Impact of Primary Site and Era of Diagnosis on Mortality in Patients with Extranodal Marginal Zone Lymphoma in the United States: A Population-Based Study

Introduction: Extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue (EMZL) is the most common subtype of MZL. EMZL can affect multiple organs throughout the body, and the management depends on the primary site of involvement and extent of disease. The most common primary site is the stomach, but other sites can be involved including the breasts, eyes, lungs, salivary glands, and skin. The last population-based survival analysis of EZML included follow-up through 2009 (Olszewski and Castillo, 2013). Given the paucity of population-based analyses of EMZL in the contemporary time-period, we sought to provide an updated analysis of disease-specific mortality and site-associated mortality using the Surveillance, Epidemiology, and End Results (SEER)-18 database.

Method: We included adult patients with EMZL diagnosed between 2000-2017 who were 18-84 years old at the time of diagnosis. We excluded patients with a history of malignancy prior to EMZL, those with missing survival times and those with central nervous system involvement. Patients were divided into two cohorts based on period of diagnosis (era-1: 2000-2008, and era-2: 2009-2017). Five-year relative survival (RS) was estimated using the Pohar-Perme method. The risk of EMZL-specific death was estimated by calculating the cumulative incidence function (CIF). The difference between CIF distributions was tested with the Pepe-Mori test. The competing risks of EMZL-specific death and death from other causes were modeled using Fine and Gray regression. All tests of differences were performed at a two-sided alpha of 0.05.

Results: We included a total of 11,597 patients with EMZL. Table 1 shows the baseline characteristics. Among the entire cohort, the median age at diagnosis was 63 years (IQR = 53-72). There was a slight female predominance (55%), and most patients were Non-Hispanic White (65%). Five-year RS was 95% (95% CI = 94-96%) during era-1 and 96% (95-97%) during era-2. There was a significant reduction in the cumulative incidence of EMZL-specific death between era-1 and era-2 (p < 0.001, Figure 1A) but not death from other causes (p = 0.35, Figure 1B). The cumulative incidence of EMZL-specific death varied according to EMZL subtype (Figure 2). In our multivariable competing risks model after adjusting for confounders (Table 2), EZMLs arising from the genitourinary (GU) tract had the worst prognosis [subhazard ratio (SHR) = 2.06, 95% CI=1.30-3.27, p=0.002], whereas EZMLs arising from the skin and connective tissue had the best prognosis [SHR = 0.66, 95% CI = 0.48 - 0.91, p = 0.01].

Conclusion: In this updated population-based analysis of EZML that includes patients treated in the contemporaneous time-period, we found that patients diagnosed with EZML in the most recent era had a significantly better prognosis. Additionally, we noted that the outcomes varied depending on primary site of involvement. Patients with EZMLs involving the GU tract had the worst prognosis, whereas patients with primary sites in the skin and connective tissue had the best prognosis. It is important to study the biology of EMZL involving the GU tract to design better therapeutic options to improve the outcomes.