Geographic Variation in the Incidence of Marginal Zone Lymphoma Subtypes and Infectious Risk Factors in the United States

Background:Helicobacter pylori is a known etiologic factor for gastric marginal zone lymphoma (MZL) of the mucosa-associated lymphoid tissue (gMALT), but data for infectious etiology in other MZL subtypes are controversial. Proposed hypothetical factors include Borrelia burgdorferi for cutaneous MZL (cMALT), Campylobacter jejuni for intestinal MZL (iMALT), and hepatitis C (HCV) for splenic (SMZL) or nodal MZL (NMZL). These associations have been largely reported in studies from Europe, but they have not been confirmed in the United States (US) (Zucca E and Bertoni F, Blood, 2016;127:2082-92). We explored geographic variation in the incidence rate (IR) of MZL and of reportable infections in the US to examine these hypotheses.

Methods: We used data from the US National Program of Cancer Registries, which has been collecting data on the incidence of all cancers from geographic areas representing 97% of the US population since 1992. We calculated age-adjusted IR (per 100 000 person-years) of MZL subtypes (gMALT, cMALT, iMALT, SMZL, and NMZL) for each US state in 2001-2016. We linked them to the 2010 US census data (to obtain demographic characteristics for each state), to state-level data on the incidence of Lyme disease (2001-2016) and of the Campylobacter infection (2016) published by the US Center for Disease Control, as well as to the published estimates of HCV prevalence (2013-2016; Rosenberg ES, et al., JAMA Netw Open, 2018;1:e186371). We examined specific etiologic hypotheses in multivariable Poisson models adjusting for median age, proportion of men, Black or Asian individuals in each state, reporting adjusted IR ratios (IRR) with 95% confidence intervals (CI). gMALT, known to not be associated with Borrelia, Campylobacter, or HCV infections, was used as a control subtype.

Results: Between 2001 and 2016, there were 5,625 cases of cMALT, 4,247 of iMALT, 29,535 of NMZL, 13,472 of SMZL, and 17,072 of gMALT reported to the US cancer registries. We observed a large variation in the IR of MZL subtypes across the US. IR varied from 0.03 to 0.22 for cMALT, 0.04 to 0.12 for iMALT, 0.32 to 0.89 for NMZL, and 0.11 to 0.47 for SMZL. For all subtypes, IR was highest in the Northeast, and lowest in the South (Table), consistent with general epidemiology of non-Hodgkin lymphoma. However, the IRR for South relative to Northeast was significantly lower for cMALT (0.42) than iMALT (0.64), NMZL (0.67), SMZL (0.68), or aggregate MZL (0.66; CI, 0.65-0.68).

In adjusted models, the IR of Lyme disease was significantly associated with increased IR of cMALT (IRR=2.51; CI, 1.63-3.87; Fig. A). Campylobacter infection was associated with iMALT (IRR=5.99; CI, 1.19-30.2; Fig. B). Additionally, a very high IR was observed in Nebraska and Utah for cMALT (states with very low incidence of Lyme disease), raising a hypothesis of other etiologic factors. In contrast, the HCV infection, most prevalent in the South, was inversely associated with the IR of NMZL and SMZL. We observed no significant association between the IR of gMALT (the control subtype with a well-established etiologic factor) and any of the studied infections, confirming the specificity of the findings.

Conclusions: Our results support the putative association between cMALT and Borellia burgdorferi infection, and between iMALT and Campylobacter infection in the US, but also suggest potential other etiologies specific to certain regions in the US. Conversely, lack of a positive association between NMZL or SMZL and HCV suggests that HCV is not a major pathogenic factor in the US. This finding is consistent with the low prevalence of HCV among NHL patients in the US as compared with Japan or Italy (Gisbert et al., Gastroenterology 2003;125(6):1723-32.; Negri et al., Int J Cancer 2004;111(1):1-8). However, the ecological results do not rule out associations at the individual level, which will require further research. Elucidating reasons for geographical variation in the incidence of MZL subtypes may ultimately provide insight into novel pathogenic mechanisms and opportunities for therapeutic approaches.

View largeDownload slide

View largeDownload slide

Close modal