Hepatitis C virus (HCV) is an RNA virus that belongs to the family of flaviviruses. It is hepatotrophic and causes hepatitis, liver cirrhosis and hepatocellular carcinoma.1 HCV infection is sometimes accompanied by autoimmune manifestations, most notably essential mixed cryoglobulinemia (EMC), which is characterized by cutaneous vasculitis, nephritis, peripheral neuropathy and clonal B-cell lymphoproliferations.2 HCV infection has also been involved in a subset of non-Hodgkin's lymphomas (NHLs), even in the absence of EMC. Higher than expected HCV prevalence among NHL patients has emerged, however, chiefly from studies in Italy3, 4 and Japan,5, 6 where intermediate to high prevalence of HCV (2–10%) can be found in the general population.
Elucidating the causes of NHL is of great importance, since the disease is one of the few neoplasms showing unexplained upward trends in incidence and mortality rates in the last decades.7, 8 We therefore conducted a systematic review of published studies on HCV infection and NHL.
We identified peer-reviewed articles on HCV and NHL through a MEDLINE and EMBASE search, personal knowledge of the authors and the references of the retrieved articles. Key search words included the terms “hepatitis” and “lymphoma.” No language or age limitation was imposed. We included articles published up to June 2003. Our primary aim was to identify all articles that presented the prevalence of HCV infection in case series of B-cell NHL patients and, if possible, in some appropriate control group. To avoid multiple inclusions of the same data, whenever the same author appeared in more than one article, we checked the place and time of recruitment of cases and, in case of overlapping, included the most recent publication only. In all groups considered (i.e., NHL, other lymphohematologic malignancies and controls) we excluded series with fewer than 20 individuals. In total, we identified 66 articles.3, 4, 5, 6, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70
Several problems of comparability emerged among the retrieved articles. With respect to the assessment of HCV infection, some studies relied on the presence of anti-HCV antibodies by enzyme-linked immunosorbent assay (ELISA) only, while others also used recombinant immunoblot assay (RIBA) for confirmation or tested for HCV RNA. First-generation ELISA had unsatisfactory sensitivity and specificity, but second- or third-generation ELISAs were used in all the studies we identified. Despite great improvements in the diagnostic test for HCV, some problems remain (most notably of specificity in respect to ELISA and sensitivity in respect to RIBA).71 Furthermore, it is not known if detectable HCV-RNA is a necessary condition for lymphomagenesis. Therefore, we classified an individual as HCV-positive if at least one of the above assays was positive.
Patients with HIV infection were explicitly excluded from most studies. Inclusion criteria for the lymphoma group varied as well; some authors included chronic lymphocytic leukemia (CLL) among NHLs, but others did not. Whenever possible, we used the definition of NHL given by the authors. When only subgroups of NHL were presented, we grouped them all as NHL. Whenever possible, we selected only B-cell NHL patients but, in some studies, the cell lineage of origin was not available. However, non-B-cell NHLs only constitute 5–10% of all NHL series.72
In respect to NHL classification, most studies used either the REAL73 or the Working Formulation74 classifications. Subcategories of NHL, however, were presented in different detail, thus hampering any consistent comparison of HCV prevalence across different histologic types. A consistent distinction between nodal and extranodal NHL, or the identification of specific extranodal sites, was also impossible in most studies.
To investigate the specificity of the association of HCV with B-cell NHL (B-NHL), we also presented the prevalence of HCV in patients with other hemolymphopoiaetic malignancies, i.e., CLL for studies that did not include them among B-NHL, Hodgkin's disease (HD), multiple myeloma (MM) and non-B-cell NHL.
HCV infection prevalence in groups other than lymphohematopoietic neoplasms was presented in several articles, generally for comparative purposes. These groups varied greatly in their definition and included patients with solid cancers, individuals undergoing various medical or surgical treatments, such as a biopsy or a colonoscopy, blood donors, or healthy subjects.
Only a few studies were designed as formal case-control studies, adequately described inclusion criteria for the control group and presented odds ratios (ORs) adjusted for age, sex and other potential confounding factors. We attempted, however, to estimate ORs for NHL by HCV infection in a few additional studies where some acceptable control group could be identified. We never used, however, patients with other lymphoproliferative diseases to compute ORs. Moreover, since the prevalence of HCV infection varies widely according to age and sex, we considered a case-control comparison adequate only in studies that presented sex- and age-adjusted ORs, or reported that cases and controls were matched by age and sex, or provided some measure of age (mean or median or distribution) and of the sex ratio in cases and controls, which allowed us to evaluate the comparability of the 2 groups. In the last 2 instances, only crude ORs could be computed.
One study compared the prevalence of NHL in HCV-positive and HCV-negative male veterans from the United States, adjusting for several confounders.69 An additional study provided the standardized incidence ratio of NHL in a cohort of HCV-infected patients compared to the general population of Osaka, Japan.70 On account of the interstudy variability in design and findings, we did not compute a pooled OR.
Table I shows the number and percentage of subjects with NHL, in the majority specified as B-cell NHL, that were positive for HCV infection in all case series with more than 20 cases by area and country. Studies that included NHL case series but did not give the prevalence of HCV infection in NHL cases separately from patients with other mixed hematologic neoplasms are shown at the end of the table. Studies were conducted in several countries of Europe, Asia and America, and the number of NHL cases included ranged from 24 to 470. Italy was the country that contributed the largest number of studies. Except for a study restricted to cutaneous B-cell lymphomas, all Italian studies showed a high prevalence of HCV infection among NHL, ranging from 8.9% to 37.1%. In total, 529 out of 2,668 (19.8%) NHL cases were HCV-positive in Italy. In Europe, HCV prevalence was also high in a study from Switzerland (9.4%), 2 studies from Spain (11.7% and 20.3%) and 2 studies from Eastern Europe, 1 from Hungary (23.8%) and 1 from Romania (29.4%). Prevalence of 1–2% was observed in most studies from central European countries, except for a French study of primary liver B-cell lymphoma, where 20.8% of cases were HCV-positive. No HCV-positive patients were observed in over 300 NHL patients from northern Europe, mainly the United Kingdom.
Table I. Prevalence of HCV Infection in Case Series of (B-Cell) Non-HODGKIN's Lymphoma Patients
Number of HCV+/total
N Italy, northern Italy; C Italy, central Italy; S Italy, southern Italy.
Includes 16 B-NHLs with involvement of liver or major salivary glands.
Reports 2 patients with HCV cirrhosis who developed NHL.
All HCV+ patients had B-NHL. Proportion of B-NHL not given.
Forty three B-NHL and 1 T-NHL.
Cases were selected from a prospective study of 48,420 individuals from California, where sera had been collected on average 21 years before diagnosis.
The four HCV-positive patients were repeatedly reactive by enzime immunoassay, but negative to recombinant immunoblot assay.
In Asia, high HCV prevalence was reported in 1 study from Saudi Arabia (21.4%), 1 from Israel (7.8%) and in the 7 studies from Japan (5.7–22.2%). The overall prevalence of HCV infection among NHL patients in Japan was 11.3%. HCV prevalence among NHL patients was lower in a study from South Korea (3.3%), 1 from Thailand (2.3%), and a small Lebanese study (0%), while it ranged between 1.1% and 13.3% in 5 studies from Turkey. Findings from North American studies differed markedly. Two studies from Canada and one from Missouri, United States, showed low prevalence of HCV infection in NHL patients (0.0–2.3%), while 3 studies conducted in New Orleans or Los Angeles found a much higher prevalence (11.5–25.0%). Finally, in a study from Brazil, 9.1% of B-NHL patients were HCV-infected (Table I).
Table II shows the specificity of the association with NHL in studies from Table I that presented findings also on at least 20 cases of CLL, HD, MM or non-B-cell NHL. For each case series, the total number of patients, the number of HCV-positive ones, the corresponding percentage and confidence interval are shown. The prevalence of HCV infection was high (over 10%) in most series of patients with CLL and MM, and in some studies it was higher than that in NHL patients. Conversely, lower prevalence of HCV infection was reported in most series of patients with HD or non-B-NHL compared to B-NHL patients, particularly in studies where the prevalence in B-NHL patients was high. The combined odds ratios of HCV positivity for patients with B-NHL compared to patients with CLL, HD, non-B-NHL and MM were 1.5 (p = 0.10), 3.2 (p = 0.01), 2.4 (p = 0.04) and 1.5 (p = 0.05), respectively.
Table II. Prevalence of HCV Infection in Various Hematological Malignancies from Studies that Included also Non-HODGKIN's Lymphomas
Chronic lymphocytic leukemia
Number of HCV-positive/total number of cases.
Percentage of HCV-positive cases with 95% confidence interval.
Table III shows the ORs for NHL from studies that included a control group that satisfied the requirements specified above. In a study from North America, the OR was not estimable, since all cases and controls were HCV-negative. In a French study, the crude OR was 0.9, whereas in other studies the OR ranged from 2.1 to 9.2. The highest OR (51.5) was reported in a subgroup of patients with extranodal NHL of the liver and salivary glands in an Italian study. Three Italian studies and a Japanese one presented OR adjusted for age, sex and a few other potential confounding factors, such as education and birthplace. These estimates ranged between 2.6 and 4.3, and the corresponding 95% confidence intervals (CIs) excluded unity. Two additional studies with different designs presented a measure of the association between HCV infection and NHL. One study compared the prevalence of NHL in 34,204 HCV-infected male U.S. veterans and 136,816 HCV-negative veterans hospitalized between 1992 and 1999. HCV infection was determined during hospital stay, and the adjusted OR was 1.2 (95% CI = 1.0–1.4). In a cohort study of 2,162 HCV-positive patients from Osaka, Japan, a standardized incidence ratio of 2.1 (95% CI = 0.6–5.4) for NHL was computed on the basis of the expected number of NHL in the general population standardized by age, sex and calendar year.
Table III. Relative Risk (RR) of (B-Cell) Non-HODGKIN's Lymphoma According to HCV Infection Status
Figure 1 shows graphically the ORs presented in Table III. Studies are grouped by design. Each line represents a study, the black square represents the OR, whereas the line through the square represents the confidence interval. The dotted vertical line represents an OR of 1, i.e., absence of association. ORs on the left of the line imply a negative association, and ORs on its right a positive one.
Our present review shows that the prevalence of HCV infection has been evaluated among over 6,000 NHL cases in Europe, America and Asia. High HCV prevalence was found especially in southern and eastern Europe, Japan and the southern United States. Before drawing any inference on the role of HCV in NHL etiology, sources of heterogeneity and possible bias in published series must be considered.
Few studies were able to estimate the difference in HCV prevalence between NHL patients and the general population of the same area. HCV prevalence varies widely between and within countries, ranging from 1–2% in most Western countries75, 76, 77 to more than 10% in Egypt,78 Italy,79, 80 South Korea81 and some areas in Japan.82 Furthermore, despite the limitations of current data on the incidence and prevalence of HCV, data on the incidence of hepatocellular carcinoma in different birth cohorts suggest that the spread of HCV may have occurred earlier in southern Europe83 and Japan84 than in the United States.85 In HCV-endemic areas, high prevalence of infection among NHL patients may be found even if no real association exists, but in countries where the prevalence of HCV is very low, or has only increased very recently, a moderate association between HCV and NHL can be undetectable in relatively small studies.
Even if HCV prevalences in NHL patients close to 20% or greater are unlikely to be due to chance, different potential biases and confounding must be considered. The longstanding interest of some working groups, notably immunologists and rheumatologists, in EMC and other manifestations of HCV infection may have led to a selective identification or referral of hemolymphopoietic neoplasms among HCV-infected individuals, who are often closely monitored over time. Furthermore, blood transfusion, one of the predominant routes of HCV transmission, has been associated with an approximately 2-fold increased risk of NHL in some studies, although later studies did not confirm the association.86 Inverted causation (i.e., acquisition of HCV as a consequence of NHL-related treatments or diagnostic procedures) is unlikely after 1990, when routine HCV screening of blood and blood products and use of disposable needles have become routine, at least in developed countries. In one case-control study,25 adjustment for history of transfusion and use of intravenous drugs (a strong risk factor for HCV infection85) did not substantially modify the OR. Some publication bias cannot be ruled out, especially in the smallest NHL series.
Most important, NHL is a heterogeneous group of diseases and the best-known infectious causes of NHLs are associated with specific subtypes of lymphoma (e.g., the retrovirus HTLV1 with acute T-cell leukemia/lymphoma, the Epstein-Barr virus (EBV) with endemic Burkitt's lymphoma and immunoblastic lymphoma in immunosuppressed patients, Helicobacter pylori with MALT-lymphoma of the stomach).87 Disaggregation of NHL subtypes is difficult in the studies included in the present review, because different classifications were used, and NHL subcategories were inconsistently or only partially reported. Some B-NHL subtypes, notably lymphoplasmacytoid lymphoma/immunocytoma,4, 25, 30 showed especially elevated HCV prevalence. In the majority of the largest studies, however, the association with HCV infection was observed in all major subtypes, including indolent and aggressive B-NHLs.15, 25, 61
The current pathogenetic hypotheses on HCV lymphomagenesis may support the implication of a broad spectrum of B-NHL. A direct oncogenic role of HCV through B-cell infection and deregulation has been proposed since the virus is lymphotropic, but this has never been proved.2 Only a subset of the neoplastic cells in HCV-associated NHLs harbor the viral genome or proteins, whereas these are frequently found in the stromal cells that surround the neoplastic ones. A more plausible pathogenic mechanism is therefore that specific B-cell clones proliferate as a consequence of the chronic antigenic stimulation sustained by HCV. Indeed, the immunoglobin variable region genes expressed by B-NHL cells from HCV-positive patients show somatic mutations indicative of an antigen selection process.2 Furthermore, the amino acid sequences of B-cell receptors in HCV-associated lymphoproliferations show a similarity with anti-HCV antibodies.2 If this is the case, an association with HCV may be shared by a broad range of lymphomas and MM, whose B cell of origin has been chronically stimulated by HCV-related antigens in germinal centerlike structures. In fact, HCV infection prevalence was high also in several case series of patients with CLL and MM, although lower than in B-NHL. Therein vigorous clonal expansion of B cells takes place and the risk of mutations and malignant transformation is high.72 Different oncogenetic mutations may occur by chance and may determine the degree of differentiation of the deriving neoplasm. Not expected to belong to this spectrum would be, however, Mantle-cell NHL, a subset of CLL and of Burkitt's lymphoma, T-cell NHL and classic Hodgkin's disease, all of which do not originate from germinal center or postgerminal center B cells.88 Indeed, some of the studies we have summarized showed lack of association between HCV and HD and T-cell NHL.
Although coinfection with HCV was frequent in HIV-infected patients in studies from the United States89, 90 and Italy,91 NHL risk was not higher than among HCV-negative HIV-positive patients. In the presence of severe immunologic impairment (CD4 < 200/mm3), a T-dependent B-cell stimulation, as predicted for HCV infection according to the chronic stimulation hypothesis, should not occur.
Differences between specific HCV genotypes (group 2) in respect to lymphomagenesis have been suggested.35, 92 In the majority of studies that considered the issue, however, the distribution of HCV genotypes among NHL patients did not clearly differ from that in controls25 or in the corresponding general population.61
Some independent support to a role of HCV infection in a fraction of B-cell NHL comes from the observation that regression of splenic lymphoma with villous lymphocytes could be achieved using interferon α and ribavirin in HCV-infected patients but not in HCV-negative ones.93 Moreover, the treatment of HCV infection with interferon α has resulted in remission of the lymphoma in one patient with immunocytoma94 and of monoclonal B-cell expansion in patients with mixed cryoglobulinemia.95 Similarly, regression of gastric mucosa-associated lymphoid tissue (MALT) NHL has been induced by treatment of Helicobacter pylori infection.96 These observations suggest the dependence of the disease process on stimulation by viral antigens.87 Thus, knowledge on viral associations may be exploited not only for prevention but also for treatment of NHL, although the extent of NHL subtypes in HCV-positive patients for which treatment of HCV infection may induce regression of the lymphoma is still uncertain.97
In conclusion, the relative risk of B-cell NHL among HCV-positive individuals versus HCV-negative individuals appears to range from 2 to 4. Epidemiologic evidence alone cannot establish whether this association is causal. Assuming, however, a relative risk of 2, the fraction of B-cell NHL attributable to HCV infection would be 10% in a country like Italy, where approximately 20% of NHL cases were found to be HCV-positive. In a country where the prevalence of HCV infection in the general population is 1%, the proportion of B-cell NHL attributable to HCV infection would be only 1% and thus could easily have been missed in studies including a few hundred NHL cases. Thus, the international variation in the association between NHL and HCV may simply be the result of a modest relative risk and a wide variation in the prevalence of the infection.
The authors thank Mrs. I. Garimoldi for editorial assistance.