Abstract
Background/Aim: Limited published real-world data describe adverse events (AEs) among patients treated for mantle-cell lymphoma (MCL). The aim of this retrospective study was to describe treatment patterns, AEs, and associated healthcare costs. Patients and Methods: Patients had two or more claims coded for MCL diagnosis, the first claim date (07/01/2012–05/31/2017) was the index date. Patients with pre-index MCL diagnosis or systemic treatment, or hematopoietic stem cell transplantation were excluded. Cohorts by regimen were followed for up to three lines of therapy. Results: Patients (n=395; median age 72 years; 31% female) were observed over a total of 576 lines of therapy, the most common being bendamustine plus rituximab; rituximab monotherapy; R-CHOP; and ibrutinib. The most frequent AEs were hypertension (40.5%), anemia (37.7%), and infection (36.1%). However, hepatotoxicity ($19,645), stroke ($18,893), and renal failure ($9,037) were associated with the highest medical costs per patient per month. Conclusion: Among patients receiving common systemic treatments for MCL, AEs occurred frequently; some imposed substantial inpatient care costs.
Mantle cell lymphoma (MCL) is a rare, but often aggressive, subtype of non-Hodgkin’s lymphoma, affecting 3% of all newly diagnosed patients with non-Hodgkin’s lymphoma (1). The incidence of MCL is increasing in the United States, with up to three times the number of men diagnosed as compared with women (1-3). The disease profile is heterogeneous, but patients usually present with advanced-stage disease (4, 5), which is associated with substantial mortality and morbidity affecting the bone marrow, gastrointestinal tract, brain, and spinal cord. Typically associated with a poor prognosis, MCL often relapses and becomes refractory to treatment. Although more aggressive therapeutic approaches are typically chosen for MCL treatment, these carry a higher risk of toxicity and secondary malignancies, which may not be tolerable among patients with MCL who are often elderly (median age at diagnosis is 68 years) and present with multiple comorbidities.
Promising developments have been built upon better understanding of the molecular pathogenesis of MCL, toward the goals of more personalized approaches to treatment. Overall, treatment-emergent adverse events (AEs) are common with MCL systemic agents, as observed through clinical trials. Novel agents such as inhibitors of Bruton’s tyrosine kinase have changed the treatment landscape for patients with MCL. Although novel agents may overcome the high-risk prognostic features of patients with MCL, the toxicity profile is still an important consideration for treatment selection. Determining the best approach for each patient with this relatively rare hematological disease is complicated by the serious morbidities associated with MCL, overlapping with many treatment-emergent AEs.
Limited data exist regarding treatment patterns and outcomes in MCL among patients treated in routine clinical practice, as opposed to randomized clinical trials. Although two other recent studies have examined treatment patterns and AEs in a real-world sample (6, 7), they included slightly older data or were limited to younger patients. Because AEs vary with MCL treatment regimens, they can be an important driver of healthcare resource utilization and costs. Recent real-world studies have explored the incremental costs associated with AEs in multiple types of cancer (8), or examined costs associated with specific types of MCL treatment (9), although these are also based on data prior to 2017. The objectives of the current study were to explore treatment patterns, evidence of AEs, and healthcare costs among a cohort of patients with newly diagnosed MCL, as observed via administrative claims data. We describe the patient characteristics, regimens received by line of therapy, rates of AEs by regimen, and healthcare utilization and costs stratified by number of incident AEs.
Patients and Methods
Study design and data source. This retrospective study spanning from 01 July 2011 to 30 June 2017, used the Optum Research Database (ORD) and linked death data from the Social Security Administration Death Master File. The ORD is a large database containing medical and pharmacy data from individuals enrolled in US commercial and Medicare Advantage with Part D health plans. In 2017, approximately 14.6 million commercial and 3.9 million Medicare Advantage enrollees with both medical and pharmacy benefits were included in the ORD. The data were de-identified and accessed in accordance with the Health Insurance Portability and Accountability Act of 1996; institutional review board approval or waiver was not required.
Study period and study patients. Figure 1 describes the study design. Data were included for commercial and Medicare Advantage plan enrollees with both medical and pharmacy benefits and at least 2 non-diagnostic medical claims with diagnosis codes for MCL (ICD 9-CM 200.4x or ICD-10-CM C83.1x) in any position, at least 7 days apart during the identification period (01 July 2012 to 31 May 2017). The first claim date was set as the index date.
Patients were required to be at least 18 years old at index date and have continuous plan enrollment for at least 12 months pre-index date (baseline period, during which no MCL diagnoses or treatments were observed) through variable follow-up of at least 1 month post-index (follow-up period, during which treatment plans and outcomes were observed). Patients were followed until the earliest of: (i) death, (ii), disenrollment from the health plan, or (iii) the end of the study period on 30 June 2017. Each time period during which any systemic regimen was administered was defined as an observed treatment line of therapy (LOT). This study examined only patients with evidence of systemic cancer therapies associated with MCL management during the follow-up period as informed by 2017 (relevant during the study period) National Comprehensive Cancer Network guidelines (10) and expert opinion. These agents included the following: alemtuzumab, bendamustine, blinatumomab, bortezomib, carboplatin, carfilzomib, carmustine, chlorambucil, cisplatin, cladribine, cyclophosphamide, cytarabine, doxorubicin, etoposide, everolimus, fludarabine, ibrutinib, idelalisib, ifosfamide, ipilimumab, lenalidomide, methotrexate, nivolumab, obinutuzumab, ofatumumab, oxaliplatin, pembrolizumab, pentostatin, pomalidomide, procarbazine, rituximab, temsirolimus, tositumomab, tositumomab I-131, venetoclax, and vincristine. The most commonly used regimens of one or more of these agents were deemed regimens of interest.
Patients were excluded if there was evidence of MCL diagnosis or at least 1 claim for MCL-directed systemic cancer therapy (Table I) during the baseline period, or hematopoietic stem cell transplant (HSCT) at any time during the study period. The exclusion of patients with evidence of HSCT was due to the AEs associated with HSCT that might overlap or influence those occurring with systemic treatments (11).
Measures and outcomes
Patient characteristics. Demographic characteristics included age as of the index year, sex, insurance type (commercial or Medicare Advantage), plan type (exclusive provider organization, health maintenance organization, indemnity, point of service, preferred provider organization), and geographic region of enrollment (per US Census designations). In addition, the index year was noted, and a baseline Quan-Charlson Comorbidity score was calculated based upon the presence of diagnosis codes on medical claims in the baseline period (12, 13).
Treatment patterns. Data regarding treatment patterns included the time to initiation of systemic cancer therapy from index date, number of LOTs received, therapy duration, received during each LOT period. A LOT period began on the date of the first observed infusion or oral prescription filled for systemic cancer therapy after the index date, and the regimen for this LOT included all agents received within 30 days of the date of administration of the first drug. The LOT period continued until the earliest of: (a) addition to or substitution of the initial 30-day regimen (LOT period end was the day prior to the start of the new agent); (b) a treatment gap of ≥60 days after the last date of all agents in the LOT period [last date for infused drugs was the latest date of administration plus 29 days, and for drugs obtained through pharmacy benefit was fill date plus (days supply −1)]; (c) death; (d) disenrollment or end of study period. Subsequent LOT periods were created based on the start of a new regimen after the end of a LOT period; up to three LOT periods were examined for each patient. Any LOT periods that ended due to disenrollment or end of the study period were considered censored but were included in the analysis.
Cohorts. For patient-level analysis, cohorts were based on regimens received in the first observed line of therapy (LOT1) after diagnosis. For analysis of LOTs, study cohorts were created based on the most common regimens received during the follow-up period regardless of LOT number, that is, patients were able to contribute more than one LOT period to a study cohort; thus, these regimen cohorts neither represent mutually exclusive patient cohorts nor are indicative of a particular LOT number/sequence (a cohort comprised any combination of first, second, or third LOTs from among all patients). However, rituximab monotherapy received during second observed LOT (LOT2) or third observed LOT (LOT3), immediately after a previous rituximab-based therapy, with an infusion frequency of no more than 1 per 56 days, was considered maintenance therapy and those LOTs were excluded from the LOT-level analysis.
AEs and costs. Diagnoses associated with AEs of interest were selected based on consultation with a clinical expert and a published study (6). The AEs of interest included anemia, arthralgia, atrial fibrillation, diarrhea, hemorrhage/bleeding, hepatotoxicity, hypertension, infection, myocardial infarction, neutropenia, renal failure, stroke, second primary cancers, and thrombocytopenia. Diagnoses related to AEs of interest identified based on ICD codes in position 1 or position 2 on medical claims during the LOT periods were considered prevalent. When an AE occurred with no code in any position on medical claims dated before the LOT period, the AEs were considered incident.
All-cause and AE-related healthcare resource utilization included inpatient stays, emergency department (ED) visits, and ambulatory (physician office or hospital outpatient) visits. All-cause and AE-related medical costs were calculated as the sum of health plan paid and patient paid costs, and adjusted to 2016 US dollars to reflect inflation. Healthcare costs were sub-divided into ambulatory costs, inpatient stay costs, ED costs, and other costs, and per patient per month (PPPM) costs were calculated. AE-related healthcare costs and utilization were determined from claims that met the AE definition, based on ICD codes.
Analyses. All study variables, including baseline and outcome measures, were analyzed descriptively. Number and percentage values were provided for dichotomous and polychotomous variables, and mean±standard deviation are provided for continuous variables. Descriptive techniques accounting for length of observation time (e.g. PPPM measures) were used where appropriate. Demographic and clinical characteristics were compared across treatment cohorts for the baseline period, using t-test or chi-square test as appropriate for the measure and distributions, with p<0.05 set as the significance level. All analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA).
Results
Patient-level results. A final sample of 395 patients met all study criteria and were included in the study (Figure 2). Accounting for 80% of patients, the most common regimens received at initiation of the follow-up period were bendamustine HCl/rituximab (BR; n=205, 52%); cyclophosphamide, doxorubicin, vincristine and rituximab with/without prednisone (designated as R-CHOP; n=51, 13%); rituximab (n=50, 13%), and ibrutinib (n=11, 3%) (Table II). Because these four regimens were most commonly used (representing 74% of all LOTs), these were identified as regimens of interest for the purpose of subsequent analyses. Other regimens were used as initial therapy for only 20% of patients (n=78). The mean age of the study population (n=395) was 71±10 years, most patients (78.5%) were over 65 years old; 30.9% of the patients were female, and 68.6% were enrolled in Medicare Advantage plans. The mean baseline Quan-Charlson Index score was 2.5±2.0 and not significantly different across initial treatment regimen cohorts. Regional distribution of patients was similar to that of the database as a whole, with a majority of patients in regions in the US known as the Midwest and South (34.7% and 38.7%, respectively). Overall, the mean duration of follow-up was 580±462 days and was not significantly different across initial treatment regimen cohorts.
The mean time to start of therapy was 72±146 days and for almost all patients (95%), systemic therapy was initiated within 12 months after diagnosis. The majority of patients (92%) received rituximab at some point during the study period (as either monotherapy or in multi-drug regimens). The next most common agents received, either as monotherapy or part of multi-drug regimens at any time during the study period, were bendamustine (62%), cyclophosphamide (26%), vincristine (26%), doxorubicin (22%), ibrutinib (17%), and bortezomib (11%).
During the study period, 39% (n=153) and 16% (n=63) of patients commenced LOT2 and LOT3 regimens, respectively. Figure 3 shows the distribution of LOT1 and LOT2 regimens among patients who started a second LOT (n=153) during the study period, and the distribution of LOT2 and LOT3 regimens among patients who started a third LOT during the study period (n=63). For patients who had BR as LOT1, 67% had no additional LOT during the study period. Similarly, for patients with LOT1 of R-CHOP, rituximab, ibrutinib, and other regimens, 55%, 56%, 81%, and 50%, respectively, had no other regimen observed during the study period.
Among all LOT periods, 37%, 39%, and 40%, of LOT1, LOT2, and LOT3 periods, respectively, were censored at disenrollment date from the health plan or the date of the end of the study period. Among the 395 patients, 576 LOT periods (across LOT1, LOT2, and LOT3) were initiated during the study period and among these, 427 observed LOTs (74% of 576) accounted for the four most common regimens received: BR (n=231, 40% of 576), rituximab (n=84, 15% of 576), R-CHOP (n=59, 10% of 576), and ibrutinib (n=53, 9% of 576).
AEs. Across all LOTs with the four regimens of interest (n=427), only 13 (3%) were free of any of the AEs of interest. Due to use of non-specific coding for second primary cancer, they were reported among 73.1% of LOTs observed, a far higher rate than expected and considered to be an artefact of the coding limitations. The most frequent AEs observed across the LOTs with the four most common regimens were hypertension (40.5%), anemia (37.7%), infection (36.1%), neutropenia (16.6%), and thrombocytopenia (13.4%) (Figure 4).
Healthcare costs and resource utilization. Among all study patients (n=395), the 12-month mean baseline PPPM medical costs were $1,492±2,514. After MCL diagnosis, the mean PPPM medical costs for the variable follow-up period was $16,117±18,902, with $1,241±2,518 for pharmacy costs, $3,957±5,634 for office visits, $6,188±8,919 for hospital outpatient visits, $158±700 for ED visits, and $4,199±14,244 for inpatient care. Among patients experiencing an AE of interest during the aggregated LOT periods, those with hepatotoxicity ($19,645±38,821), and stroke ($18,893±45,400) experienced the highest PPPM medical costs, driven by inpatient care. Mean all-cause PPPM costs associated with renal failure ($9,037±20,649), atrial fibrillation ($5,751±18,638), and anemia ($5,097±17,042) were also among the five most costly (Figure 5).
Among all patients (n=395), all-cause healthcare utilization PPPM visits were as follows: 3.4 office visits, 2.7 hospital outpatient visits, 0.27 ED visits, and 0.13 inpatient visits. Per patient per month estimates for AE-related inpatient stays were highest for renal failure (0.27), hepatotoxicity (0.24), myocardial infarction (0.20), and atrial fibrillation (0.15) (Figure 6).
Discussion
While data regarding outcomes for patients receiving chemoimmunotherapy or novel agents are available from clinical trials, there are limited data in the real-world setting during the novel agent era. These data provide contemporary real-world evidence describing changing treatment patterns, AEs, and costs associated with systemic treatment of MCL in routine practice. This study shows that AEs are common during treatment and expensive. Therefore, as new treatments and combinations are being developed and recommended, there must also be focus on management of AEs to achieve the best outcomes for patients.
Chemoimmunotherapy was still the most common treatment regimen during the study time period until 2017, with ibrutinib being more common in LOT2 and LOT3, for which it was approved in MCL by the Food and Drug Administration in 2013. AEs and healthcare costs were significant overall and varied by treatment regimen. Hypertension was the most common AE, while hepatoxicity, stroke, and renal failure were the most costly AEs. Inpatient costs were higher for non-hematologic AEs than previous studies indicated.
Goyal et al. published a healthcare claims analysis among patients with MCL of similar age and comorbidity burden as the current study and included the same four most common treatments, yet they reported fewer patients receiving BR and more patients receiving R-CHOP than the current study (6). Furthermore, our study included a higher proportion of patients with Medicare Advantage insurance. In their study, neutropenia was the most common AE (20.7%), followed by dehydration (13.7%), anemia (12.8%), fever (11.3%), and thrombocytopenia (8.5%). Renal failure was reported among 4.7% and atrial fibrillation among 3.6% of treatment periods. In the current study, the most common AE was hypertension (40.5%), followed by anemia (37.7%), infection (36.1%), neutropenia (16.6%), and thrombocytopenia (13.4%). Renal failure was observed among 10.3% and atrial fibrillation among 11.9% of patients. Although Goyal et al. reported AEs associated with specific therapies, statistical comparisons by regimen were not performed in either study, due to the small sample size (6). Notably, in addition to regimen, age and health of patients appear to be important factors in AE profiles during treatment. Another observational MCL study by Kabadi et al. examined patients younger than 65 years of age, which represents a smaller proportion of patients with MCL nationwide than this study (7). AE burden was similar to that observed in this study. Healthcare costs associated with AEs in the current study were predominantly attributable to inpatient care, with the highest medical cost estimates for hepatotoxicity ($19,645 PPPM), stroke ($18,893 PPPM), and renal failure ($9,037 PPPM). Data regarding costs associated with AEs during cancer treatment for MCL patients in routine practice are rare; comparability is limited by differing patient characteristics, lengths of observation, and definitions of incident AEs (i.e. absence of AE prior to treatment initiation, primary, or secondary versus any position in medical claim when identifying AE). Wong et al. demonstrated incremental costs among patients with lymphatic and hematopoietic cancer of $10,046 for neutropenia/leukopenia and $6,794 for anemia, as grouped within an anti-neoplastic treatment episode (8). The current study found PPPM values within the same order of magnitude (anemia $5,097 and neutropenia $3,222), but comparisons are hindered by varying treatment period lengths. Within Goyal et al., inpatient utilization was found to be the main driver of all-cause and MCL-specific costs, which rose proportionally with the number of AEs reported among all regimens, but costs associated with specific AEs were not reported (6). Kabadi et al. observed that mean healthcare costs rose with the number of AEs reported, with the highest proportion being for pharmacy costs, which may have been due to the more intensive regimens used for younger patients such as R-CHOP (7). These intensive regimens are associated with more frequent AEs and associated healthcare resource use and costs. These varying real-world samples provide a basis for comparing AE profiles and associated costs attributable to existing and new regimens, toward the goal of optimizing treatment response but at lesser cost in harm from AEs and their care burden. As other novel agents have been approved (e.g. acalabrutinib and zanubrutinib) or are commercially available (e.g. venetoclax) and treatment patterns change, understanding the AE profiles of agents will be important in helping to determine treatment choice.
Limitations
With the increasing incidence of MCL, development of the MCL prognostic index (14), and more individualized chemoimmunotherapeutic approaches proposed in recent years, results from observational studies are increasingly important. However, although healthcare claims represent a good source for real-world analyses among large samples, certain limitations apply to their use. Claims are collected for payment rather than research and do not include all data needed to describe study samples and determine or explain interventional outcomes. Specifically, unavailability of prognostic variables (e.g. deletion of 17p), which might influence treatment choice and outcomes, is an important limitation. Due to limitations in claims coding for secondary cancer, the rates observed in the current study were artificially high based upon nonspecific coding; thus, they were removed from the analysis. Although an algorithm was created to exclude LOT periods for rituximab maintenance therapy, claims data do not include information to identify these conclusively and so rates of rituximab monotherapy use may not reflect real-world use. Results of the current study are specific to commercial and Medicare Advantage patients; thus, results may not be generalized to patients insured by other types of plans or with no insurance coverage.
Some study design factors may have limited the interpretations of findings in the current study. Only a specific set of AEs were examined, which might have affected the magnitude of costs observed for patients with multiple AEs but their choice was based upon those commonly observed in clinical trials and they were similarly observed in the only two recent comparable real-world studies available. Censoring 40% of LOT periods at the end date of the current study might also have hindered comparisons between studies as not all AEs or costs relative to a treatment regimen were captured. In addition, our exclusion of patients who received HSCT may have resulted in lower observed rates of AEs (11). It is unclear whether HSCT-related AEs influenced the type and frequency of AEs identified as treatment-related AEs in comparable studies. Small sample sizes restricted the ability to compare regimens for treatment-incident AEs; thus, no conclusions can be drawn about costs being influenced by the regimen. More observational studies with larger samples and consistent design will clarify the scope of treatment-emergent AEs.
Conclusion
Under the most common treatment regimens received for real-world MCL management, AEs were common and associated with substantial costs, predominantly for inpatient care. This study contributes data that are valuable in improving the understanding of real-world AE rates and AE-associated costs. As uptake of novel agents in MCL management is expected to increase, additional research is needed to describe real-world treatment patterns and AEs involved in treatment of MCL.
Acknowledgements
Medical writing assistance was provided by Caroline Jennermann, an employee of Optum.
Footnotes
↵Authors’ Contributions
SK, SB, LL, and TO designed the study, drafted the article, and approved the final version.
Conflicts of Interest
This study was funded by AstraZeneca. Authors Dacosta, Byfield, and Le are employees of Optum, and were contracted to perform the study. Authors Kabadi and Olufade are employees of AstraZeneca; Drs. Kabadi and Olufade own stock in AstraZeneca. Portions of this work were part of a podium presentation at AMCP Nexus 2018, October 22-25, 2018, Orlando, FL, USA.
- Received December 9, 2020.
- Revision received December 23, 2020.
- Accepted January 7, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.