Chavan, Shweta S.

Abstract Introduction Over the last 15 years gene expression profiling (GEP) has been used to define myeloma molecular subgroups and to determine clinical prognosis. Two major molecular subgroup classifications have been used: the UAMS which determines 7 subgroups and the TC classification based on the presence of IgH translocations and expression of D group cyclins. For prognosis, although a number of different GEP signatures have been defined, the widely used GEP70 identifies 15% of patients with high risk (HR) disease who have a median PFS and OS of 1.75 and 2.83 years. An ideal classification system would identify clinically relevant subgroups with distinct etiology and biology using standardized techniques. We have examined a large group of patients characterized at multiple genetic levels to optimize the diagnostic approach of newly diagnosed patients going forward. Materials and methods Study subjects included 1349 cases enrolled in Total Therapy trials (median follow up 7.5 years). Gene expression profiling was used to determine GEP70 risk status, molecular subgroup by UAMS and TC classifications, and to devise a new and extended TC classification (TC10). Interphase FISH associated with IgH translocations and 1q+ and 17p- were used to build GEP proxies. Data from mutational analysis generated by the FoundationOne targeted sequence panel was also incorporated. Results were validated on the UK MRC MyelomaIX and Hovon65/GMMG-4 studies. Results An initial agnostic analysis of GEP data using sparse k-means clustering verified the existence of TC based groups. Six groups were identified that corresponded overwhelmingly with known TC subgroups; CCND1-t(11;14), D1-HRD, D2-HRD, MMSET, MAF/CCND2, and CCND3. Further comparisons between the molecular subgroup and TC classifier revealed that the UAMS 7 subgroups clustered strongly within one predominant TC group: CD-1 and CD-2 to t(11;14), HY to D1, LB and PR to D2, MF to t(14;16) or t(14;20), and MS to t(4;14). As the UAMS molecular subgroups are largely contained within the TC framework, we aimed to extend the TC by developing the TC10. To extend the known TC subgroups, unsupervised clustering was applied to the 3 largest subgroups [t(11;14), D1, and D2] to determine the strongest single divisor within each respective subgroup. The dominant feature within the t(11;14) cases was CD20 expression, while the D1 and D2 subgroups both split according to RRAS2. CD20 is associated with PAX5 and VPREB3 expression, and RRAS2 is associated with decreased PTP4A3 and increased TNFAIP3 and BIRC3 expression. RRAS2 activation within D1 subgroup and CD20 activation within t(11;14) cases corresponds to an increased time to response to induction therapy suggesting they constitute important biological subgroups. The TC10 combines the known etiologic subgroups of the TC with functionally relevant subdivisions to create 10 novel subgroups: t(11;14) CD20+/-, D1: RRAS2+/-, D2: RRAS2+/-, t(4;14), t(14;16), t(14;20), and t(6;14). Analysis of mutational data revealed that RRAS2 and CD20 activation within the D1, D2, and t(11;14) subgroups reduced the number of mutations in the MAPK pathway. Further mutational analysis revealed that median mutational load was highest in t(14;16) and lowest in D2: RRAS2+ subgroups. The GEP70 score identifies 15% of patients with HR disease and is specific for this purpose. In an analysis of risk assessment methods, we compared GEP detected adverse lesions [t(4;14), t(14;16), t(14;20), 17p- and 1q+] with the GEP70 and revealed that GEP70 HR identified samples have lower OS rates than cases with more than one adverse lesion (validated in external sets). GEP70 HR segregates non-uniformly across molecular subgroups as over 40% of all HR cases are found in the TC10 t(4;14), t(14;16), and t(14;20) subgroups. GEP70 HR cases also have a higher mutational load than low risk cases. Furthermore, GEP70 HR is uniquely associated with 1q+ and 17p- as cases with at least one of these adverse lesions are 4.9 times as likely to be GEP70 HR as cases with neither. Conclusion GEP profiling has a central role in simplifying and standardizing the molecular subgroup designation and risk stratifying of MM patients. The GEP70 risk score reliably identifies HR cases and outperforms FISH in risk assessment, even in validation data sets. The TC10 provides a classification system that improves upon previous methods by defining both etiological and functionally meaningful subgroups. Disclosures Stein: University of Arkansas for Medical Sciences: Employment. Davies:University of Arkansas for Medical Sciences: Employment; Celgene: Consultancy; Janssen: Consultancy; Millenium: Consultancy; Onyx: Consultancy. Heuck:University of Arkansas for Medical Sciences: Employment; Celgene: Consultancy; Janssen: Other: Advisory Board; Millenium: Other: Advisory Board; Foundation Medicine: Honoraria. Weinhold:University of Arkansas for Medical Sciences: Employment; Janssen Cilag: Other: Advisory Board. Chavan:University of Arkansas for Medical Sciences: Employment. Thanendrarajan:University of Arkansas for Medical Sciences: Employment. Epstein:University of Arkansas for Medical Sciences: Employment. Yaccoby:University of Arkansas for Medical Sciences: Employment. Zangari:University of Arkansas for Medical Sciences: Employment; Novartis: Research Funding; Onyx: Research Funding; Millennium: Research Funding. van Rhee:University of Arkansa for Medical Sciences: Employment. Kaiser:Janssen: Honoraria; Amgen: Consultancy, Honoraria; Celgene: Consultancy, Honoraria, Research Funding; BristolMyerSquibb: Consultancy; Chugai: Consultancy. Sonneveld:Janssen-Cilag, Celgene, Onyx, Karyopharm: Honoraria, Research Funding; novartis: Honoraria. Goldschmidt:Takeda: Consultancy, Membership on an entity's Board of Directors or advisory committees; Onyx: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Bristol-Myers Squibb: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Janssen-Cilag: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Celgene: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Millenium: Honoraria, Research Funding, Speakers Bureau; Amgen: Consultancy, Membership on an entity's Board of Directors or advisory committees; Chugai: Honoraria, Research Funding, Speakers Bureau; Novartis: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau. Barlogie:University of Arkansas for Medical Sciences: Employment. Morgan:Bristol Myers Squibb: Honoraria, Membership on an entity's Board of Directors or advisory committees; MMRF: Honoraria; CancerNet: Honoraria; Takeda: Honoraria, Membership on an entity's Board of Directors or advisory committees; University of Arkansas for Medical Sciences: Employment; Weismann Institute: Honoraria; Celgene: Honoraria, Membership on an entity's Board of Directors or advisory committees.

Abstract Introduction The spectrum and frequency of mutations in newly diagnosed and relapsed MM has been reported and exhibits a pattern which is distinct from other peripheral lymphoid disorders. A number of data sources have suggested that MM is not a single disease, but rather a collection of molecularly diverse entities which present as malignancies of plasma cells with different clinical courses and response to therapy. To test this hypothesis we examined the spectrum of common mutations and their associations with gene expression in etiological subgroups defined by a simplified TC classification based on translocation subgroup and the deregulation of a D group cyclin. Materials and methods We examined a set of gene expression data from 907 newly diagnosed MM patients and classified them according to an updated TC model. An additional set of 482 cases underwent both sequencing by the FoundationOne Heme targeted panel and gene expression profiling. Mutational data were analyzed for significant associations with gene expression data within five primary TC subgroups: D1, D2, CCND (translocated 11q13 or 6p21), MMSET, and MAF. Results Gene expression signatures of GEP70 risk status, chromosomal aberrations [1q+, 1p-, 13q-, 17p-, HRD], proliferation index, NF-kB activation, and BCL2/MCL1 ratio were not distributed evenly across the TC subgroups, consistent with distinct biological differences amongst groups. Having noted differential expression patterns between subgroups, we hypothesized that distinct patterns of mutation may also exist across TC subgroups. Mutations in RB1 and CDKN2C, seen primarily in the D2 and MMSET subgroups, were mutually exclusive yet both associated with increased proliferation and HR status. Thus two paths to extreme proliferation emerge through either ahomozygous deletion of CDKN2C (with low expression of CDKN2C) or an RB1 alteration (with high expression of CDKN2C). Mutations in KRAS, NRAS, and BRAF were not evenly distributed across subgroups and were significantly inversely associated with an NF-kB signature. RRAS2 was also significantly inversely associated with MAPK mutations in the D1 and D2 subgroups, while genes encoding sprouty-related proteins, SPRED1 and SPRED2, were positively associated with MAPK mutations in the CCND-11q13 subgroup. Overall, MAPK mutations were most significantly associated with elevated expression of DKK1, a known Wnt antagonist, and DUSP6, a known inhibitor of the pathway. These gene expression patterns were primarily localized in the D1, D2, and CCND-11q13 subgroups. In contrast, the MMSET and MAF subgroups had a unique patterns of expression not seen in the D1, D2, or CCND subgroups. A gene set enrichment analysis showed that the DNA replication and cell cycle pathways were significantly enriched in the MMSET and MAF subgroups in the presence of MAPK pathway mutations. Within the MMSET subgroup, MAPK mutations were positively associated with GEP70 HR, proliferation index, and membership of the UAMS-PR subtype, while being negatively associated with FGFR3 expression. This result indicates that the reliance on FGFR3 signaling as an oncogenic driver is lost in the presence of a MAPK activating mutation. Conclusions In MM proliferative signals are delivered via the RAS and NF-kB pathways and activation of these two pathways appears to offer mutually exclusive pathways to disease progression since the majority of cases exhibit a reciprocal relationship between these two signaling pathways. The MMSET and MAF subgroups lack a strong association between NF-kB and MAPK signaling pathways, which may indicate that their initiating translocation event and subsequent genetic patterns provide a unique background in which MAPK alterations accelerate progression. As mutational interactions are differential across TC subgroups, we propose a comprehensive approach to MM classification that includes the etiologic designation by simplified TC subgroups and the subsequent use of genetic and biological markers characteristic of acquired features associated with disease progression, such as MAPK or NF-kB activation. Distinct patterns of RNA expression are associated with DNA mutations in myeloma when contextualized by etiologic subgroups. This indicates that the cellular background in which a mutation occurs has a distinct impact on downstream expression patterns-this observation is particularly relevant to the MMSET and MAF subgroups. Disclosures Pawlyn: Celgene: Consultancy, Honoraria, Other: Travel Support; Takeda Oncology: Consultancy. Morgan:Univ of AR for Medical Sciences: Employment; Takeda: Consultancy, Honoraria; Celgene: Consultancy, Honoraria, Research Funding; Bristol Meyers: Consultancy, Honoraria; Janssen: Research Funding.