Serum microRNA Signature Predicting Poor Therapeutic Response to Bortezomib-Based Therapy and Clinical Outcome in Newly Diagnosed Multiple Myeloma: A Result of miRNA Profiling By Deep Sequencing

Introduction

Bortezomib-based regimens are the preferred induction therapy for multiple myeloma (MM) in resource limited regions. Response to bortezomib is inconsistent across patients, and impact of disease heterogeneity on treatment failure remains an unpredictable challenge. Circulating/cell-free miRNA (cfmiRNA) have been shown to contribute to disease prognostication and response monitoring in relapse/refractory setting, serving as an auxiliary biomarker. However, data on real world utility of cfmiRNA in prediction of therapeutic response in newly diagnosed MM (NDMM), treated without transplant is scarce.

Aim

To identify MM-specific cfmiRNA signature and investigate its utility in prediction of initial therapeutic response (ITR) to Bortezomib-based regimens in NDMM. Further, we aimed to study the kinetics of cfmiRNA during induction therapy as a tool for disease monitoring and predicting clinical outcome.

Methods

We prospectively enrolled 141 NDMM patients diagnosed and treated at Tata Memorial Centre, India. Small RNA sequencing for miRNA profiling was performed in 65 NDMM and 10 age-matched control and validated by quantitative PCR (qPCR). cfmiRNA levels were evaluated by qPCR at diagnosis, at the end of 3 cycles (follow up1-FU1) and 6 cycles (follow up2-FU2) of chemotherapy. Changes in cfmiRNA expression at FU1 and FU2 was represented as fold change (FC) relative to baseline. Samples were grouped as high or low expressors using median FC of individual miRNAs as cut-off. Patients received bortezomib-based treatment. ITR was monitored at the end of FU2.

Results

Median age of patients was 55 years (range:27-81 years; M:F-2:1). Median follow-up was 35 months (1-71 months). ITR post FU2 included 7.09% CR, 43.97% VGPR, 33.33% PR, 2.84% SD, 2.84% PD and 9.93% deaths during initial therapy. BM clonal plasma cells (CPC) were detected in 134/141 patients (median, range:97.8%, 1.5-100%). Small RNA sequencing identified 36 miRNAs differentially expressed (DEMs) in MM serum (FC≥2; p-adj<0.05) at diagnosis, the greater majority being upregulated (31/37). DEMs with Log₂FC≥2 was confirmed by qPCR (n=21). There was significant correlation between miRNA levels measured by RNA-seq and qPCR (R²=0.78; p<0.001), with a near-constant bias of 0.6361 for miRNA with Log₂FC≥4 (n=13; Upper, lower limits of agreement:2.463, -1.191). Results were validated in additional 76 NDMM and 10 control samples.

Of the 13 DEMs, high expression (Log₂FC>6.01; p=0.0047) of miR-4665-5p at diagnosis was significantly associated with poor ITR at FU2 (OR:2.92; p=0.01), however, no association with OS or PFS was observed.

We studied changes in cfmiRNA levels at FU1 and FU2 for therapeutic monitoring. Univariate analysis identified 5 miRNA - miR-92b-5p, miR-4775, miR-103a-2-5p, miR-4665-5p and miR-365a-3p significantly associated with poor ITR at FU1 and FU2. Multivariate logistic regression yielded a model (Hosmer-Lemeshow p=0.9) with high expression of miR-92b-5p (Log₂FC>2.04; p=0.006; OR:5.9; p<0.0001) and miR-103a-2-5p (median Log₂FC=1.65; p=0.0329; OR:2.54; p=0.0106) at FU2 significantly associated with poor ITR (p=0.0049). A combination of these miRNAs could differentiate good and poor responders with area under ROC curve of 0.87, 83.64% sensitivity and 75% specificity. Among these, miR-92b-5p demonstrated an association with poor PFS (27 vs. not reached; HR:2.4; p=0.0005) and OS (52 months vs. not reached; HR:2.88; p=0.0056). Univariate analysis also showed high risk cytogenetics, high LDH and CPC>90% of total plasma cells to be associated with poor PFS. Expression of miR-92b-5p at FU2 was independently associated with poor PFS (HR:1.4; p=0.0005) on Cox proportional hazards regression.

Conclusion: We studied the kinetics of serum miRNA and demonstrated that changes in levels of miR-92b-5p and miR-103a-2-5p are useful in prediction of poor response to Bortezomib-based regimen. Increase in levels of miR-92b-5p at follow up time points post induction was found to be an indicator of poor ITR and shorter PFS. The role of miR-92b-5p in the biology of MM is yet to be evaluated. Herein, we highlight the utility of cfmiRNA assays as non-invasive biomarker in prediction of initial therapeutic response to Bortezomib-based protocol. Determining clinical relevance of these findings could expand applications of cfmiRNA as a biomarker.

No relevant conflicts of interest to declare.