Disparities in real-world treatment patterns of hypomethylating agents among patients with MDS in the United States Open Access

Myelodysplastic syndromes (MDSs) are the most common myeloid malignancies in the United States.¹ MDS comprises a heterogeneous collection of clonal hematopoietic stem cell disorders characterized by ineffective hematopoiesis and a clinical course characterized by transfusion dependency, infections, bleeding complications, and risk of leukemic transformation. The hypomethylating agents (HMAs) azacitidine and decitabine are the recommended treatment options for patients with higher-risk MDS. HMAs have been associated with hematologic improvements; enhanced quality of life; delayed progression to acute myeloid leukemia; and, additionally, improved survival with azacitidine when compared with conventional care regimens.^2,3 However, the real-world outcomes of patients with MDS treated with HMAs in the United States have been consistently inferior to that reported in clinical trials, with several US population registry studies showing modest to no improvement in survival over the past 2 decades despite increasing use of HMAs over that time.^4-8 

Several population-level studies have identified low rates of HMA use^7,9-16 and shorter than recommended treatment duration^9-13,16 as key treatment-related factors driving poor real-world outcomes of patients with MDS in the United States. Patients with MDS are typically a vulnerable demographic cohort, ∼83% are aged ≥65 years, who often have a high prevalence of multiple comorbidities at diagnosis.^17,18 Furthermore, the logistical and clinical demands of outpatient HMA regimens, marked by repetitive chemotherapy cycles over prolonged periods, present significant challenges, particularly because of the high number of outpatient visits, the risk of cytopenia-related complications such as bleeding and infections often necessitating hospitalization, and the long-term need for transfusions. We hypothesize that factors beyond disease-related prognostic variables, that is, demographic or socioeconomic disparities and inconsistent adherence to established clinical practice guidelines, may significantly influence both initiation and persistence of HMA therapy. Given that HMAs remain the mainstay of treatment for higher-risk MDS since their approval in 2004, and considering the pervasive limitations in pursuing allogeneic hematopoietic cell transplantation in most patients because of high transplant-related vulnerabilities,¹⁷ a key strategy for improving outcomes involves identifying clinical and nonclinical barriers to appropriate HMA use. The primary objective of this study, therefore, was to examine treatment patterns of HMA administration among MDS Medicare beneficiaries by addressing 2 key questions: (1) which demographic, clinical, and socioeconomic factors are associated with the likelihood of initiating HMA, and (2) whether these factors also influenced the duration of HMA treatment.

This retrospective cohort study used 2011 to 2014 national Medicare data to examine treatment patterns in patients diagnosed with MDS. Claims for Medicare Parts A, B, and D included dates of service; International Classification of Diseases, ninth revision clinical modification (ICD-9-CM) diagnoses codes; and services provided based on ICD-9-CM procedure codes (in inpatient and outpatient files), Healthcare Common Procedure Diagnosis Coding System codes, and National Drug Codes. We used the 2013 Area Health Resources File, publicly available through the Health Resources and Services Administration website (accessed on 3 September 2019) to obtain county-level characteristics including poverty, educational attainment, rurality, and availability of health care resources. The study was approved by the Cleveland Clinic and Case Western Reserve University review board and the privacy board of the Centers of Medicare and Medicaid Services. This study was registered at Clinical Trials.gov (NCT02863458).

Our study included fee-for-service Medicare beneficiaries aged ≥66 years with at least 1 inpatient or 2 outpatient MDS claims within a 12-month period between 2012 and 2013, using ICD-9-CM codes (supplemental Table 1) to confirm diagnoses. To ensure incident cases, we used pre-2012 claims for a 1-year look-back and patients were followed up through 2014 for up to 24 months after diagnosis. Those without full Medicare Part A and B coverage or enrolled in managed care were excluded because of incomplete data. The final selection of the study cohort is shown in Figure 1.

Baseline individual-level demographic and clinical variables included age categories (aged 66-74, 75-84, and ≥85 years), sex, race (Non-Hispanic White, Non-Hispanic Black, and other), frailty (based on a validated claims-based frailty index),¹⁹ numbers of cytopenias (any combination of claims for anemia, neutropenia, and thrombocytopenia within 16 weeks preceding or after the Index Date of Diagnosis [IDD]), receipt of bone marrow biopsy as part of diagnostic evaluation for MDS (within 1 year preceding or after the IDD), and transfusion burden (TB). We assessed TB at diagnosis based on criteria adapted from the International Working Group 2018 recommendations for MDS, with low TB, moderate TB, and high-TB categories defined as 1 to 3, 4 to 7, and ≥8 unique dates of transfusions, respectively, during the 16 weeks before or after IDD.²⁰ In addition, as a proxy for complexity of health care needs, we accounted for residence in a nursing home in the 6 months preceding the IDD. County-level variables included percent of individuals with incomes below 100% of the federal poverty level; educational attainment based on the percent of adults (aged >25 years) without a high school diploma; rurality, based on the rural-urban continuum code (RUCC), grouped as RUCC 1 (counties in metropolitan areas with ≥1 million population), RUCC of 2 or 3 (counties in metropolitan areas of 250 000 to <1 000 000 population, and <250 000 population, respectively), and RUCC of >3 (nonmetropolitan counties); and availability of general internal medicine subspecialists per 100 000 population including for general internal medicine and internal medicine subspecialties. All continuous county-level variables (poverty, educational attainment, and availability of providers) were grouped in quartiles. Using pharmacy claims data, the index date of HMA was defined as the first date of HMA treatment initiation and the agent used on that date was defined as the index HMA. All subsequent claims for HMA treatment were then collected on treated patients till discontinuation.

Our primary outcome of interest was receipt of HMA. Secondary outcomes were patterns of HMA use and duration of HMA use (persistence of therapy). We developed a novel claims-based algorithm to identify HMA treatment patterns as depicted in Figure 2. Consistent with the guidelines, a 28-day treatment period was considered as 1 treatment cycle. The start date of the first treatment cycle for any patient corresponded to the date of the first index HMA claim for that patient, and tabulation of the number of treatment cycles received was based on the consecutive 28-day treatment periods the patient continued to receive treatment. The definition of complete or incomplete treatment cycle and treatment gaps or interruptions was adapted from US Food and Drug Administration product label recommended dosing schedule,^21,22 methodologic criteria used in previous published literature,^11-15,18 guidelines from the National Comprehensive Cancer Network²³ and consistent with dosing schedule used in contemporary clinical practice. A complete cycle of azacitidine was defined as presence of at least 4 azacitidine claims within any consecutive 28-day treatment period. A complete cycle of decitabine was defined as presence of at least 3 claims in a 28-day treatment period. If there were <4 azacitidine claims or <3 decitabine claims in any consecutive 28-day treatment period, then it was defined as incomplete cycle. If no treatment claims were identified within any consecutive 28-day treatment period, it was defined as treatment gap or interruption. HMA treatment discontinuation was defined by 1 of the following: the patient’s death, end of study period (indicated by the last HMA claim occurring within 3 months of the study’s end), or other reasons (indicated by the last HMA claim occurring >3 months before the study’s end).

Descriptive information included demographic, clinical, and county-level characteristics for the entire study cohort and stratified by treatment received: those who did not receive any HMA, and patients treated with either azacitidine or decitabine. We generated data visualization plots showing the cumulative proportion of complete cycles, incomplete cycles, and discontinuation rates, separately for azacitidine- and decitabine-treated MDS cohorts. Similar plots were generated displaying cumulative proportions of HMA discontinuation either because of death (from any cause), end of study period (31 December 2014), or because of any other reasons. Multivariable logistic regression models were then used to identify demographic, clinical, and socioeconomic factors that were significantly associated with receipt of any HMA. Additionally, we analyzed factors associated with receiving “suboptimal” HMA therapy (defined as receiving 1-3 cycles) vs minimal recommended therapy (≥4 cycles). We used SAS version 9.4 for data management, and R version 3.6.3 for statistical analyses.

Demographic, clinical and county-level characteristics of patients with MDS by receipt of HMA treatment are shown in Table 1. Our study population included 49 154 patients with MDS. Of those, 41 219 (84%) did not receive any HMA. Among patients who received HMA, 5796 (73%) received azacitidine, and the remainder received decitabine. The HMA-treated group was predominantly White (91.3%); most had 2 or 3 cytopenias (71.5%); a bone marrow biopsy as part of the diagnostic evaluation was almost universal (97.8%); 71% were transfusion dependent at diagnosis; 50% were from census tract regions classified as medium-high to high poverty level; 50% had low educational attainment (without a high school diploma); 80.4% were from counties with >250 000 population (less rural); and were in general, less frail (97.2% were prefrail and mildly frail). A small fraction of the treated patients were nursing home residents (3%). Approximately half of the treated patients with MDS resided in areas in which the availability of general internal medicine physicians and subspecialists was low.

Figure 3 shows the treatment patterns of patients with MDS by HMA type. Of 5796 patients with MDS who were treated with azacitidine, we observed a steady decrease in the proportion of patients remaining on treatment over time, with the sharpest drop in the first 6 cycles. By the end of cycle 4, 34% had discontinued treatment and the corresponding number at the end of 6 cycles was ∼50% (Figure 3A). Of all patients who were initiated on azacitidine, only 69% of patients had a complete first treatment cycle and that proportion declined to 60% by treatment cycle 6, before stabilizing between 50% and 55% for subsequent treatment cycles for the remainder of the study period. The proportion of treated patients who received complete cycles in the first 4 treatment periods was 51.1%, and the corresponding number for the first 6 treatment cycles was 41% (Table 2). Similar treatment patterns were observed with decitabine (Figure 3B). The first 4 treatment cycles were completed in 44.1% of patients, and the corresponding number for the first 6 treatment cycles was 32.8% (Table 2).

Figure 4 shows the temporal pattern of discontinuation of HMA over the entire study period and the underlying reasons for discontinuation. The 2 reasons that accounted for most HMA treatment discontinuation were death and causes that could not be easily identified in the claims data set (other than death or end of study period). For the azacitidine-treated cohort (Figure 4A), mortality was high and increased over time, from 10% in the second cycle to 23% by cycle 6, and thereafter, approximately a third (∼30%-37%) of the patients died in each of the remaining treatment cycles beyond cycle 9. Likewise, a steadily growing proportion of patients discontinued HMA treatment for reasons that were not clearly identifiable in the claims data set, from 6% in cycle 2, to 44% in cycle 24. For decitabine (Figure 4B), the patterns were very similar to that of azacitidine, although the percentage of those who died was somewhat higher in the decitabine group. Only a small fraction of patients (∼5%) remained on treatment with either azacitidine or decitabine at the end of 24 treatment periods.

Table 3 shows the results of the multivariable logistic regression model predicting receipt of any HMA (azacitidine or decitabine). Advancing age was associated with lower odds of receiving an HMA. Compared with patients aged 65 to 74 years, the odds of receiving HMA was significantly lower in the 75 to 84 years age group (adjusted odds ratio [aOR], 0.81; 95% confidence interval [CI], 0.76-0.86) and even lower in patients aged ≥85 years (aOR, 0.41; 95% CI, 0.38-0.44). Females had significantly lower odds of being treated with an HMA (aOR, 0.81; 95% CI 0.77-0.86). Among race/ethnic categories, Blacks and those of “other” race/ethnicity had lower odds of receiving HMA than White patients (aOR, 0.70 [95% CI, 0.62-0.80] and aOR, 0.78 [95% CI, 0.68-0.90], respectively). Surrogate clinical markers of higher disease risk such as cytopenia affecting at least 2 cell lineages, and any level of transfusion dependence (low, medium, or high) at the time of diagnosis were associated with higher odds of receiving HMAs. Compared with patients with no or 1 cytopenia, the odds of receiving an HMA was twice as high among those with 2 to 3 cytopenias (aOR, 2.08; 95% CI, 1.96-2.20); and compared with those who were transfusion independent, those with low and medium/high dependence had 3.3 to 6.1 times higher odds of receiving HMAs, respectively. The odds of receiving HMAs among those who underwent bone marrow biopsy was 12 times higher than those who did not (aOR, 12.61; 95% CI, 10.78-14.74). Being a nursing home resident (aOR, 0.55; 95% CI, 0.47-0.63) was associated with lower odds of receiving HMAs. Compared with patients who are not frail, higher levels of frailty were associated with lower odds of receiving HMAs in a dose-response fashion, those who were prefrail, mildly frail, or moderately/severely frail had 0.67 (95% CI, 0.59-0.76), 0.32 (95% CI: 0.28-0.37), and 0.19 (95% CI, 0.13-0.20) times the odds of receiving HMA, respectively. Poverty, educational attainment, rurality, and availability of specialty physicians did not predict for receipt of HMAs.

Next, using logistic regression, we analyzed the variables that predicted receipt of at least 4 completed cycles of HMA (recommended minimal duration of therapy) compared with 1 to 3 cycles (suboptimal therapy). In adjusted analyses, the specific HMA agent used predicted for treatment duration. Compared with azacitidine, patients on decitabine had lower odds of completing ≥4 cycles (aOR, 0.70; 95% CI, 0.62-0.78). Clinical indicators of the number of treatment cycles received (or longer treatment duration) were cytopenias and transfusion dependence. When compared with patients with 0 to 1 cytopenia, those with 2 to 3 cytopenias had lower odds of receiving ≥4 HMA cycles (aOR, 0.69; 95% CI, 0.61-0.78). With transfusion independence as the referent category, patients with low transfusion dependence at diagnosis had 31% higher odds of receiving ≥4 HMA cycles (aOR, 1.31; 95% CI, 1.16-1.48) but we did not observe any significant association between medium or high transfusion dependence and the number of HMA cycles. Nursing home residents had lower odds of receiving ≥4 HMA cycles (aOR, 0.64; 95% CI, 0.46-0.90). Compared with patients who are not frail, higher levels of frailty were associated with lower odds of receiving HMA, those who were mildly frail, or severely/moderately frail had 0.61 (95% CI, 0.48-0.78) and 0.50 (95% CI, 0.34-0.75) times the odds to receive HMAs, respectively. Poverty, educational attainment, rurality, and availability of specialty physicians did not predict for receipt of number of HMA cycles received.

In this retrospective US cohort analysis of HMA treatment patterns in 49 154 patients with MDS, we investigated demographics, clinical characteristics, and socioeconomic factors associated with receipt of HMAs and, additionally, predictors of duration of therapy in those who received HMAs. Our analyses revealed age-, sex-, and race-related disparities in the receipt of HMA even after adjusting for MDS-related clinical characteristics that are markers of disease severity (cytopenias and TB at diagnosis), frailty, and known socioeconomic determinants of disparity such as poverty, educational attainment, rurality, and availability of specialty physicians. Patients aged ≥85 years, females, and minoritized racial groups (including Black patients and other races) had significantly lower odds of receiving HMAs. Notably, only 16% of the entire MDS cohort received HMAs, a surprisingly low proportion given that ∼46% of patients presented with bicytopenia or pancytopenia and 43% were transfusion dependent at diagnosis, both clinical surrogates of higher-risk disease. We observed high rates of early treatment; 66% of patients on azacitidine and 40% on decitabine received ≤4 treatment cycles. A striking 18% of patients received only 1 cycle of HMA and only 50% of patients remained on treatment for the first 6 cycles. Furthermore, a high proportion of patients received incomplete treatment cycles. Depending on the HMA used, only 40% to 50% of patients received the entirety of the first 4 treatment cycles (complete cycles) and the corresponding number was 30% to 40% for the first 6 treatment cycles. Taken together, these findings reveal that real-world HMA treatment patterns differ significantly from those reported in clinical trials or academic settings and often fall short of evidence-based guidelines. This is demonstrated by low rates of HMA use, treatment biases favoring younger White males, high rates of early discontinuation, and subtherapeutic dosing, all indicative of suboptimal or incomplete treatment.

Considering 30% to 40% of all patients with newly diagnosed MDS have higher-risk disease for which HMA would be recommended,^24,25 HMA use was low in our study, a finding consistent with previous studies that have reported similarly low rates (13%-18%) of HMA use in patients with newly diagnosed MDS (regardless of risk category).^14,15,25 Although this study was not designed to explore the reasons underlying age-, sex- and race-related disparities in receipt of HMA, findings from several previous studies offer plausible explanations for the patterns observed in our cohort. Concerns about high treatment vulnerability, and patient preference likely contributed to the low use of HMAs in older age cohorts. Females with MDS have a more favorable prognosis compared with males, driven largely by a less aggressive mutational profile, higher concentration of lower-risk International Prognostic Scoring System categories, and fewer competing risks of death, factors that may influence physicians toward delaying initiation of HMA therapy.^26,27 Similarly, less severe disease biology and clinical presentation in Black patients, as reported in limited series, could potentially explain less HMA administration.^28-31 Beyond biological differences, nonclinical factors may also play an important role. A body of literature on cancer disparities has consistently shown that implicit bias among oncologists can lead to less aggressive, less appropriate, and less timely treatment for Black patients, even when clinical presentations are comparable with those of White patients with same cancers.^32-34 Patient-related factors, such as race-related attitudes or beliefs, may further influence treatment decisions.³⁵ Importantly, even after adjusting for socioeconomic determinants, HMA use remained disproportionately low among Black patients, an unexpected finding given the well-documented role of socioeconomic factors in driving treatment disparities. This could be explained by ecologic fallacy, a known limitation in registry-based studies, in which aggregate socioeconomic indicators at the geographic level (county in this case) are used as proxies for individual-level data.³⁶ We note that county-level characteristics may not have been sufficiently granular to capture race-related socioeconomic differences.

Of note, bone marrow biopsy might have a gatekeeper effect in HMA treatment decision-making³⁷ as evidenced by the significantly higher odds of HMA use among patients who underwent this diagnostic procedure. Nearly all patients (∼98%) treated with HMA received a bone marrow biopsy as part of diagnostic evaluation for MDS. In contrast, among 41 219 patients with MDS who did not receive HMA in this study, up to 37% (n = 15 000) did not receive a bone marrow biopsy. This raises the concerning possibility of missed treatment opportunities in this cohort, given that ∼40% of these patients presented with bicytopenia or pancytopenia and were transfusion dependent at diagnosis (clinical surrogates for higher-risk MDS), a diagnosis further supported by the absence of any other claims-identifiable hematologic condition, including malignancies. We observed a dose-response relationship between frailty and HMA use; patients with higher levels of frailty had lower odds of receiving HMAs. Similarly, residency in nursing homes was associated with lower odds of receiving HMAs. The decision to not treat these select MDS cohorts is probably reflective of appropriate treatment decision guided by high treatment-related vulnerabilities expected in these patients with complex health care needs.

The high early discontinuation rates of HMA seen in our study are comparable with those reported in other contemporaneous studies that used different data sources and used different methodologies to define treatment discontinuation, 33% to 45% receiving <4 cycles^9-11,13 and 41 to 69% receiving <5 or <6 cycles.^11-13,18 This is contrary to the recommended duration of treatment for optimal HMA response, defined as receiving at least 4 to 6 cycles of HMA in the absence of clear progression or unacceptable toxicity.^2,3,21,22 In our adjusted analyses, clinical factors were the most important determinants of duration of therapy; presence of bicytopenia or pancytopenia and poor health status as determined by frailty and residency in a nursing home were associated with lower odds of receiving ≥4 cycles of HMA. Age, sex, race, and socioeconomic determinants of disparity had no impact on treatment duration. Ecologic fallacy can possibly explain why we did not observe any impact of socioeconomic differences on HMA treatment duration,³⁶ although most MDS cases in the United States are diagnosed and managed in community settings.¹⁰ In addition to nonpersistence of HMA therapy, another concerning finding was that the treatment cycles in the first 4 to 6 months of HMA treatment were incomplete in ∼50% of the patients. Clearly, this is expected to lead to adverse outcomes as shown in our study in which ∼40% of the entire HMA-treated cohort died or progressed to acute myeloid leukemia within 90 days of last index HMA treatment (Table 2), a finding consistent with previous studies showing high rates of disease progression off treatment. Considering most of our HMA-treated cohort were appropriate candidates for HMA therapy (∼28% had mild or higher frailty)), poor tolerability to treatment would not be expected. Given the low discontinuation rates (∼5%) because of hematologic adverse events reported in clinical trials^2,3 and diminishing incidence of cytopenias with continued treatment,^2,3 treatment-related complications cannot explain the observed HMA treatment patterns in our cohort. Plausible explanations include insufficient experience of the treating physician with HMA use¹³; wide discordance in disease burden perception among patients and physicians, with the latter favoring early discontinuation¹⁵; logistical challenges; physician knowledge gaps and biases; and possibly toxicities.

Our study has several strengths. First, by using Medicare claims files as our data source, we ensured a large nationally representative MDS study cohort because Medicare is the primary insurer for ∼97% of the US population aged ≥65 years, a demographic cohort with the highest prevalence of MDS.¹ Second, by linking Medicare claims files with county-level data, we created a comprehensive data set that enabled robust analysis of treatment patterns and the impact of demographics, clinical characteristics, and socioeconomic factors on treatment patterns. Third, although our data span from 2011 to 2014, the findings remain timely and relevant, reflecting current treatment practices. This is supported by the modest real-world survival gains observed^4-6 despite an increase in HMA use (from 11% during 2004-2007 to 21.5% around 2016), indicating that the factors influencing HMA use and duration identified in this study continue to be applicable. Our results supplement findings of other studies indicating not just low use of HMA but possibly inadequate treatment (shorter treatment duration and or incomplete cycles) as potential reasons for poor outcomes of these patients in real world.

As with other published MDS studies based on administrative claims-based data sets,^6,7,9,10,12 this study shares similar inherent limitations. First, cytogenetic or molecular data of patients with MDS are not present in the Medicare claims data set^6,7,11,12,14; therefore, we cannot comment on the appropriateness of the decision to initiate HMA in these patients based on disease. Our analyzed HMA-treated cohort were very likely high-risk MDS, a determination that can be reasonably made considering that all had received diagnostic bone marrow biopsy, had no evidence of other hematologic malignancy, had received previous MDS therapies (Table 2), had clinical indicators of disease severity such as 2 or 3 cytopenias and transfusion dependence, and demonstrated high rates of disease progression, scenarios in which HMA treatment would be justified in routine clinical practice. Second, we were unable to ascertain the reasons for early discontinuation of HMA therapy or incomplete treatment cycles beyond the identified factors shown in Table 2.

In conclusion, in our population-level analysis of HMA use in patients with MDS in the United States, we identified age-, sex-, and race-related disparities in the receipt of HMA favoring young White males. Additionally, we report high discontinuation rates and suboptimal treatment (incomplete cycles), portending a poor outcome for these patients. Our work highlights the need to shift our approach beyond disease-related factors to recognize and narrow disparities in HMA treatment patterns in routine clinical practice to improve outcomes.

S.M.K. was supported by research grants from the National Cancer Institute (P30 CA043703 and U01CA284198), the Centers for Disease Control and Prevention (U48 DP005030-05S1 and U48 DP006404-03S7); National Institutes of Health (UH3-DE025487 and R01 AG074946-01); and American Cancer Society (132678-RSGI-19-213-01-CPHPS); and by contracts from Cleveland Clinic Foundation, including a subcontract from Celgene Corporation. A.S. was supported by the Winn Career Development Award (American Association for Cancer Research [AACR]), VeloSano Impact Award, VeloSano collaborative grant, North American Neuroendocrine Tumor Society (NANETS) grant, and the Clinical and Translational Science Collaborative (CTSC) pilot award from Case Western Reserve University. This work was funded by Bristol Myers Squibb, formerly Celgene (investigator sponsored non-therapeutic trial, Grant number—CELG1611SM) and Velosano 2020 Impact Award for Rare Diseases (Cleveland Clinic Taussig Cancer Institute) to S. M. The funding agencies had no role in the design of the study, collection and analysis of data, and the decision to publish.

Contribution: S.M. and W.D. were responsible for study conception and design, provision of study material, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; A.T.G., H.E.C., A.S., A.S.A., A.J., F.A., M.K.M.A., J.M.J., S.B., and P.C. were responsible for study conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript; and M.A.S. and S.M.K. were responsible for study conception and design, provision of study material, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript.

Conflict-of-interest disclosure: S.M. reports serving on advisory board of, consulting for, and receiving honoraria from, Celgene/Acceleron, Bristol Myers Squibb (BMS), Novartis, Recordati Rare Diseases (formerly EUSA), Blueprint Medicines, Genentech, AbbVie, Aplastic Anemia and MDS International Foundation, and CCM Biosciences; and receiving research support from BMS, Novartis, and Jazz Pharmaceuticals. M.A.S. reports serving on advisory board of, consulting for, and receiving honoraria from, BMS, Agios, Keros, and Kurome. J.M.J. reports serving on advisory board of, consulting for, and receiving honoraria from, Autolos, Sobi, Jazz, Cardinal Health, OncLive, and MJH Life Sciences. H.E.C. reports serving on advisory board of, consulting for, and receiving honoraria from, CTI Biopharma, BMS, Stemline, Kura, and Novartis; participation in speakers bureau for Jazz, Novartis, BMS, Stemline, and Agios; receiving research support from Celgene; and serving on the data safety monitoring board for Astex, AbbVie, Takeda, and Syndax. M.K.M.A. reports serving on advisory board of, consulting for, and receiving honoraria from, Syndax. A.S.A. reports serving on advisory board of, consulting for, and receiving honoraria from, Astra Zeneca, Kite, Servier, Takeda, and Pfizer; and received research support from Amgen, Kura, Seattle Genetics, Pfizer, Beam, Immunogen, AbbVie, Incyte, OBI, Kite, and Servier. P.C. reports serving on advisory board of, consulting for, and receiving honoraria from, AbbVie, ADC Therapeutics, Autolus, BMS, Novartis, Recordati Rare Disease, Synthekine, Sobi, and Takeda; and received research support from AbbVie, ADC Therapeutics, Genentech, Genmab, and Recordati Rare Diseases. A.J. reports serving on advisory board of, consulting for, and receiving honoraria from, Sobi, Takeda, Servier, Novartis, Rigel, and Geron; and reports research support from Novartis and Servier. F.A. reports serving on advisory board of, consulting for, and receiving honoraria from, BMS, Celgene, Caribou Sciences, GI Innovation, and Poseida Therapeutics; and reports research support Allogene Therapeutics, Celgene, BMS, and Caribous Sciences. A.T.G. reports serving on advisory board of, consulting for, and receiving honoraria from, GlaxoSmithKline, Rain Oncology, Pharma Essentia, AbbVie, Disc Medicine, Agios, Keros, and Karyopharm. The remaining authors declare no competing financial interests.

Correspondence: Sudipto Mukherjee, Department of Hematology and Medical Oncology, Taussig Cancer Institute, Cleveland Clinic, 10201 Carnegie Ave, Cleveland, OH 44195; email: mukhers2@ccf.org.