Predicting outcomes in multiple myeloma: do we really need another model?

Dytfeld et al. report their single-institution prospective phase II clinical trial using induction therapy consisting of bortezomib, liposomal doxorubicin, and dexamethasone (VDD) in 40 patients with newly diagnosed multiple myeloma [1]. The overall response rate at the completion of six cycles included 85% partial remission (PR) and 57.5% at least very good partial remission (VGPR). In addition, the authors have developed a model which can predict the likelihood of achieving a VGPR based upon responses following two cycles of VDD therapy. The model incorporates the following biochemical parameters: ≥90% reduction in involved free light chain, free kappa/lambda ratio normalization, and ≥90% reduction in serum paraprotein production. The achievement of ≥VGPR was associated with a superior progression-free survival (PFS) and overall survival (OS) compared to those patients whose response was <VGPR. The authors opine that use of this predictive model for achievement of a ≥VGPR after two cycles of therapy may allow for optimizing individual treatment strategies ultimately resulting in improved survival.

The preliminary results of VDD have been previously reported by Jakubowiak et al. [2]. The PR and ≥VGPR rates (85% and 57.5%, respectively) are comparable to other combinations incorporating novel agents with or without an anthracycline: lenalidomide and dexamethasone (Rd); bortezomib and dexamethasone (Vd); lenalidomide, bortezomib, liposomal doxorubicin, and dexamethasone (RVDD); and bortezomib, cyclophosphamide, and dexamethasone (VCD) [3–6]. However, although not compared directly in a phase III setting, the regimen with the deepest response rate is the combination of lenalidomide, bortezomib, and dexamethasone (RVD): partial response (PR) and ≥VGPR are 100% and 67%, respectively [7]. If the goal of therapy is to achieve at least a VGPR, then the RVD regimen may be the preferred choice of treatment. Further, extramedullary toxicity is not a concern (i.e. cardiac monitoring, hand/foot syndrome), except for the potential risk of venothromboembolic events. Thus, the addition of an anthracycline has no advantage over the more commonly used triple therapies of RVD and VCD.

There are a number of well-established prognostic models in myeloma. The International Staging System has been universally accepted [8]. In addition, risk stratification utilizing cytogenetic abnormalities is routinely incorporated into the initial patient evaluation [9]. However, neither of these models predicts for disease response. This study utilizes a predictive model based upon three routinely performed biochemical tests for disease response: involved free light chain, kappa/lambda ratio, and serum M protein [1]. When incorporating all three of these parameters into their model after two cycles of treatment, Dytfeld and colleagues demonstrated a sensitivity and specificity exceeding 90% to predict for a ≥VGPR at the end of the planned six cycles of therapy. Although serum M reduction as a single parameter was also predictive of VGPR, the combination yielded a significantly greater sensitivity and specificity. Since all three of these parameters are readily available and routinely included in response evaluation, this is a simple model to predict the likelihood of achieving a VGPR at the completion of induction therapy. Thus, after two cycles of therapy, clinicians could utilize this model to modify their treatment regimen to increase the likelihood of achieving a VGPR. Other prognostic models predict PFS and/or OS but are not response-based, giving credence to the uniqueness of this model.

The primary endpoint of the current study was the impact of response to induction therapy on survival. They observed that patients achieving a ≥VGPR to induction therapy, with or without consolidative autologous transplant, were associated with an improved PFS and OS. Again, ≥VGPR was predicted by biochemical paraprotein analysis after two cycles of therapy. Achievement of at least a VGPR or a near complete remission post-transplant has been shown to be predictive for outcome in a number of studies (see review by Chanan-Khan and Giralt [10]). Fewer studies have correlated an improvement in PFS and OS by the depth of pre-transplant response [11,12]. One study and one meta-analysis demonstrated the importance of both pre-transplant and post-transplant depth of response and the impact on survival; the latter included 21 studies consisting of 4940 patients [13,14]. In contrast, inability to achieve at least a PR to novel agents pre-transplant is associated with an inferior outcome (PFS 13.1 vs. 22.1 months and OS 30.4 vs. 73.5 months, respectively) [15].

In summary, Dytfeld and colleagues’ model is predictive of achieving a ≥VGPR after completion of induction therapy based upon routine laboratory parameters evaluated after two cycles of treatment [1]. The depth of response, both pre- and post-transplant, is associated with improved PFS and OS. Thus, in theory, this model is of clinical significance: in patients whose biochemical parameters after two cycles of therapy are not predictive to achieve a ≥VGPR, this allows for the option to modify the therapy. However, given the depth of response achieved with the current combination regimens with ≥VGPR achieved in >60–70% of patients, this model really has minimal impact on routine patient management. Further, the best use of the model may be for those patients not predicted to achieve ≥VGPR after two cycles of treatment to hasten proceeding to autologous transplant rather than modifying the initial regimen, since high-dose therapy provides the single best treatment for their disease.

Potential conflict of interest: A disclosure form provided by the author is available with the full text of this article at www.informahealthcare.com/lal.