Chronic viral hepatitis, HIV infection and Non-Hodgkin lymphomas in West Africa, a case-control study

Abstract

Non-Hodgkin lymphomas (NHL) are underestimated causes of cancer in West Africa where chronic viral hepatitis and HIV are endemic. While the association with HIV infection has already been characterized, limited information is available on the association between chronic viral hepatitis and NHL in sub-Saharan Africa. A case-control study was conducted in referral hospitals of Abidjan (Cote d’Ivoire) and Dakar (Senegal). Cases of NHL were matched with controls on age, gender and participating site. The diagnosis of NHL relied on local pathological examination completed with immunohistochemistry. HIV, HBV and HCV serology tests were systematically performed. A conditional logistic regression model estimated the associations by the Odds Ratio (OR) with their 95% confidence interval (CI). A total of 117 NHL cases (Abidjan n = 97, Dakar n = 20) and their 234 matched controls were enrolled. Cases were predominantly men (68.4%) and had a median age of 50 years (IQR 37-57). While Diffuse Large B-cell lymphoma were the most reported morphological type (n = 35) among mature B-cell NHL, the proportion mature T-cell NHL (30%) was high. The prevalence figures of HBV, HCV and HIV infection were 12.8%, 7.7% and 14.5%, respectively among cases of NHL. In multivariate analysis, HBV, HCV and HIV were independently associated with NHL with OR of 2.23 (CI 1.05-4.75), 4.82 (CI 1.52-15.29) and 3.32 (CI 1.54-7.16), respectively. Chronic viral hepatitis B and C were significantly associated with NHL in West Africa. Timely preventive measures against HBV infection and access to curative anti-HCV treatment might prevent a significant number of NHL.

1 |. BACKGROUND

Non-Hodgkin lymphomas (NHL) are important and underestimated causes of cancer in resource-constraints settings such as sub-Saharan Africa (SSA) where the lack of appropriate diagnostic capacity prevents any accurate estimates of the true burden of these malignant conditions. Aside the lack of access to immunohistochemistry (IHC) and molecular tests needed to correctly ascertain its diagnosis, NHL can be easily confounded with active tuberculosis, highly prevalent in SSA leading to missed or delayed NHL diagnosis.¹ Despite these diagnostic challenges, NHL is on the rise and ranked among the six most important cancer in SSA in terms of incidence and mortality.^2-4 Etiologic factors leading to NHL are largely unknown but several infectious agents that are highly prevalent in SSA such as Epstein-Barr Virus (EBV) or Human T-Cell Lymphotropic Virus-1 (HTLV-1) have been characterized in the causal pathway to this malignant condition.⁵ The association between HIV and NHL is also well recognized as three NHL types are AIDS-classifying diseases (Diffuse Large-B cell NHL, Primary Central Nervous system NHL and Burkitt Lymphomas).⁶ Previous reports have showed a clear association between HIV and NHL in SSA prior to the scale-up of antiretroviral treatment (ART) but limited information is available now that HIV-infected persons are increasingly accessing to ART.^7,8 As the occurrence of this malignancy is significantly related to the level of immune suppression,^9,10 an updated knowledge on NHL characteristics occurring in HIV-infected patients accessing ART in SSA is needed.

Chronic infection with hepatitis C virus (HCV) has been classified as “carcinogenic agents with sufficient evidence in humans” but hepatitis B virus (HBV) infection remains considered by the International Agency for Research on Cancer (IARC) in its classification as an “agent with limited evidence in humans with regards to the risk of NHL.”^5,11 While growing evidences support a positive association between HBV infection and NHL in Europe and South East Asia, no information is available from SSA, a region highly endemic for HBV.¹² Given the public health importance of chronic viral hepatitis and HIV in West Africa, it is of major importance to improve our understanding on the potential impact of these infectious agents on the risk of NHL.

2 |. METHODS

A case-control study has been initiated in referral hospitals of Abidjan (Cote d’Ivoire) and Dakar (Senegal). In these two referral hospitals, hematology wards were in charge to prospectively pre-include every patient with a suspected diagnosis of lymphoma. Case definition of NHL relied on a first histological examination performed by local referent pathologists in the university hospitals of Dakar and Abidjan. Paraffin-embedded biopsies from patients suspected from NHL were sent for a confirmatory diagnostic based on IHC centrally performed abroad in France (CERBA, Europe). Confirmed NHL were subsequently classified into subtypes according to the 2008 WHO classification and based on the InterLymph hierarchical classification of lymphoid neoplasms for epidemiologic research.¹³ A control group (two controls for one case of confirmed NHL) of patients seeking care in the same referral hospitals were enrolled in wards not dedicate to infectious diseases including orthopedic surgery, cardiovascular disease and endocrinology wards. All included controls needed to have a clear primary diagnosis not related to any malignant condition or infectious disease that motivated their access to care. The controls were matched with cases of NHL based on their age (±2 years), gender and participating referral hospital.

2.1 |. Biological measurements

All cases and controls were systematically tested for HBV, HCV and HIV infection. Serological tests comprised screening for anti-HCV antibody, anti-HIV antibody, HBs antigen and anti-HBc antibody. Participants that were positively screened for HBs antigen, anti-HCV antibody and anti-HIV antibody underwent viral load measurements for their respective HBV, HCV or HIV infections. All serology and viral load testing were locally performed in the biology unit of Abidjan and Dakar participating referral hospitals. Serological tests were performed by Enzyme-Linked Immuno Assay test while Polymerase Chain Reactions allowed viral load measurements in Abidjan and Dakar referral laboratories under “Roche Cobas® AmpliPrep/Taqman 48” and “Hitachi Cobas® e411,” respectively. To prevent any artefactual observed association between carriage of HBs antigen and NHL, all serological measures were performed on blood collected prior to any chemotherapy initiation as HBV infection can be reactivated in 14%-50% of patients undergoing chemotherapy for NHL.¹⁴

2.2 |. Statistical analysis

Comparisons between cases and matched controls or according to serologic status were performed. Categorical variables, represented as absolute values (percentages) were compared by the Chi-square test or exact Fisher test when appropriate. Continuous variables, represented as median, ±interquartile range (IQR) were compared based on Wilcoxon test. A conditional logistic regression model was used to measure the association between NHL and available factors of interest. Analyses were computed with the PHREG procedure by using the discrete logistic model and forming a stratum for each matched set of one cases and two controls in the SAS 9.4 software (SAS Institute Inc., North Carolina). A stepwise descending procedure was then applied to select available factors that were significantly associated with NHL. A P-value of <.05 was considered for statistical significance. Odds ratio (OR) estimates were reported with their 95% confidence interval (CI).

3 |. RESULTS

From November 2017 to April 2020, 191 patients presenting with a suspected lymphoproliferative disorder were preincluded and approached to participate in the present study. They underwent biopsy for a first local histological examination by local pathologist in Dakar and Abidjan. Of these, 30 were discarded based on the direct examination and one participant was withdrawn because of a subsequent refusal to participate. Of the 160 participants eligible for a confirmatory diagnosis with IHC, 10 died or were loss to follow-up prior to their completion of study participation. The 150 remaining biopsies were systematically transferred for a second reading combined with IHC and based on this second analysis, 33 (22.0%) were subsequently excluded.

A total of 117 NHL cases (Abidjan n = 97, Dakar n = 20) and their 234 matched controls were finally enrolled. Cases were predominantly men (68.4%) and had a median age of 50 years (IQR 37-57). Mature B-cell lymphomas represented 70.0% of NHL cases with diffuse large B-cell lymphoma being the most reported morphological type (n = 35) (Table 1).

TABLE 1.

Demographical, clinical and histological characteristics of participants with non-Hodgkin lymphoma according to a positive HIV, HBV or HCV serology in Côte d’Ivoire and Senegal (n = 117)

	NHL total (n = 117)	Positive HIV serology (n = 17)	P a	Positive HBV serology (n = 15)	P a	Positive HCV serology (n = 9)	P a
	N(%)	N(%)		N(%)		N(%)
Sex			2 × 10−4		.74		.27
Women	37 (31.6)	12 (70.6)		4 (26.7)		1 (11.1)
Men	80 (68.4)	5 (29.4)		11 (73.3)		8 (88.9)
Age, median (IQR)	50 [37–57]	44 [33–2]	.22	43 [39-53]	.14	55 [52-57]	.10
Participating site			.54		.99		.35
Abidjan (Côte d’Ivoire)	99 (83.2)	15 (88.2)		13 (86.7)		9 (100.0)
Dakar (Senegal)	20 (16.8)	2 (11.8)		2 (13.3)		0 (0.0)
Tobacco useb			.71		.21		.45
No	94 (79.0)	14 (82.4)		10 (77.7)		8 (88.9)
Yes	25 (21.0)	3 (17.6)		5 (33.3)		1 (11.1)
Alcohol use (AUDIT-Cc)			.74		.99		.80
No, moderate	102 (85.7)	15 (88.2)		13 (86.7)		7 (57.1)
Hazardous drinking	17 (14.3)	2 (11.8)		2 (13.3)		2 (42.9)
NHL subtypesd	N	n (%)	–	n (%)	–	n (%)	–
B-cell	82	14 (17.1)		10 (12.2)		6 (7.3)
Diffuse large B-cell	35	5 (14.3)		5 (14.3)		3 (8.6)
Burkitt	4	1 (33.3)		0 (0.0)		0 (0.0)
Nodal marginal zone	5	1 (20.0)		0 (0.0)		0 (0.0)
Mantle cell	13	1 (7.7)		1 (7.7)		0 (0.0)
Follicular	4	1 (25.0)		1 (25.0)		0 (0.0)
Small lymphocytic	9	1 (11.1)		2 (22.2)		1 (11.1)
Lymphoplasmacytic	6	0 (0.0)		0 (0.0)		2 (33.3)
T-cell	35	3 (8.6)		5 (14.3)		3 (8.6)
T-cell N/K	2	0 (0.0)		1 (50.0)		0 (0.0)
Cutaneous T-cell lymphoma	1	0 (0.0)		0 (0.0)		0 (0.0)
Angioimmunoblastic T-cell	3	0 (0.0)		1 (33.3)		0 (0.0)
Precursor lymphoblastic	3	0 (0.0)		0 (0.0)		0 (0.0)
Peripheral T-cell	3	1 (33.3)		0 (0.0)		1 (33.3)
Anaplastic large-cell	2	0 (0.0)		0 (0.0)		0 (0.0)

Abbreviations: NHL, non-Hodgkin lymphoma; N/K, natural killer.

^aP-values from Fisher’s exact test comparing NHL with a positive HIV, HBV and HCV serology vs NHL without positive serology, respectively.

^bAny reported present or past tobacco use.

^cHazardous drinking defined based on an AUDIT-C score ≥4 in men and ≥3 in women.

^dNHL subtypes presented in a hierarchical model according to the InterLymph hierarchy of 2008 WHO classification.

The prevalence of HIV infection was 14.5% in NHL cases versus 5.6% in control participants (P = .005) (Table 2). Cases of NHL infected with HIV were predominantly women (78.6%) and had a median age of 44 [33-52] years. Of the 17 HIV-infected cases of NHL, 12 (70.6%) were already known as HIV-infected prior to their NHL diagnosis and 11 (64.7%) in care and treated for their HIV infection. An undetectable HIV viral load was reported in 17 (56.7%) of the 30 HIV-infected participants. The proportion of undetectable HIV viral load was 47.1% in HIV-infected cases of NHL versus 69.2% in HIV-infected control participants (P = .22). In multivariate analysis, HIV was independently associated with NHL with an adjusted OR of 3.32 (1.54-7.16) (Table 3).

TABLE 2.

General characteristics of cases of NHLs and their corresponding controls, 2017-2020

	Controlsa (n = 234)		Cases (n = 117)
	n	(%)	n	(%)	P
Referral hospitals					–
Abidjan, Cote d’Ivoire	194	(82.9)	97	(82.9)
Dakar, Senegal	40	(17.1)	20	(17.1)
Age class (years)					–
<40	68	(29.1)	35	(29.9)
40-49	54	(23.1)	23	(19.6)
50-59	67	(28.6)	34	(29.1)
≥60	45	(19.2)	25	(21.4)
Gender					–
Women	74	(32.6)	37	(32.6)
Men	160	(68.4)	80	(68.4)
Education					.85
No school	32	(13.7)	18	(15.4)
Primary school	52	(22.2)	27	(23.1)
Secondary and over	150	(64.1)	72	(61.5)
Tobacco use					.30
No smoking	171	(73.1)	93	(79.5)
Present/past smoker	63	(26.9)	24	(20.5)
Alcohol useb					.14
No, moderate drinking	189	(79.9)	101	(86.3)
Hazardous drinking	49	(20.1)	16	(13.7)
HBs antigen					.10
Negative	218	(93.2)	102	(87.2)
Positive	16	(6.8)	15	(12.8)
Anti-HBc antibody					.63
Negative	74	(31.6)	40	(34.2)
Positive	160	(68.4)	77	(65.8)
Anti-HCV antibody					.01
Negative	229	(97.9)	108	(92.3)
Positive	05	(2.1)	9	(7.7)
Anti-HIV antibody					.005
Negative	221	(94.4)	100	(95.5)
Positive	13	(5.6)	17	(14.5)

Abbreviations: HBs, hepatitis B surface; HBc, hepatitis B core; HCV, hepatitis C virus; HIV, human immunodeficiency virus.

^aControls matched to cases on gender, age (±2 years) and referral hospital.

^bHazardous drinking defining based on an AUDIT-C score ≥ 4 in Men and ≥ 3 in women.

TABLE 3.

Factors associated with NHL in Abidjan (Côte d’Ivoire and Dakar [Senegal]), 2017-2020

	Unadjusted analysis			Adjusted analysis
	n/Na,b	OR (95% CI)	P	aOR (95% CI)	P
Alcohol usec			.11
No/moderate use	101/288	1		–
Hazardous drinking	16/63	0.58 (0.29-1.14)
Anti-HIV antibody			.009		.002
Negative	100/321	1		1
Positive	17/30	2.61 (1.27-5.38)		3.32 (1.54-7.16)
HBs antigen			.06		.03
Negative	102/320	1		1
Positive	15/31	2.03 (0.96-4.32)		2.23 (1.05-4.75)
Anti-HBc antibody			.56
Negative	40/114	1		–
Positive	77/237	0.84 (0.48-1.49)
Anti-HCV antibody			.03		.008
Negative	108/337	1		1
Positive	9/14	3.60 (1.21-10.74)		4.82 (1.52-15.29)

Abbreviations: HBs, hepatitis B surface; HBc, hepatitis B core; HCV, hepatitis C virus; HIV, human immunodeficiency virus.

^aMultivariate analysis matched on age, sex and referral center.

^b(n) Number of NHL cases/(N) Total number of patients per specified categories.

^cDeclared alcohol use during the past year assessed using the short version of the alcohol use disorder identification test (AUDIT-C); score ≥4 in men and ≥3 in women was used to define hazardous drinking.

The prevalence of a positive HBs antigen was 12.8% in NHL cases versus 6.8% in control participants (P = .10). A total of 31 (8.8%) participants reported being vaccinated against HBV prior to study participation. Of the 31 participants with a positive HBs antigen, only three (9.7%) knew their HBV status prior to study participation. The proportion of HBs antigen positive participants with a detectable HBV viral replication was 80% in cases of NHL and 68.7% in controls (P = .68). In multivariate analysis, HBV was independently associated with NHL with an adjusted OR of 2.23 (1.05-4.75).

The prevalence of a positive HCV serology was 7.7% in NHL cases versus 2.1% in control participants (P = .01). Of the 14 participants positive for anti-HCV antibody, 4 (28.6%) had a detectable HCV viral load, 2 of them being NHL cases. Only 12 participants (3.4%) knew their HCV status prior to study participation. In multivariate analysis, HCV was independently associated with NHL with an adjusted OR of 4.82 (1.52-15.29). No cases of coinfections with HIV, HCV or HBV were reported among cases of NHL. Among controls, only one participant harbored a positive HIV and HCV serology.

4 |. DISCUSSION

Our present findings suggest a significant association between chronic HBV infection and NHL in West Africa. The strength of this association was quite similar to previous case-control studies conducted so far outside SSA, in other regions with “intermediate and high HBV prevalence.”^12,15 Mechanisms leading to a higher risk of NHL are still investigated. Contrarily, to lymphocyte-transforming viruses such as EBV or HTLV-1, HBV was initially suspected to increase the risk of developing NHL through chronic immune stimulation and site-specific inflammation.¹⁶ Recent findings have identified that HBV DNA integration was a common phenomenon in NHL cells, suggesting a direct carcinogenic role of HBV in the occurrence of this malignant condition.¹⁷ The accumulated epidemiologic evidences of the association between HBV and NHL combined with the growing mechanistic explanations in the role of HBV towards the occurrence of NHL might enable to reconsider the current IARC classification of HBV towards NHL from Group 2A to Group 1 (carcinogenic to humans). The possibility that naturally acquired HBV infection might be associated with NHL was recently reported from a large case-control study conducted in china.¹⁸ In our studied population, the proportion of positive anti-HBc antibody/negative HBs antigen serology was high and equally distributed in both cases and controls, not supporting any relationship between naturally HBV immunization and NHL.

Significant associations were also reported between NHL and a positive HCV serology has previously reported elsewhere.⁵ In West Africa, while the overall prevalence of HCV infection estimates ranged from 1% to 3% in the adult general population, some important intraregional variations have been reported with settings harboring HCV prevalence over 10%.^19,20 Our assessment of chronic HCV infection relied on the combination of HCV serology with PCR detection of HCV replication. However, due to our limited sample size, we were not able to formally assess the relationship between actively replicating HCV infection and NHL. As previously reported, relying solely on HCV serology significantly overestimate the proportion of active HCV infection stressing the need to systematically confirm positive HCV serology through PCR or alternative biomarkers such as the core HCV antigen.²¹

A similar association between HIV and NHL was reported compared to our previous case referent study on cancer and HIV conducted a decade ago in West Africa with a prevalence of 14.5% of HIV-infected among participants diagnosed with NHL (adjusted OR of 3.6 [1.9-6.8]).⁸ In the ART era, NHL is now the leading cause of cancer among HIV-infected people living in high-income countries.¹⁰ While depleted immunologic status and uncontrolled HIV viral replication have been identified as major predictors of NHL, it remains highly associated with HIV infection despite early ART initiation and controlled HIV replication.^9,22 Kaposi sarcoma is currently the leading cause of cancer in HIV-infected persons in SSA but growing evidence suggest a steep decrease in its incidence among PLHIV initiating ART.²³ We can therefore anticipate the growing importance of the two other AIDS-classifying cancers, namely cervical cancer and NHL among PLHIV as this population ages and as ART continues to expand. While cervical cancer is amenable to prevention strategies through immunization and screening, NHL is currently not considered through preventive strategies. It is therefore important to expand initiatives that enhance early NHL diagnosis^24,25 and access to adapted curative and palliative treatments in resource-constraints settings.^26,27

A particularly high proportion of T-cells NHL were reported in this present case series, higher than the proportion of T-cells NHL reported from Western Europe/Northern America (9%) and Southern Africa (15%).²⁸ A previous international review of pathological findings of NHL cases showed a higher proportion of T-cell NHL in developing countries compared to developed countries, although no data from West Africa were available.²⁹ In the particular context of West Africa, the high burden of HTLV-1 infection could partly explain the high proportion of mature T-cell lymphomas observed in our case series. Infection with HTLV-1 has been identified as the causative agent of adult T-cell leukemia and lymphoma since 1980 and classified by IARC as carcinogenic to humans.^5,30 A recent meta-analysis on the burden of HTLV-1 infection in SSA reported an estimated HTLV-1 prevalence of 3.19% (95% CI 2.36-4.12%) in the adult general population.³¹ Infection with HTLV-1 is usually acquired vertically from mother to child. The lifetime risk of developing an adult T-cell leukemia or lymphoma among adults infected with HTLV-1 during the perinatal period has been estimated as high as 25%.³² These elements support the need to document the true burden of HTLV-1 in adult T-cell NHL occurring in this highly endemic region.

4.1 |. Limitations

The cross-sectional nature of study design precludes any inferential link between the studied infectious agents and the occurrence of NHL. However, it is widely accepted that in SSA, the great majority of HBV infection are acquired through mother-to-child transmission or during early childhood (<5-years old), leaving limited doubts on the anteriority of chronic HBV infection compared to NHL. Interestingly, no co-infection between HIV and chronic viral hepatitis was reported in NHL cases. Adjusting for HIV infection which shares same routes of transmission with HBV and HCV, provides additional evidence that the reported association between chronic viral hepatitis and NHL in West Africa is unlikely to be mediated by HIV infection. Our control group of participants recruited in healthcare settings might also have introduced some bias as potentially over representing some infectious conditions. However, the reported HBV and HCV prevalence among our control group is quite similar to regional estimates from the general adult population in Wes Africa that ranged between 6%-8% and 1%-3%, respectively.^20,33 Of note, the reported HIV prevalence was slightly higher than UNAIDS HIV prevalence estimates, potentially underestimating the magnitude of this association, lower than the ones previously reported in our setting or more recently in Rwanda.^8,34 HBs antigen was found to be significantly associate with NHL only in multivariate analysis. However, this is not unusual given that univariate and multivariate models ask different questions. Univariate model asks about the crude relationship between LNH and the presence of HBs antigen, uncontrolled for other key co-factors included in the adjusted model such as HIV and HCV serology. Therefore, one of the possible explanations is that HIV and HCV infection have a strong effect (which is the case here), so without adjusting for these covariates, unexplained variability is large and weaker effects cannot be seen, while after adjusting for these two parameters, weaker effects are not anymore small compared to the remaining unexplained variability. Finally, our limited sample size did not allow any stratified analyses according to NHL sub-types.

5 |. CONCLUSIONS

Chronic viral hepatitis B and C were significantly associated with NHL in West Africa. Timely preventive and treatment measures against HBV infection and access to curative anti-HCV treatment might prevent a significant number of NHL in the region. In a context of increasing access to ART, NHL remains highly associated with HIV infection in West Africa. The persistently high burden of NHL in HIV-infected persons supports the need to enhance diagnostic capacities and access to adapted chemotherapies in this high-risk population.

What’s New?

While hepatitis B virus (HBV) infection is positively associated with non-Hodgkin lymphoma (NHL) in some regions of the world, whether HBV increases risk of NHL remains uncertain. In this case-control study based on data from referral hospitals in sub-Saharan Africa, the authors describe a significant association between HBV infection and NHL. In West Africa, NHL was significantly associated with chronic infection with HBV and hepatitis C virus (HCV). The findings suggest that the implementation of measures to prevent HBV infection, along with efforts to ensure access to anti-HCV treatments, could greatly reduce NHL in settings with high HBV prevalence.

Abbreviations:

ART

antiretroviral treatment

CI

confidence interval

EBV

Epstein-Barr virus

HBV

hepatitis B virus

HCV

hepatitis C virus

HTLV-1

human T-cell lymphotropic virus-1

IARC

International Agency for Research on Cancer

IHC

immunohistochemistry

IQR

interquartile range

NHL

non-Hodgkin lymphomas

OR

odds ratio

SSA

sub-Saharan

DATA AVAILABILITY STATEMENT

The data that support the findings of our study are available from the corresponding author upon reasonable request.