Portal de Periódicos da CAPES

Survival of patients with marginal zone lymphoma in the United States: A population‐based cohort study (2000 to 2017)

2021; Wiley; Volume: 96; Issue: 4 Linguagem: Inglês

10.1002/ajh.26103

1096-8652

John L. Vaughn, Laura C. Pinheiro, Adam J. Olszewski, Narendranath Epperla,

Colorectal and Anal Carcinomas

Marginal zone lymphoma (MZL) is an indolent Non-Hodgkin lymphoma (NHL) that originates from the marginal zone of B-cell follicles. Treatment of MZL has advanced over the past two decades. In the early 2000s, the use of rituximab for MZL improved the outcomes of both treatment-naïve and refractory patients.1 Rituximab was subsequently adopted alone or in combination with chemotherapy for the treatment of MZL. In the mid-2000s, bendamustine-rituximab (BR) was studied in indolent NHLs and was found to be non-inferior to rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone with a better toxicity profile.2 In the early 2010s, several other regimens showed promising results including chlorambucil and rituximab, and novel agents such as ibrutinib, idelalisib, and lenalidomide (alone or in combination with rituximab). Despite the advances in treatment for MZL, there are limited data showing improvement in outcomes at the population level in the US. Hence, we sought to determine whether the disease-related survival of patients with MZL improved at the population level using the Surveillance, Epidemiology, and End Results (SEER) database. We hypothesized that patients with MZL diagnosed during the most recent period were less likely to die from their disease. The SEER-18 database includes data from 18 state and local cancer registries. We included adults aged 18–84 years who were diagnosed with MZL between 2000–2017. MZL cases were identified using the third edition of the International Classification of Disease for Oncology histology codes 9689, 9699, 9760, and 9764. The primary outcome was relative survival (RS). Secondary outcomes included the crude probability of MZL-specific death in the presence of competing risks [also known as the MZL-specific cumulative incidence function (CIF)] and overall survival (OS). RS was defined as the ratio of all-cause to expected survival following diagnosis of MZL and was estimated using a model-based approach rather than a nonparametric approach. MZL-specific CIF was defined as the probability of dying from MZL following diagnosis of the disease in the presence of the competing risk of death from other causes. OS was defined as the probability of survival following diagnosis of MZL. The primary exposure was the period of diagnosis, which was labeled as “period 1” for patients diagnosed between 2000–2008 and “period 2” for patients diagnosed between 2009–2017. We categorized the period of diagnosis as a binary variable to test the primary study hypothesis after adjustment for potential confounders in a multivariable model. The time periods were selected based on the results of prior research and the chronology of treatment advances.3 However, we also included the year of diagnosis as a continuous variable in univariable models to estimate changes in the study outcomes over calendar time. Covariates for our study included age at diagnosis, sex, race, Ann Arbor stage at diagnosis, B symptoms at diagnosis, county median household income, and MZL subtype. We used Royston-Parmer models to model the study outcomes.4 Missing data were handled with complete case analysis. Statistical tests were performed at a two-sided alpha of 0.05. All statistical analyses were performed using Stata version 16.1 (StataCorp LLC, College Station, TX) with the stpm2 and stpm2cr packages. Additional method details are included in the Supplementary Method S1. There were 27 403 eligible patients identified. Among these patients, 5206 (19%) were excluded due to a history of prior malignancy, 84 (0.3%) were excluded for primary sites in the CNS, 13 (0.05%) were excluded for diagnosis by autopsy or death certificate, 3319 (12%) were excluded for missing survival times (loss to follow-up), and 419 (2%) were excluded for lacking histologic confirmation. The final study cohort consisted of 18 362 patients with MZL. Table S1 shows patient characteristics. Missing data were common with 46% of patients having at least 1 missing variable. The most frequently missing variable was B symptoms (36%) followed by Ann Arbor stage (15% missing), median household income (6% missing), and race (2% missing). Median follow-up was 7.3 years [95% confidence interval (CI), 7.25–7.5 years] by reverse Kaplan–Meier method. Patient characteristics stratified by the period of diagnosis are shown in Tables S2–S5. For patients with MZL, 5-year RS increased from 0.89 (95% CI, 0.87–0.91) in 2000 to 0.94 (95% CI, 0.94–0.95) in 2012 (Figure S1). RS estimates for each period of diagnosis and MZL subtype are included in Table S6. On multivariable analysis, we found that patients with MZL diagnosed during period 2 had a significantly lower excess hazard rate than those diagnosed during period 1 [adjusted excess hazard ratio (aEHR), 0.72; 95% CI, 0.63–0.83; P < .001] (Table 1). In addition to increasing age (the effect of which was modeled continuously and nonlinearly), the following covariates were associated with increased excess mortality: male sex (aEHR, 1.35; 95% CI, 1.18–1.54), Black race (aEHR, 1.46; 95% CI, 1.16–1.83 versus White), advanced stage (aEHR, 3.68; 95% CI, 2.89–4.69), B symptoms (aEHR, 3.41; 95% CI, 2.56–4.55), and nodal MZL (aEHR, 2.45; 95% CI, 1.90–3.17 versus extranodal MZL) (Table 1). In our exploratory RS models for different MZL subtypes, the aEHRs for period of diagnosis were 0.74 (95% CI, 0.50–1.09) for SMZL, 0.74 (95% CI, 0.58–0.95) for EMZL, and 0.72 (95% CI, 0.59–0.87) for NMZL. The results of our models for specific MZL subtypes are shown in Tables S7–S9. In our competing risks analysis using cause-of-death classification by SEER, there were also improvements in the crude probability of MZL-specific death. The 5-year MZL-specific CIF decreased from 0.12 (95% CI, 0.11–0.13) in 2000 to 0.07 (95% CI, 0.07–0.08) in 2012 (Figure S2), indicating that patients diagnosed at the end of the study period were approximately half as likely to die from their disease. On multivariable analysis, patients diagnosed during period 2 had a 31% lower adjusted subdistribution hazard rate for MZL-specific death (Table 1). In addition to improvements in RS and MZL-specific CIF, there were also improvements in OS over the study period. For patients with MZL, 5-year OS increased from 0.77 (95% CI, 0.75–0.79) in 2000 to 0.84 (95% CI, 0.83–0.85) in 2012 (Figure S3). The results of the multivariable analysis for OS are presented in Table 1, showing that patients diagnosed during period 2 had an 18% lower adjusted hazard rate for all-cause mortality. A sensitivity analysis using multiple imputation with chained equations instead of complete case analysis showed that our multivariable RS model was robust to the presence of missing data (Table S10). Using both relative and cause-specific approaches, we found that patients with MZL diagnosed during the most recent period were less likely to die from their disease. There are several possible explanations for this finding. The use of rituximab for MZL in the early 2000s likely played an important role since rituximab was widely adopted and incorporated into chemotherapy regimens throughout the 2000s, which was associated with improved survival.5 It is also possible that patients benefited from the abandonment of toxic fludarabine-based regimens that were supplanted by more tolerable and efficacious alternatives such as BR. Advances in supportive care including wider adoption of antibiotic prophylaxis may have also played a role.6 The potential impact of novel agents such as lenalidomide-based regimens and ibrutinib is less clear as the results of the clinical trials with these agents were not published until a few years ago. In contrast, we did not find evidence of a stage shift or decrease in disease severity within the limitations of our study design. Our exploratory analyses identified multiple prognostic factors. As with the previous study by Olszewski and Castillo, we found that older age, male sex, and Black race were associated with lower survival for patients with MZL.3 We also found that patients who resided in counties with higher median household incomes had improved survival. To our knowledge, this is the first study to show that socioeconomic status may have a prognostic impact on patients with MZL, though similar findings have been reported in other lymphomas. Additional research is needed to assess the impact of race and socioeconomic status on the survival of patients with indolent lymphomas. These findings could be incorporated into future prognostic indices for patients with MZL. Limitations of our study include the observational study design, lack of data regarding laboratory studies and treatment regimens, and potential bias from patients who were lost to follow-up. Notwithstanding these limitations, our study offers important insight into the survival of patients with MZL in the US outside the setting of clinical trials. In conclusion, we conducted a population-based cohort study to understand the survival of patients with MZL in the US over the past two decades. We found that patients diagnosed during the most recent time period were less likely to die from their disease after adjustment for potential confounders. These data suggest that patients with MZL have benefited from therapeutic advancements outside the setting of clinical trials, and will serve as a benchmark for future research. As we continue to develop new treatments for these patients, it will be important to assess their survival in the real world using population-based cancer registry data. This work was presented in part at the American Society of Hematology Annual Meeting, December 2020. Study design: JLV and NE. Data collection: JLV. Data analysis: JLV. Data interpretation: All authors. Manuscript preparation: All authors. Manuscript approval: All authors. JLV declares no relevant conflict of interest. LCP declares no relevant conflict of interest. AO declares no relevant conflict of interest. NE has no relevant COI pertaining to the manuscript. Data openly available in a public repository Appendix S1: Supporting Information Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.