Definition of a Prognostic Risk-Classifier Based on SNPs Array Analysis of Copy Number Alterations in Newly Diagnosed Transplant-Eligible Multiple Myeloma Patients

BACKGROUND. The genomic heterogeneity that characterizes all Multiple Myeloma (MM) patients (pts) is the consequence of an evolutionary process operating during the pre-neoplastic phase of the disease. Whatever genomic event occurs, it accounts for the de-regulated expression of either one of the cyclin D or other key regulators of cell cycle progression. Therefore, a deficient regulation of cell cycle control is considered as a hallmark of MM. Moreover, among the MM genomic heterogeneity, several lesions are recurrent and their presence significantly impacts the prognosis.

AIM. To deeply characterize the MM clone of newly diagnosed MM pts by means of a genomic approach, aimed to identify Copy Number Alterations (CNAs) affecting clinical outcomes.

PATIENTS AND METHODS. We analyzed 247 pts who were enrolled in the EMN02/HO95 phase III study and were eligible for randomization to receive standard dose intensification therapy with bortezomib-melphalan-prednisone (VMP) (n=93) or high dose intensification therapy with autologous stem cell transplantation (ASCT) (n=154) after a V-based induction. Genomic data were obtained from all the pts, whose highly purified CD138+ BM plasma cells were profiled by SNPs array. Any CNA ≥50kb was evaluated. At a median follow-up of 34 months (m), the estimated progression-free (PFS) and overall survival (OS) was 63% and 85%, respectively.

RESULTS. In order to identify genetically homogeneous subgroups of pts, we first dissected the genomic complexity observed in the overall population. As expected, most pts had a complex genomic background, while occasionally few of them (23/247, 9%) carried only one of the following alterations: CN losses (L) on chr. 13q, CN gains (G) on chr.1q, hyperdiploidy and any IgH translocations. Remarkably, these 4 alterations were also the most frequently observed across the overall pt population, and their combination identified 16 subgroups harbouring a progressively more complex genomic profile. This suggests that these 4 alterations - either combined or alone - are both necessary and sufficient to describe the genomic background of the whole population. Indeed, all the other observed CNAs have a lower frequency and were randomly scattered among the 16 subgroups.

When the clinical outcomes of these subgroups were analyzed, it was found that those with inferior outcomes carried at the same time both CN-L on chr.13q and CN-G on chr.1q, regardless of others genomic lesions. Based on this, we re-stratified pts according to the absence or presence of 1 or both of these lesions, to define the following 3 groups: group 1 (64 pts), carrying both CN-L on chr.13q and CN-G on chr.1q; group 2 (94 pts), carrying either CN-L on chr.13q or CN-G on chr.1q; group 3 (89 pts), lacking both these lesions.

Providing that pts included in these groups were correctly randomized in a 1:1 ratio to receive either VMP or ASCT, we observed that the 34-m estimates of PFS was significantly shorter in group 1 (48%, 61% and 72% respectively; p=0.0021). OS rates were 72% in group 1 vs 88% in group 2 and 91% in group 3 (p=0.0001).

PFS hazard ratio (HR) of pts in group 1 (HR=2,51, 95%CI: 1.48-4.25, p=0.0006) was comparable to that of pts carrying either del(17p) (HR=2.09, 95%CI: 1.20-3.65) or t(4;14) (HR=1.84, 95%CI: 1.14-2.99), resulting as an independent factor in predicting PFS in a Cox multivariate analysis (both from del(17p) and t(4;14), p=0.01). PFS benefit correlated with ASCT in comparison with VMP was statistically significant in groups 2 and 3, while only a trend was observed in group 1 (table 1).

The genomic background of pts belonging to group 1 was characterized by the presence of several genomic aberrations, which overall severely compromise genes involved in the control of cell cycle progression (Rb1, CKS1B, MDM4, MCL1, genes coding for the DREAM-complex components, YAP1, MYC, FANCA).

CONCLUSIONS. Based on an evolutionary rationale, we proposed a simply, CNAs-based risk-classifier, which was able to segregate our cohort of pts with newly diagnosed MM into 3 well-populated groups, thus recapitulating most of the genomic aberrations commonly described in MM. The 3 groups have different clinical outcomes and PFS benefit imparted from ASCT as compared to VMP. According to the observed genomic background, a strong de-regulation of cell cycle control might be responsible of the dismal outcome of pts carrying both CN-L on chr.13 and CN-G on chr. 1q.

Supported by AIRC, Fond. del Monte Bo e Ra

View largeDownload slide

View largeDownload slide

Close modal