Relevance of Bone Marrow Biopsies for Response Assessment in US National Cancer Institute National Clinical Trials Network Follicular Lymphoma Clinical Trials

PURPOSE

Bone marrow biopsies (BMB) are performed before/after therapy to confirm complete response (CR) in patients with lymphoma on clinical trials. We sought to establish whether BMB add value in assessing response or predict progression-free survival (PFS) or overall survival (OS) outcomes in follicular lymphoma (FL) subjects in a large, multicenter, multitrial cohort.

METHODS

Data were pooled from seven trials of 580 subjects with previously untreated FL through Alliance for Clinical Trials in Oncology (Alliance) and SWOG Cancer Research Network (SWOG) completing enrollment from 2008 to 2016.

RESULTS

Only 5/580 (0.9%) had positive baseline BMB, CR on imaging, and subsequent positive BMB (P < .0001). Therefore, BMB were irrelevant to response in 99% of subjects. A sensitivity analysis of 385 FL subjects treated on an Eastern Cooperative Oncology Group study was included. In the Eastern Cooperative Oncology Group cohort, 5/385 (1.3%) had BMB that affected response assessment. Since some subjects do not undergo confirmatory BMB, we performed a landmark survival analysis from first radiologic CR with data from 580 subjects from Alliance and SWOG. Of subjects with CR on imaging (n = 187), PFS and OS were not significantly different among those with negative BMB to confirm CR (n = 47) versus those without repeat BMB (n = 140; PFS: adjusted hazard ratio, 1.10, 95% CI, 0.62 to 1.94, log-rank P = .686; OS: hazard ratio, 0.59, 95% CI, 0.23 to 1.53, log-rank P = .276).

CONCLUSION

We conclude that BMB add little value to response assessment in subjects with FL treated on clinical trials and we recommend eliminating BMB from clinical trial requirements. BMB should also be removed from diagnostic guidelines for FL except in scenarios in which it may change management including confirmation of limited stage and assessment of cytopenias. This would reduce cost, patient discomfort, resource utilization, and potentially remove a barrier to trial enrollment.

INTRODUCTION

Clinical trial requirements can be burdensome and deter patients from participation. Initiatives through ASCO focus on simplifying such protocols in a patient-centered approach to encourage enrollment.^1-3 In clinical practice of follicular lymphoma (FL), utility of bone marrow biopsies (BMB) is controversial. The National Comprehensive Cancer Network (NCCN) guidelines for FL recommend BMB and aspiration in certain circumstances including to confirm stage I-II disease in patients being considered for radiotherapy or in those who require investigation of cytopenias. They note that BMB can be omitted for patients being observed without treatment.⁴ The European Society for Medical Oncology Clinical Practice Guidelines advise that BMB and aspiration should be done in all patients with newly diagnosed FL, particularly those with suspected early-stage disease.⁵ Response criteria for clinical trials in lymphoma require that BMB be performed at baseline and then repeated to confirm complete response (CR) in those with positive baseline BMB.^6-8

CONTEXT

• Key Objective

• Are bone marrow biopsies (BMB) relevant for response assessment in follicular lymphoma (FL) clinical trials?

• Knowledge Generated

• In 99% of subjects with FL enrolled on National Clinical Trials Network clinical trials, response is unchanged on the basis of BMB results. Among subjects with complete response on imaging, there is no progression-free survival or overall survival difference in those with negative confirmatory BMB versus those who do not undergo the procedure.

• Relevance (S. Lentzsch)

• BMB should be removed from diagnostic guidelines for FL except in scenarios in which it may change management including confirmation of limited stage and assessment of cytopenias.*

  *Relevance section written by JCO Associate Editor Suzanne Lentzsch, MD, PhD.

We hypothesized that only rarely do subjects have a positive baseline BMB, CR on imaging, and then a positive subsequent BMB—the only scenario in which BM assessment could change response determination. We then investigated 99 subjects with FL treated on clinical trials at a single institution and found that 1.0% had a BMB that could have affected response assessment.⁹ We performed a similar analysis of the randomized GALLIUM clinical trial, which enrolled untreated subjects with FL to obinutuzumab plus chemotherapy versus rituximab plus chemotherapy followed by obinutuzumab or rituximab maintenance.¹⁰ We found that only 5/1,202 subjects (0.4%) had BMB that affected response assessment when computed tomography (CT)–based International Working Group (IWG) 2007 response criteria were used.¹¹ The GALLIUM trial required fluorodeoxyglucose-positron emission tomography (PET) imaging in the first 170 enrolled subjects, and PET was optional in subsequent subjects. PET was performed in 282 patients with positive/indeterminate baseline BMB. Two hundred thirteen of these patients underwent confirmatory BMB. Of 213 subjects with positive or indeterminate baseline BMB who underwent PET and repeat BMB after treatment, BMB were relevant for response assessment in a maximum of 10 (4.7%, five positive BMB and five indeterminate). We sought to confirm these results in subjects with untreated FL enrolled on National Cancer Institute National Clinical Trials Network (NCTN) trials. Our goals were to validate findings of preliminary studies to foster efforts to simplify future clinical trial requirements for subjects who may be discouraged from participation in clinical trials, and to change practice guidelines requiring BMB in the majority of patients with FL. In addition to encouraging patient participation in clinical trials and minimizing patient discomfort, this effort would also decrease cost and utilization of resources.

METHODS

We identified all clinical trials completing enrollment in the modern therapeutic era from 2008 to 2016 by the Alliance for Clinical Trials (Alliance and legacy Cancer and Leukemia Group B [CALGB]) and SWOG Cancer Research Network (SWOG) of subjects with untreated FL for which BMB results at baseline and during response assessment, and best response by imaging, were available. All studies were approved by the institutional review board at each participating site, and informed consent forms were signed by all subjects enrolled on the trials. For each study, and for all studies combined, we calculated the proportion of subjects with positive baseline BMB, CR on imaging after treatment, and positive repeat BMB using the total number of subjects enrolled as the denominator. These are the only subjects in whom BMB would affect response assessment. The majority of studies used IWG 2007 guidelines with five of eight protocols stating that BMB morphology and immunohistochemistry were required to be negative (with goal > 20 mm core) but that bone marrow aspirate was not required to be negative to confirm CR. Those five protocols indicated that a small population of clonal lymphocytes by flow cytometry in the bone marrow aspirate was considered a CR until data became available demonstrating a clear difference in patient outcome. Similarly, IWG 2007 guidelines indicate that histologically normal bone marrows with a small (< 2%) B-cell population detected by flow cytometry should be considered normal.⁷ The SWOG and Eastern Cooperative Oncology Group (ECOG) studies did require both biopsies and aspirates to be negative to confirm CR.

Statistical analysis was conducted by the Alliance Statistics and Data Center. We tested against the null hypothesis that this proportion was ≥ 10%, versus the alternative hypothesis that this proportion was < 10% (the threshold below which BMB would be considered inconsequential for response assessment), using a one-sided exact binomial test. With 500 subjects, we would have 99% power to reject that this proportion is > 10% when the true proportion is 5% using a binary test of a single proportion with one-sided alpha = .025. The power calculation was not a design feature of the study. This is a secondary analysis using existing data from completed NCTN clinical trials. Response criteria were CT-based. Imaging was not used to assess for BM involvement. BMB were unilateral in six clinical trials and bilateral in two.

Because confirmatory BMB were not completed in all indicated subjects, landmark survival analyses compared progression-free survival (PFS) and overall survival (OS) of subjects with CR on imaging and negative BMB versus subjects with CR on imaging without repeat BMB. This analysis was performed in subjects with a positive baseline BMB. Subjects with CR on imaging were categorized as having negative repeat BMB or no repeat BMB within 60 days of the first CR on imaging. PFS was defined from time of CR to progression or death. OS was defined as time of first CR to death. The time-to-event end points were calculated using a landmark analysis approach from first radiologic CR and estimated using Kaplan-Meier, and compared using log-rank tests, as well as univariate and multivariate Cox models adjusted for age, sex, stage, and Follicular Lymphoma International Prognostic Index (FLIPI) score, and stratified by treatment arm within study. An ECOG trial was analyzed separately as a sensitivity analysis, as it included only one time point for follow-up bone marrow assessment (13 weeks), in contrast to the Alliance and SWOG studies with repeated evaluations of response by imaging and BMB.

To investigate whether cytopenias were associated with positive findings of FL on initial BMB, we identified subjects enrolled on the five Alliance clinical trials whose baseline blood counts were available and met any of the following criteria: absolute neutrophil count < 1.0 × 10⁹/L, hemoglobin < 12 g/dL, or platelet count < 100 × 10⁹/L. We compared baseline BMB results of subjects with and without one or more cytopenias by these criteria.

RESULTS

We identified seven studies meeting inclusion criteria that completed enrollment of a total of 580 subjects with FL, the majority of whom had advanced stage disease, from 2008 to 2016 through Alliance and SWOG (Table 1). One patient with no baseline BMB result was excluded. Five studies were phase II and one each were phase I and phase III. Six of the studies used IWG 2007 criteria for response assessment and one used study-defined criteria. Subject characteristics are listed in Table 2. Median age was 55 years (51% male, 96% stage III-IV, and 88% grade I-II). FLIPI scores were low in 20%, intermediate in 46%, and high risk in 35%. Chemotherapy-based regimens were administered to 67% of subjects. Baseline BMB were positive in 55%. Of subjects enrolled on Alliance clinical trials with available information (n = 191), there was no association between subjects with one or more cytopenia at baseline and positive initial BMB results (P = .8465; Appendix Table A1 [online only]).

TABLE 1.

Summary of Studies

TABLE 2.

Subject Characteristics

Only 5/580 (0.8%) FL subjects in Alliance and SWOG trials had positive baseline BMB, CR on imaging, and subsequent positive BMB (P < .0001; Appendix Table A2, online only). Of 344 subjects with a CR on imaging after treatment, 1.5% (5/344) had BMB that altered response assessment. See Appendix Figures A1A and A1B (online only) for CONSORT diagrams for Alliance and SWOG, and ECOG trials.

The landmark survival analysis from time of first radiologic CR was performed in subjects with previously untreated FL enrolled on Alliance and SWOG trials. Of subjects with CR on imaging (n = 187), PFS and OS were not different among subjects with negative BMB within 60 days of CR on imaging (n = 47) versus subjects without repeat BMB within 60 days of imaging (n = 140; PFS: HR_adj = 1.10, 95% CI, 0.62 to 1.94, log-rank P = .686; OS: HR = 0.59, 95% CI, 0.23 to 1.53, log-rank P = .276; Fig 1 and Appendix Table A2).

FIG 1.

Landmark survival analysis. For subjects with positive BMB at baseline and subsequent CR on imaging, (A) progression-free survival and (B) overall survival were not different between those patients who had negative BMB within 60 days of the first CR on imaging (n = 47) and those who did not undergo repeat BMB within 60 days (n = 140). BM, bone marrow; BMB, bone marrow biopsy; CR, complete response; KM, Kaplan-Meier; NE, not evaluable.

Two clinical trials conducted by ECOG that met inclusion criteria were identified (E4402 and E2408). E2408 was excluded because the data were not available at the time of our data collection. A sensitivity analysis was conducted on the 385 subjects with untreated FL on RESORT (E4402), a phase III trial of rituximab extended schedule versus retreatment at progression. Because imaging results were mandated only at an interim point (and therefore the results are not necessarily reflective of best response), we were unable to combine the analysis with that from the Alliance and SWOG trials. Characteristics of subjects on the ECOG trial were similar to those enrolled on Alliance and SWOG trials (Appendix Tables A3 and A4, online only). Of 385 subjects enrolled on the ECOG trial, only five (1.3%) had BMB that affected response assessment (Appendix Fig A1B).

DISCUSSION

In conclusion, we establish that BMB do not affect response assessment in NCTN clinical trials enrolling subjects with untreated FL. We analyzed two different data sets with close to 1,000 total subjects, and the results of BMB changed response assessment in only 10 subjects (about 1%). On average, in the primary data set, one subject's response assessment was altered for every 116 subjects required to undergo BMB. A significant percentage of subjects with CR on imaging did not undergo confirmatory BMB (99/198, 50% in Alliance/SWOG trials; and 22/47, 47% in the ECOG trial). Although reasons for the omissions were not recorded, we presume that both patient and physician preference may have been the primary contributing factors (Appendix Tables A5 and A6, online only).

The landmark survival analysis was performed to compare outcomes in subjects who did not undergo BMB with those who did have confirmatory BMB, and we found no difference in PFS and OS. Therefore, BMB do not enable identification of distinct PFS/OS outcomes in FL patients with positive findings at baseline.

In another subtype of lymphoma, classical Hodgkin lymphoma, BMB were formerly required as part of diagnostic workup beginning with the guidelines established by the Committee on Hodgkin's Disease Staging Classification in 1971.¹² With incorporation of CT imaging into practice guidelines, recommendations were changed in 1989 to include BMB as a requirement in only those HL patients with stage III-IV disease or stage II disease with high-risk features in whom bone marrow involvement would change management.¹³ When PET-CT became routinely used in staging of HL, this modality was established to accurately assess for bone marrow involvement in that disease.¹⁴ A retrospective study of 454 patients with HL concluded that BMB do not alter risk assessment or treatment decisions in these patients, and the procedures were subsequently eliminated from staging requirements.^8,15-17 In diffuse large B-cell lymphoma (DLBCL), multiple studies have determined that PET-CT accurately reveals bone marrow involvement.^18-20 Therefore, NCCN guidelines state that BMB are not necessary in DLBCL if PET-CT detects bone disease.⁴ Our findings from a retrospective review of the GOYA study indicate that BMB have minimal impact on response assessment in patients with DLBCL as well.¹¹

Guidelines for FL still require routine staging BMB, although the procedures are inconsistently done in clinical practice.^4,5 NCCN guidelines do specifically note that the procedures can be eliminated in those who are being followed without treatment.⁴ Our study indicates that BMB are unnecessary for response assessment and do not affect PFS or OS.

BMB in FL are still useful in select circumstances including for confirmation of stage I disease in which radiotherapy can be administered with curative intent. The stage I FL patient population was studied prospectively in the National LymphoCare Study, and those who underwent rigorous staging including imaging and BMB had longer PFS compared with those who did not.²¹ In addition, in FL patients with significant cytopenias, BMB can help determine the etiology. Interestingly, in our study, there was no association found between baseline cytopenias and positive findings on BMB.

Limitations of our study include its retrospective nature and response criteria variation in NCTN FL clinical trials. The CT-based IWG 2007 criteria were the response criteria in place at the time when the trials included in our study were performed. Current clinical trials primarily use the PET-based Lugano criteria. The two do not differ with respect to BMB recommendations; therefore, we expect our findings and conclusions to also apply when Lugano criteria are used to determine response. Our prior investigation of the GALLIUM data set found that a minimal number of subjects with FL who underwent PET imaging had alteration in response assessment on the basis of BMB. Because of differences in timing of BMB requirements, we were unable to include FL subjects from the ECOG trial in the primary analysis; however, findings with this cohort were similar to those in the primary group. We cannot rule out the influence of tumor bulk in the patient population studied in our analysis, but we think it is unlikely that this would affect our findings. The broad inclusion criteria are an advantage of our study, which make our conclusions applicable to the wide variety of patients with FL receiving frontline therapy. We note that bone marrow assessments may have been limited by technical issues including size of core, as well as lack of central review. In addition, a significant number of subjects did not undergo confirmatory BMB. This may indicate the hesitancy of both patients and physicians regarding invasive procedures, especially when the results would not change management. The landmark survival analysis confirms there is no difference in PFS or OS in subjects with CR on imaging regardless of BMB results.

Less invasive diagnostic tests including liquid biopsies are currently being investigated in lymphomas. Circulating tumor DNA levels of BCL2/IgH rearrangement and V(D)J immunoglobulin sequences correlate with PFS in FL.^22-25 In mantle cell lymphoma, higher levels of circulating tumor DNA are associated with bone marrow involvement.²⁶ These types of assays may eventually provide a surrogate for bone marrow involvement in lymphomas and become part of standard staging and monitoring of lymphomas including FL.

BMB requirements may discourage participation in clinical trials and add pain, expense, and time without providing necessary information in enrolled FL subjects. Furthermore, for 99% of patients with FL (including 96% with advanced stage disease), we show no benefit to marrow assessment on either prognosis or response assessment, validating previously published smaller experiences. Based upon these results, BMB should be eliminated from diagnostic guidelines in FL, and no longer incorporated as response assessments in clinical trials for patients with FL. This strategy is consistent with initiatives including the American Board of Internal Medicine Foundation's Choosing Wisely Campaign, as well as joint efforts by ASCO and Friends of Cancer Research to maximize value of medical interventions and modernize eligibility criteria and therefore make clinical care and trial enrollment more patient-focused.^1-3,27

APPENDIX

FIG A1.

(A) Flow diagram for Alliance and SWOG trials. (B) Flow diagram for the ECOG trial. BMB, bone marrow biopsy; CR, complete response; ECOG, Eastern Cooperative Oncology Group.

TABLE A1.

Baseline BMB Results in Patients With Normal Versus Decreased Blood Counts in Alliance Studies With Data Available (N = 191)

TABLE A2.

Proportion of Subjects With Positive BMB at Baseline, CR on Post-Therapy Imaging, and Positive Repeat BMB

TABLE A3.

Subject Characteristics (Alliance and SWOG v ECOG studies)

TABLE A4.

Subject Characteristics by Study

TABLE A5.

Univariable and Multivariable Cox Proportional Hazards Models

TABLE A6.

Characteristics of Patients With Positive Baseline BMB and Complete Response on Imaging

DATA SHARING STATEMENT

Data sharing requests should be made to sar2014@med.cornell.edu for deidentified data. Such requests will be considered by the study team after publication following review and approval of proposals, and with appropriate data-sharing agreements in place.

AUTHOR CONTRIBUTIONS

Conception and design: Sarah C. Rutherford, Jun Yin, Levi Pederson, Sumithra J. Mandrekar, John P. Leonard

Provision of study materials or patients: Mazyar Shadman, Hongli Li, Michael L. LeBlanc, Vaishalee Kenkre, Fangxin Hong, Kristie A. Blum, Travis Dockter, Peter Martin, Sin-Ho Jung, Barbara Grant, Cara Rosenbaum, Chaitra Ujjani, Paul M. Barr, Joseph M. Unger, Bruce D. Cheson, Nancy L. Bartlett, Brad Kahl, Jonathan W. Friedberg, Sumithra J. Mandrekar, John P. Leonard

Collection and assembly of data: Sarah C. Rutherford, Jun Yin, Levi Pederson, Gabriela Perez Burbano, Mazyar Shadman, Hongli Li, Michael L. LeBlanc, Vaishalee P. Kenkre, Fangxin Hong, Kristie A. Blum, Travis Dockter, Barbara Grant, Chaitra Ujjani, Joseph M. Unger, Bruce D. Cheson, Nancy L. Bartlett, Brad Kahl

Data analysis and interpretation: Sarah C. Rutherford, Jun Yin, Levi Pederson, Gabriela Perez Burbano, Betsy LaPlant, Mazyar Shadman, Vaishalee P. Kenkre, Kristie A. Blum, Peter Martin, Sin-Ho Jung, Cara Rosenbaum, Paul M. Barr, Joseph M. Unger, Nancy L. Bartlett, Brad Kahl, Jonathan W. Friedberg, Sumithra J. Mandrekar, John P. Leonard

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Relevance of Bone Marrow Biopsies for Response Assessment in US National Cancer Institute National Clinical Trials Network Follicular Lymphoma Clinical Trials

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).