Development and validation of a prognostic staging system for primary plasma cell leukemia

Abstract

Background: The existing risk models for multiple myeloma (MM) are suboptimal for the stratification of patients with primary plasma cell leukemia (pPCL), a rare and peculiar MM. In this study, we aimed to develop a staging system for pPCL defined as the presence of ≥ 5% circulating plasma cells (CPC) according to the new diagnostic criteria, utilizing one of the largest series of patients with pPCL.

Methods: This multicenter retrospective study included 340 patients with pPCL (the training cohort) from 25 centers nationwide in China. The prognostic impact of baseline characteristics and cytogenetic abnormalities was evaluated. Univariate and multivariate analyses were conducted to identify variables predicting overall survival (OS) to develop a staging system. Its performance was then validated in an independent cohort (n = 80). Genome-wide DNA and RNA sequencing were performed to explore the molecular basis for inter-stage clinical heterogeneity.

Results: Del(17p), t(4;14), and t(14;16), but not 1q+, were verified as high-risk cytogenetic abnormalities (HRCAs) of pPCL. HRCA, elevated LDH, and thrombocytopenia had the highest impact on OS and were used to create a simple algorithm, stratifying patients with pPCL into stages I, II, and III, with median OS of 54.1, 24.0, and 5.4 months (II vs. I: HR, 1.986; 95% CI, 1.034-3.814; P = 0.0394; III vs. II: HR, 3.206; 95% CI, 1.757-5.852; P = 0.0001) in the training cohort and 62.1, 31.6, and 21.8 months (II vs. I: HR, 2.013; 95% CI, 0.954-4.251; P = 0.0664; III vs. II: HR, 2.694; 95% CI, 1.136-6.392; P = 0.0245) in the independent validation cohort. The accuracy (c-index 0.711) was higher than other models. Moreover, patients with different stages had highly diverse genomic and transcriptomic aberrations.

Conclusions: We propose a pPCL-specific staging system based on LDH, thrombocytopenia, and cytogenetic abnormalities, which warrants further validation, particularly in a prospective setting.

Keywords: Biological heterogeneity; High-risk cytogenetics; Overall survival; Primary plasma cell leukemia; Staging system.

Fig. 1

Baseline characteristics and PFS according to level of CPCs. All patients with pPCL (n = 113; the training cohort) and NDMM (n = 1,610). Top, percentage of CPCs (blue, < 5%; green, 5–19%; red, ≥ 20%); middle, baseline characteristics; bottom, PFS. CPCs, circulating plasma cells; LDH, lactate dehydrogenase; ULN, upper limit of normal; ISS, International Staging System; R-ISS, revised International Staging System; PLT, platelet; CA, cytogenetic abnormality; HRCA, high-risk cytogenetic abnormality [defined as del(17p), t(4;14), or t(14;16)]; SRCA, standard-risk cytogenetic abnormality; 1q+, 1q gain/amplification

Fig. 2

Algorithm for the staging system (PSS). (A) Univariate analysis on OS in the training cohort (n = 113). (B) Multivariate analysis of the variables significantly associated with OS. (C) An algorithm for the PSS. M, male; F, female; Y, yes; N, no; β2-MG, β2-microglobulin; ALB, albumin; LDH, lactate dehydrogenase; CPCs, circulating plasma cells; CA, cytogenetic abnormality; HR, high risk; SR, standard risk; ULN, upper limit of normal; FISH, fluorescent in situ hybridization

Fig. 3

Survival according to the staging system. Kaplan-Meier curves for OS (A) and PFS (B) in the training cohort (n = 113), and for OS (C) and PFS (D) in the validation cohort (n = 80)

Fig. 4

Prediction of OS by the staging system. Univariate (A) and multivariate analyses (B) on OS in the training cohort (n = 113). CsCa, serum corrected calcium; Hb, hemoglobin; PLT, platelet; β2-MG, β2-microglobulin; ALB, albumin; LDH, lactate dehydrogenase; ULN, upper limit of normal; CPCs, circulating plasma cells; BMPCs, bone marrow plasma cells; CA, cytogenetic abnormality; HRCA, high risk CA [defined as del(17p), t(4;14), or t(14;16)]; SRCA, standard risk CA

Fig. 5

Subgroup analysis according to treatment. Kaplan-Meier curves of OS in pPCL patients who received induction with doublet (proteasome inhibitor/PI or immunomodulatory agent/IMiD plus dexamethasone) (A) and triplet (PI and IMiD plus dexamethasone) (B), or with (C) and without (D) autologous stem cell transplant (ASCT)

Fig. 6

Comparison between the staging system and other models. (A) Stratification of patients with pPCL by the ISS, R-ISS, and PSS in the training cohort. (B) Migration from the ISS to the PSS. (C) Migration from the R-ISS to the PSS. (D) Decision curves (DCA) of 2-year survival

Fig. 7

Genomic and transcriptomic analyses of CD138⁺ tumor cells in paired PB and BM samples. (A) Heatmap for gene expression profile. (B) Volcano plots of differentially expressed genes (DEGs) by comparing PB with BM (red, up-regulated; blue, down-regulated). (C) GO analysis of PB-specific DEGs. (D) Venn diagrams of PB-specific DEGs (upper, up-regulated; lower, down-regulated). (E) Heatmap of PB-specific DEGs. (F) qPCR analysis of selected PB-specific genes in CD138⁺ cells from additional patients with pPCL. (G) Immunohistochemical staining in a representative patient with NDMM (EM, extramedullary). (H) Kaplan-Meier curves of OS for patients with NDMM in the Hanamura cohort (n = 542). (I) Chromosomal distribution of mutated genes. (J) Venn diagram of mutated genes (color fonts indicate missense mutations: red - mutated genes in all three patients; blue - mutated genes in one or two of three patients)