Initial immunoglobulin M ‘flare’ after rituximab therapy in patients diagnosed with Waldenstrom macroglobulinemia

An Eastern Cooperative Oncology Group Study

Authors


  • The current study was conducted by the Eastern Cooperative Oncology Group (Dr. Robert L. Comis, Chairman).

Abstract

BACKGROUND

The goal of the current study was to characterize the initial upsurge in immunoglobulin M (IgM) levels after treatment with rituximab in patients with Waldenstrom macroglobulinemia (WM).

METHODS

As part of a Phase II Eastern Cooperative Oncology Group study, 72 patients were treated with rituximab (375 mg/m2 weekly for 4 weeks) between April 2000 and January 2002. IgM levels in these patients were measured at five separate time points so that any temporal changes that occurred could be characterized.

RESULTS

Of the 54 patients for whom the relevant IgM measurements were available, 29 (54%) experienced an increase in IgM levels between baseline and the first scheduled postbaseline time point. At 2 months, 13 of 22 evaluable patients (59%) continued to have elevated IgM levels, and at 4 months, elevated IgM levels persisted in 4 of 15 evaluable patients (27%). Overall, a nonlinear trend characterized by an initial increase in IgM levels followed by a decrease in these levels was observed (P < 0.0001).

CONCLUSIONS

Treating physicians should be aware that an IgM ‘flare’ may occur in up to 54% of patients treated with rituximab; however, most of these patients experience a decrease in IgM levels within 4 months after the initiation of therapy. Therefore, patients should not be discouraged from continuing to receive this potentially effective therapeutic agent, as responses to rituximab may develop slowly. Longer follow-up will reveal whether patients who experience an upsurge in IgM levels have poorer overall survival or shorter times to progression compared with patients who do not experience this IgM flare. Factors predicting an initial increase in IgM levels could not be identified. Cancer 2004. © 2004 American Cancer Society.

Waldenstrom macroglobulinemia (WM) is a low-grade lymphoproliferative disorder characterized by the presence of lymphoplasmacytic infiltrates in bone marrow and by the presence of the monoclonal protein immunoglobulin M (IgM) in serum.1–4 The options currently available for the treatment of this malignancy include alkylating agents,5 purine nucleoside analogs,6–8 and rituximab.9, 10 In recent years, rituximab has become a key therapeutic option for patients with WM, due to its selective activity and tolerability, as well as to the absence of late myelodysplasia and leukemia in rituximab recipients.9, 11 Reported response rates in patients who receive rituximab for previously untreated or recurrent WM range from 30% to 75%.9, 12{FNO}

Previous reports have demonstrated that some patients treated with rituximab or cladribine may develop transient increases in monoclonal protein levels.9, 13 Dimopoulos et al.9 found that this phenomenon occurred 15-30 days after the initiation of treatment in a small number of patients, including both those who experienced responses and those who did not. Nonetheless, previous reports describing this phenomenon involved relatively small patient cohorts, did not specify the prevalence or the duration of the observed IgM upsurges, and did not examine the correlation between elevated IgM levels and overall response. Better understanding of the IgM ‘flare’ phenomenon is critical, because such knowledge can prevent patients with WM from having a potentially effective treatment regimen discontinued prematurely. Therefore, the objectives of the current study were to characterize this phenomenon, to assess its correlation with response to therapy, and to identify factors that predict its development.

MATERIALS AND METHODS

As part of a prospective, two-stage Phase II Eastern Cooperative Oncology Group (ECOG) study (E3A98), 72 patients with WM were treated with rituximab (375 mg/m2 weekly for 4 weeks) between April 2000 and January 2002. After receiving rituximab intravenously over the course of 4 hours on Days 1, 8, 15, and 22, patients were followed every 3 months for the next 2 years, every 6 months for 3 years thereafter, and then once yearly. Inclusion criteria included age ≥ 18 years and an ECOG performance status of 0-3. Diagnoses were confirmed by the presence of 1) bone marrow lymphoplasmacytosis (> 10% lymphoplasmacytic cells) and measurable disease, defined by quantitative IgM monoclonal protein levels > 1000 mg/dL; 2) impaired bone marrow functioning due to infiltration by lymphoplasmacytic lymphoma; or 3) WM accompanied by clinically significant anemia (hemoglobin concentration < 11 g/dL), bulky lymphadenopathy, or symptoms attributable to hyperviscosity. Eligible patients also were required to have received no more than two previous treatment regimens and to have received no anti-CD20 therapy.

A complete response was indicated by a serum M-spike of zero, the complete absence of IgM in serum, the complete absence of any IgM light chain in urine (as detected by immunofixation analysis), and the observation of < 5% lymphocytes on bone marrow biopsy. An objective response was indicated by a reduction of > 50% in IgM levels as measured using quantitative immunoglobulin analysis. Finally, a minor response was indicated by a reduction of 25-50% in IgM levels as measured using quantitative immunoglobulin analysis and a 25-50% decrease in urine light chain excretion, if present at levels > 150 mg per 24 hours before treatment. The response evaluation process involved 2 measurements made 4 weeks apart from one another.

IgM flare was defined as any increase (relative to baseline) in IgM levels that occurred before the end of therapy. An alternative analysis, in which IgM flare was defined as an increase in IgM levels of ≥ 25% relative to baseline, also was performed. Temporal changes were ascertained by measuring IgM levels at 5 different time points: Time Point 0 (baseline), Time Point 1 (termination of rituximab therapy—i.e., 1 month from the start of treatment), Time Point 2 (2 months from the start of treatment), Time Point 3 (3 months from the start of treatment), and Time Point 4 (4 months from the start of treatment).

The current prospective study was approved by the institutional review board at each participating institution and was conducted in accordance with the ethical guidelines mandated by the Declaration of Helsinki.

Statistical Analysis

Descriptive statistics were used to characterize the current study cohort, as well as to characterize IgM levels and the observed temporal changes in these levels. The Fisher exact test14 was used to test for associations between various dichotomous clinical characteristics and IgM flare, and the Wilcoxon rank-sum test15 was used to evaluate differences in continuous clinical variables between patients who experienced an IgM flare and those who did not. In addition, estimates of overall and progression-free survival were obtained using the method of Kaplan and Meier,16 and differences in survival between patients who experienced an IgM flare and those who did not were assessed using the log-rank test.17

Overall survival was calculated from the time of study registration to the time of death. Each patient was censored at the most recent date on which he or she was known to be alive. Progression-free survival was defined as the time from study registration to progression or death. Patients who did not experience progression were censored at the most recent date on which they were known to have stable disease or an enduring response to treatment.

Aside from these survival analyses, a repeated measures analysis was performed to model and test for significant changes in IgM levels over time. In this analysis, both linear and nonlinear IgM response profiles were fitted to the study data; to account for the repeated measurements made within each patient, a compound symmetric covariance structure was assumed. Contrasts were used to ascertain pairwise differences between IgM levels at baseline and IgM levels at postbaseline time points.18

RESULTS

Patient Characteristics

The median patient age at enrollment was 68 years (range, 44-89 years). There were 38 male patients (53%). Twenty-eight patients (41%) had previously received single-agent chemotherapy, 11 (16%) had previously received multiagent chemotherapy, and 32 (46%) had previously received alkylator-based chemotherapy. In total, 72 patients were enrolled in the study. Data on IgM levels at baseline and/or Time Point 1 were not available for 18 of 72 patients (25%); of the remaining 54 patients, 29 (54%) had elevated IgM levels (relative to baseline) at Time Point 1 (i.e., 1 month from the start of rituximab therapy) (Fig. 1). By Time Point 2, nine of the 22 evaluable patients who had elevated IgM levels at Time Point 1 (41%) had experienced a decrease in IgM levels to below baseline, whereas 13 patients (59%) continued to have elevated IgM levels; data on IgM levels at Time Point 2 were unavailable for 7 patients. Only 2 of the 7 evaluable patients who had elevated IgM levels at Time Point 2 (29%) continued to have elevated levels at Time Point 3, whereas the remaining 5 patients (71%) saw their IgM levels decrease to below baseline. By Time Point 4, only 4 of the 15 patients for whom data were available (27%) had elevated IgM levels.

Figure 1.

Serially measured immunoglobulin M (IgM) levels in patients who experienced an initial IgM ‘flare’. Twenty-nine patients (54%) experienced an increase in IgM levels between baseline and Time Point 1. By Time Point 4, however, 11 of the 15 patients for whom data were available (73%) had IgM levels that were below baseline levels.

To explore the possibility of a delayed flare phenomenon, we performed a separate analysis of patients whose IgM levels had initially decreased. Twenty-five of 54 patients (46%) experienced a decrease in IgM levels between baseline and Time Point 1. By Time Point 2, three of the 21 evaluable patients who had reduced IgM levels at Time Point 1 (14%) had experienced an increase in IgM levels to above baseline. At Time Point 3, all 4 patients for whom data were available had IgM levels that were reduced relative to baseline, and at Time Point 4, only 1 of 17 patients for whom data were available (6%) had IgM levels that exceeded those measured at baseline.

When IgM flare was defined as an increase in IgM levels of ≥ 25% relative to baseline, 14 of 54 patients (26%) were found to have experienced an IgM flare by Time Point 1. At Time Point 2, seven of the 9 evaluable patients who had experienced an initial IgM flare (78%) continued to have elevated IgM levels, whereas the remaining 2 patients (22%) did not. At Time Point 3, elevated IgM levels persisted in 1 of 2 evaluable patients (50%), and at Time Point 4, IgM levels were found to be elevated in 1 of the 3 patients for whom data were available (33%) (Table 1). Only 1 of the 54 patients in the current study (2%) had an initial increase in IgM levels of ≥ 50% relative to baseline.

Table 1. Median Immunoglobulin M Levels for Patients Who Experienced an Increase of ≥25%a in Immunoglobulin M Levels and for Patients Who Did Not
Time point≥25% increaseDecrease or <25% increase
No. of patientsMedian IgM level (mg/dL)Range (mg/dL)No. of patientsMedian IgM level (mg/dL)Range (mg/dL)
  • IgM: immunoglobulin M.

  • a

    Relative to baseline.

01437751160–51404046771125–13,880
11460252590–8420404540630–15,700
2953902840–6830344049566–12,400
3243602950–577092284466–13,600
4345103050–5520293390322–1290

Follow-Up Measurement of IgM Levels

Median IgM levels throughout the course of treatment are presented in Tables 2 and 3. Twenty-nine patients (54%) experienced an increase in IgM levels between baseline and Time Point 1 (Fig. 2). Among these patients, the median IgM level was 4440 mg/dL (range, 1160-13,880 mg/dL) at baseline and increased to 5710 mg/dL (range, 1510-15,700 mg/dL) at Time Point 1 (i.e., 1 month from the start of treatment). The median within-patient difference between IgM levels at baseline and IgM levels at Time Point 1 was 1150 mg/dL. At 2 months, 13 of the 22 patients for whom data were available (59%) continued to have elevated IgM levels, and at 4 months, elevated IgM levels were documented once more in 4 of the 15 patients for whom data were available (27%). Six patients experienced increases of ≥ 2000 mg/dL in their IgM levels between baseline and Time Point 1. Among these 6 patients was an individual whose IgM level increased by 2200 mg/dL between baseline and Time Point 1; this patient's IgM level fell to below baseline after 5 months, and the patient continued to experience disease remission throughout 17 additional months of follow-up.

Table 2. Median Immunoglobulin M Levels for Patients Who Experienced an Immunoglobulin M ‘Flare’ and for Patients Who Did Not
Time pointFlareNo flare
No. of patientsMedian IgM level (mg/dL)Range (mg/dL)No. of patientsMedian IgM level (mg/dL)Range (mg/dL)
  1. IgM: immunoglobulin M.

02944401160–13,8802546261125–9900
12957101510–15,700254042630–9690
2224840873–12,400213460566–7900
374630689–13,60041632.5466–2284
4154510510–12,900172908322–7090
Table 3. Median Immunoglobulin M Levels for the Trial Cohort as a Whole
Time pointNo. of patientsMedian IgM level (mg/dL)Range (mg/dL)
  1. IgM: immunoglobulin M.

07042701125–13,880
1564795630–15,700
2544054566–12,400
3132950466–13,600
4403220322–12,900
Figure 2.

Serially measured immunoglobulin M (IgM) levels in 72 patients treated with weekly rituximab. Temporal changes were characterized by measuring IgM levels at five separate time points. Data presented are log-transformed IgM levels.

The overall effect of time on IgM levels was found to be significant (P < 0.0001), with repeated measures analysis revealing a nonlinear trend characterized by an initial increase and a subsequent decrease in IgM levels (Fig. 2). The P value associated with this nonlinear (quadratic) trend was 0.009.

Within the context of our model, we used contrasts to test for pairwise differences between IgM levels at baseline and IgM levels at postbaseline time points. No significant difference was found between IgM levels at baseline and IgM levels at Time Point 1 or Time Point 2. In contrast, significant differences were noted between IgM levels at baseline and IgM levels at Time Point 3 (P = 0.002) and also between IgM levels at baseline and IgM levels at Time Point 4 (P < 0.0001). These findings suggest that IgM levels decreased nonlinearly between baseline and Time Points 3 and 4. Based on our model, it appears that before these decreases, there was an initial increase in IgM levels; however, the existence of a statistically significant increase between baseline and Time Point 1 could not be established.

A significant association was found between response to therapy and the IgM flare phenomenon (P < 0.001). Using criteria that are described elsewhere in the current report, responses were characterized as being complete, objective, or minimal. Twenty-eight percent of all patients who experienced increases in their IgM levels had responses to therapy, compared with 80% of all patients who did not experience this IgM flare. Nonetheless, there was no statistically significant difference in terms of overall (P = 0.7) or progression-free survival (P = 0.3) between patients who experienced an IgM flare and those who did not (Figs. 3, 4). Furthermore, although there was a significant association between IgM flare and response to therapy (P < 0.001), one potential confounder is the fact that responses themselves were defined according to percent changes in IgM levels.

Figure 3.

Overall survival. Patients who experienced an immunoglobulin M ‘flare’ (dotted line) did not differ from patients who did not (solid line) in terms of overall survival (log-rank test: P = 0.73).

Figure 4.

Progression-free survival (PFS). Patients who experienced an immunoglobulin M ‘flare’ (dotted line) did not differ from patients who did not (solid line) in terms of PFS (log-rank test: P = 0.37).

Serum viscosity was measured at the start of therapy, with no follow-up measurements made during the course of treatment. No association was found between baseline serum viscosity and IgM flare (P = 0.95). Thus, with the goal of identifying factors that could predict the occurrence of this phenomenon, we subsequently analyzed baseline IgM levels, previous therapy, and percentage of malignant cells in bone marrow for potential associations with IgM flare; however, no significant associations were found (P = 0.4, P = 0.1, and P = 0.2, respectively).

DISCUSSION

Although previous reports have described the occurrence of an upsurge in IgM levels in patients treated with rituximab or cladribine,9, 13 many questions regarding this IgM flare remain unanswered. The objectives of the current study were to better characterize this phenomenon, to investigate its potential association with response to therapy, and to identify factors that could predict its occurrence. Our findings suggest that among patients receiving rituximab, IgM ‘flares’ are relatively common, occurring in up to 54% of cases, and can persist for up to 4 months after the start of therapy. A smaller proportion of patients (14%) may experience an increase in IgM levels only after an initial decrease, although the mechanism underlying this delayed flare phenomenon is not well understood.

Elevated IgM levels during rituximab therapy are not indicative of treatment failure. After 4 months of follow-up, IgM levels had decreased to below baseline in 73% of all patients who initially had elevated levels. Treating physicians should be aware of the delayed nature of responses to rituximab therapy and should not be discouraged if patients experience an initial upsurge in IgM levels. Slow responses to cytotoxic therapy have also been documented in patients with WM.5

Response rates were poorer for patients who experienced an initial IgM flare compared with those who did not (28% vs. 80%). Long-term follow-up is necessary to determine whether patients in the former group have a shorter time to progression compared with patients in the latter group. Previous therapy, percentage of malignant cells in bone marrow, baseline serum viscosity, and baseline IgM level were not associated with the occurrence of the IgM flare phenomenon in the current study. Conversely, elevated IgM levels may have an effect on serum viscosity, as Dimopoulos et al.9 documented two cases in which patients with increased IgM levels subsequently developed hyperviscosity and required plasmapheresis. Because viscosity measurements were performed only before the start of treatment in the current study, we were not able to ascertain hyperviscosity or the need for therapeutic plasmapheresis. Nonetheless, no patient who experienced an initial increase in IgM levels had his or her treatment modified secondary to this increase.

Treating physicians must be aware that although the IgM flare phenomenon may be observed in up to 54% of patients treated with rituximab, most of these patients will see their IgM levels decrease within 4 months after the start of therapy. Thus, physicians should not be discouraged from continuing to administer this potentially effective therapeutic agent, as responses may develop relatively slowly. Factors predicting the occurrence of the IgM flare phenomenon could not be identified, but additional follow-up will reveal whether this phenomenon is associated with poorer overall and progression-free survival.

Ancillary