Original Research Highlight article
Hepatitis B infection outcome is associated with novel human
leukocyte antigen variants in Ghanaian cohort
Kwesi Z Tandoh1,2 , Kwadwo A Kusi1,2,3, Timothy N Archampong4, Isaac Boamah5 and
Osbourne Quaye1,2
1West African Centre for Cell Biology of Infectious Pathogens, College of Basic and Applied Sciences, University of Ghana, Legon LG54,
Ghana; 2Department of Biochemistry, Cell and Molecular Biology, College of Basic and Applied Sciences, University of Ghana, Legon
LG54, Ghana; 3Department of Immunology, Noguchi Memorial Institute for Medical Research, University of Ghana, Legon LG 581,
Ghana; 4Department of Medicine and Therapeutics, School of Medicine and Dentistry, University of Ghana, Accra 4236, Ghana;
5Department of Microbiology, School of Medicine and Dentistry, University of Ghana, Accra Box 4236, Ghana
Corresponding authors: Osbourne Quaye. Email: oquaye@ug.edu.gh; Kwesi Z Tandoh. Email: kztandoh@st.ug.edu.gh
Impact statement Abstract
Genetic association studies can determine Chronic hepatitis B infection is an important medical problem in sub-Saharan Africa. With
the effect size of gene loci on disease increasing concerns of dwindling access to needed care, increasing cost of treatment, and
outcomes. In the arena of HBV infections,
HLA alleles that associate with HBV out- rising prevalence of dire outcomes like liver cirrhosis and hepatocellular cancer, the need to
comes can be used in clinical management determine the genetic associations underpinning hepatitis B virus persistence or clearance
decisions. This potential translational utility
can shape the future management of HBV in a population comes to the fore. Genetic association studies have suggested a variation in
infections by identifying at-risk individuals human leukocyte antigen alleles associated with hepatitis B virus outcome along geo-ethnic
and tailoring medical interventions
accordingly. This precision medicine motif lines. We investigated the association of human leukocyte antigen alleles to hepatitis B virus
is currently only a nascent idea. However, it outcome against this backdrop. We used targeted next generation sequencing to type the
has stakes that may well override the cur-
human leukocyte antigen class I and II alleles of 173 study participants. These comprised of
rent “wait and see” approach of clinical
management of HBV infections. Here, we 92 cases with chronic hepatitis B infection and 81 healthy controls with serological evidence
have identified HLA alleles associated with of naturally cleared hepatitis B virus infection. We have identified human leukocyte antigen
HBV outcome in a Ghanaian cohort. Our
findings support the motif that HLA alleles alleles associated with hepatitis B virus clearance and persistence for the first time in a
associate with HBV outcome along geo- Ghanaian population. The class 1 allele C*16:01 (odds ratio (OR)¼3.4, confidence interval
ethnic lines. This buttresses the need for
further population pivoted studies. In the (CI)¼ 1.6–7.0, P-value¼ 0.01) was associated with hepatitis B virus persistence. Four class I
long term, our findings add to efforts alleles and one class II allele: A*34:02 (OR¼ 0.1, CI¼ 0.04–0.2, P-value¼ 3.4e-05), A*74:01
towards the development of an HLA
molecular-based algorithm for predicting (OR¼ 0.3, CI¼ 0.2–0.7, P-value¼ 0.0135), B*13:02 (OR¼0.04, CI¼ 0.01–0.2, P-value¼
HBV infection outcomes. 0.000172), C*08:04 (OR¼ 0.06, CI¼ 0.01–0.2, P-value¼7.83e-05), and DRB1*08:04
(OR¼ 0.2, CI¼ 0.03–0.27, P-value¼ 0.000252) were associated with hepatitis B virus clearance. Our data show that previously
reported human leukocyte antigen alleles associations to hepatitis B virus outcome are not found in this Ghanaian study. This
study has therefore identified human leukocyte antigen types that are associated with either hepatitis B virus persistence or
clearance and highlights the importance of geo-ethnic pivoted studies in determining the genetic associations to acute hepatitis B
virus infection outcome.
Keywords: Human leukocyte antigen, chronic hepatitis B infection, association, genetics, virology, immunology
Experimental Biology and Medicine 2020; 245: 815–822. DOI: 10.1177/1535370220921118
ISSN 1535-3702 Experimental Biology and Medicine 2020; 245: 815–822
Copyright ! 2020 by the Society for Experimental Biology and Medicine
816 Experimental Biology and Medicine Volume 245 May 2020
...............................................................................................................................................................
Introduction Materials and methods
Chronic hepatitis B (CHB) is still important in the disease Study participants recruitment
morbidity and mortality books of many countries, especial-
ly those in sub-Saharan Africa. In 2015, the World Health Ninety-two chronic HBV cases were recruited from the
Organization (WHO) estimated that 257 million people Department of Medicine’s Liver/Gastroenterology clinic.
were living with CHB worldwide; 887,000 deaths were This clinic is run every Tuesday at the old medical block
reported in 2015 from CHB-related cirrhosis and hepatocel- of KBTH. All cases were verbally counseled on the study
lular cancer (HCC). As of 2016, only 4.5 million (16.7%) of and informed consent was documented. Cases were iden-
persons diagnosed with CHB were on treatment (WHO tified as patients with at least sixmonths history of persis-
Hepatitis B fact sheet, 2017). tent hepatitis B surface antigen (HBsAg) positivity. Cases
The plurality of outcomes of CHB ranges from were not stratified according to outcome and included
asymptomatic chronic infection to chronic hepatitis B dis- asymptomatic CHB, persistently active CHB, liver cirrho-
ease, hepatocellular cancer, and liver cirrhosis.1 Acute sis, and hepatocellular cancer from CHB. A data collection
hepatitis B virus (HBV) infection in adults results in viral tool was used to document demographic details, general
persistence in 5–10% of cases.2 The determinants of out- medical history, and some laboratory results.
come following HBV infection are not well understood.3 Eighty-one controls were recruited from replacement
Age, sex, and immune status have been reported to be asso- blood donors of the KBTH blood bank. Replacement
ciated with outcome following HBV infection.4 The blood donors are family or friends of patients who have
immune response has been conceptually implicated as received blood transfusions that must be replaced. It is a
vital to the determination of outcome following HBV system utilized by the KBTH to maintain stocks in the
infection. blood bank. All controls were verbally counseled on the
The key immunological determinant is thought to be study and informed consent was documented. Controls
the potency of the CD8þ T cell response. This leads to have had previous hepatitis B infection and have cleared
clearance of the virus (resolution) if sufficient; or persis- the virus. Controls were identified as replacement donors
tence of the virus (chronic hepatitis B) if not sufficient. with HIV (human immunodeficiency virus) and HBsAg-
This is undergirded by the adequacy of the HLA- negative serological tests, and IgG positive for HBcAb (hep-
restricted antigen presentation to CD4þ T helper cells, atitis B core antibody) and/or HBsAb (hepatitis B surface
and failure or success to mount an effective antibody antibody). Measurement of HBsAg, HBcAb, and HBsAb
response to the core and surface antigens on HBV.1,5–10 was done using an immunochromatographic-based rapid
VR
HLA proteins execute the vital function of presenting for- diagnostic kit (Dia Spot ) and following themanufacturer’s
eign antigens to CD8þ cytolytic T cells (class 1) and instruction. Over 200 replacement donors were screened.
CD4þ helper T cells (class 2).11 HLA genes are the most A data collection tool was used to document demographic
polymorphic loci known in the human genome. Thus, details.
they serve as ideal candidates for genetic association Exclusion criteria for recruitment included participants
studies into the determinants of outcome following who refused to consent for the study, cases with diagnosed
acute HBV infection.3 co-morbidities such as alcoholic liver disease, hepatitis C,
Previous studies have investigated the association and HIV infections.
of HLA alleles with HBV persistence or clearance.12–14
Thursz et al.15 demonstrated that MHC class II Sample collection and HLA assignment
allele DRB*1302 was associated with reduced risk of A buccal swab was taken for each participant. Buccal swabs
chronic HBV infection in the Gambia. This finding were shipped to GeorgetownUniversity Medical Center for
seems to vary among chronic HBV-infected patients in HLA typing by targeted next generation sequencing.
different parts of the world. Wang et al.14 proposed Isolation of DNA using magnetic beads (KingFisher Flex
that these seeming discordant conclusions can be Purification Instrument, ThermoFisher) was followed by
brought closer to harmony by conducting HLA polymor- long-range amplification of HLA loci in separate polymer-
phism association studies in different ethnic groups around ase chain reactions.16 The class I (HLA-A, -B, -C) amplicons
the world. included all exons and introns including portions of the 50
With the rising importance of precision medicine, and 30 untranslated regions. The DRB1 amplicon includes
it will benefit future clinical management of CHB to exons 2–3. The DQB1 amplicon includes exons 2–4. DRB1
be able to predict the outcome of CHB individually. and DQB1 analysis includes only exons; introns are not
This will require a molecular prediction algorithm that evaluated.
includes population-specific independent factors that are The pooled amplicons were purified and sheared to an
predictive of CHB outcome. One important step to this average size of 700 base pairs by sonication. A DNA library
goal will be to identify population-specific HLA alleles was constructed using an Accel-NGS 2S Plus DNA library
associated with HBV persistence and clearance. This kit (Swift Biosciences, Ann Arbor, MI, USA). DNA frag-
study sought to interrogate the association between HLA ments were tagged with one unique dual index combina-
alleles and HBV infection outcome in a Ghanaian study tion (Swift Biosciences). Libraries from 96 individuals were
population. combined, purified, and 600–1200 bp DNA fragments
Tandoh et al. Association of novel HLA variants to acute HBV outcome 817
...............................................................................................................................................................
selected. The pooled libraries were sequenced simulta- HLA haplotype frequencies for both cases and controls
neously in a single 500 cycle (V2) paired-end run using an were determined using Hapl-o-Mat.19 An expectation max-
Illumina MiSeq (Illumina, San Diego, CA, USA). imization (EM) algorithm was used that resolved for phase
Data analysis used the AssignTM for TruSightTM HLA ambiguity in the HLA data. The algorithm uses a maxi-
software (version 2.1.0.934, Illumina Inc., San Diego, CA, mum likelihood estimation to compute the most probable
USA). Sequencing data were interpreted using the July 2018 set of haplotypes that explain the unphased genotype
20
IPD-IMGT/HLA database 3.33.0.17 input. The HLA haplotype frequencies from the
Ghanaian study population (Supplementary Table 1) were
Data analysis compared with other populations from the US National
Marrow Donor Program (NMDP) using principal compo-
Data cleaning was done using Excel. Exploratory data anal- nents analysis (PCA).21 The first principal component
ysis was done in R version 3.6. Association analysis was accounted for over 90% of the variance in the Ghanaian
done using the python script for HLA association analysis, haplotype frequencies (Supplementary Figure 1). The first
PyHLA.18 Odds ratios were calculated in R using both 2 2 and second PCAs demonstrate that the Ghanaian haplotype
contingency tables and univariate logistic regression. frequencies do not have close genetic proximity to any group
Multivariate logistic regressionmodels used all HLA alleles in the NMDP database (Supplementary Figure 2).
with a mean allele frequency greater than 5% as indepen- Tests for Hardy Weinberg equilibrium were done in
dent variables. Haplotype construction and analysis were Arlequin version 3.5.2.2.22 HLA haplotype frequencies for
done using the software Hapl-o-Mat.19 cases met with Hardy Weinberg equilibrium expectations.
HLA haplotype frequencies for controls did not meet
Results Hardy Weinberg equilibrium expectations (Supplementary
Figure 3).
Characteristics of study participants recruited
A total of 173 study participants were recruited; 92 cases HLA alleles associated with HBV persistence
and 81 controls. Both groups were similar in gender distri- Two alleles were associated with HBV persistence follow-
bution (Chi-square¼ 0, P-value¼ 1). However, the case ing 2 2 contingency table and univariate logistic regres-
group was significantly older (median age¼ 36) compared sion analysis (Table 2, Figure 1). The class I HLA allele,
to the controls group (median age¼ 33, Mann–Whitney C*16:01 (OR¼ 3.4, CI: 1.6–7.0, P-value¼ 0.005) and the
U test statistic¼ 4428, P-value¼ 0.03). Both groups also dif- class II HLA allele, DQB1*05:01 (OR¼ 3.3, CI¼ 1.5–7.5,
fered in ethnicity distribution (Fisher’s test, P-value¼ 0.03) P-value¼ 0.03). The risk of both alleles seemed additive,
(Table 1). Class I and II HLA distribution between the two as increasing the presence of the allele was associated
groups was similar (Chi-square, P-value¼ 0.62 and 0.85, with increasing odds of HBV persistence (cases)
respectively, Table 1). (Cochrane Armitage trend test (CATT) C*16:01¼3.25,
P-value¼ 0.001); DQB1*05:01¼2.73, P-value¼ 0.006)
Table 1. Characteristics of study participants recruited. (Supplementary Table 8).
For the class I HLA allele C*16:01, 39 study participants
Characteristic Case Control P-value had this allele. Five were homozygous—all cases. C*16:01
was heterozygous in 24 cases and 10 controls. A multivar-
Number (n) 92 81
Gender (n, %) iate logistic regression analysis of the effect size of C*16:01
Male 52 (30) 69 (40) 5.33e-05a on HBV outcome was done using all HLA alleles as inde-
Female 40 (23) 12 (7) pendent variables. This showed C*16:01 was associated
Age 36 33 0.03b with HBV persistence (OR¼3.1, CI: 1.5–6.5, P-value¼ 0.01).
Ethnicity (n, %) However, when the model was adjusted for age and eth-
Akan 49 (30.6) 46 (26.6) 0.003a
nicity as covariates, the association of C*16:01 allele
Ewe 10 (6.9) 14 (8.1)
Ga-Adangbe 11 (7.5) 20 (11.6) with HBV persistence was lost (OR¼ 0.21, CI¼ 0.09–0.48,
Others 14 (8.1) 1 (0.6) P-value¼ 0.00002) (Table 3, Figure 2).
HLA Class I (n) For the class II HLA allele DQB1*05:01, 33 study partic-
A 184 156 0.62c ipants had this allele, 5 were homozygous—all of whom
B 184 152 were cases. DQB1*05:01 was heterozygous in 17 cases and
C 182 158 8 controls. A multivariate logistic regression analysis of the
HLA Class II (n)
effect size of DQB1*05:01 on HBVoutcome was done using
DPB1 80 160 0.85c
DQB1 182 160 all HLA alleles as independent variables. This, however,
DRB1 184 160 did not identify DQB1*05:01 as an allele associated with
HBV persistence or clearance.
aFisher’s exact test for count data.
bMann–Whitney U test.
cChi-square test. HLA alleles associated with HBV clearance
Note: Statistically significant P-values are in bold. The HLA Class I and II data
presented here represent count summaries of the respective alleles for cases Four class I and four class II alleles were associated with
and controls. HBV clearance following 2 2 contingency table analysis
818 Experimental Biology and Medicine Volume 245 May 2020
...............................................................................................................................................................
Table 2. Alleles associated with HBV outcome following 2 2 contingency table analysis.
Number
Allele Case Control Odds ratio (OR) Confidence interval P-value
C*16:01 29 10 3.4 (3.3) 1.6–7.0 0.005 (0.004)
DQB1*05:01 27 8 3.3 (2.9) 1.5–7.5 0.03 (0.02)
A*34:02 6 31 0.14 (0.12) 0.06–0.34 2.57e-05 (6.86e-6)
A*74:01 15 32 0.38 (0.3) 0.2–0.34 3.24e-02 (0.0008)
B*13:02 2 27 0.05 (0.04) 0.01–0.22 2.02e-06 (3.53e-05)
C*08:04 3 27 0.08 (0.07) 0.02–0.27 6.28e-06 (2.01e-5)
DQB1*04:02 18 40 0.36 (0.25) 0.2-0.64 4.30e-03 (5.4e-5)
DQB1*06:02 22 43 0.41 (0.2) 0.3-0.8 2.33e-02 (2.4e-8)
DRB1*08:04 17 39 0.31 (0.24) 0.2-0.6 1.60e-03 (5.25e-5)
DRB1*15:03 20 43 0.38 0.2-0.7 4.60e-03
*Results for univariate logistic regression outputs placed in brackets for ORs and p-values
Univariate logistic regression analysis
C*16:01
DQB1*05:01
A*34:02
A*74:01
B*13:02
C*08:04
DQB1*04:02
DQB1*06:02
DRB1*08:04
DRB1*15:03
1 2 3 4 5 6 7
Odds ratio
Figure 1. HLA alleles with association to HBV outcome following univariate logistic regression analysis. HLA alleles with odds ratios less than 1 are associated with
HBV clearance; and alleles with odds ratios greater than 1 are associated with HBV persistence. (A color version of this figure is available in the online journal.)
HLA Alleles
Tandoh et al. Association of novel HLA variants to acute HBV outcome 819
...............................................................................................................................................................
and univariate logistic regression analysis (Table 2, Class I HLA alleles associated with HBV clearance
Figure 1). Following multivariate logistic regression, all The class I alleles, A*34:02 (OR¼ 0.14, CI: 0.06–0.34,
four class I and only one class II allele remained significant- P-value¼ 2.57e-05); A*74:01 (OR¼ 0.3, CI: 0.2–0.7,
ly associated with HBV clearance (Table 2, Figure 1). P-value¼ 0.01), HLA supertype association A0323; B*13:02
(OR¼ 0.05, CI: 0.01–0.22, P-value¼ 2.02e-06), and C*08:04
(OR¼ 0.08, CI: 0.02–0.27, P-value¼ 6.28e-06) were associat-
ed with HBV clearance. The effect of all these HLA alleles
seemed to be additive, as decreasing the presence of this
Table 3. Alleles associated with HBV outcome following multivariate allele was associated with increasing odds of HBV persis-
logistic regression analysis. tence (CATT: A*34:02, z¼ 5, P-value¼ 5.4e-7; A*74:01,
Allele Odds ratio (OR) Confidence interval P-value z¼ 3.2, P-value¼ 0.002; B*13:02, z¼ 5.5, P-value¼ 4.4e-8;
C*08:04, z¼ 5.2, P-value¼ 1.49e-7) (Supplementary
C*16:01 3.1 1.5–6.5 0.01 Table 8).
A*34:02 0.1 0.04–0.2 3.35e-05
Thirty-seven study participants had the A*34:02 allele.
A*74:01 0.3 0.2–0.7 0.0135
B*13:02 0.04 0.01–0.2 0.0002 None were homozygous for it. A*34:02 was heterozygous
C*08:04 0.07 0.01–0.2 7.83e-5 in 6 cases and 31 controls with the allele. Forty-seven study
DRB1*08:04 0.2 0.1–0.4 0.00003 participants had the A*74:01 allele; only one case was
homozygous for it. A*74:01 was heterozygous in 14 cases
*An additive model with Bonferroni’s correction for multiple testing was used.
and 32 controls with the allele. Of the 29 study participants
Multivariate logistic regression analysis adjusted for covariates additive model
C*16:01
A*34:02
A*74:01
B*13:02
C*08:04
p < 0.05
DRB1*08:04
0.5 1.0 1.5 2.0 2.53.03.54.0
Odds ratio
Figure 2. HLA alleles with association to HBV outcome following multivariate logistic regression analysis. All HLA alleles with allele frequency greater than 5% were
used as independent variables in this model. HLA alleles with odds ratios less than 1 are associated with HBV clearance; and alleles with odds ratios greater than 1 are
associated with HBV persistence. (A color version of this figure is available in the online journal.)
HLA Alleles
820 Experimental Biology and Medicine Volume 245 May 2020
...............................................................................................................................................................
with the B*13:02 allele, nonewere homozygous for it. B*13:02 with the allele. Sixty-five study participants had the
was heterozygous in 2 cases and 27 controls with the allele. DQB1*06:02 allele. Four cases and one control were homo-
Of the 30 study participants with the C*08:04 allele, none zygous for this allele. DQB1*06:02 was heterozygous in 18
were homozygous for it. C*08:04 was heterozygous in cases and 42 controls with the allele. Of the 56 study par-
3 cases and 27 controls with the allele. Therefore, these pro- ticipants with the DRB1*08:04 allele, none were homozy-
tective alleles were more frequent in controls than cases. gous for it. DRB1*08:04was heterozygous in 17 cases and 39
A multivariate logistic regression analysis of the effect controls with the allele. Of the 30 study participants with
size of the HLA alleles was done using all HLA alleles as the DRB1*15:03 allele, none were homozygous for it.
independent variables. This showed A*34:02 (OR¼ 0.1, CI: DRB1*15:03 was heterozygous in 3 cases and 27 controls
0.04–0.2, P-value¼ 3.4e-5); A*74:01 (OR¼ 0.3, CI¼ 0.2–0.7, with the allele. Therefore, these class II protective alleles
P-value¼ 0.0135); B*13:02 (OR¼ 0.04, CI¼ 0.01–0.2, P-val- were also more frequent in controls than cases.
ue¼ 0.000172); and C*08:04 (OR¼ 0.06, CI¼ 0.01–0.2, P-val- A multivariate logistic regression analysis of the effect
ue¼ 7.83e-05) were still associated with HBV clearance. size of the class II HLA alleles was done using all HLA
When the model was adjusted for age, sex, and ethnicity alleles as independent variables and adjusted for age, sex,
as confounding covariates, the association of A*34:02, and ethnicity as confounding covariates. This showed
A*74:01, B*13:02, and C*08:04 alleles with HBV clearance DQB1*04:02, DQB1*06:02, and DRB1*15:03 alleles were no
was still protective (Table 3, Figure 2). longer associated with HBV clearance. Only DRB1*08:04
(OR¼ 0.2, CI¼ 0.03–0.27, P-value¼ 0.000252) were still
Class II HLA alleles associated with HBV clearance associated with HBV clearance in the multivariate logistic
The class II alleles, DQB1*04:02 (OR¼ 0.36, CI: 0.2–0.64, model (Table 3, Figure 2).
P-value¼ 4.3e-3), DQB1*06:02 (OR¼ 0.4, CI: 0.3–0.8, P-val-
ue¼ 2.33e-2), DRB1*08:04 (OR¼ 0.3, CI: 0.2–0.6, P-val- Haplotype analysis shows A*34:02B*13:02,
ue¼ 1.6e-3), and DRB1*15:03 ((OR¼ 0.38, CI: 0.2–0.7, A*34:02DRB1*08:04, and A*74:01C*08:04 are
P-value¼ 4.6e-3) were associated with HBV clearance. associated with HBV clearance
The effect of all these HLA alleles seemed to be additive, Supplementary Table 2 shows the top 10 most common one
as decreasing the presence of this allele was associated with field haplotypes among all study participants.
increasing odds of HBV persistence (CATT: DQB1*04:02, Supplementary Tables 3 and 4 show the most frequent hap-
z¼ 3.7, P-value¼ 0.0002; DQB1*06:02, z¼ 3.1, P-value¼ lotypes for cases and controls, respectively. The most fre-
0.002; DRB1*08:04, z¼ 4.2, P-value¼ 3.2e-5; DRB1*15:03, quent one field haplotype among all study participants
z¼ 5.4, P-value¼ 6.7e-8). were A*34B*13C*08DQB1*06DRB1*08 (16.25%) and
Fifty-eight study participants had the DQB1*04:02 allele. A*74B*53C*04DQB1*04DRB1*15 (16.25%). Both hap-
Two cases and one control were homozygous for this allele. lotypes were found exclusively among the controls. The
DQB1*04:02 was heterozygous in 16 cases and 39 controls most common haplotype among the cases were A*30
B*42C*07DQB1*04DRB1*03 (2.71%) and A*02B*53
Table 4. HLA haplotypes significantly associated with outcome (clear- C*04DQB1*06DRB1*15 (1.6%).
ance) following acute HBV infection. Supplementary Table 5 shows the top 10 most frequent
Case Control two field haplotypes for all study participants.
Haplotype number number OR P-value Supplementary Tables 6 and 7 show the most frequent hap-
lotypes for cases and controls, respectively. The most
A*34:02B*13:02 0 27 0.01 0.00000000007991*
 common two field haplotypes among all study participantsA*34:02 DRB1*08:04 1 28 0.02 0.0000000005546*
 were A*34:02B*53:01C*04:01DQB1*06:02DRB1*15:03A*74:01 C*08:04 0 26 0.01 0.0000000002136*
(7.9%) and A*74:01B*13:02C*08:04DQB1*04:02
*Statistically significant. DRB1*08:04 (7.6%). Both haplotypes were found exclusive-
ly among the controls. The most common two field haplo-
type among cases was A*30:01B*42:01C*17:01
Table 5. HLA loci zygosity for all study participants.
DQB1*04:02DRB1*03:02 (3.4%).
Loci Zygosity Case (number) Control (number) Total
A Homozygous 7 1 8
Table 6. Association between HLA loci heterozygosity and HBV
Heterozygous 85 78 163
B Homozygous 9 3 12 persistence.
Heterozygous 83 76 159 HLA loci Odds Ratio P-value
C Homozygous 15 5 20
Heterozygous 77 74 151 A 0.2 0.9
DPB1 Homozygous 17 43 60 B 0.9 0.8
Heterozygous 20 23 43 C 0.4 0.05
DQB1 Homozygous 22 6 28 DPB1 0.3 0.06
Heterozygous 69 74 143 DQB1 0.3 0.005*
DRB1 Homozygous 7 3 10 DRB1 0.5 0.3
Heterozygous 85 77 172
*Statistically significant.
Tandoh et al. Association of novel HLA variants to acute HBV outcome 821
...............................................................................................................................................................
Next, we investigated the relationship between haplo- C*16:01 was significantly associated with HBV persis-
types and HBVoutcome. We used HLA alleles with signif- tence. This has never been reported. The HLA molecule
icant association with HBV outcome in a multivariate encoded by this allele may be poor at antigen presentation
logistic regression adjusted for age, sex, and ethnicity as to CD8þ T cells; therefore, its association with HBV persis-
covariates (Table 3). Combinations of these HLA alleles tence. However, functional studies are yet to implicate this
were used to form two, three, and four block haplotypes. allele in poor antigen presentation to CD8þ T cells.
These haplotypes were then tested for association with out- The molecules encoded by A*34:02, A*74:01, B*13:02,
come after HBV infection using a univariate logistic model. and C*08:04 are likely to be very efficient at viral epitope
Haplotypes A*34:02B*13:02, A*34:02DRB1* 08:04 and
 presentation to CD8þ T cells. This may explain their asso-A*74:01 C*08:04 were the only haplotypes significantly ciation with HBV clearance. Functional studies are yet to
associated with HBV clearance (Table 4). validate this. This is the first study to report on this associ-
ation. This may be on account of this allele’s frequency in
Heterozygosity at class II locus DQB1 is associated
other populations compared to the Ghanaian population.
with reduced odds of HBV persistence
The association of heterozygosity with outcome follow-
Next, we interrogated the association between heterozy- ing HBV infection was significant for only the DQB1 loci.
gosity at an HLA class loci and outcome following HBV This was similar to findings in the Gambian cohort15 and a
infection. Heterozygosity at an HLA loci is conceptually Caucasian cohort.3 The inability to detect a significant asso-
associated with a higher odd for antigen recognition, bind- ciation between heterozygosity in class I HLA loci and HBV
ing, and presentation to immune cells.24,25 This in turn may outcomemay stem from inadequate statistical power of this
lead to a higher odds of HBV clearance. Therefore, we study design.
tested the hypothesis that HLA loci heterozygosity will be We recognize some limitations of this study. This study
associated with increased odds of HBV clearance or only focused on HLA alleles as the predictors of outcome
reduced odds of HBV persistence. following HBV infection. Factors such as other genetic
Table 5 summarizes the zygosity for the HLA loci typed determinants (cytokine genes and expression profiles, for
in all the study participants. The Cochran–Mantel–
example), nutritional status, and socioeconomic strata were
Haenszel test of independence showed that there was an
completely ignored. Our application of Bonferroni correc-
association between zygosity and outcome of HBV infec-
P ¼ tion in the multivariate analysis may have excluded other-tion across all HLA loci ( -value 2.2e-16; Woolf test
P-value¼ 0.6). Table 6 shows the summary of the effect wise significant allele associations. This may have
size of the association between HLA loci heterozygosity increased type 2 errors. Our reporting of novel HLA asso-
and HBV persistence. Heterozygosity at all HLA loci was ciations requires further studies to investigate biological
associated with a reduced odds of HBV persistence plausibility of these alleles. Compared to other studies,
(OR< 1). However, only heterozygosity at the class 2 our relatively small sample size may have limited the
HLADQB1 loci was significantly associated with a reduced power of the experimental design to detect allele associa-
odds of HBV persistence (OR¼ 0.3, P-value¼ 0.005). tions. Wemay therefore have missed significant alleles with
low frequency on account of this. Replication of this study
Discussion in different populations in Ghana and elsewhere will help
to validate the findings. Notwithstanding these limitations,
Genetic association studies (candidate gene) are powered
the data generated here bring into perspective the poten-
to identify the effect size of a genetic locus on a disease. The
tially important role of HLA type differences as determi-
current body of scientific evidence suggests that HLA allele
nants of HBV infection outcome.
types influence HBV infection outcome along geo-ethnic
14 In summary, our data show that HLA class I moleculeslines. This tacit consensus in the field of HLA/HBV asso-
ciations highlights the need for population-tailored study are significantly associated with HBV infection outcome in
designs. Here, we show the importance of this paradigm. the Ghanaian population. This suggests that the role of
We set out to determine the HLA alleles associated with cytolytic CD8þ T cells may be pivotal to acute HBV infec-
HBV outcome in a Ghanaian study population. We have tion persistence or clearance. Further functional studies to
identified HLA alleles never reported as associated with interrogate this hypothesis are needed. The HLA heterozy-
outcomes following acute HBV infection. gous advantage is most significant at the class II HLA
The HLA class I allele C*16:01 was associated with a DQB1 loci. Overall, the data highlight the importance of
3-fold increase in HBV persistence. A*34:02, A*74:01, geo-ethnic pivoted studies in determining the genetic fac-
B*13:02, and C*08:04 were associated with HBV clearance. tors associated with acute HBV infection outcome.
Overall, the HLA class 1 loci had the most significant asso-
ciations similar to the findings by Thio et al.3 This implicates AUTHOR’S CONTRIBUTIONS
the role of CD8þ T cells as pivotal in determining outcome
following acute HBV infection. This finding also agrees KZT, KAK, and OQ conceived the study; KZT, TA and IB per-
with a study that showed that chimpanzees infected with formed study participants recruitment, data and sample col-
HBV could not clear the virus after CD8þ Tcell depletion.8 lection; KZT, KAK, and OQ analyzed the data and drafted the
Of the class II HLA alleles, only DRB1*08:04 was signifi- manuscript; OQ and KAK supervised the study. All authors
cantly associated with HBV clearance. critically reviewed and edited the manuscript.
822 Experimental Biology and Medicine Volume 245 May 2020
...............................................................................................................................................................
ACKNOWLEDGEMENTS 2. Zampino R, Boemio A, Sagnelli C, Alessio L, Adinolfi LE, Sagnelli E,
Coppola N. Hepatitis B virus burden in developing countries. World J
The authors acknowledge the contribution of Mr. Derrick Gastroenterol 2015;21:11941
Tetteh and Mr. Kamara Ustaz who helped with sample and 3. Thio CL, Thomas DL, Karacki P, Gao X, Marti D, Kaslow RA, Goedert
data collection respectively; Nurses Joyce Konadu Asamoah JJ, Hilgartner M, Strathdee SA, Duggal P. Comprehensive analysis of
and Patience Attipoe; and Ms. Yvett Quartey (Records unit) of class I and class II HLA antigens and chronic hepatitis B virus infection.
the KBTH Department of Medicine’s gastro-enterology/liver J Virol 2003;77:12083–7
4. Hyams KC. Risks of chronicity following acute hepatitis B virus infec-
clinic who helped in facilitating the recruitment of case study
tion: a review. Clin Infect Dis 1995;20:992–1000
participants. The authors also acknowledge the help of Mr.
5. Baumert TF, Thimme R, von Weizs€acker F. Pathogenesis of hepatitis B
Raymark Asare-Danso and Mr. Edward Lartey Okine of the virus infection. World J Gastroenterol 2007;13:82
KBTH Blood bank for their help in facilitating the recruitment 6. Maini MK, Boni C, Lee CK, Larrubia JR, Reignat S, Ogg GS, King AS,
of control study participants. The authors would also like to Herberg J, Gilson R, Alisa A. The role of virus-specific CD8þ cells in
thank Carolyn Hurley, Kanthi Kariyawasam, and Lihua Hou liver damage and viral control during persistent hepatitis B virus infec-
of Georgetown University Medical Center in Washington DC tion. J Exp Med 2000;191:1269–80
for typing the HLA alleles. 7. Pungpapong S, Kim WR, Poterucha JJ. Natural history of hepatitis B
virus infection: an update for clinicians. Mayo Clin Proc 2007;967–75
8. Thimme R, Wieland S, Steiger C, Ghrayeb J, Reimann KA, Purcell RH,
DECLARATION OF CONFLICTING INTERESTS
Chisari FV. CD8þ Tcells mediate viral clearance and disease pathogen-
The author(s) declared no potential conflicts of interest with esis during acute hepatitis B virus infection. J Virol 2003;77:68–76
9. Webster GJ, Reignat S, Maini MK, Whalley SA, Ogg GS, King A, Brown
respect to the research, authorship, and/or publication of this
D, Amlot PL,Williams R, Vergani D. Incubation phase of acute hepatitis
article.
B in man: dynamic of cellular immune mechanisms. Hepatology
2000;32:1117–24
ETHICAL APPROVAL STATEMENT 10. Lee WM. Hepatitis B virus infection. N Engl J Med 1997;337:1733–45
11. Williams TM. Human leukocyte antigen gene polymorphism and the
Ethical clearance for this study was obtained from the histocompatibility laboratory. J Mol Diagn 2001;3:98–104
Institutional Review Boards of the Noguchi Memorial 12. Yan Z-H, Fan Y,Wang X-H, Mao Q, Deng G-H, Wang Y-M. Relationship
Institute of Medical Research (NMIMR, Study Number: 074/ between HLA-DR gene polymorphisms and outcomes of hepatitis B
17–18) and Korle-Bu Teaching Hospital (KBTH, Reference viral infections: a meta-analysis. World J Gastroenterol 2012;18:3119
Number 00080/2018); and the research and development 13. Zhang Z, Wang C, Liu Z, Zou G, Li J, Lu M. Host genetic determinants
office of the National Blood Services of Ghana (Research of hepatitis B virus infection. Front Genet 2019;10:696
Protocol NBSGRD/11110/11). 14. Wang L, Zou Z-Q, Wang K. Clinical relevance of HLA gene variants in
HBV infection. J Immunol Res 2016;2016:9069375
15. ThurszMR, Kwiatkowski D, Allsopp CE, Greenwood BM, Thomas HC,
FUNDING
Hill AV. Association between an MHC class II allele and clearance of
hepatitis B virus in the Gambia. N Engl J Med 1995;332:1065–9
This work was supported by funds from aWorld Bank African
16. Hou L, Enriquez E, Persaud M, Steiner N, Oudshoorn M, Hurley CK.
Centres of Excellence grant (ACE02-WACCBIP: Awandare) Next generation sequencing characterizes HLA diversity in a registry
and a Wellcome/African Academy of Sciences DELTAS population from The Netherlands. HLA 2019;93:474–83
Africa grant (DEL-15–007: Awandare). Kwesi Z. Tandoh was 17. Robinson J, Halliwell JA, Hayhurst JD, Flicek P, Parham P, Marsh SG.
supported by WACCBIP-World Bank ACE Master’s fellow- The IPD and IMGT/HLA database: allele variant databases. Nucleic
ship. The DELTAS Africa Initiative is an independent funding Acids Res 2015;43:D423–D31
scheme of the African Academy of Sciences (AAS)’s Alliance 18. Fan Y, Song Y-Q. PyHLA: tests for the association between HLA alleles
for Accelerating Excellence in Science in Africa (AESA) and and diseases. BMC Bioinform 2017;18:90
supported by the New Partnership for Africa’s Development 19. Scha€fer C, Schmidt AH, Sauter J. Hapl-o-Mat: open-source software for
HLA haplotype frequency estimation from ambiguous and heteroge-
Planning and Coordinating Agency (NEPAD Agency) with
neous data. BMC Bioinform 2017;18:284
funding from the Wellcome Trust (107755/Z/15/Z:
20. Ameen R, Al Shemmari SH, Marsh SGE. HLA Haplotype Frequencies
Awandare) and the UK government. The views expressed in and Genetic Profiles of the Kuwaiti Population. Med Princ Pract
this publication are those of the author(s) and not necessarily 2020;29:39–45
those of AAS, NEPAD Agency, Wellcome Trust or the UK 21. Gragert L, Madbouly A, Freeman J, Maiers M. Six-locus high resolution
government. HLA haplotype frequencies derived from mixed-resolution DNA
typing for the entire US donor registry. Hum Immunol 2013;74:1313–20
ORCID IDS 22. Excoffier L, Laval G, Schneider S. Arlequin (version 3.0): an integrated
software package for population genetics data analysis. Evol Bioinform
Kwesi Z Tandoh https://orcid.org/0000-0002-1628-2845 Online 2007;1:47–50
Osbourne Quaye https://orcid.org/0000-0002-0621-876X 23. Sidney J, Peters B, Frahm N, Brander C, Sette A. HLA class I super-
types: a revised and updated classification. BMC Immunol 2008;9:1
24. Parham P, Ohta T. Population biology of antigen presentation by MHC
SUPPLEMENTAL MATERIAL
class I molecules. Science 1996;272:67–74
25. Carrington M, Nelson GW, Martin MP, Kissner T, Vlahov D, Goedert JJ,
Supplemental material for this article is available online.
Kaslow R, Buchbinder S, Hoots K, O’Brien SJ. HLA and HIV-1: hetero-
zygote advantage and B* 35-Cw* 04 disadvantage. Science
REFERENCES 1999;283:1748–52
1. McMahon BJ. Natural history of chronic hepatitis B – clinical implica-
tions. Medscape J Med 2008;10:91 (Received March 2, 2020, Accepted April 1, 2020)