Hepatitis C infection and risk of malignant lymphoma

Authors


Abstract

The association between hepatitis C virus (HCV) infection and risk of malignant lymphoma remains controversial, perhaps due to small-sized studies and low prevalence of HCV in the general population. On the basis of a large Danish-Swedish population-based case-control study, 2,819 lymphoma patients and 1,856 controls of second-generation Danish-Swedish origin were screened for HCV infection using an enzyme-linked immunosorbent assay and a confirming recombinant immunoblot assay (RIBA) test. Positive samples were tested with real-time PCR for the presence of HCV RNA. The association between HCV infection and risk of malignant lymphoma was assessed by logistic regression. When intermediate RIBA test results were interpreted as positive, anti-HCV antibody positivity was associated with a nonsignificant increased risk of non-Hodgkin lymphoma (NHL) overall (odds ratio (OR) = 2.2; 95% confidence interval (CI) 0.9–5.3; n = 20 cases), of B-cell lymphomas combined (OR = 2.4 [1.0–5.8]; n = 20) and of lymphoplasmacytic lymphoma (OR = 5.2 [1.0–26.4]; n = 2). No patients with T-cell or Hodgkin lymphoma were HCV-positive. A more conservative definition of HCV positivity (disregarding intermediate RIBA results) resulted in an OR = 1.6 (0.3–8.5; n = 5) for NHL overall. When the definition was further restricted to require HCV RNA positivity, OR was 1.7 (0.2–16.2; n = 3) for NHL overall. Our findings from a population with a low prevalence of HCV suggest a positive association between HCV and risk of NHL, in particular of B-cell origin. © 2008 Wiley-Liss, Inc.

A number of viral infections have been linked to the development of subtypes of malignant lymphoma, including Epstein–Barr virus (EBV), human herpesvirus 8 (HHV-8), human immunodeficiency virus (HIV) and human T-cell lymphotrophic virus type 1 (HTLV-1).1 More recently, evidence has accumulated to suggest a role also for hepatitis C virus (HCV), a single-stranded RNA virus,2 in lymphoma development.

Infection with HCV is primarily associated with an increased risk of liver cirrhosis and hepatocellular carcinoma.3 Its putative role in the development of certain lymphoproliferative diseases became apparent when chronic HCV infection was recognized as the principal cause of mixed cryoglobulinaemia (MC), particularly type II.4 MC, a low-grade lymphoproliferative disorder, can evolve into non-Hodgkin lymphoma (NHL).5 Rearrangement of the anti-apoptotic bcl-2 gene via t(14;18) translocation is the most common chromosomal translocation found in lymphoid malignancies, especially follicular lymphoma (FL) and diffuse large B-cell lymphoma (DLBCL), and likely represents an early event in lymphomagenesis.6 The translocation has been found in peripheral mononuclear cells in a large fraction of patients with type II MC, and also in about 35% of patients with chronic HCV infection.7

Recent systematic reviews and meta-analyses of previous studies have shown striking geographic and demographic variation in the association between HCV infection and risk of NHL with high relative risks in Southern European populations, but not in Northern Europe or North America.8–10 The discrepant findings may reflect several factors, such as wide variation in HCV prevalence in the general population between countries and even between different regions in the same country,11–13 the use of a variety of control populations (population-based, hospital-based and blood-donor-based controls) not necessarily representative of the source population that gave rise to the cases, lack of statistical power in some studies especially for NHL subtypes, and, finally, heterogeneity of HCV detection methods.

Large-scale, population-based studies are needed to confirm or refute the proposed association between HCV and NHL. If a causal association exists, antiviral treatment of the HCV infection could inhibit t(14;18) translocations that potentially lead to lymphocyte transformation, and also may entail regression of selected low-grade HCV-associated lymphomas.2, 14 Antiviral treatment could also prevent NHL development through other mechanisms, such as abatement of chronic antigen stimulation.2 To this end, we investigated HCV infection and the risk of NHL overall, of specific NHL subtypes, and of Hodgkin lymphoma (HL) in a large, population-based Scandinavian case-control study.

Subjects and methods

Study population

The study was based on data and serum samples collected in a nation-wide Danish-Swedish case-control study (Scandinavian Lymphoma Etiology study, SCALE) from 1999 to 2002, previously described in detail Ref.15. The SCALE study population encompassed the entire Danish population aged 18 to 74 years in the period June 1, 2000, to August 30, 2002, and the corresponding Swedish population in the period from October 1, 1999, to April 15, 2002. Eligible cases in SCALE were those with a primary, newly diagnosed, and morphologically verified malignant lymphoma. Cases and controls were required to speak Danish or Swedish, and to have no history of organ transplantation, HIV infection or prior hematopoietic malignancy. Cases were identified through a rapid case ascertainment network including all departments diagnosing and treating malignant lymphomas in both countries (39 departments in Denmark and 118 in Sweden). Continuous collaboration with the Danish National Pathology Registry and with the 6 regional cancer registries in Sweden ensured complete reporting through the network. Controls were randomly sampled from the entire Danish and Swedish populations using continuously updated, computerized population registers. A subset of controls was sampled every 6 months during the study period, and was frequency-matched within each country on the expected distribution of cases of malignant lymphoma, by sex and age in 10-year intervals.

Overall, 3,673 incident cases with malignant lymphoma (3,055 with NHL and 618 with HL) and 3,187 controls were enrolled and interviewed by telephone. The participation rates were 81% among identified eligible NHL cases, 91% among HL cases, and 71% among controls. The main reason for nonparticipation was early death among cases (n = 279), whereas unwillingness was most frequent (n = 718) among controls.15 The telephone interviews followed a structured questionnaire covering known and suspected risk factors for malignant lymphoma. Furthermore, all study subjects were asked to provide a blood sample, to which 85% of the participating cases and 65% of the participating controls consented. When possible, blood specimens were obtained from lymphoma patients prior to initiation of treatment. All specimens were stored at −80°C. For the present analyses, the study population was restricted to individuals who donated blood and were born in Denmark and Sweden, to parents also both born in Denmark or Sweden.

The study was approved by regional ethics committees in both countries. Informed consent was obtained from each participant before the interview and blood sampling.

Histopathological classification

All included cases in the SCALE study were classified according to the current World Health Organization (WHO) classification of lymphoid tumors.6

In Denmark, tumor material from cases was reviewed within the Danish Lymphoma Group Registry (LYFO).16 Ten percent of all incident cases in Denmark are continuously randomly chosen and reviewed by expert hematopathologists. Additionally, LYFO-approved senior hematopathologists performed the primary evaluation of the diagnostic tumor specimens of all but 20% of the study cases.

In Sweden, all cases were histopathologically evaluated by one of six senior hematopathologists/cytologists affiliated with the study. The original diagnostic tumor slides were reviewed for all but 1.5% of cases, for whom the written results of the primary morphological and immunohistochemical investigation were evaluated.

Information on lymphoma topography and location (nodal/extranodal) was obtained through LYFO in Denmark and 6 regional lymphoma registries in Sweden.

Exposure assessment

All serum samples were screened for the presence of anti-HCV antibodies with a third-generation indirect enzyme-linked immunosorbent assay (ELISA-3; AxSym HCV version 3.0, Abbott Laboratories, Abbott Park, Illinois) to qualitatively detect antibodies to the core, NS3, NS4 and NS5 antigens of HCV. ELISA-positive samples were confirmed by using a third-generation recombinant immunoblot assay (RIBA-3; Chiron RIBA HCV 3.0 SIA, Chiron Corporation, Emeryville, CA), detecting recombinant c33c and NS5 antigens and synthetic 5-1-1, c100 and c22 peptides to HCV. The analyses were performed at the Department of Virology, Statens Serum Institut, accredited by the Danish Accreditation and Metrology Fund (DANAK) according to ISO 17025.

Additionally, a qualitative real-time (RT) PCR test for the presence of HCV RNA was carried out in all ELISA-positive serum samples at the Department of Virology, Statens Serum Institut. RNA was extracted from 100 μl plasma using the Roche MagNA (Total Nucleic Acid Isolation Kit, Roche Applied Science, Indianapolis), according to the manufacturer's instructions. A 103-base-pair region in the 5′UTR was amplified in a two-step real-time RT-PCR using TaqMan Gold RT-PCR Kit (Applied Biosystems, Foster City, CA). In the RT reaction we used sequence-specific HCV5UTR-R1primer (5′-CACTCGCAAGCACCCTATCA-3′)17 in a final concentration of 1,000 nM for cDNA synthesis in a 40 μl reaction containing 5.5 mM MgCl2. RT reactions were incubated in 25°C for 10 min, then in 48°C for 30 min and 95°C for 5 min. Ten microliter of the cDNA was amplified in 50 μl PCR reactions using 400 nM forward primers HCV-5UTR-Fw1 (5′-ACTACTGTCTTCACGCAGAAAGC-3′) and HCV-5UTR-Fw2 (5′-ACTTCTGTCTTCACGCGGAAAGC-3′) and reverse primer HCV-5UTR-Rv(5′-GGTTCCGCAGACCACTATGG-3′). The PCR reactions contained 5.5 mM MgCl2 and 5% dimethyl sulfoxide (Sigma). Amplicons were detected using 200 nM of each of the specific MGB probes (Applied Biosystems) HCV-5UTR-MGB1 (5′-[FAM]-CACTCATACTTACGCCATG-[TAMRA]-3′) and HCV-5UTR-MGB2 (5′-[FAM]-CACTCGTACTTACGCCATG-[TAMRA]-3′). PCR reactions were incubated according to the PCR kit manufacturer's instructions using an ABI7000 instrument (Applied Biosystems) in 9600 emulation mode. The lower detection limit of the assay is ∼1,000 IU/ml.

We defined HCV infection as a positive test for anti-HCV antibodies (positive ELISA test in combination with a positive or intermediate RIBA test), with or without detection of HCV RNA. Individuals with intermediate RIBA results (i.e., the detection of antibodies against only one HCV epitope) were included in our study as HCV-infected, because the sensitivity of RIBA tests to fully detect antibodies against different HCV epitopes may be suboptimal in immunocompromised individuals.18–20

Statistical analyses

Associations between HCV infection and risk of NHL overall, specific NHL subtypes and HL were evaluated using unconditional logistic regression. To estimate relative risks, we calculated odds ratios (OR) with 95% confidence intervals (CI) with initial adjustment for the matching variables sex, age in 10-year intervals, and country of residence. Additional adjustment for history of autoimmune disease, allergy, mononucleosis or venereal disease, previous HIV test, previous blood transfusion, travel history (outside Europe or to Southern Europe for more than 2 months) conferred only marginal changes in OR estimates, whereas the addition of educational level changed estimates more than 10%. Accordingly, educational level (≤9, 10–12, or ≥13 years) was added to the multivariable model. All statistical tests were two-sided likelihood-ratio tests. Effect modification by age group (<55 or ≥55 years) or sex was tested by introducing an interaction term in the multivariable model. Confidence intervals were calculated based on the Wald test. p-Values less than 5% were considered statistically significant.

To test the effect of lymphoma treatment on serological results we defined two case subgroups according to whether the serum sample was taken before or after initiation of treatment. We assessed potential heterogeneity of the effect of HCV seropositivity on lymphoma risk in a polytomous regression model, with the aforementioned two lymphoma subgroups and control status as outcomes. Each parameter in the model was the log OR of a lymphoma subgroup outcome compared with control status for the exposure levels compared with a reference level, and heterogeneity was tested with a Wald test.

Results

Overall, 2,819 patients with malignant lymphoma and 1,856 controls, representing all participants who donated blood and were of at least second-generation Danish/Swedish origin, were included in the analysis. The distribution of cases and controls by age, sex, country of residence, educational level and HCV infection status is given in Table I.

Table I. General and disease characteristics of controls and patients with Hodgkin lymphoma, non-Hodgkin lymphoma (NHL) overall, and NHL subtypes
CharacteristicsControls, N (%)HL, N (%)All NHLs, N (%)NHL subtypes
DLBCL, N (%)CLL, N (%)FL, N (%)MCL, N (%)MZL, N (%)LPL, N (%)TCL, N (%)
  • NHL, non Hodgkin lymphoma; HL, Hodgkin lymphoma; DLBCL, diffuse large B-cell lymphoma; CLL, chronic lymphocytic lymphoma; FL, follicular lymphoma; MCL, mantle cell lymphoma; MZL, marginal zone lymphoma; LPL, lymphoplasmacytic lymphoma, TCL, T-cell lymphoma; HCV, hepatitis C virus.

  • 1

    Anti-HCV antibody positivity defined as a positive enzyme-linked immunosorbent assay test and positive or intermediate recombinant immunoblot assay test; three HCV positive cases were of B-NHL subtype, not otherwise classified.

  • 2

    Numbers may not add up due to missing information.

Study participants overall1,856 (100)466 (100)2,353 (100)596 (100)577 (100)454 (100)123 (100)87 (100)94 (100)160 (100)
Age (years)          
 18–44216 (12)307 (66)235 (10)95 (16)15 (3)46 (10)3 (2)4 (4)4 (4)45 (28)
 45–54324 (17)50 (11)442 (19)113 (19)101 (18)107 (24)22 (18)15 (17)14 (15)25 (16)
 55–65580 (31)56 (12)784 (33)180 (30)193 (33)169 (37)45 (37)32 (37)36 (38)42 (26)
 65–74736 (40)53 (11)892 (38)208 (35)268 (46)132 (29)53 (43)36 (41)40 (43)48 (30)
Median (range)61 (18–75)36 (17–74)61 (18–74)60 (19–74)63 (30–74)58 (22–74)63 (37–74)62 (26–74)62 (28–74)57 (18–74)
Sex          
 Male1,013 (55)252 (54)1,423 (60)364 (61)374 (65)217 (48)93 (76)47 (54)62 (66)97 (61)
 Female843 (45)214 (46)930 (40)232 (39)203 (35)237 (52)30 (24)40 (46)32 (34)63 (39)
Country of residence          
 Denmark735 (40)198 (42)813 ((35)206 (35)217 (38)174 (38)43 (35)42 (48)42 (45)60 (38)
 Sweden1,121 (60)268 (58)1,540 (65)390 (65)360 (62)280 (62)80 (65)45 (52)52 (55)100 (62)
Education level (years)2          
 ≤9522 (28)67 (14)815 (35)208 (35)195 (34)150 (34)47 (39)29 (33)39 (41)47 (30)
 10–12522 (28)263 (56)1,005 (43)253 (43)239 (42)206 (46)47 (39)43 (49)42 (45)72 (45)
 ≥13443 (24)134 (29)509 (22)131 (22)134 (24)91 (20)27 (22)15 (17)13 (14)40 (25)
Anti-HCV antibodies1          
 Negative1,849 (99.6)466 (100)2,333 (99)591 (99)571 (99)450 (99)123 (100)87 (100)92 (98)160 (100)
 Positive7 (0.4)20 (1)5 (1)6 (1)4 (1)2 (2)

In total, 84 individuals (57 NHL patients [2% of all tested], 6 HL patients [1% of all tested] and 21 control subjects [1% of all tested]) had a positive ELISA test for anti-HCV antibodies. In subsequent confirmatory analyses with RIBA, 57 of these were negative and 27 were either positive (n = 7) or intermediate (n = 20), including 20 NHL patients (0.7% of all tested) and 7 control subjects (0.4% of all tested). HCV RNA was detected in 4 of the 27 anti-HCV-antibody positive serum samples (from three NHL patients and one control subject).

HCV infection, defined as a positive ELISA test and a positive or intermediate RIBA test for anti-HCV antibodies, was associated with a statistically nonsignificantly increased risk of NHL overall (OR = 2.2; 95% CI 0.9–5.3), and a significantly increased risk of all B-cell lymphomas combined (OR = 2.4; 95% CI 1.0–5.8). Among the major NHL subtypes, HCV infection was statistically significantly associated with an increased risk of lymphoplasmacytic lymphoma (LPL; OR = 5.2; 95% CI 1.0–26.4; n = 2), and nonsignificantly associated with chronic lymphocytic leukemia (CLL; OR = 2.7; 95% CI 0.9–8.4; n = 6), DLBCL (OR = 2.4; 95% CI 0.8–7.8; n = 5), and FL (OR = 2.7; 95% CI 0.8–10.0; n = 4). HCV infection was not found in patients with mantle cell lymphoma (MCL), marginal zone lymphoma (MZL), T-cell lymphoma or Hodgkin lymphoma (HL). There was no variation in the association between HCV infection and risk of NHL overall by sex (pinteraction = 0.1) or age (pinteraction = 0.9). Likewise, timing of blood sampling relative to treatment among lymphoma cases had no impact on the observed associations between lymphoma risk and HCV seropositivity (p = 0.6).

In a more conservative approach, defining HCV-positive as someone with a positive ELISA test and a positive RIBA test, we found the following associations with HCV positivity: NHL overall (OR = 1.6; 95% CI 0.3–8.5; n = 5 cases and 2 controls), B-cell lymphomas combined (OR = 1.8; 95% CI 0.3–9.2; n = 5 cases), FL (OR = 3.2; 95% CI 0.4–24.2; n = 2 cases) and CLL (OR = 4.7; 95% CI 0.8–28.9; n = 3 cases). Restricting the analyses further to individuals who tested positive for both anti-HCV antibodies and HCV RNA, the OR for NHL overall was 1.7 (95% CI 0.2–16.2; n = 3 cases and 1 control).

Table II shows the characteristics of all HCV-infected cases and controls by age, sex, country of residence and presence of anti-HCV antibodies (by ELISA or RIBA) and HCV RNA. One of the HCV-infected cases had primary extranodal lymphoma, one had combined nodal/extranodal disease, and the remainder with available information all had nodal disease. Eight patients and 3 control subjects reported potential behavioral risk factors of HCV infection, including previous HIV testing (although HIV-positive individuals were excluded from the entire SCALE study), history of venereal disease, and/or blood transfusion before 1991 (when mandatory HCV screening of blood donors was introduced).

Table II. Characteristics of Controls and Non-Hodgkin Lymphoma Patients with Serological Evidence of Hepatitis C Virus Infection
PatientAge group (yrs)SexCountryELISARIBAHCV RNAControl/CaseNodal/ExtranodalPotential risk factor/marker
  • ELISA, enzyme-linked immunosorbent assay; RIBA, recombinant immunoblot assay; M, male; F, female; DK, Denmark; S, Sweden; IM, intermediate; LPL, lymphoplasmacytic lymphoma; CLL, chronic lymphocytic leukemia; FL, follicular lymphoma; DLBCL, diffuse large B-cell lymphoma; B-NHL, B-cell non-Hodgkin lymphoma not otherwise classified; HCV, hepatitis C virus; HIV, human immunodeficiency virus; NK, not known.

  • 1

    Mandatory HCV tests among blood donors in Sweden and Denmark were officially implemented in 1991.

160–64MDK+IMLPLNodal 
255–59FDK+IMCLLNK 
365–69MDK+IMCLLNK 
450–54MDK+IMFLNodal 
565–69FDK+IMDLBCLNodal + extranodal 
665–69MDK+IMDLBCLNodal 
750–54FDK++CLLNKPrevious history of venereal disease
865–69MDK+IMControl  
930–34FS+IMControl  
1045–49MS+++Control Previous history of venereal disease
Previous HIV test
1155–59MS+IMControl Previous HIV test
1255–59MS+IMControl  
1365–69MS+IMControl  
1470–74MS++Control Previous HIV test
 Blood transfusion before 19911
1555–59FS+++FLNodalPrevious HIV test
1640–44FS+IMDLBCLNodalPrevious HIV test
1765–69FS+IMDLBCLNodal 
1850–54MS+++CLLNodalPrevious history of venereal disease
1960–64FS+IMCLLNodalBlood transfusion before 19911
2065–69MS+IMFLNodal 
2160–64FS+IMB-NHLNK 
2265–69MS+IMDLBCLNodal 
2340–44MS+++FLNodalPrevious HIV test
2465–69MS+IMB-NHLNodal 
2565–69MS++CLLExtranodalPrevious HIV test
 Blood transfusion before 19911
2670–74MS+IMLPLNodal 
2765–69MS+IMB-NHLNKPrevious history of venereal disease

Discussion

In a large, population-based case-control study in Denmark and Sweden, countries with a low prevalence of HCV in the general population,21 we found some evidence of a role for HCV in the etiology of lymphoma.

Positive associations between HCV infection and NHL risk have consistently been reported in studies conducted in Italy,22–25 Japan26, 27 and Spain,28, 29 but not in France30, 31 or Canada.32, 33 The inconsistency is likely explained in part by differences in the study design,10 and low statistical power in some of these studies.34 In two recent large cohort studies of HCV-infected persons from low-prevalence areas, Sweden35 and the US,36 however, significantly increased risks of NHL were observed (standardized incidence ratio (SIR) = 1.89; 95% CI 1.10–3.03 in Sweden, and adjusted hazard ratio = 1.28; 95% CI 1.12–1.45 among US veterans). Similarly, a large, European multi-center case-control study of lymphoid malignancies found an elevated NHL risk of 1.82 (95% CI 1.13–2.91) in HCV RNA-positive persons.37 These latter results are in line with the magnitude of association observed in our study.

The majority of infectious agents linked to lymphoma etiology have been associated primarily with specific subtypes of lymphoma, e.g., Helicobacter pylori with gastric mucosa-associated lymphoid-tissue (MALT) lymphoma, HTLV-1 with adult T-cell leukemia/lymphoma,1 HHV-8 with primary effusion lymphoma,38 and EBV with Burkitt lymphoma, NK/T-cell lymphomas and HL.39, 40 In contrast, all major NHL subtypes, including indolent and aggressive NHLs and CLL, have been reported to be potentially associated with HCV in various studies.23, 28, 35, 37, 41 We found evidence of HCV infection in patients with DLBCL, CLL, FL and LPL, but not T-cell lymphoma or HL. Thus, our results are in agreement with the literature.35, 37, 42 A specific association with risk of LPL was also reported by the aforementioned US veteran study, in which the relative risk was 2.76 (95% CI, 2.01–3.79).36

Several clinical and laboratory study results support the hypothesis of an association between HCV and NHL. In a small series of patients with splenic MZL, Hermine et al.,2 reported complete tumor remission after antiviral treatment in all HCV-infected patients, whereas no tumor regression was attained in HCV-negative patients. Furthermore, the time course of HCV clearance mirrored the time to remission of NHL, and HCV relapse was associated with recurrence of NHL.2 Antiviral treatment also resulted in tumor regression in a few patients with HCV-associated MALT lymphoma of the oral cavity and of the salivary glands, immunocytoma, or a monoclonal B-cell expansion.43–46 These observations are analogous to those for H. pylori infection and gastric MALT lymphoma, in which antibiotic eradication of the causative infection leads to tumor regression.47 Autoimmune diseases such as type II MC and Sjögren's syndrome have been associated with both HCV infection and NHL risk.48–50 In type II MC and HCV-associated NHLs, a strikingly similar set of immunoglobulin region gene features has been identified in B lymphocytes. These findings suggest an ongoing process of somatic mutation, possibly due to chronic HCV infection.51 Additionally, a potential role of the t(14;18) translocation in the pathogenesis of HCV-associated lymphomagenesis is supported by our finding that two out of three HCV RNA-positive cases were FL, and forty-five percent of the total anti-HCV-positive cases were FL and DLBCL, the NHL subtypes in which the anti-apoptotic bcl-2 gene mutation is most frequent.

HCV-driven chronic antigenic stimulation triggering proliferation of B-cell clones is the presumed mechanism for HCV-associated pathogenesis of lymphomas, including CLL.52 Such a mechanism would favor lymphoma subtypes originating from germinal center or postgerminal center B cells.6 An increased mutation frequency in BCL-6 and p53 has been demonstrated in HCV-infected B-cell lines, and also in HCV-associated lymphomas.53 BCL-6 and p53-mutations, in turn, have been associated with DLBCL and a subset of FLs and CLLs.54 Moreover, as HCV is not known to be oncogenic and does not integrate into host genomes, induction of mutations in tumor suppressor genes could be a potential alternative pathogenic mechanism in lymphomagenesis following chronic but possibly also acute HCV infection.3

Strengths of our study include its large size, population-based design, rapid case ascertainment, uniform classification of NHL subtypes according to the WHO,6 the capacity to evaluate associations with several subtypes of NHL, and the use of a battery of well-established, rigorous tests for serological detection of HCV. Blood samples from a high proportion of both lymphoma cases and controls were available for the detection of anti-HCV-positive and HCV RNA-positive subjects. One potential limitation of our study is the possible underestimation of HCV prevalence among cases, if immunocompromised lymphoma patients failed to mount a detectable antibody response to HCV infection, though we found a similar prevalence of anti-HCV positivity in patients' blood samples obtained before and after initiation of lymphoma treatment.

The prevalence of HCV infection was 0.1% (1/735) in Danish and 0.5% (6/1,121) in Swedish population-based control subjects. These prevalence figures are in agreement with estimates for both countries ranging from 0.1 to 0.2% in Denmark and 0.03 to 0.5% in Sweden.13, 55, 56 We evaluated the possibility of biases related to blood sampling among cases and controls by comparing risk factors for HCV infection in cases and controls who did and did not provide blood samples. While cases did not differ, more controls with a history of blood transfusion and men reporting sex with other men consented to blood sampling, thereby likely resulting in overestimated HCV prevalence. On the other hand, male controls reporting previous HIV testing were less inclined to blood sampling than other men, which may have resulted in underestimated HCV prevalence. Overall, therefore, we cannot rule out participation bias among the controls, but suspect that the opposite effects observed among the controls would cancel out.

Diagnostic tests with suboptimal sensitivity could have led to further underestimation of de facto HCV-infected individuals. Antibody assays have greatly improved since the first generation became available,57 and ELISA-3 and RIBA-3 have high sensitivity and specificity.57 Although we cannot completely rule out an underestimation of active HCV infection due to low sensitivity of the HCV PCR test, we believe that any such bias would be non-differential between cases and controls.

In conclusion we observed some evidence for a role of HCV in the etiology of lymphoma. However, the magnitude of association was modest. Further studies are needed to focus on the biological mechanisms of HCV-related lymphomagenesis and contributing host and viral factors.

Acknowledgements

The authors thank Mrs. Charlotte Appel and Mrs. Leila Nyrén for excellent project coordination, Mrs. Kirsten Ehlers (LYFO), the personnel at the regional lymphoma registers in Sweden for help with data collection, and all doctors and nurses who participated in our rapid case ascertainment system. The funding sources were not otherwise involved in the study.

Ancillary