Cancer Causes & Control

, Volume 20, Issue 6, pp 981–987

Viral and non-viral risk factors for non-Hodgkin’s lymphoma in Egypt: heterogeneity by histological and immunological subtypes

Authors

  • Lenka Goldman
    • Lombardi Cancer CenterGeorgetown University
  • Sameera Ezzat
    • National Liver Institute
  • Nadia Mokhtar
    • National Cancer InstituteCairo University
  • Amany Abdel-Hamid
    • National Cancer InstituteCairo University
  • Nathan Fowler
    • MD Anderson
  • Iman Gouda
    • National Cancer InstituteCairo University
  • Soheir Abdel Latif Eissa
    • National Cancer InstituteCairo University
  • Mohamed Abdel-Hamid
    • Minia University
    • Lombardi Cancer CenterGeorgetown University
Original Paper

DOI: 10.1007/s10552-009-9316-0

Cite this article as:
Goldman, L., Ezzat, S., Mokhtar, N. et al. Cancer Causes Control (2009) 20: 981. doi:10.1007/s10552-009-9316-0

Abstract

Objective

Non-Hodgkin’s lymphomas (NHL) are etiologically heterogeneous malignancies. In Egypt, we previously reported an association of increased NHL risk with chronic hepatitis C virus (HCV) infection. Our present aim is to assess the association between HCV infection and histological subtypes of NHL.

Methods

We conducted a case–control study at the National Cancer Institute of Cairo University. Cases with NHL (n = 486) were matched to controls (n = 786) who were orthopedic patients from the same referral regions. Participants provided a blood sample for HCV markers (anti-HCV, HCV RNA) and answered a questionnaire on possible risk factors. Case–control differences were assessed by odds ratios and 95% confidence intervals from logistic regression analysis.

Results

Cases with diffuse large B cell lymphoma (n = 146), chronic lymphocytic leukemia (n = 58), marginal zone lymphoma (n = 24), follicular lymphoma (n = 23), and mantle cell lymphoma (n = 16) were recruited. HCV RNA prevalence was 27% in controls and 26%–48% in the NHL subgroups: it was associated (p < 0.001) with diffuse large B cell, marginal zone, and follicular lymphomas with odds ratios of 3.2, 4.4, and 3.3, respectively.

Conclusion

HCV is a risk factor for diffuse large B cell, marginal zone, and follicular lymphomas in Egypt.

Keywords

Non-Hodgkin’s lymphomaHepatitis C virusEgyptEpidemiologyRisk factors

Abbreviations

NHL

Non-Hodgkin’s lymphoma

HCV

Hepatitis C virus

CLL

Chronic lymphocytic leukemia

WHO

World Health Organization

OR

Odds ratio

CI

Confidence interval

NCI

National Cancer Institute

EIA

Enzyme-linked immunoassay

RT-PCR

Reverse transcription-polymerase chain reaction

Introduction

Hepatitis C virus (HCV) is a worldwide health problem, and the World Health Organization (WHO) estimates that 170 million people are currently infected with the hepatitis C virus. The prevalence of the disease ranges between 1% of the population in Europe to over 5% in Africa [1]. Approximately 8–10 million people in Egypt, or 12–15% of Egyptians, have serological evidence of HCV infection [2, 3] although this proportion is even higher among subgroups such as rural males [4]. This high rate of infection may be secondary to prior large scale treatment of schitosomiasis carried out in several areas of the country between 1920 and 1980. An account of these mass campaigns using parenterally administered antimony salt revealed a high potential for spread of blood borne pathogens [5]. Although the incidence of HCV is lower in the younger population, exposure continues, likely secondary to exposure to improperly screened blood products, invasive medical procedures, and the receipt of injections from “informal” health care providers.

HCV has been suggested as a cause or contributing factor in the etiology of B-cell non-Hodgkin’s lymphoma (NHL). HCV is a lymphotrophic virus, capable of replicating within the B cells and of triggering malignant transformation, although the carcinogenic mechanism remains to be fully elucidated [68]. Some but not all previous epidemiological studies have reported statistically significant associations between serological markers of HCV and increased risk of NHL as a whole, particularly in regions of moderate to high prevalence of the virus [9]. Few studies have reported risk estimates for HCV in separate subtypes of NHL.

Previously, as part of a larger and ongoing study of the malignant complications of chronic HCV infection in Egypt, where the virus is highly endemic, we reported the association of HCV with increased risk of B-cell NHL as a whole (OR = 2.3, 95% CI 1.5–3.5) [10]. The association was statistically significant and robust to potential confounders such as age, sex, and region of residence, but at that time, the sample size was not large enough to examine the HCV association in histological and immunological subtypes of NHL. Our intent in the current study is to reassess the HCV–NHL association in a larger study population, to assess its association with different subtypes of NHL, since the different subtypes are thought to have distinct etiologies, and to evaluate the associations of NHL risk with additional potential environmental and medical risk factors.

Materials and methods

Case population

Cases with NHL were recruited between October 1999 and March 2004 from patients attending the outpatient clinic at the National Cancer Institute (NCI), a major referral center in Cairo that is affiliated to Cairo University. Patients were eligible to participate if they were over the age of 17, first diagnosed with cancer ≤6 months prior to interview, no history of prior cancer, and physically and mentally capable of understanding and completing the interview. Their classification as confirmed cases was subject to final diagnoses involving pathology data and medical records review according to the WHO classification system for lymphomas. Subjects whose records were not located, or who were found to have a diagnosis other than NHL, were dropped from the study. There were no exclusions of persons with autoimmune diseases or organ transplantation. Details of recruitment were previously reported [10].

Selection of controls

Control subjects free from cancer were sampled from the Kasr El Aini Faculty of Medicine Orthopaedic Hospital in Cairo, Egypt. They were frequency-matched to the case group by rural versus urban birthplace, gender, and five-year age category. The rationale for selecting fracture patients was to obtain a representative sample of the source population of the cases by region, since both hospitals draw patients from the same area, and by HCV infection status, since HCV positivity and fracture are likely to be independent [10]. Potential controls had to be ≥18 years old, as well as physically and mentally capable of participating.

Interview procedures

The institutional review boards at each of the participating institutions approved the study protocol. Written or witnessed oral informed consent was obtained from each participant. Trained research assistants administered a standardized Arabic-language questionnaire in a face-to-face interview that lasted 30 min. The questionnaire asked about birthplace, residency and employment histories, alcohol and smoking histories, exposures to pesticides and other industrial or agricultural substances (solvents, dyes, paints, adhesives, pesticides, and herbicides), education, and medical history (diabetes, blood transfusion or donation, bilharzia, tuberculosis, and liver problems). On completion of the interview, a specimen of whole blood was collected.

Laboratory assays

Within 4 h of collection, the blood was separated and the serum was divided into aliquots and stored at −80°C. Samples were later thawed and tested for anti-HCV antibody by Abbott HCV enzyme-linked immunoassay (EIA) 3.0 (Abbott Park, IL, USA) according to the manufacturer’s instructions. Samples were initially tested for HCV RNA by direct nested reverse transcription-polymerase chain reaction (RT-PCR) as described previously [11]. Samples that tested negative by RT-PCR and positive by EIA were re-tested by conventional RT-PCR, which included an RNA purification step.

In all cases where formalin-fixed tissue from NHL cases was available at the NCI (365 cases, 75% of all cases), immunophenotyping for B- and T-cell markers was performed in the Department of Pathology, using pan-B (CD-20) and pan-T (CD-45) monoclonal antibodies with the DAKO EnVision System (Code No. K4006, DAKO, Carpinteria, CA, USA). The extraction results did yield different results in 51 subjects (changed from positive to negative). There were 17 samples that were negative for anti-HCV and positive for HCV RNA.

Statistical analyses

Characteristics of cases and controls were compared either by Pearson’s χ2 test (for categorical variables) or by a t-test (for continuous variables). Odds ratios (OR) and 95% confidence interval (CI) were calculated for each subtype separately, using unconditional logistic regression models with adjustment for matching on age, birthplace, and gender. All tests were two-tailed, and the statistical analyses were performed using SAS, version 9.1 (Cary, NC).

Results

From 1,567 contacted subjects with provisional diagnoses of NHL, 1,094 were eligible and 966 agreed to participate (88.3%). From these participants, 486 were confirmed as NHL at the time of the statistical analysis. A total of 1,022 control subjects were recruited; the participation rate in the controls was 76.9%.

Table 1 shows the frequency of each NHL subtype. The majority were of diffuse large cell type (54.9%), of which 146 (30.0%) were confirmed to be of B cell lineage by immunohistochemistry (IHC). Other common subtypes were chronic lymphocytic leukemia (11.9%), follicular lymphoma (6.3%), and mantle cell lymphoma (3.3%). We combined the T cell lymphoma subgroups—precursor T cell, peripheral T cell, and not otherwise specified T cell—into one group, T cell lymphoma (n = 23, 6.6%). We also combined mucosa associated lymphoid tissue lymphoma with nodal marginal zone lymphoma into one group (n = 24, 6.6%) according to the World Health Organization classification system. The remaining subgroups had very few cases to be included in statistical analysis. The unclassified lymphoma subgroup and the cutaneous lymphoma not otherwise specified subgroup was excluded from further analysis.
Table 1

Prevalence of NHL subtypes among the NHL patients, Egypt 1999–2004

NHL subtype

n

%

Diffuse large cell—histopathology diagnosis only

121

24.90

Diffuse large B cell—proven by IHC

146

30.04

Chronic lymphocytic leukemia

58

11.93

Unclassified lymphoma

53

10.91

Follicular lymphoma

23

4.73

MALTa lymphoma

19

3.91

Mantle cell lymphoma

16

3.29

Peripheral T cell

11

2.26

Anaplastic large cell

9

1.85

T cell, not otherwise specified (NOS)

8

1.65

Nodal marginal zone lymphoma

5

1.03

Precursor T lymphoblastic lymphoma

4

0.82

Cutaneous lymphoma NOS

6

1.23

Lymphoplasmacytoid lymphoma

3

0.62

Burkitt’s lymphoma

1

0.21

Plasma cell myeloma

2

0.41

Precursor B lymphoblastic leukemia

1

0.21

Total

486

 

aMALT Mucosa associated lymphoid tissue lymphoma

Table 2 shows the distributions of age, sex, marital status, and rural birthplace for the largest NHL subgroups. Cases with mantle cell lymphoma, chronic lymphocytic leukemia, or diffuse large B cell lymphoma had mean ages that were similar to those of controls (within five years). T cell lymphoma cases were notable for having a mean age nearly 15 years younger than controls (35.5 vs. 50.3 years, respectively). Cases and controls did not differ by percentage of males, except for the T cell lymphoma group, which had the lowest proportion of males (37.5% vs. 66.4% in controls, p = 0.012). This case group, along with mantle cell lymphoma, also differed from controls in marital status. Higher proportions of cases with rural birthplace compared to controls were characteristic of diffuse large B cell and chronic lymphocytic leukemia.
Table 2

Social and demographic characteristics of major NHL subtypes, Egypt 1999–2004

 

n

Average age

Male

Rural birthplace

n (%)

p

n (%)

p

Diffuse large B cell

146

46.7

90 (61.6)

0.249

73 (50.3)

0.2824

Chronic lymphocytic leukemia

58

54.6

34 (58.6)

0.217

31 (53.5)

0.7975

T cell lymphoma

24

35.5

9 (37.5)

0.012

13 (59.1)

0.7164

Marginal zone lymphoma

24

44.2

15 (62.5)

0.677

13 (54.2)

0.9212

Follicular lymphoma

23

44.6

11 (52.2)

0.150

12 (52.2)

0.7747

Mantle cell lymphoma

16

49.7

8 (50.0)

0.165

10 (62.5)

0.5602

Controls

786

50.3

522 (66.4)

 

308 (39.2)

 
The prevalence of viral hepatitis markers among the NHL subtypes is described in Table 3. Both anti-HCV and HCV-RNA measures are dichotomous variables. Statistically significant associations with past and current HCV infections were observed for diffuse large B cell (OR = 2.6, 95% CI 1.8–3.9, and OR = 3.2, 95% CI 2.1–4.7, respectively) and marginal zone lymphomas (OR = 3.4, 95% CI 1.4–8.5, and OR = 4.4, 95% CI 1.8–10.6, respectively). All models were adjusted for sex, age, and birthplace. In addition, current HCV infection was associated with a significantly elevated risk of follicular lymphoma (OR = 3.3, 95% CI 1.3–8.0). Among all the analyzed subtypes, there was no group for which only anti-HCV and not HCV RNA positivity was associated with NHL risk. For all cases combined, the OR was 1.9 (95% CI 1.4–2.6) and 2.4 (95% CI 1.8–3.2) for past and current HCV infections, respectively. Odds ratios for blood transfusion were as follows: diffused B = 0.6 (0.3–1.0), leukemia = 1.4 (0.7–2.7), T-cell = 0.3 (0.1–1.5), Mucosa-Associated Lymphatic Tissue Lymphomas = 1.9 (0.8–4.8), follicular = 0.2 (0.02–1.4), and Mantle = 2.0 (0.7–5.9). Since no association between NHL subtype and blood transfusion was significant, we did not include these results in the paper.
Table 3

Prevalence and odds ratios of serum anti-HCV in major NHL subtypes, Egypt 1999–2004

 

Anti-HCVa

n

%

ORb

95% CI

p-value

Controls

n = 786

283

37.4

1.0

ref.

  

Diffuse large B cell

n = 146

79

54.9

2.6

1.8

3.9

<0.0001

Chronic lymphocytic leukemia

n = 58

24

41.38

1.1

0.6

2.0

0.6829

T cell lymphoma

n = 24

8

34.8

1.7

0.6

4.7

0.3173

Marginal zone lymphoma

n = 24

14

58.3

3.4

1.4

8.5

0.0076

Follicular lymphoma

n = 23

9

40.9

1.5

0.6

3.8

0.3928

Mantle cell lymphoma

n = 16

5

31.3

0.8

0.2

2.3

0.6186

NHL cases listed

n = 296

139

47.0

1.9

1.4

2.6

<0.0001

 

HCV RNAc

n

%

ORb

95% CI

p-value

Controls

n = 786

180

23.8

1.0

ref.

  

Diffuse large B cell

n = 146

65

44.5

3.2

2.1

4.7

<0.0001

Chronic lymphocytic leukemia

n = 58

19

32.8

1.5

0.9

2.8

0.1552

T cell lymphoma

n = 24

6

26.1

1.7

0.6

5.0

0.3605

Marginal zone lymphoma

n = 24

12

50.0

4.4

1.8

10.6

0.0010

Follicular lymphoma

n = 23

10

43.5

3.3

1.3

8.0

0.0096

Mantle cell lymphoma

n = 16

3

18.8

0.7

0.2

2.6

0.6171

NHL cases listed

n = 296

115

38.9

2.4

1.8

3.2

<0.0001

aAnti-HCV antibody is considered a measure of past HCV infection

bAdjusted for age, gender, and rural birthplace

cHCV-RNA is a measure of current HCV infection

When the polytomous analysis was performed to analyze the heterogeneity by NHL subtypes, the odds ratios for HCV infection remained elevated for diffused large B cell (OR = 2.0 95% CI 1.4–2.9 for anti HCV, and OR = 2.6, 95% CI 1.8–3.7 for HCV RNA). The marginal zone lymphoma odds ratios for anti HCV stayed elevated as well at 2.3 (95% CI 1.1–5.4), but HCV RNA became insignificant (OR = 3.2 95% CI 0.2–2.6). The results for follicular lymphoma were no longer significant in the polytomous regression (OR = 0.8 95% CI 0.3–2.2 for anti HCV).

We also examined the possible associations between the major NHL subgroups and behavioral, occupational, environmental, and medical characteristics where the exposed group consisted of at least 10 subjects. Smoking, treatment of shistosomiasis, diabetes, exposures to pesticides, or other chemicals were not significantly associated with any case subtypes nor were any of these odds ratios >1.5. Shoveling grains, which was included in the questionnaire as a possible marker of aflatoxin exposure, was associated with elevated risk of follicular lymphoma (OR = 2.2, 95% CI 1.2–4.3). Even after adjustment for sex, age, and urban vs rural birthplace, positive association with growing rice was seen for diffuse large B cell lymphoma (OR = 5.1, 95% CI 2.3–11.2). Dog ownership was associated with increased risk for mantle cell lymphoma (OR = 7.5, 95% CI 1.7–32.4, p-value 0.0069), while cat owners had increased risk of chronic lymphocytic leukemia (OR = 3.5, 95% CI 1.1–11.3).

Discussion

The present study is among the largest investigations of HCV and NHL and its subtypes to date. Our results suggest that past and present HCV infections are associated with diffuse large B cell, marginal zone, and follicular B-cell lymphomas (Table 4). Birthplace in rural areas and rice cultivation (possible markers for past exposure to HCV) were associated with diffuse large B cell lymphoma. Certain environmental factors, such as shoveling grains (a possible marker for mycotoxins) and pet ownership, were associated with follicular and mantle cell lymphomas.
Table 4

Odds ratios of agriculture exposures in major NHL subtypes, Egypt 1999-2004

 

Growing rice

Shoveling grains

ORa

95% CI

p

ORa

95% CI

p

Controls

1

ref.

  

1

ref.

  

Diffused large B cell

5.1

2.3

11.2

<.0001

0.9

0.6

1.3

0.4487

Chronic lymphocytic leukemia

3.1

1.0

9.9

0.0529

1.0

1.0

1.8

0.9754

T cell

3.2

0.2

48.8

0.3974

1.6

0.7

3.7

0.2332

Marginal zone

1.7

0.3

10.0

0.6

1.3

1.3

2.8

0.5504

Follicular

2.2

0.5

9.6

0.3050

2.2

1.2

4.3

0.0166

Mantle cell

0.8

0.1

7.5

0.8535

0.9

0.3

0.3

0.8434

 

Owning cats

Owning dogs

ORa

95% CI

p

ORa

95% CI

p

Controls

1

ref.

  

1

ref.

  

Diffused large B cell

2.5

1.0

6.2

0.0471

1.3

0.5

3.4

0.5650

Chronic lymphocytic leukemia

3.5

1.1

11.3

0.0407

1.8

0.5

6.5

0.3698

T cell

4.1

0.8

20.7

0.0836

3.0

0.6

14.7

0.1750

Marginal zone

1.9

0.2

15.6

0.5577

Follicular

4.6

0.9

23.3

0.0652

Mantle cell

2.6

0.3

22.2

0.3851

7.5

1.7

32.4

0.0069

aUnadjusted

De Re et al. [12] concluded that specific B clone cells proliferate as a consequence of a chronic antigen stimulation exerted by HCV-associated antigens, and thus, HCV infected lymphomas appear to provide a plausible example of lymphoma initiation by chronic antigen stimulation. This association between HCV and NHL has been reported by several studies [5, 1315]. Gisbert et al. [13] observed that HCV prevalence in NHL patients is approximately 15% higher than in the general population. Although few studies have looked at the association between HCV and various NHL subgroups, a link between lymphomas of B cell lineage and HCV has been suggested [16, 17]. While our finding of a significant association between active HCV infection and diffuse large B cell lymphomas is consistent with the results of several European and American studies, it is contradicted by others. These differences seem to be largely geographical-based: studies conducted in Romania, Hungary, Italy, and California report the mean incidence of HCV infection in patients with NHL to be between 19 and 22%, much higher than that in the general population [1821]. However, studies conducted in Germany, France, Spain, Turkey, Switzerland, and Canada reported a low prevalence of HCV infection in patients with B cell NHL [16, 17, 2226]. Subtypes of NHL were examined separately in only a few reports, but our results are consistent with the recent pooled analysis of seven studies reported by de Sanjose et al. [27]; our study and theirs reported associations between HCV and diffuse large B-cell and marginal zone lymphomas. Also, a report from a low HCV prevalence region (Sweden and Denmark) found a positive association between HCV and B-cell NHL [28]. Seve et al. [29] observed a significantly high prevalence of HCV antibodies in MALT lymphoma patients (odds ratio 9.87; 95% CI 2.59–37.69). In our study, a comparable association was observed in Egypt with a somewhat lower OR of 4.0 (95% CI 1.3–11.9).

Among other possible medical risk factors for NHL are adult-onset diabetes [30], a history of blood transfusions [31], and certain medications, e.g., antibiotics in general [32], sulfonamides and cimetidine [33], while factors associated with a lower risk of NHL include certain vaccines, non-steroidal anti-inflammatory drugs, and allergies to plants and animals [33]. A previous study of the association of NHL with blood transfusions (a possible marker for transfusion-acquired HCV) was much stronger in follicular and small lymphocytic subtypes than with diffuse large cell NHL [31]. In contrast, we did not observe any significant associations with self-reported medical history factors.

Regarding the possible association of cigarette smoking history with the NHL subtypes, we found no statistically significant case–control differences. On the other hand, Besson et al. [34] concluded that an association is observed between smoking and NHL among women only, although in the total population a relationship was suggested between smoking and follicular NHL. Stagnaro et al. [35] similarly observed associations between increased risks for follicular lymphoma among smokers of blond (OR = 2.1, 95% CI 1.4–3.2) and mixed (OR = 1.8, 95% CI 1.1–3.0) types of tobacco.

Associations of pesticides and increased NHL prevalence have been reported in several studies from diverse regions [32, 36, 37]. However, none of these studies examined the association between the individual NHL subgroups. We found no such associations in our study population, although it is possible that the association we report between rice cultivation and large B cell lymphoma is a marker for chemical exposures.

While this study has a relatively large sample size, we recognize its limitations. Most of the NHL subtypes were rare in this population, with exception of diffuse large B-cell NHL. Although we confirmed HCV status by serological testing, all other variables were self-reported. It is possible that some subjects with low viral load were misclassified as HCV RNA negative due to the low sensitivity of RT PCR in such instances. This would tend to reduce the magnitude of the odds ratios we reported; therefore, it is possible that we underestimated the true association of HCV and NHL. We confirmed each individual diagnosis by an expert pathology review that included immunohistochemical testing in the majority cases (75%). Our questionnaire was extensive and covered medical and environmental exposures. Finally, it is possible that HCV, rather than a causal factor itself, may be a marker for immunosuppression or immune deficiency that could underlie both the chronic infection and the malignancy.

In conclusion, this study demonstrated an association between potential risk factors and certain subtypes of NHL. Diffuse large B cell, marginal zone, and follicular lymphomas were strongly associated with HCV infection. Several of the examined environmental exposures also demonstrated positive relationships among the NHL subgroups.

Acknowledgments

This study was supported by NIH Grant R01CA85888. The authors thank Karen Cowgill for her contributions to the study design and supervision of the initial years of field work. The Department of Pathology at the National Cancer Institute of Cairo University provided outstanding institutional support for the recruitment of cases and the confirmation of their diagnoses. Likewise, we are grateful to the Kasr El Aini Faculty of Medicine for supporting the recruitment of controls.

Copyright information

© Springer Science+Business Media B.V. 2009