Comprehensive Biologic Characterization of 99 Multiple Myeloma Patients Using Cytomorphology, FISH, Gene Expression Profiling and Mutation Screening Leads to Important Clinical and Therapeutic Insights

Introduction:

Multiple myeloma (MM) is a genetically heterogeneous disease regarding both chromosomal abnormalities (CA) and dysregulated gene expression. Lately, several gene mutations (mut) have been identified further contributing to the genetic complexity. However, data on parallel assessment of morphologic, cytogenetic and molecular genetic parameters is scarce.

Aim:

Integration of morphological and extensive genetic information in an MM cohort to improve understanding of the disease biology and classification, providing a basis for evaluation of the most suitable therapy for MM patients (pts) by this holistic approach for future studies.

Methods:

We investigated 99 newly diagnosed MM pts (46 female, 53 male; median age 69 years, range 43 - 88). Plasma cell (PC) in bone marrow by cytomorphology ranged from 10 to 96% (median 54%). PC were enriched by magnetic-activated cell sorting targeting CD138 (median purity 95%) before interphase FISH was performed to detect hyperdiploidy, del(13q), del(17p), t(4;14), t(11;14), t(14;16), t(14;20), t(6;14), 1q gain, del(12p) and MYC rearrangements. Purified samples were further analyzed by next generation sequencing (NGS) using a comprehensive 36-gene panel targeting genes previously described mutated in MM. Library was prepared by TSCA-LI Multiple Myeloma Panel (Illumina, San Diego, CA). Gene expression profiling (GEP) was performed using Affymetrix HG U133 Plus 2.0 arrays. The MMprofiler assay algorithms were used to calculate the SKY92 signature classification into standard/high risk groups (Kuiper et al., 2012).

Results:

The frequencies of CA detected by FISH were consistent with published data. According to R-ISS high risk (hr) CA was defined by del(17p), t(4;14) and/or t(14;16) (28/98 pts, 29%). All other cases (71%) were standard risk (sr) (Palumbo et al., 2015).

First, GEP were analyzed in relation to the CA risk groups. Cluster analysis revealed the majority of hr CA pts clustering together with overexpression of genes including ROBO1, CCNB2, FGFR3, WHSC1, DSG2 and PBX1, consistent with prior publications on hr GEP signature (Shaugnessy et al., 2007; Zhan et al., 2006). However, 9 pts assigned hr by CA clustered together with sr CA cases. Thus, in 10% of our pts GEP clusters would not be concordant with the risk classification by CA.

In addition, the expression data were also analyzed based on the SKY92 signature. In consistency with published data this analysis assigned 16 pts (17%) as hr (Kuiper et al., 2012). Interestingly, out of the 9 hr CA pts mentioned above which clustered with sr CA 8 pts were also assigned sr by SKY92 classifier. Further, regarding the hr CA group, 8/28 pts (29%) also revealed a hr SKY92 signature. These patients may need further attention. Further, regarding the sr CA group, 7/65 (11%) revealed a hr SKY92 signature.

Focusing on NGS, we found 115 mut (with mut load ≥10%) in 67/93 pts (72%; range 0-5) affecting 17 genes. Most commonly mut genes were NRAS (26%), KRAS (21%), TLR4 (11%), BRAF (8%), FAM46C (8%) and TP53 (7%). No difference in mut frequency between hr and sr CA pts was observed. However, association of FAM46Cmut with hyperdiploidy as well as CCND1mut with t(11;14) could be corroborated (12% vs 0%, p=0.095; 9% vs 0%, p=0.056, respectively). Besides, FAM46Cmut was associated with del(17p) (23% vs 5%, p=0.058) and a strong association of KRASmut to 1q gain was found (32% vs 11%, p=0.029), while KRASmut and NRASmut were mutually exclusive of del(12p) and t(4;14). Of note, 3/6 TP53mut pts concomitantly harbored del(17p) detected by FISH. According to CA 2/3 of these pts without del(17p) would have been classified sr. Thus, molecular data might improve risk classification.

Considering biological pathways (pw) connected to currently used therapeutics (e.g. vemurafenib, bortezomib), we found the following mut frequencies in genes associated with the respective pw: 46% in MAPK/ERK-pw, 16% in NFkB-pw, 2% in HOXA9-pw and 15% associated with RNA processing. Interestingly, we found mut in IKZF1 (6%), IKZF2 (6%), IKZF3 (1%) and IRF4 (4%), which represent critical targets of the immunomodulatory drug/CRBN mediated anti-tumor activity.

Conclusion:

1) A comprehensive analysis of MM based on cytogenetic, gene expression and mutation data may lead to the identification of new biologic subgroups. 2) Molecular mutations should be further evaluated to allow precision medicine approaches including respective pathway components.