Early Prediction of Treatment Response in Newly Diagnosed Multiple Myeloma

Background: The introduction of several novel agents has led to an improvement in response rates and survival outcomes in patients with newly diagnosed multiple myeloma. Despite significant therapeutic advances a subset of patients with multiple myeloma do not achieve a deep response to first-line treatment and biomarkers are needed to identify those at risk.

Methods: We studied 472 patients with newly diagnosed multiple myeloma who were treated in clinical practice (n = 280, training cohort) or on prospective clinical trials (n = 192, validation cohort) at Mayo Clinic between 12/2003 and 12/2015. All patients had measurable disease (M-spike ≥ 1.0 g/dL) and received treatment with novel agents. Serum M-spike and free light chains (FLC) were measured before and after the first treatment cycle. The outcome of interest was the best response to first-line treatment (evaluated using the International Myeloma Working Group Uniform Response Criteria). Failure to achieve a deep response was defined as not achieving a very good partial response or better. The serum parameters of interest were the relative decrease in M-spike and absolute free light chain difference (ΔFLC). The Wilcoxon signed-rank test was used to compare the serum parameters before and after the first treatment cycle. Logistic regression models were used to assess the associations between the change in serum parameters after the first treatment cycle and best response to first-line treatment. P-values below 0.05 were considered statistically significant.

Results: The median age at diagnosis in the training and the validation cohort were 67 (32 - 94) and 66 years (41 - 86), respectively. One hundred eighty one (65%) and 103 patients (54%) were male, respectively. The three most common regimens in the training cohort were lenalidomide + dexamethasone, lenalidomide + cyclophosphamide + dexamethasone, and bortezomib + lenalidomide + dexamethasone. In the validation cohort, the three most common regimens were carfilzomib + thalidomide + cyclophosphamide, lenalidomide + cyclophosphamide + dexamethasone, and ixazomib + cyclophosphamide + dexamethasone. In the two cohorts, the median baseline M-spike decreased from 3.1 g/dL (1.0 - 10.0) to 1.7 g/dL (0.0 - 6.6) and 3.2 g/dL (1.0 - 8.5) to 1.4 g/dL (0.0 - 4.5) after the first treatment cycle, respectively (p < 0.001 for both comparisons). The median baseline ΔFLC decreased from 31.1 mg/dL (0.0 - 2269.7) to 5.4 mg/dL (0.0 - 785.8) and from 23.3 mg/dL (0.1 - 3579.6) to 4.3 mg/dL (0.0 - 1139.3) after the first treatment cycle, respectively (p < 0.001 for both comparisons). M-spike reduction < 50% during the first treatment cycle was associated with failure to achieve a deep response: OR 4.54 (95% CI 2.72 - 7.58, p < 0.001) in the training cohort and OR 6.68 (95% CI 3.42 - 13.03, p < 0.001) in the validation cohort. Patients with reduction in M-spike < 10% during the first treatment cycle experienced failure to achieve a deep response in 86% (25/29) and 100% (10/10), respectively. ΔFLC reduction < 50% during the first treatment cycle was associated with treatment failure to achieve a deep response: OR 4.83 (95% CI 2.83 - 8.25, p < 0.001) in the training cohort and OR 5.09 (95% CI 2.37 - 10.92, p < 0.001) in the validation cohort. Patients with reduction in ΔFLC < 10% during the first treatment cycle experienced failure to achieve a deep response in 83% (34/41) and 91% (19/21), respectively.

Conclusions: Early changes in M-spike and ΔFLC are strong predictors of response to treatment. We established and prospectively validated two readily available biomarkers that can identify patients at risk for treatment failure and may inform treatment decisions in newly diagnosed multiple myeloma.