The Challenges of Epidemiologic Research in Non-Hodgkin Lymphoma

The incidence of non-Hodgkin lymphoma (NHL) increased by more than 80% between 1975 and 1991 in the United States, representing one of the largest increases of any cancer.¹ While the overall NHL incidence rates began to stabilize in the late 1990s, the temporal trends varied by histologic subtype with the incidence of T/NK [natural killer]–cell lymphomas increasing 4% per year during 1992-2001 compared with 0.5% per year for B-cell lymphoma.² Part of the overall steep increase in NHL incidence over the last several decades is due to human immunodeficiency virus (HIV) infection, although the incidence of NHL has been increasing irrespective of HIV status.³ Other pathogenic associations with NHL include viruses and bacteria (eg, human T-cell lymphotropic virus type 1,⁴ human herpesvirus 8,⁵ hepatitis C,⁶ and Helicobacter pylori⁷ ) and medications that cause extreme immunodeficiency. However, the etiology of NHL remains largely unknown. Furthermore, the significant geographic differences (eg, the incidence of NHL is 3-4 times more common in the United States vs Japan)⁸ and varying incidences by sex, age, and race in population registry data are unexplained.^{2 ,9}

Epidemiologic studies of NHL risk factors, such as genetic, infectious, environmental, racial, and lifestyle factors, are critical for improving understanding of the pathogenesis of NHL. Part of the scientific challenge in NHL epidemiologic research relates to the multiple histologic and clinicopathologic subtypes that encompass the broad category of NHL. Among the more than 25 classified NHL histologic subtypes, there is some mechanistic commonality of lymphomagenesis, but it has become apparent that significant etiologic heterogeneity exists.^{10 - 11}

T-cell lymphomas represent approximately 10% to 12% of all NHLs diagnosed in Western countries.^{2 ,12} The World Health Organization/European Organization for Research and Treatment of Cancer classification recognized 8 distinct clinicopathologic peripheral T-cell NHLs.^{13 - 14} Anaplastic large-cell lymphoma (ALCL), systemic type, represents the third most prevalent peripheral T-cell lymphoma and accounts for approximately 2% to 3% of all NHLs.¹⁵ Systemic ALCL must be differentiated from the cutaneous variant, primary cutaneous CD30+ ALCL, which has an excellent outcome with 5-year disease-free survival of better than 95%.¹⁶ In contrast, systemic ALCL is an aggressive NHL typically presenting with advanced-stage disease, systemic symptoms, and frequent extranodal disease. Systemic ALCL may be subdivided based on the tissue expression of the tyrosine kinase anaplastic lymphoma kinase (ALK) protein, created from the balanced nonrandom chromosomal translocation t(2;5)(p23;q35), although ALK (at chromosome 2p23) may fuse with different gene partners.¹⁷ The determination of ALK positivity is important as it denotes a favorable prognosis with reported 5-year overall survival of approximately 70% compared with 49% for ALK-negative patients.¹⁸ There are no clearly defined risk factors specifically associated with systemic ALCL. Some reports have shown that Epstein-Barr virus may be important in the pathogenesis of ALCL,¹⁹ although other studies have not supported this association.²⁰

In this issue of JAMA, de Jong and colleagues²¹ examined the association of breast implants with systemic ALCL. Among a Dutch population-based database, the authors identified 389 women with primary breast lymphoma, 11 of whom had a pathologic diagnosis of systemic ALCL (all ALK-negative). Five of these 11 primary breast ALCL patients had prior breast implants (saline filled, silicone textured). Of note, lymphoma was detected in each of the fibrous capsules of the breast prostheses. Through a nested matched case-control study, de Jong and colleagues identified 1 to 5 breast lymphoma controls matched on age and year of diagnosis. One patient in the matched control group had primary breast lymphoma (follicular lymphoma histology). The odds ratio for the association between ALK-negative ALCL of the breast and breast prosthesis was 18.2 (95% confidence interval [CI], 2.1-156.8).

The major strength of this report is the comprehensive population database with identification of all cases of primary breast lymphoma, thereby decreasing selection bias, and selection of control cases from the same cohort. Limitations include overascertainment and uncontrolled confounders in that women who receive cosmetic breast implants have been shown to differ from other women in several characteristics, which may be associated with risk of NHL, including increased access to medical care, higher socioeconomic status, different residency patterns, and bearing children at young ages.²² Also the wide CIs for the odds ratios suggest low statistical power. Similar cases of breast implant–related primary breast lymphoma have been reported in the literature,²³ although not all series have confirmed this association.²⁴

Primary breast lymphoma represents a distinct lymphoma clinicopathologic subtype. Outside the context of this report, the vast majority of primary breast lymphomas are of B-cell origin, with diffuse large B-cell lymphoma (DLBCL) representing the most frequent diagnosis. The reported outcomes of primary breast lymphoma are fair with 5-year disease-free survival rates of 55% to 65% following breast-conserving surgery and combined chemotherapy and radiation therapy.^{25 - 26} The central nervous system appears to be an unusual sanctuary site for primary breast lymphoma, especially among those of DLBCL histology. Clinical follow-up of the primary breast ALCL cases reported by de Jong and colleagues was not available; it would be interesting to know if these cases were associated with conventional ALK-negative ALCL outcomes.

In terms of plausible biological mechanisms regarding the association of breast implants and ALK-negative primary breast lymphoma, de Jong and colleagues hypothesized either an immunologic response (direct or indirect) related to the prosthesis, direct toxic damage from the silicone components, or both. Recent evidence of autoimmune disease contributing to the risk of T-cell ALCL has been reported.²⁷ Presence of celiac disease (odds ratio, 24.0; 95% CI, 8.8-65) and psoriasis (odds ratio, 2.25; 95% CI, 1.00-5.06) have been associated with increased risk of T-cell systemic ALCL, suggesting a possible pathogenic mechanism of chronic antigenic stimulation with local antigenic drive, ultimately leading to the development of lymphoma. Although 3 of the 5 index cases reported by de Jong et al had lymphoma diagnosed 1, 3, and 4 years following placement of bilateral breast implants, an important issue is whether these latency periods were long enough for development of lymphoma attributable to the prostheses.

Continued epidemiologic research is critically needed to piece together the puzzle of NHL etiology, and the report by de Jong et al highlights the importance and the challenges of NHL epidemiologic research. These findings must be considered preliminary and hypothesis-generating and are not strong enough to definitively conclude that breast implants predispose women to NHL. However, given that silicone is immunogenic, further evaluation of breast implant–related lymphoma is warranted, particularly by studies with statistical power, sufficient follow-up, and information on other factors. Furthermore, other lifestyle factors such as alcohol consumption, chemical exposure (eg, pesticide), sun exposure, and diet need to continue to be investigated as factors that modify (increase or decrease) the risk of NHL.^{28 - 30} Additional studies based on specific NHL histologies are needed,¹³ and they should evaluate multiple etiologic factors together as it is unlikely that 1 or 2 risk factors will yield definitive associations. It is more likely that a compilation of infections, autoimmunity, genetic host susceptibility, and environmental and lifestyle factors will identify the most robust predictors of lymphoma risk.