Stay tuned for coverage from the 2015 ASCO Annual Meeting. Today, a SERMO Oncologist shares their thoughts on new genomics research presented by Marc Ladanyi from Memorial Sloan-Kettering Cancer Center. All opinions are his own.
Marc Ladanyi from Memorial Sloan-Kettering Cancer Center (MSKCC) presented his insitution’s solid tumor genomic profiling program, with an emphasis on NSCLC, as part of the education session on personalized lung cancer therapy.
The program is called MSK-IMPACT. They conduct genomic profiling of advanced solid cancers and try to identify “actionable mutations.” They test formalin-fixed paraffin-embedded samples. They need 20% viable tumor cells in the biopsy and compare against normal DNA (e.g. from blood). Germline variants are filtered out. There is a 3-week turnaround time.
The genes on the panel initially included 341 genes, 33 introns, and the TERT promoter. 69 more have been added recently. 85% of samples yield results. They have processed 5000 cases since January 2014. The top tumors are (no surprise) breast, NSCLC, colon, prostate, and sarcoma.
He focused on 243 NSCLC samples. 93% of samples were successful. The most common mutations were TP53 (60%), EGFR (30%), KRAS (20%), and KEAP1 (18%). Of 227 lung cancer successful samples, they found 23 targetable lesions, e.g. EML4-ALK. He also mentioned MET mutations causing exon 14 skipping, which cause a loss of ubiquitin ligase activity, leading to too much EGFR expression. This is treatable with crizotinib or cabozantinib.
They are on pace for 6000 samples per year and are trying to expedite the work flow. They have made the EGFR and ALK tests faster with IHC methods. He ended with a quote, “Rapid generation of big datasets will drive further discoveries.”
I was hoping to hear the rate of actionalbe mutations in other solid tumors. Typically, it is less than 5% for such studies. The rate of 10% is not surprising for NSCLC, since EGFR mutations and ALK mutations occur with that tumor at about that frequency. Most of the NSCLC actionable mutations were ones we already test for routinely in adenocarcinoma/non-squamous histology, so I am not sure how much the panel adds to current diagnostic algorithms. As for the big datasets quote, I am not sure what the big datasets will bring–but the best data is generated when a specific question is asked, when there is a hypothesis. I am uncertain of the impact of data mining for its own sake, and I am not quite as optimistic about the benefits of “big data” in the absence of a guiding hypothesis.
If you are an M.D. or D.O. in the US or UK, please join us on SERMO.