Targeted Deep Sequencing in Untreated Progressive CLL Helps to Better Dissect the Genomic Landscape and Shows Novel Associations between Mutations, Prognostic and Clinical Parameters: Data from E1912, a Randomized Phase III Study of the ECOG-ACRIN Cancer Research Group

Next-generation sequencing studies have significantly improved our understanding of the chronic lymphocytic leukemia (CLL) genetic landscape, tumor architecture and allowed the identification of associations between specific mutations and clinical outcome. Hence, mutations in TP53, NOTCH1, SF3B1, BIRC3, EGR2 and MYD88 are known to be associated with differential prognosis. Despite the work performed on these genes, several additional putative tumor-driver genes are recurrently mutated in CLL, but their clinical impact remains unclear.

In order to gain further understanding, we performed targeted deep sequencing in baseline samples of a large CLL cohort (n=436) participating in the North American Intergroup study (NAIG) designated as E1912 and led by ECOG. This trial is a randomized Phase III of Ibrutinib (Ibr) and Rituximab (R) vs FCR in previously untreated CLL. We screened 25 putative driver genes recurrently mutated in CLL, and 2 genes (BTK and PLCG2) associated with Ibr resistance. The CD5+/CD19+ B lymphocyte population was flow-sorted. All coding regions were amplified using customized oligos. Samples were run on an IonTorrent sequencer. Variants were called using IonTorrent Somatic VariantCaller. VCF files were annotated using BioR. Somatic variants with a Mapping Quality <20 or read depth <10X were removed. Variants of interest were inspected using IGV. The mutational status was analyzed for association with FISH, IGHV mutation status, expression of CD38, CD49d and ZAP70 (all 3 determined at entry to the trial). We were not able to evaluate associations to clinical responses for the different arms as this trial remains blinded for clinical outcome until 2018.

Deep sequencing allowed us to analyze the CLL genomics landscape in more detail. An average coverage of 970X depth was obtained, detecting 35% of mutations in minor subclones (<10% of reads). Therefore, mutation rates for most genes were higher than previous published whole exome and whole genome sequencing studies. Overall, 350 (80%) patients (pts) carried at least one putative driver gene mutated, 205 (47%) carried 2 or more and 89 (20%) 3 or more mutated genes (range 0-7). The most commonly mutated gene was NOTCH1 (22%), followed by SF3B1 (20%), ATM (18%), BIRC3 (11%), XPO1 (10%) and POT1 (9%). Mutations in the RAS pathway were found in 13% of cases (KRAS 7%, BRAF 6%, NRAS 2%). Overall, 25% of cases with activating mutations in the RAS pathway (BRAF, KRAS, NRAS), showed >1 mutation affecting one or more genes of this pathway (range 2-5 mutations). Multiple mutations affecting the same gene (>2 mutations in tumor suppressor genes or >1 mutation in activating genes) were also recurrently found in BIRC3, ATM, TP53, MED12, SF3B1, XPO1, EGR2, DDX3X, and NOTCH1. Overall, the existence of genetic subclones showing convergent evolution was observed in 15% of CLL pts. Even using deep sequencing, no BTK or PLCG2 mutations were found in progressive but untreated CLLs. We found multiple associations between mutations and known prognostic and clinical parameters. Mutations in NOTCH1 (p=0.002), BIRC3 (p<0.001), MED12 (p<0.001), KRAS (p<0.001), and BRAF (p=0.001) were enriched in patients with trisomy 12; ATM (p<0.001) and EGR2 (p=0.002) in cases with deletion 11q and MYD88 (p=0.008) with deletion 13q. Mutations in SF3B1, NOTCH1, XPO1, MED12, BRAF were associated with ZAP70 expression (p<0.05). CD38 expression was significantly higher (p<0.05) in pts with mutations in SF3B1, NOTCH1, BIRC3, EGR2, KRAS, BRAF and CD49d expression in cases with mutations in NOTCH1, BIRC3, EGR2, KRAS, and BRAF .

This study, performed on a large cohort of untreated but progressive CLL, provides novel data about the mutation status of driver genes and their associations with clinical and prognostic parameters. By using deep sequencing we found that mutation rates were higher than previous studies for most genes, including key cancer-related pathways (i.e. RAS). We previously described the existence of convergent evolution in CLL (Blood 125:492-8). In this study we showed that this evolutionary mechanism is a common event in CLL, found in 15% of these trial pts and involving most driver genes. Additionally, our data reinforces that BTK and PLCG2 mutations are not found in Ibr-naive pts. Finally, this study provides the genetic groundwork for subsequent prospective comparative analyses of the clinical outcome for CLL treated with Ibr-R vs FCR in this phase 3 NAIG trial.