Long-Term Follow-up Identifies Double Hit and Key Mutations As Impacting Progression Free and Overall Survival in Multiple Myeloma

Introduction: The study of multiple myeloma (MM) genomics has identified many abnormalities that are associated with poor progression free survival (PFS) and overall survival (OS). Copy number abnormalities have been extensively studied in many datasets with long follow-up, however, the prognostic impact of mutations have not been extensively studied and available datasets have generally had a relatively short follow-up of 22-25 months, with one dataset being up to 5.4 years. These analyses have identified a range of mutations that are associated with prognosis, making it important to extend these observations in larger studies with robust diagnostic technologies.

Methods: Samples from newly diagnosed MM patients enrolled in Total Therapy trials (n=199) were sequenced on a targeted panel consisting of 140 genes and additional regions of interest for copy number, as well as tiling of the Ig and MYC loci for detection of translocations. Samples were sequenced to a median depth of 452x using 2x75 bp paired end reads. Reads were aligned to hg19 and mutations called using Strelka and filtered with fpfilter. Translocations were called by Manta, and copy number determined by read depth ratio and loss of heterozygosity comparison with a patient matched non-tumor sample. Additional copy number data were generated by ultra-low pass whole genome sequencing (median 0.5x). Events in <2% of patients were not considered for further analysis. Risk groups including international staging system (ISS), revised-ISS, IMWG risk groups, and Double Hit MM (biallelic TP53 or amp1q with ISS III) were defined.

Results: The median follow-up for this dataset was 8 years, with a median PFS of 6 years and OS of 11 years. The median age was 60.6 years and risk groups were comparable to other studies with 29.1% of patients with ISS III and 20% with high IMWG risk status. In a univariate analysis the markers with highest hazard ratios (HR) for PFS were Double Hit (9%, HR 5.2; 95% CI 2.79-9.76), abnormal BIRC3 (5%, 2.89; 1.32-6.32), ISS III (29%, 2.88; 1.65-5.02), mutation BRAF (11%, 2.26; 1.3-3.93), mutation LRP1B (6%, 2.23; 1.39-3.58), mutation DIS3 (9%, 2.2; 1.22-3.97), bi-allelic inactivation CYLD (10%, 2.04; 1.01-4.10), and high IMWG risk (20%, 2.01; 1.29-3.13). For OS the markers with highest HR were ISS III (5.21; 2.46-11.07), mutation KMT2C (3%, 4.4; 1.37-14.14), t(14;16) (4%, 3.83; 1.38-10.62), mutation EGR1 (4%, 3.58; 1.28-10.00), Double Hit (3.24; 1.65-6.40), mutation BRAF (2.89; 1.57-5.33), mutation LRP1B (2.49; 1.19-5.24), rearrangements surrounding MYC (46%, 2.49; 1.50-4.11), and high IMWG risk (2.11; 1.26-3.53).

In a multivariate analysis for PFS Double Hit (HR 4.37, 95% CI 2.31-8.26), loss of BIRC2/3 (5%, 3.95; 1.69-9.21); mutation LRP1B (3.21; 1.53-6.72), mutation DIS3 (2.44; 1.31-4.53), ISS III (2.29; 1.22-4.32), mutation BRAF (2.28; 1.24-4.18) contributed to the model. For OS, ISS III (3.15;1.40-7.06); 1q21 amp (6%, 2.988; 1.01-8.86); mutation LRP1B (2.90; 1.33-6.35), Double Hit (2.51; 1.05-6.01), deletion CDKN1B (10%, 2.44; 1.15-5.16), and mutation BRAF (2.25; 1.13-4.48) contributed to the model.

Conclusion: We confirm the clinical relevance of Double Hit risk status that constitutes 9% of patients; median PFS of 2 vs. 7 years (P<0.0001), and OS 3 vs. 13 years (P=0.0003). With long follow-up and deep sequencing additional mutated genes associated with adverse outcome were identified including BRAF (11%), DIS3 (9%), LRP1B (6%) and KMT2C (3%). Further, inactivation of NF-κB regulators (CYLD, BIRC2/3) were associated with poor PFS or OS. Patients with a BRAF mutation had a median PFS of 2 vs. 7 years (P=0.003), and OS of 6 vs 13 years (P=0.0004), indicating a potential useful intervention for BRAF inhibitors.