua]on of LM because of their noninvasive nature and convenience to pa]ents. However, MRI findings are nega]ve in 25%50% of pa]ents [3, 4], and unequivocal findings may only appear in latestage disease when the pa]ent is already debilitated. CSF cytologic analysis provides diagnos]c confirma]on of LM but is associated with a rela]vely low sensi]vity (approximately 50% on the first lumbar puncture) and is highly examinerdependent. Improved diagnos]c tools are required to facilitate early diagnosis. To this end, we explored whether sufficient quan]ty and quality of DNA can be isolated from CSF for genomic study and whether the CSF pellet or CSF supernatant, would be more suitable for detec]ng cfDNA. We used an inhouse sequencing assay, MSKIMPACT [5], to interrogate 341 clinically relevant cancer genes in tumor derived cfDNA from 53 pa]ents. Results of CSF cfDNA were compared to standard CSF cytopathologic analysis from that same CSF sample and with MRI findings performed at the same ]me. When possible, we compared CSF cfDNA with DNA from tumor ]ssue (primary tumor and nonCNS sites) to determine similari]es and differences in gene]c altera]ons between these different compartments. Acknowledgements Introduc5on Methods References Our study demonstrates that genomic analysis of CSF, using a sufficiently sensi]ve and comprehensive plaiorm, may be useful to facilitate diagnosis of tumor in the CNS, monitor the evolu]on of the cancer genome during treatment of CNS cancers, guide the choice of secondline agents, and perhaps iden]fy pathways that are uniquely associated with cancer spread to the central nervous system. Center of Molecular Oncology, Department of Pathology & Department of Neurology Targeted Capture, Sequencing & Genomic Analysis Captures all proteincoding exons of 341 cancerassociated genes Sequence pairend reads (2x100) on HiSeq 2500 Analyse genomic data using methods described previously [5]. Extrac]on of cfDNA. Centrifuged at 10,000 g for 30 min at 4℃ to remove residual precipitated cellular components QIAamp Circula]ng Nucleic Acid Kit Cerebrospinal Fluid Collec]on and Prepara]on. Lumbar Puncture Centrifuged at 1,000 x g, 4°C for 5 min to separate supernatent & pellet Image 1: Comparison of tumorderived DNA from CSF cell pellet and supernatant. (A) Schema]c of separa]on of CSF pellet and supernatant. Cellular DNA is isolated from the pellet, and cfDNA is isolated from the supernatant. (B) Variant allele frequencies for known muta]ons in CSF cfDNA and pellet DNA. (C) Log2 ra]os of normalized sequence coverage for target exons in CSFcfDNA and pellet DNA for pa]ent 8 . Greater than 10fold amplifica]on of HER2 was observed in CSFcfDNA, whereas HER2 amplifica]on was barely detectable in pellet DNA. (D) Evidence of EML4ALK gene fusion in CSF cfDNA and pellet DNA for pa]ent 6. Readpairs suppor]ng the fusion (red) are visualized using the Integra]ve Genomics Viewer. Results Muta5ons detected in most pa5ents with LM disease Image 2: (A) Schema]c showing paTerns of CNS involvement in pa]ents with solid tumors. (B) Percentage of pa]ents for which highconfidence soma]c altera]ons were detected by MSKIMPACT. Pa]ents are grouped according to the presence or absence of intraparenchymal brain metastases and leptomeningeal metastases (LM) Drugresistance mechanisms in CSF in pa5ents with CNS relapse Image 3: Summary of genomic profiling results from CSF and other tumor sites in pa]ents who developed progressive CNS disease during treatment with the indicated kinase inhibitors. Tumor evolu5on in pa5ents with primary brain tumors. Image 4: Tumor evolu5on in pa5ents with primary brain tumors. (A) Spa]al and temporal heterogeneity between samples obtained at diagnosis, at recurrence and from CSF in pa]ent 42 with recurrent glioblastoma. CSF cfDNA harbors a PTEN R130* muta]on (VAF=0.25), while resec]on #2 harbors a PIK3CA H1047R muta]on (VAF=0.441). (B) CSF molecular profile for a pa]ent 45 with anaplas]c oligodendroglioma contains the IDH1 R132H muta]on and 1p/19q dele]on found in ]ssue resec]on #2, as well as 454 nonsilent soma]c muta]ons. 448 SNVs represent C>T/G>A muta]ons demonstra]ng temozolomideinduced mutagenesis. A B 1. Le Rhun, E., S. Taillibert, and M.C. Chamberlain, Carcinomatous meningi]s: Leptomeningeal metastases in solid tumors. Surg Neurol Int, 2013. 4(Suppl 4): p. S26588. 2. Wasserstrom, W.R., J.P. Glass, and J.B. Posner, Diagnosis and treatment of leptomeningeal metastases from solid tumors: experience with 90 pa5ents. Cancer, 1982. 49(4): p. 75972. 3. Clarke, J.L., et al., Leptomeningeal metastases in the MRI era. Neurology, 2010. 74(18): p. 144954. 4. Freilich, R.J., G. Krol, and L.M. DeAngelis, Neuroimaging and cerebrospinal fluid cytology in the diagnosis of leptomeningeal metastasis. Ann Neurol, 1995. 38(1): p. 517. 5. Cheng, D.T., et al., Memorial Sloan Ke-eringIntegrated Muta4on Profiling of Ac4onable Cancer Targets (MSKIMPACT): A Hybridiza4on CaptureBased NextGenera4on Sequencing Clinical Assay for Solid Tumor Molecular Oncology. J Mol Diagn, 2015. 17(3): p. 25164. Pilot: CSF cell pellet vs. CSF cfDNA