Unresolved issues in the comparison of therapies and determination of responses in Waldenström macroglobulinemia

Waldenström macroglobulinemia (WM) is an uncommon clinico-pathological syndromic manifestation of lymphoplasmacytic lymphoma. In the more common sub-types of non-Hodgkin lymphoma (NHL) there are internationally accepted robust and reproducible response criteria [1]. As with other forms of NHL, WM can lead to tissue infiltration resulting in the clinical, laboratory, and radiological manifestations of lymphadenopathy, hepato-splenomegaly, and peripheral blood cytopenias. In common with the standard response criteria for NHL, evaluation of reduction or complete resolution of these manifestations can be determined using routine clinical evaluation, laboratory, imaging, and bone marrow histologic evaluation criteria.

However, WM potentially has two unique additional manifestations which complicate response assessment. First, by definition, all cases of WM are associated with a monoclonal immunoglobulin M (IgM) paraprotein. Extrapolating from observations and conventions derived from studies in multiple myeloma (MM), in WM, while there is modest utility in comparing monoclonal IgM levels across patients as a measure of relative tumor burden, serial measurements of the monoclonal IgM paraprotein level in an individual patient are accepted as a quantitative measure of disease burden. While in MM disease progression can occur in the absence of a paraprotein rise, characterized by an increase of the free light chains in the serum only (‘light chain escape’), particularly after a long course of disease with several lines of treatment [2], to our knowledge this has not been reported for WM. In the relatively rare case of WM-transformation to a high-grade lymphoma, this can occur either with or without increasing levels of the pre-existing paraprotein [3]. Thus, appropriately quantitative measurement of the serum monoclonal IgM is included in response criteria for WM [4].

While Rourke et al. state, but do not present data proving, the close correlation of total IgM levels by nephelometry and monoclonal IgM level by protein electrophoresis in their institution [5], it cannot be assumed that this is true for all laboratories, and thus we support the recommendation of the International Consensus Response Criteria that protein electrophoresis only be used for response assessment in WM [4]. Considering further extrapolation from the field of MM, it will likely be very useful to also explore the impact of normalization of free serum light chain levels on quality of response and time to event criteria in patients with WM.

The second unique additional manifestation of WM is due to direct sequelae of the monoclonal IgM, either physical or immunological. Such manifestations can be protean, including tissue amyloid deposition, peripheral neuropathy, cryoglobulinemia, cold agglutinins, or hyperviscosity. Many of these typically resolve with control, even if incomplete, of the WM, but the manifestations of amyloidosis and peripheral neuropathy are very slow to resolve, and may never do so completely, despite morphologic, radiologic, and laboratory ‘complete remission’ of the underlying WM. This is problematic for determination of complete response (CR) in WM according to the proposed criteria of Rourke et al., who recommend that CR requires resolution of ‘signs or symptoms attributable to WM.’ How are the persistent manifestations of amyloid and peripheral neuropathy to be handled in this context? We propose that resolution of these manifestations is not essential for classification of a response as CR, provided all other criteria are met.

At baseline in the setting of significant bone marrow infiltration, peripheral blood cytopenias are considered a manifestation of the underlying WM. Particularly following fludarabine-based therapy, post-treatment cytopenias are relatively common [6]. Again, interpretation of such ‘signs or symptoms’ of WM as peripheral blood cytopenias need to be interpreted cautiously, and must not be attributed to persistent or unresponsive WM in the setting of otherwise CR. As resolution of such manifestations is not required for partial response (PR), it is uncommon that this will complicate response determination. It is reasonable to exclude the requirement for normalization of peripheral blood counts as a manifestation of WM in determination of CR, akin to recent revisions in CLL response criteria [7]. In line with generally accepted recommendations for indolent lymphomas, there is currently no evidence for a role of F18-fluorodeoxyglucose positron emission tomography (FDG-PET) scanning in WM, although this modality seems to be useful in response assessment of MM [8,9].

In our opinion, to enhance the speed with which we evaluate new agents in WM, it is crucial to focus on the statistical power of clinical trials to answer the question under investigation. This will be best achieved through multi-national collaboration. Such studies would also open the opportunity to prospectively define clinical risk factors in WM, as the quoted international prognostic scoring system for WM (IPSS-WM) was established in a retrospective analysis [10]. It is also important to note that the IPSS-WM was established based on the outcome of patients with previously untreated WM, and therefore does not necessarily apply to pretreated patients with the precision suggested by Rourke et al. The prospective evaluation of risk factors in pretreated patients would be another important objective of future clinical trials, as nowadays almost all patients undergo several lines of treatment during their course of disease.

Another important issue that needs to be addressed is the timing of response assessment. Differing from other lymphomas and also from MM, responses in WM can be delayed by many months, as it has been shown particularly after fludarabine-based treatment [11]. Therefore, we recommend that in WM responses are assessed at 3, 6, and 12 months post-treatment in order not to miss the best response achieved. Like in MM, this response should be confirmed in a repeat assessment any time after the initial detection of response.

When the outcome of different treatment options for WM is discussed, clear distinctions should be made between different endpoints, most importantly between response rates (RRs) on the one hand and progression-free survival (PFS), response duration (RD), or overall survival (OS) on the other hand. This is particularly important as, with nucleoside analogs like fludarabine, responses seem to be particularly durable even in previously treated patients [11,12], whereas with other alternatives such as alkylator-based combinations or new agents like bortezomib there appears to be a discrepancy between relatively high RR, and rather short RD/PFS. This may be due to the differential reduction of paraprotein level to an apparently greater degree than the reduction of organ infiltration [13,14].

For the treatment of WM with the anti-CD20 antibody rituximab, it has been shown that also patients achieving only a minor response with a paraprotein reduction by 25–50% benefit with regard to PFS when compared to patients achieving less than a 25% reduction [15]. However, given the already known differences in correlations of responses of paraprotein, tissue infiltration, and response durations after different treatments, such a beneficial effect of a minor response cannot be assumed on a general basis, but still needs to be shown for other treatment regimens. We think that the comparison of PFS or RD amongst different treatment options has much more relevance for patients, and the evidence suggests that in WM the RRs correlate with PFS or OS to varying extents, depending on treatment agents and schedules [11–21].

Nevertheless, RRs do have relevance in WM, as patients who achieve a response usually also experience improvement in disease-related symptoms such as fatigue or signs of hyperviscosity. For treatment with single-agent fludarabine, Rourke et al. report an ‘average’ RR of 34% (see their Figure 1). Although the method used is not stated, it appears to be a simple mean of the response rates for included studies, not accounting for the varying sample sizes. Unfortunately, some relevant studies of fludarabine have been omitted, one with exclusively untreated patients, and in the other trial all but one of the patients were treatment-naive. Both studies report high response rates of 79% and 86%, respectively [20,21]. If these trials are included, the mean RR amongst the five available studies of single-agent fludarabine would be 53% instead of the 34% of the three trials Rourke et al. refer to. Also, where possible, such a ‘meta-analysis’ of RR reported in different trials should distinguish between treatment-naive and previously treated patients for each treatment option under investigation to ensure comparability across agents.

Regarding the definition of different response categories, Rourke et al. recommend the inclusion of the additional response classes near complete remission (nCR) and very good partial remission (VGPR), according to the existing response criteria in MM. However, differing from MM, the significance of nCR or VGPR for PFS or OS has not yet been shown in WM. Studies evaluating these questions should be awaited before additional response categories are used routinely. We agree with Rourke et al. that all future clinical trials in WM should strictly apply the latest definition of response criteria, and should at least also report PFS and OS in order to improve the comparability of chronically scarce data. Time to next treatment may be of interest as well, but is much more confounded by different factors than PFS, e.g. patients' and clinicians' choice, availability of drugs, etc.

Evidence from trials of the much commoner and therefore better studied MM, but also of other low-grade lymphomas, should serve as guidance for building hypotheses that can only be proven or refuted in well-designed clinical trials in patients with WM.

The one recently published, large randomized trial in WM does provide some guidance in the choice of therapies for WM [22]. This trial showed major clinical benefit in response rate, PFS, and OS from the addition of rituximab to conventional chemotherapy (CHOP; cyclophosphamide, doxorubicin, vincristine, prednisone). Rituximab is associated with a substantial risk of ‘IgM-flare,’ which can be especially clinically problematic in patients with high baseline paraprotein levels, and thus must be used with appropriate precautions and monitoring [23]. With these caveats, rituximab should be a component of the primary therapy of all patients with WM requiring treatment, and, in all but the most frail or elderly patients where cytotoxic chemotherapy is felt to be associated with prohibitive risks, this should be in combination with one of the active chemotherapy agents/regimens [24]. Results of the recently completed large randomized trial of chlorambucil versus fludarabine [25] will clarify the relative utility of alkylating agents and nucleoside analogs and will both guide the choice of preferred class of initial chemotherapy agent and inform the design of future international cooperative trials to enhance our evidence base available to select optimal initial therapy for patients with WM.