Impact of Gain and Amplification of 1q in Newly Diagnosed Multiple Myeloma Patients Receiving Carfilzomib-Based Treatment in the Forte Trial

Background

Copy-number alterations of chromosome 1q are frequently found in multiple myeloma (MM) and are associated with poor prognosis. Recently, it has been demonstrated that the number of 1q copies correlates with a high-risk behavior (BA Walker et al, Leukemia 2019, TM Schmidt et al, Blood Cancer J 2019), but no data are available in carfilzomib-treated patients (pts). Here we aim at dissecting the role of Gain1q (3 copies of 1q) vs amplification 1q (Amp1q, ≥4 copies of 1q) in carfilzomib-treated NDMM pts enrolled in the randomized FORTE trial (NCT02203643).

Methods

Fluorescence in situ hybridization (FISH) in CD138+ purified bone marrow plasma cells (BMPCs) was centralized and performed at baseline. Two hundred BMPC nuclei from each sample were scored. The cut-off level for Gain1q was 10% of nuclei with ≥3 copies of 1q (mean plus 3 standard deviations of 1q alterations in BMPC from 15 healthy donors). The cut-off for Amp1q was 20% of nuclei with ≥4 copies of 1q.

In the FORTE trial, transplant-eligible NDMM pts were randomized to receive carfilzomib (K) lenalidomide (R) dexamethasone (d) induction followed by autologous stem-cell transplantation (ASCT) and KRd consolidation (KRd_ASCT), 12 KRd cycles (KRd12) or K-cyclophosphamide(C)-d induction, followed by ASCT and KCd consolidation (KCd_ASCT). After consolidation, pts were further randomized to receive KR vs R maintenance.

Results

A total of 474 pts were enrolled. Median follow-up from 1^st randomization was 45 months (m). Evaluation of 1q by FISH was missing in 70 pts (15%), while in 4 pts (1%) FISH was present but the number of 1q copies was not evaluable. Among evaluable pts, chromosome 1q was normal in 219 (55%) pts, Gain1q was found in 129 (32%) pts, while Amp1q in 52 (13%). Gain1q- and Amp1q-positive pts were well distributed among treatment arms.

Baseline characteristics associated with Amp1q, compared to Gain1q, were LDH >upper limit of normal (P=0.002) and low hemoglobin (P=0.029) and platelets (P=0.044).

Best response to therapy was not significantly different in Normal 1q vs Gain1q vs Amp1q groups (≥very good partial response rates: 85% vs 84% vs 77%; stringent complete response rates: 52% vs 50% vs 38%). Best overall minimal residual disease negativity by flow cytometry (sensitivity 10^-5) pre-maintenance was also not significantly different (55% vs 55% vs 44%, respectively).

In a multivariate analysis adjusted for treatment and Revised International Staging System (R-ISS), the risk of progression/death was significantly higher in the presence of Gain1q vs Normal 1q (HR 1.65, 95% CI 1.14-2.37, P=0.007) and the highest in the presence of Amp1q as compared to both Normal 1q (HR 3.04, 95% CI 1.99-4.65, P<0.001) and Gain1q (HR 1.84, 95% CI 1.21-2.81, P=0.004; Fig. 1A).

Median progression-free-survival (PFS) was not reached in the Normal 1q group, while Gain1q (53 m) and especially Amp1q (21.8 m) groups performed very poorly.

The presence of Amp1q vs Normal 1q (HR 5.88, 95% CI 3.10-11.17, P<0.001) and Gain1q (HR 3.13, 95% CI 1.73-5.68, P<0.001) predicted a lower overall survival as well (Fig. 1B).

Subgroup analysis on the presence/absence of concomitant high-risk features was performed. Gain1q predicted a lower PFS compared to Normal 1q in the presence of concomitant standard-risk features (ISS 1, ISS 2, standard-risk cytogenetics) but not in the presence of high-risk disease (ISS 3, high-risk cytogenetics). On the other hand, the worse prognosis of Amp1q pts was confirmed across all subgroups.

A subgroup analysis according to the upfront treatment received was performed. Interestingly, treatment with KRd_ASCT completely abrogated the risk conferred by Gain1q (HR 1.25 vs Normal 1q, 95% CI 0.58-2.7, P=0.565), while Amp1q-positive pts still showed a very poor outcome (median PFS 17 m, HR 6.03 vs Normal 1q, 95% CI 2.78-13.1, P<0.001).

In KCd_ASCT and KRd12-treated pts, the 3 groups performed similarly to the overall population.

Conclusion

This is a first report on the prognostic role of the number of 1q copies in carfilzomib-treated NDMM pts. Having ≥4 copies of 1q universally predicts a very poor PFS and OS despite the use of a 2^nd generation proteasome inhibitor upfront. On the other hand, KRd_ASCT completely abrogated the PFS disadvantage conferred by 3 copies of 1q.

RNA sequencing on representative samples of Normal 1q vs Gain1q vs Amp1q is in progress to explore differentially expressed genes in Amp1q pts that could be exploited in future treatment strategies.