Onsets of progression and second treatment determine survival of patients with symptomatic Waldenström macroglobulinemia

Abstract

Few reports assess prognosis during follow-up of patients with symptomatic Waldenström macroglobulinemia (WM). In 121 WM patients treated between 1993 and 2016, we analyzed the prognostic role during the clinical course of the initial International Prognostic Scoring System for WM (IPSSWM). Then, we assessed onset of response, progression, and second treatment initiation coded as time-dependent covariates. High-risk IPSSWM was an adverse prognostic factor for survival after first treatment initiation (SAFTI). Nevertheless, the corresponding Dxy concordance index obtained in multiple landmark analyses decreased from 0.24 to 0.08 during the first 6 years, in accordance with a departure from the proportional hazard assumption. By contrast with onset of response (whatever its level), onset of progression and initiation of second-line treatment retained prognostic value for SAFTI (P = .02 and P = .006, respectively). These findings were confirmed in cause-specific Cox models for deaths related to WM, but not for unrelated deaths. Time to progression after first-line treatment and time to initiation of second-line treatment had no prognostic value for survival after these 2 events. These results were confirmed in an independent series of 119 patients homogeneously treated with chemoimmunotherapy. Finally, after second-line and third-line treatment, onset of progression had significant prognostic value for subsequent risk of related death only. Thus, taking initial IPSSWM and delayed response to treatment into account, only onset of progression and second treatment initiation provided additional prognostic information for SAFTI. Therefore, progression-free survival or time to next treatment may be satisfactory surrogate end points of SAFTI in WM.

Graphical abstract

Figure 1.

Definition of response, progression, and next treatment initiation coded as time-dependent covariates. Any time-dependent covariate is coded 0 until the occurrence of the event and 1 after, as shown in the figures. (A) Blue curves depict the evolution of SMIC in case of response (solid line) or in case of response followed by progression (solid line followed by dashed line). Time of onset of response, time of onset of failure (or progression), and time of next treatment initiation are available for computation as a time dependent covariate coded 0 or 1 as shown in the figure (respectively blue, green, and red lines). (B) Red curves depict the evolution of SMIC in a patient who failed to respond. In this case, the time of next treatment initiation for any criteria is similar to the time of progression (red and green lines). tdc, time-dependent covariate.

Figure 2.

Competing risk analyses in 115 patients with available SMIC monitoring (69 patients have died). (A) Cumulative incidence of response categories after first-line treatment in patients with symptomatic WM. (B) Cumulative incidence of progression after first-line treatment (87 progressions were recorded, and 21patients died before the occurrence of progression). (C) Cumulative incidence of second treatment initiation after first-line treatment (64 patients required a second-line treatment, and 27 patients died before receiving second-line treatment). (D) Cumulative incidences of related and unrelated death (the competing risk was death unrelated or without definitive conclusion and death unrelated or without definitive conclusion, respectively).