Etiology of non-Hodgkin lymphoma: A review from epidemiologic studies

Abstract

Non-Hodgkin lymphoma (NHL) contributes to significant cancer burden and mortality globally. In recent years, much insight into the causes of NHL has been gained by evaluating global differences through international collaboration and data pooling. NHL comprises different subtypes that are known to behave differently, exhibit different prognoses, and start in distinct cell types (B-cell, T-cell, and NK-cell, predominantly), and there is increasing evidence that NHL subtypes have different etiologies. Classification of NHL can be complex, with varying subtype frequencies, and is a consideration when evaluating geographic differences. Because of this, international pooling of well-executed epidemiologic studies has conferred power to evaluate NHL by subtype and confidence with minimal misclassification. Given the decreasing burden in some regions while cases rise in Asia, and especially China, this report focuses on a review of the established etiology of NHL from the epidemiologic literature in recent decades, highlighting work from China. Topics covered include demographic patterns and genetic determinants including family history of NHL, as well as infection and immunosuppression, lifestyle, environment, and certain occupational exposures contributing to increased disease risk.

1. Introduction

Non-Hodgkin lymphoma (NHL) is the 11th most common cancer diagnosis and the 11th leading cause of cancer death in the world. In 2020, there were 544,352 people diagnosed with NHL and 259,793 deaths globally. Although the overall global burden of NHL has risen 139.7% since 1990,¹ its incidence varies greatly between countries (Fig. 1). In 2020, China had the largest number of cases, accounting for approximately 17.1% of global NHL cases, followed by the US with 13.5%. While NHL deaths and disability adjusted life years (DALY) have decreased worldwide since 1990, NHL is a major challenge in regions of the world where case numbers are on the rise, as well as in key demographics that have significant disparities in survival.^1,2 Most NHL malignancies evolve from mature B lymphocytes and to a lesser degree from T lymphocytes or natural killer (NK) cells.^3,4 The frequency of NHL subtypes varies by region,⁵ with about 85% B cell lymphomas and about 15% T cell lymphomas in western countries and an inverse breakdown elsewhere.⁶ Geographic variations have been exploited to better understand causes, and epidemiologic investigation of NHL overall as well as by subtype has been pursued in etiologic studies in many countries using both case-control and cohort designs.

Fig. 1.

Geographic variation in Non-Hodgkin lymphoma incidence.

For the past two decades, the International Lymphoma Epidemiology Consortium (InterLymph) has undertaken multiple collaborative research projects pooling data across studies.^5,7 These pooled studies point to many causes of NHL and NHL subtypes,8, 9, 10 as does the broader epidemiologic literature on NHL. Contributors to NHL risk include genetic determinants, infection and immunosuppression, lifestyle and environmental factors, and occupational exposures.^9,11, 12, 13 Although the understanding of NHL risk factors has improved, there is still a growing literature on the important factors involved in the etiology of NHL, especially in countries with increasing incidence such as China.

Here, we provide a short overview of the broad risk factor categories relevant to the etiology of NHL, especially those indicated by InterLymph as well as in the expanding body of epidemiologic studies from China. We present brief evidence in support of the key factors contributing to an increased disease risk in recent decades and highlight some evidence from China.

2. Age

The literature shows that the risk for NHL increases with age, with highest incidence among those over the age of 80 and an increasing percentage in incident number among most age groups since 1990¹. This global increase in risk with age is illustrated in Fig. 2. In general, an earlier age at NHL diagnosis is often reported for patients in developing countries.⁶

Fig. 2.

Global variation in age-specific incidence of Non-Hodgkin lymphoma.

3. Geographical variation

Age adjusted incidence rate for NHL varies significantly between countries.¹⁴ In 2020, the highest rates were reported by Israel (13.3 per 100,000), Slovenia (13.2 per 100,000), and Australia (13.1 per 100,000), compared with 12.1 per 100,000 in the US, 9.6 per 100,000 in Japan, 6.4 per 100,000 in the Republic of Korea, 4.3 per 100,000 in China, and 6.4 per 100,000 in Thailand. NHL incidence is rising in 17 regions; the greatest observed increase is in East Asia, where incidence increased 343.8% between 1990 and 2019. Twenty-eight countries/territories showed decreasing trends, with the lowest estimated annual percent change occurring in Zimbabwe, followed by Kazakhstan and the US.¹ Although GLOBOCAN does not report NHL by subtype,^15,16 there is significant variation in distribution by subtype in developing regions which report a lower frequency of B cell lymphoma and a higher frequency of T and NK cell lymphoma than their developed counterparts.⁶

4. Gender

Men have higher rates of incidence and mortality from NHL globally17, 18, 19 (Fig. 3), and there is an unexplained male predominance in almost all subtypes.¹⁷ This varies from 1.5:1 overall to 3:1 for Burkitt lymphoma (BL), mantle cell lymphoma (MCL), hairy cell leukemia (HCL), and Waldenström macroglobulinemia (WM).¹⁷ Follicular lymphoma (FL) and Marginal Zone lymphoma (MZL) rates are similar in both sexes.¹⁷ Additionally, the sex distribution of NHL varies globally with a significantly higher number of males diagnosed in the developing world compared to the developed world.⁶ Given the significant heterogeneity of NHL, future investigation by disease subtype could further explore gender as a risk factor.²⁰

Fig. 3.

Global variation in Non-Hodgkin lymphoma incidence and mortality stratified by gender.

5. Familial and genetic susceptibility

Studies have observed clustered NHL incidences in families with shared genetic susceptibility and environmental exposures. A 2007 InterLymph²¹ pooled analysis of 10,211 NHL cases and 11,905 controls found that those who reported first-degree relatives with NHL were themselves at greater risk [odds ratio (OR) = 1.5, 95% CI: 1.2 to 1.9]. This risk was highest for those reporting a brother with NHL (OR = 2.8, 95% CI: 1.6 to 4.8), and was consistent for all NHL subtypes evaluated; however this was not true of opposite sex siblings.²⁰

In a series published in J Natl Cancer Inst Monogr in 2014, InterLymph investigators reported medical history, lifestyle, and occupational risk factors in addition to family history. Evaluating diffuse large B-cell lymphoma (DLBCL), the investigators found that family history of NHL increased DLBCL risk (OR = 2.0, 95% CI: 1.5 to 2.5)²²; FL risk was increased in those with a first-degree relative with NHL (OR = 2.0, 95% CI: 1.6 to 2.5)²³; chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) risk was associated with family history of any hematological malignancy (OR = 2.2, 95% CI: 1.8 to 2.7)²⁴; MZL was associated with family history of hematologic cancer (OR = 1.9, 95% CI: 1.4 to 2.6) and of NHL (OR = 2.8, 95% CI = 1.3 to 6.0)²⁵; peripheral T cell lymphomas (PTCLs) were associated with a family history of hematologic malignancies (OR = 1.9, 95% CI: 1.3 to 2.8)²⁶; a hematological malignancy among first-degree relatives was associated with a twofold increased risk of MCL (OR = 2.0, 95% CI: 1.4 to 2.8), with a stronger association in men (OR = 2.2, 95% CI: 1.4 to 3.4) than women (OR = 1.6, 95% CI: 0.8 to 3.2)²⁷; lymphoplasmacytic lymphoma/Waldenström macroglobulinemia (LPL/WM) risk was increased with hematologic malignancy in a first-degree relative (OR = 1.6, 95% CI: 1.0 to 2.6)²⁸; mycosis fungoides and Sézary syndrome (MF/SS) was associated with family history of multiple myeloma (OR = 8.5, 95% CI: 3.3 to 21.8)²⁹; and family history of hematological malignancy was associated with an increased risk of adult acute lymphocytic leukemia (ALL) (OR = 2.6, 95% CI: 1.2 to 5.5).³⁰

Moderate associations between several genes and the development of NHL have been identified.¹⁷ The tumor necrosis factor (TNF) gene is involved in inflammatory response and B cell apoptosis.³¹ TNF gene variants illustrate a modest increase in NHL risk overall, with a slightly stronger association with T cell lymphoma and MCL.³²  An InterLymph investigation also found an association between a polymorphism in TNF (−308 G) and increased risk of NHL (P for trend = 0.005), particularly for the main histological subtype DLBCL (OR = 1.3, 95% CI: 1.1 to 1.5) for GA genotype and for AA genotype (OR = 1.6, 95% CI: 1.2 to 2.3; P for trend <0.0001), but not for FL.³³ Certain gene polymorphisms involved in the regulation of interleukin 10 (IL10), a cytokine closely related to the proliferation and survival of B cells, were associated with a modest increase in risk of DLBCL and MCL.³² InterLymph investigators also reported an association between IL10–3575 T/A polymorphism and increased NHL risk (P for trend = 0.02), notably for DLBCL (P for trend = 0.006).³³ The investigators concluded that “common polymorphisms in TNF and IL10, key cytokines for the inflammatory response and Th1/Th2 balance, could be susceptibility loci for NHL” and additional subtypes.^34,35 In a case-control study in China, the investigators also reported that a polymorphism in TNF (TNFAIP8 rs1045241C>T) may contribute to NHL susceptibility in the Chinese population.³⁶ Interleukins including IL1B and IL6 have been found to play a key role in other studies,³⁷ indicating a role in immune cell proliferation. Studies of other genetic variants have yielded negative or inconsistent findings.^17,38,39 These included investigation of polymorphisms in the human leukocyte antigen (HLA) region,40, 41, 42, 43 the proapoptotic BCL2L11 gene,⁴² the EOMES gene (protein involved in defense against viral infections),^44,45 and in genes involved in height,⁴⁶ lipid traits,⁴⁷ interaction with smoking metabolism (NAT- N-acetyltransferase enzymes),⁴⁸ and obesity.⁴⁹

A series of InterLymph genome-wide association studies (GWAS) considered novel and established loci associated with risk of DLBCL, FL, CLL/SLL, and MZL. In the DLBCL GWAS, five independent single nucleotide polymorphisms (SNPs) in four loci achieved genome-wide significance marked by rs116446171 at 6p25.3 (EXOC2; P = 2.33 × 10⁻²¹), rs2523607 at 6p21.33 (HLA-B; P = 2.40 × 10⁻¹⁰), rs79480871 at 2p23.3 (NCOA1; P = 4.23 × 10⁻⁸) and two independent SNPs, rs13255292 and rs4733601, at 8q24.21 (PVT1; P = 9.98 × 10⁻¹³ and 3.63 × 10⁻¹¹, respectively). These data provided substantial new evidence for genetic susceptibility and pointed to pathways involved in immune recognition and immune function in the pathogenesis of DLBCL.⁵⁰ In the FL GWAS, a number of loci associated with FL were identified, providing evidence that multiple common variants beyond the more known HLA region including 11q23.3 (rs4938573, P = 5.79 × 10⁻²⁰) near CXCR5; 11q24.3 (rs4937362, P = 6.76 × 10⁻¹¹) near ETS1; 3q28 (rs6444305, P = 1.10 × 10⁻¹⁰) in LPP; 18q21.33 (rs17749561, P = 8.28 × 10⁻¹⁰) near BCL2; and 8q24.21 (rs13254990, P = 1.06 × 10⁻⁸) near PVT1⁴¹. In the CLL GWAS, the InterLymph investigators identified ten independent associated SNPs in nine new loci at 10q23.31 [ACTA2 or FAS (ACTA2/FAS), P = 1.22×10⁻¹⁴], 18q21.33 (BCL2, P = 7.76×10⁻¹¹), 11p15.5 (C11orf21, P = 2.15×10⁻¹⁰), 4q25 (LEF1, P = 4.24×10⁻¹⁰), 2q33.1 [CASP10 or CASP8 (CASP10/CASP8), P = 2.50×10⁻⁹), 9p21.3 (CDKN2B-AS1, P = 1.27×10⁻⁸), 18q21.32 (PMAIP1, P = 2.51×10⁻⁸), 15q15.1 (BMF, P = 2.71×10⁻¹⁰), and 2p22.2 (QPCT, P = 1.68×10⁻⁸), as well as an independent signal at an established locus (2q13, ACOXL, P = 2.08×10⁻¹⁸). The proximity of several of these loci to genes involved in apoptosis suggested a plausible underlying biological mechanism for CLL.⁵¹ In the MZL GWAS, the investigators identified two independent loci near BTNL2 (rs9461741, P = 3.95 × 10⁻¹⁵) and HLA-B (rs2922994, P = 2.43 × 10⁻⁹) in the HLA region. This offered the first evidence that genetic variation in the major histocompatibility complex influences MZL susceptibility.⁴³

6. Immune system abnormalities

Some autoimmune disorders are recognized as risk factors for NHL, with variation by subtype. Chronic immune stimulation due to infections, autoimmune inflammation, and/or suppression of regulatory cells is known to increase lymphoma risk.⁵² The incidence of NHL in patients with primary immune deficiencies is high, ranging from 12% to 25% depending on the degree of immune suppression.⁵² An InterLymph pooled analysis of self-reported autoimmune conditions and risk of NHL overall and by subtype (including 29,423 participants in 12 case-control studies) reported an increase in DLBCL risk with B cell activating autoimmune diseases (OR = 2.4, 95% CI: 1.8 to 3.1)²² and for the MZL subtype EMZL (OR = 6.4, 95% CI: 4.2 to 9.7); NMZL (OR = 7.8, 95% CI: 3.3 to 18.3); and SMZL (OR = 4.3; 95% CI: 1.5 to 12.1).²⁵ However, history of autoimmune disease was not associated with FL²³ or MCL.²⁷ The investigators also reported that Sjögren syndrome was associated with a 6.5-fold increased risk of NHL, a near 1000-fold increased risk of parotid gland MZL (OR = 996, 95% CI: 216 to 4596), and with DLBCL and FL.⁵³ In the same report, the investigators further found that systemic lupus erythematosus was associated with a 2.7-fold increased risk of NHL and an increased risk for the DLBCL and MZL subtypes. Additionally, hemolytic anemia was associated with DLBCL. T cell NHL risk was increased for participants with celiac disease and psoriasis. Results for rheumatoid arthritis were heterogeneous between studies. Furthermore, inflammatory bowel disorders, type 1 diabetes, sarcoidosis, pernicious anemia, and multiple sclerosis were not associated with risk of NHL or NHL subtypes. In an analysis of the association between diabetes and the risk of NHL using data from the Shanghai Men's Health Study and the Shanghai Women's Health Study, the investigators found that patients had a higher risk of incident NHL after diabetes diagnosis (HR = 2.00, 95% CI = 1.32–3.03).⁵⁴

7. Blood transfusions

Interest in blood transfusion as a causal link to NHL revolves around several biologic mechanisms: oncogenic viruses, transfusion-associated immune suppression, and engraftment of lymphoma cells from a donor with subclinical lymphoma.⁵⁵ In a 2019 InterLymph pooled analysis of self-reported blood transfusion and risk of NHL⁵⁶ using data from 13 case-control studies (including 10,805 cases and 14,026 controls), history of any transfusion among non-Hispanic whites was inversely associated with NHL risk for men (OR = 0.7, 95% CI: 0.6 to 0.8) but not women (OR = 0.9, 95% CI: 0.8 to 1.0). Among men, associations were stronger in hospital-based studies (OR = 0.6, 95% CI: 0.4 to 0.7) than population-based studies (OR = 0.8, 95% CI: 0.7 to 1.0). However, in the InterLymph subtype analyses published in J Natl Cancer Inst Monogr, investigators reported that DLBCL was inversely associated with previous blood transfusion (OR = 0.7, 95% CI: 0.6 to 0.8)²² as were FL (OR = 0.8, 95% CI: 0.7 to 0.9),²³ and CLL/SLL (OR = 0.8, 95% CI: 0.7 to 0.9).²⁴ A case-control study in Connecticut investigating NHL only found an increased risk of NHL for allogeneic blood transfusion to treat anemia.⁵⁷ While the evidence suggests that there is no association between blood transfusion and risk of NHL, these studies were unable to assess confounding by indication or possible recall bias.

8. Infections

There is strong evidence for the association between infections and NHL, as it has been consistently reported that defects in immunity or immune stimulation can result in viral mediated B-cell transformation and/or replication.⁵⁸ Meta-analyses have previously revealed an association between hepatitis C (HCV) infection and increased risk of NHL,⁵⁹ and investigators estimate that the fraction of NHL attributable to HCV could reach 10% in countries with high prevalence of infection.⁵⁹ Meta-analysis on infection as a risk factor specifically revealed a 2.5 fold increase in NHL risk (OR = 2.6, 95% CI: 2.2 to 2.9) in those infected with hepatitis B (HBV).⁶⁰ A 2021 InterLymph pooled analysis of self-reported history of infections and NHL risk from 17 case-control studies (12,585 cases and 15,416 controls)⁶¹ found that history of infectious mononucleosis (IM) was associated with an excess risk of NHL in 16 studies (OR = 1.3, 95% CI: 1.0 to 1.6) while history of measles or whooping cough was associated with an approximate 15% reduction in risk of NHL. An earlier InterLymph analysis evaluating NHL and HCV reported that HCV infection was detected in 172 NHL cases (3.6%) and in 169 (2.7%) controls (OR = 1.8, 95% CI: 1.4 to 2.3). Subtype-specific analyses found an association between HCV and MZL (OR = 5.3, 95% CI: 2.5 to 11.3),²⁵ DBCL (OR = 2.0, 95% CI: 1.5 to 2.8),²² LPL/WM²⁸ and lymphoplasmacytic lymphoma (OR = 2.6, 95% CI: 1.1 to 5.8), and CLL/SLL²⁴, but not FL.^23,62 The investigators concluded that there was little evidence of an association between NHL risk and infection in general,⁶¹ but there were some associations for specific subtypes. These data relied on self-report of some conditions that might be hard to recall depending on the data collection format, which could bias the findings.

Epstein–Barr virus (EBV) was associated with B cell lymphoma, T cell lymphoma, and NK cell lymphoma.⁶³ An InterLymph pooled analysis exploring the interaction between NHL, IM history, and immune-related genotypes from 12 case–control studies (7926 NHL patients and 10,018 controls) found evidence of an interaction effect between IM history and two variants on T cell lymphoma risk: rs1143627 in interleukin-1B (IL1B) (P_interaction = 0.04, OR_interaction = 0.09, 95% CI: 0.01 to 0.87) and rs1800797 in interleukin-6 (IL6) (P_interaction = 0.03, OR_interaction = 0.08, 95% CI: 0.01 to 0.80). The authors concluded that genetic risk variants in IL1B and IL6 might affect the association between IM and T-cell lymphoma through a pathway decreasing risk of immune cell proliferation.³⁷ EBV and HBV are endemic in China with a seroprevalence of 90%⁶⁴ and 7.2%⁶⁵ respectively, which may play a crucial role in the disease burden of lymphoma.⁶⁶ A pooled study³⁰ based on three prospective cohorts from China and Singapore confirmed that an increased risk of NHL was related to EBV infection (OR = 2.17) and HBV infection (OR = 2.16).⁶⁷

As the incidence and distribution of viruses differ between East Asian and Western populations, investigators have further evaluated the associations between NHL risk and serologic markers of multiple viral infections in pre-diagnostic blood in three population-based prospective cohorts in Shanghai and Singapore. An increased NHL risk was observed for higher compared to lower antibodies against the HBV-HBc and HBe antigens.⁶⁸

9. Smoking

Smoking is of particular interest for NHL risk as it may alter immunity and has been shown to induce t(14; 18) translocation that results in overproduction of the anti-apoptosis protein bcl-2⁶⁹. An InterLymph pooled-analysis including 12 case-control studies of cigarette smoking revealed a dose–response relationship with increasing NHL risk in heavy or long duration smokers.⁷⁰ Stratification by NHL subtype revealed the most consistent association with smoking in FL (OR = 1.3, 95% CI: 1.1 to 1.5).⁷⁰

The authors suggest that even if not relevant to all subtypes, cigarette smoking should be added to the few modifiable NHL risk factors identified. Subtype specific analyses found that smoking was associated with specific types of DLBCLs.²⁹ For FL, longer duration of cigarette use elevated the risk specifically in women,²⁸ and smoking showed a protective effect (OR = 0.9, 95% CI: 0.8 to 1.0) for CLL/SLL²⁷. For PTCL, smoking 40 or more years increased risk (OR = 1.9, 95% CI: 1.4 to 2.6),²⁶ and duration of cigarette smoking increased risk of LPL/WM (OR = 1.5, 95% CI: 1.0 to 2.1) for ≥ 40 years versus nonsmokers²⁴ and for MF/SS (OR = 1.6, 95% CI: 1.0 to 2.3).²³ However, MCL risk was not associated with tobacco use.²²

10. Alcohol

Multiple epidemiologic studies have examined the association of alcohol in NHL risk, with inconsistent results.⁷¹ While the relationship is unclear,⁷² one popular hypothesis is that alcohol produces carcinogenic metabolites and inactivates tumor suppressor genes.⁷³ Evidence also indicates that light to moderate alcohol intake improves immunocompetence,⁷² and it is also possible that antioxidants found in wine can improve inflammation and endothelial function.⁷⁴ In 2005, InterLymph reported on risk due to alcohol consumption using nine case-control studies (6,492 cases and 8,683 controls). The investigators reported that alcohol drinkers had a lower risk than non-drinkers (OR = 0.8, 95% CI: 0.8 to 0.9), and that compared with non-drinkers ORs were lower for current drinkers (OR = 0.7, 95% CI: 0.6 to 0.8) versus former drinkers (OR = 1.0, 95% CI: 0.8 to 1.1).⁷¹ Lifetime alcohol consumption was associated with a decreased risk of DLBCL (OR = 0.6, 95% CI: 0.4 to 0.8 for >400 kg versus nondrinker)²² as well as a decreased risk of FL in women (OR = 0.8, 95% CI: 0.7 to 0.9).²³ Similarly, a decreased risk was found for MZL with consumption of any alcohol (OR = 0.5, 95% CI: 0.3 to 0.8) and lower consumption of wine (ORs < 0.45) compared with nondrinkers²⁵ and PTCL (ever drinker) (OR = 0.6, 95% CI: 0.5 to 0.8).²⁶ Lower risk estimates were recorded for Burkitt's lymphoma (OR = 0.5, 95% CI: 0.3 to 0.8),⁷⁵ while consumers of alcohol had an increased risk of B cell acute lymphoblastic leukemia (OR = 2.9, 95% CI: 1.2 to 7.0).³⁰ In summary, alcohol consumption might be associated with lower overall risk of NHL, and this risk appears to vary by subtype. However, a limitation of the findings is that harmonization of alcohol consumption details was a challenge across the InterLymph studies.

11. Body mass index (BMI)

Obesity (BMI ≥ 30 kg/m²) is increasing in prevalence worldwide⁷⁶ and is an NHL risk factor of concern because of its impact on immune function and inflammatory response.⁷⁷ A meta-analysis of 16 prospective studies showed an estimated 7% increase in risk of NHL per 5 kg/m² increment in BMI (RR = 1.1, 95% CI: 1.0 to 1.10),⁷⁷ while a 2008 pooled investigation of case-control studies found no association between severe obesity (BMI ≥ 40 kg/m²) and NHL risk overall (OR = 1.0, 95% CI: 0.7 to 1.4) or for most NHL subtypes. However, the association between severe obesity and DLBCL (OR =1.8, 95% CI: 1.2 to 2.6) is notable.⁷⁸ In an updated pooled investigation of case-control studies published in 2015, obesity was found to be associated with DLBCL risk (OR = 1.3, 95% CI: 1.0 to 1.7). While no association with obesity was observed between FL and CLL/SLL⁴⁹, an investigation of rarer subtypes24, 25, 26, 27 found an increased risk of MF/SS (OR = 1.6, 95% CI: 1.0 to 2.4)²⁹ .

12. Sun exposure

Sun exposure has been postulated as a risk factor for NHL due to the immunosuppressive effects of ultraviolet radiation.^79,80 The evidence on the association between sun exposure, vitamin D intake, and NHL has evolved in the past two decades as more recent studies have reported that sun exposure is associated with a lower risk of NHL subtypes.81, 82, 83, 84 In a 2008 pooled analysis of 10 studies (8,243 cases and 9,697 controls) the risk of NHL decreased significantly with increasing recreational sun exposure (OR = 0.8, 95% CI: 0.6 to 0.9). An InterLymph investigation of DLBCL reported that higher recreational sun exposure illustrated an inverse risk (OR = 0.8, 95% CI: 0.7 to 0.9)²² and similar protective findings were reported for CLL/SLL²⁴. However, no association was reported for FL²³ or MCL.²⁷ The investigators subsequently concluded that increased recreational sun exposure may protect against NHL,⁸⁵ but the risk varied by subtype. However, exposure misclassification in estimating the impact of sun exposure was significant as there was great variation in sun exposure assessment across InterLymph studies – for example, most InterLymph studies did not assess lifetime sun exposure. This is a limitation in the literature on this topic.

13. Hair dye

Use of hair dyes has been investigated as a potential risk factor for NHL86, 87, 88 as some dyes contain the suspected carcinogens phenylenediamines and paraphenylenediamine.^89,90 Greater duration of hair dye use and the use of dark-colored dyes are associated with elevated risks of NHL.^91,92 Specifically, in a meta-analysis published in 2005⁸⁷, the investigators reported a slight increase in the risk of NHL for ever-users of hair dye (RR = 1.2, 95% CI: 1.1 to 1.4). The increased risk was observed for some subtypes including mediastinal DLBCL²² and MZL.²⁵ As such, this evidence of risk should be investigated further. A key consideration in future work is that many carcinogens were removed from hair dyes in the 1980s following an International Agency for Research on Cancer (IARC) monograph publication on the risks posed by carcinogens⁹³ as well as subsequent reports specifically addressing occupational and personal use of hair dyes.⁹⁴ Careful consideration of the timing of dye use is therefore key to understanding their impact.

14. Diet

Studies exploring the possible relationship between NHL risk and consumption of animal-derived foods have yielded conflicting results.95, 96, 97, 98, 99, 100 A 2007 report from the World Cancer Research Fund/American Institute for Cancer Research noted that most published papers reported an increased NHL risk with greater consumption of meat, milk, and dairy products. A literature review and meta-analysis published in 2016,¹⁰¹ found a statistically significant association between red meat consumption and the risk of NHL. The investigators found no significant association between the consumption of either processed or white meat and NHL, while the association between dairy and the risk of NHL was borderline significant. However, there was no statistically significant association between the consumption of eggs and the risk of NHL (SRR 1.2, 95% CI: 0.9 to 1.6, I² = 78%) based on 5775 cases from ten independent studies.¹⁰¹ In an evaluation of dietary patterns and NHL risk in the Multiethnic cohort that included vegetables, fruit and milk, and fat and meat consumption patterns, no significant association was found between any of these patterns and the overall risk of NHL.¹⁰² However, the vegetable pattern was inversely associated with risk in White women with a hazard ratio of 0.6 (P for trend = 0.04), while the fat and meat pattern was associated with a five-fold increased risk of FL in men of all races (P for trend = 0.03).

15. Occupational exposures

Many occupations have been considered as risk factors in the NHL epidemiologic literature, with an emphasis on exposure to pesticides and organic solvents such as benzene and trichloroethylene. A 2016 InterLymph analysis of NHL and occupation (10,046 cases and 12,025 controls) confirmed associations between NHL and farming occupations (field crop/vegetable farm workers OR = 1.3, 95% CI: 1.1 to 1.5; general farm workers OR = 1.2, 95% CI: 1.0 to 1.4), as well as occupations such as women's hairdressers (OR = 1.3, 95% CI: 1.0 to 1.7), charworkers/cleaners (OR = 1.2, 95% CI: 1.0 to 1.4), spray-painters (OR = 2.0, 95% CI: 1.3 to 3.3), electrical wiremen (OR = 1.2, 95% CI: 1.0 to 1.5), and carpenters (OR = 1.4, 95% CI: 1.0 to 1.9). Associations were also observed for DLBCL and CLL/SLL in women's hairdressers and for DLBCL and peripheral T cell lymphoma in textile workers. The investigators concluded that farming, hairdressing, and textile industry-related exposures may contribute to NHL risk.¹⁰³ In a more recent report¹⁰⁴ InterLymph investigators further evaluated occupational insecticide use and risk of NHL, finding that use of organophosphate insecticides was associated with increased risk of NHL overall with subtype analyses specifically suggesting an association between diazinon and FL. Carbaryl was associated with an increased risk of all NHL, and the strongest associations were found with T cell NHL for ever-use and longer duration. NHL was not associated with other insecticides, and some inverse associations were estimated in relation to historical dichloro-diphenyl-trichloroethane (DDT) use.¹⁰⁴ They also reported on the association between occupational exposure to trichloroethylene and risk of NHL and found an impact on the risk of FL, but not NHL overall or other subtypes. When they restricted their analysis to the most likely exposed study subjects, risk of NHL overall, FL, and CLL were elevated and increased by duration of exposure. The Interlymph investigators concluded that these findings support the hypothesis of an increase in risk of specific NHL subtypes associated with occupational exposure to trichloro ethylene (TCE).¹⁰⁵ In three prospective cohorts in China and Singapore¹⁰⁶ the investigators reported an increased risk of NHL among those with higher organochlorine pesticide levels (specifically hexachlorocyclohexane). While this association was highly consistent across the three cohorts, no significant associations were observed for other organochlorines including dichlorodiphenyl dichloroethene (DDE). Additionally, large cohort studies in urban Shanghai populations found an association between occupational benzene exposure levels and increased NHL risk.⁶⁸

16. Air pollution

Epidemiological studies of the relationship between air pollution and NHL risks are limited with mixed results.¹⁰⁷ Earlier studies using residence as a proxy variable for environmental pollution concluded there was no association between NHL risk and residence in urban or heavily polluted areas.^108,109 The American Cancer Society Cancer Prevention Study II (ACS CPS-II) initially reported that ambient exposures to fine particulate matter (PM_2.5), nitrogen dioxide (NO₂), or ozone (O₃) were not associated with NHL risk.¹¹⁰ In addition, two studies based on the Danish Cancer Registry also found no association between NHL and PM_2.5, NO₂, or O₃^111,112. However, several recent studies observed an increased NHL risk associated with ambient air pollution. One study based on the National Health Interview Survey observed significant association between ambient PM_2.5 exposure and NHL risk with a hazard ratio of 1.5 (95% CI: 1.1 to 2.0) per 10 μg/m³ increase.¹¹³ Another analysis of the Danish Cancer Registry concluded that some constituents of PM_2.5 increased NHL risk, which included primary carbonaceous particles (OR = 1.03, 95% CI: 1.00 to 1.07 per interquartile range) and secondary organic aerosols (OR = 1.5, 95% CI: 1.1 to 2.1 per interquartile range).¹¹⁴ Some of these mixed results can be attributed to the use of varied ambient exposure models and modeling approaches as well as dramatically different environmental contexts where NHL cases in these studies reside.

17. Pesticides and other organic pollutants

Many studies have evaluated the association between NHL risk and exposures to organic pollutants including benzene, specific pesticides, and chlorinated insecticides.¹¹⁵ As we have summarized some results of exposures to organic pollutants in the occupational exposures section, this section will focus on non-occupational exposure studies.

Organochlorine pesticide is one of the most studied organic pollutants in relation to NHL. In a case-control study nested within a Danish population-based prospective cohort, researchers observed increased NHL risk associated with DDT measured in tissue (OR = 1.4, 95% CI: 1.1 to 1.7, per interquartile range).¹¹⁶ A Canadian population-based study also reported increased NHL risk associated with pesticide chlordane (OR = 2.7, 95% CI: 1.7 to 4.2, highest versus lowest quartile)¹¹⁷ based on pesticides or pesticide metabolites measured in plasma. This conclusion was supported by a Swedish study (measured in plasma) and a US study (measured in adipose tissue) (OR = 6.8, 95% CI: 2.3 to 20.0 and OR = 3.4, 95% CI: 1.9 to 6.2, respectively).^118,119 In contrast, among eight studies on hexachlorobenzene that relied on self-report, only one observed an association with NHL.¹¹⁵ The association between pesticides and NHL was largely limited to NHL cases with the t(14;18) chromosomal translocation as demonstrated by two epidemiological studies.^120,121 This translocation appears to be one step in the progression of a normal cell to a cancer cell; however it is unclear whether pesticides cause the t(14;18) translocation or whether it is generated during the course of malignant transformation as a result of the developing genomic instability that arises during disease progression.¹¹⁵

There are fewer studies on specific insecticides, and the findings are mixed. Two studies on organochlorine insecticides reported positive association with NHL^117,122 while three other studies reported null association.^116,123,124 Moreover, analysis of the ACS CPS-II nutrition cohort suggested that ambient benzene exposure was associated with NHL subtypes including T cell lymphoma in both sexes and FL in males.¹²⁵

Although the epidemiological evidence for certain organic pollutants and NHL is growing, little is known about the biological/toxicological mechanisms by which these compounds may contribute to this disease. Future studies are needed to address this gap.

18. Evidence from Chinese studies

To our knowledge, only a few studies have investigated the risk factors for NHL in the Chinese population, and almost all of them focused on the association between NHL and previous diseases. Summarizing the reports above, we note that one study that pooled participants from two population-based cohorts in Shanghai observed increased NHL risk associated with type 2 diabetes.⁵⁴ A possible association between H. pylori and NHL was also seen in a large multi-regional cross-sectional study in China.¹²⁶ A Chinese case-control study concluded that HBV infection and chronic inflammation contributed to an increased NHL risk.¹²⁷ Chinese studies have identified SNPs associated with an increased risk of B cell and T cell NHL, but none of these SNPs are in immunoregulatory genes,^128,129 whereas one study reports a role of immunoregulatory gene variants in NHL etiology in the Chinese population.¹³⁰ Overall, the body of literature investigating potential risk factors in the Chinese population appears to be very limited. Large cohort studies are needed to investigate risk factors specific to this population.

19. Summary

Epidemiologic investigation of NHL etiology has had significant support in recent decades. InterLymph has been especially productive and has been effective at integrating individual studies in a manner that confers strong evidence. It is increasingly clear that established risk factors such as age, immune modulation from autoimmune disease, and infections account for a small percentage of the overall disease burden. An increasing focus on NHL subtypes, especially in pooled studies, may expand our insight. Likewise, more investigation into emerging exposures with integrated genetic factors and comprehensive medical records is needed. Given the growing disease burden in China, more research in this area of the world to understand its unique exposures is also warranted.

Declaration of competing interest

The authors declare that they have no conflict of interests.