Lack of association between HLA-A, -B and -DRB1 alleles and the development of SARS: a cohort of 95 SARS-recovered individuals in a population of Guangdong, southern China.

Severe acute respiratory syndrome (SARS), caused by infection with a novel coronavirus (SARS-CoV), was the first major novel infectious disease at the beginning of the 21st century, with China especially affected. SARS was characterized by high infectivity, morbidity and mortality, and the confined pattern of the disease spreading among the countries of South-East and East Asia suggested the existence of susceptible factor(s) in these populations. Studies in the populations of Hong Kong and Taiwan showed an association of human leucocyte antigen (HLA) polymorphisms with the development and/or severity of SARS, respectively. The aim of the present study was to define the genotypic patterns of HLA-A, -B and -DRB1 loci in SARS patients and a co-resident population of Guangdong province, southern China, where the first SARS case was reported. The samples comprised 95 cases of recovered SARS patients and 403 unrelated healthy controls. HLA -A, -B and -DRB1 alleles were genotyped using polymerase chain reaction with sequence-specific primers. The severity of the disease was assessed according to the history of lung infiltration, usage of assisted ventilation and occurrence of lymphocytopenia. Although the allelic frequencies of A23, A34, B60, DRB1*12 in the SARS group were slightly higher, and A33, -B58 and -B61 were lower than in the controls, no statistical significance was found when the Pc value was considered. Similarly, no association of HLA alleles with the severity of the disease was detected. Thus, variations in the major histocompatibility complex are unlikely to have contributed significantly to either the susceptibility or the severity of SARS in the population of Guangdong.

Introduction

Severe acute respiratory syndrome (SARS) originated in southern China in November 2002, reached Hong Kong in February 2003, and thereafter spread rapidly to 29 countries/regions on five continents. By the end of the epidemic, the global cumulative patient total was 8098 with 774 deaths. Seven Asian countries/regions were among the top 10 populations affected, China and Hong Kong accounting for 87% of all cases and 84% of deaths from SARS (Lam ). SARS was mostly confined to the populations of South‐East and East Asia (Mainland Chinese, Hong Kong residents, Vietnamese, Singaporeans and Taiwanese). The disease is caused by infection with a novel single‐stranded RNA coronavirus (Peiris ) and starts with an influenza‐like illness characterized by non‐specific, systemic symptoms, followed by the rapid development of an unspecific bronchopneumonia associated with lower respiratory tract or gastrointestinal symptoms. Most patients recovered after 1 or 2 weeks, but some developed an acute respiratory distress syndrome (Zhong ). The confined geographical and ethnic spread pattern of the disease suggested the possibility of genetic variations that had influenced susceptibility to infection with SARS‐CoV and the development of the disease itself.

Recent reports have suggested that a single nucleotide polymorphism (SNP) in the 3′ UTR of the OAS‐1 and MxA promoter region is associated with host susceptibility to SARS infection in the Han Chinese population (Hamano ; He ). Also, an association study that examined polymorphisms in the cytokine response genes [interferon (IFN)‐gamma, tumour necrosis factor (TNF)‐alpha and interleukin (IL)‐10] showed that the 874 A/T IFN‐gamma SNP was a risk factor for SARS infection, although no association was seen with either TNF‐alpha or IL‐10 polymorphisms (Chong ). Chan ) reported that individuals who were homozygous for CLEC4M (a binding receptor for SARS Co‐V) were less susceptible to SARS (Chan ). Conversely, association studies failed to demonstrate that ACE2 SNPs were implicated in SARS (Itoyama ), and that the ACE I/D polymorphism was neither directly related to increased susceptibility to SARS‐coronavirus infection nor the clinical outcome of SARS patients (Chan ).

The human leucocyte antigen (HLA) system is widely used to investigate the aetiology of infectious diseases and autoimmune disorders. Studies in Hong Kong (Ng ) and Taiwan (Lin ) populations showed an association of HLA‐B*0703–DRB1*0301 with the development of SARS, and HLA‐B46 with the severity of SARS, respectively. However, the data on which these reports were based were limited in number, and no comparable data from Mainland China have been available. In this study, samples from 95 SARS patients and 403 healthy controls were typed for HLA‐A, ‐B and ‐DRB1 alleles to determine if an association existed between HLA variations and SARS.

Subjects and methods

Subjects

The cohort of 95 SARS‐recovered individuals comprised patients admitted to the Guangdong Provincial TCM Hospital in Guangzhou (82 cases) and the Peking University Shenzhen Hospital in Shenzhen (13 cases) between March and June 2003. All patients were diagnosed according to the World Health Organization (WHO) case definition of probable SARS case (WHO, 2003). All of the SARS subjects were ethnic Chinese and none were biological relatives (only the first infected patient in a family was included if several family members had been diagnosed). The age range of the patients was 19–72 years, and the male:female ratio was 0.76. All 95 SARS patients showed radiographic evidence of lung infiltrates and were seropositive for SARS‐CoV antibody during the SARS episode. A sex‐ and age‐matched healthy control group of 403 healthy volunteers resident in the same geographical area were also genotyped. The mean age of the controls was 45.6 years, and the male:female ratio was 0.73. Informed written consent was obtained from all tested subjects.

The hospital records of the 95 SARS cases indicated that 63 patients (66.3%) had lymphocytopenia (< 1 × 109 L−1), with two patients (2.11%) showing diffuse infiltration of both lungs. Auxiliary ventilation was applied by mask in 30 cases (31.6%), with intubation additionally used in two cases (2.1%). Lymphocytopenia, lung involvement and auxiliary ventilation were considered as indicative of disease severity in our study.

HLA allele typing

Peripheral blood samples were collected in EDTA vacutainers, with genomic DNA prepared by standard phenol/chloroform/isoamyl alcohol extraction and ethanol precipitation for immediate storage at –20 °C. DNA quantity and quality were assessed by spectroscope (Pharmacia Biotech, Uppsala, Sweden) absorbance at 260 and 280 nm. The identification of HLA‐A, ‐B and ‐DRB1 alleles in all subjects was performed by DNA typing, using polymerase chain reaction with sequence‐specific primers (PCR‐SSP) based on the protocol of Bunce ), with slight modification to correspond with serological typing results. The conversion was based on the WHO nomenclature table of correspondence (Marsh ).

Briefly, PCR mixtures consisted of 67 mm Tris base pH 8.8, 16.6 mm ammonium sulphate, 2 mm magnesium chloride, 0.01% (v/v) Tween 20, 200 pmol of each dNTP, 1–4 pmol of each allele‐specific primer, 0.1 pmol of DRB1 control primers or 2 pmol of APC control primers, between 0.1 and 0.01 pg DNA and 0. 1875 U Taq polymerase. The primer concentrations in the primer mixes were empirically titrated with positive and negative samples to ensure maximum efficiency. PCR amplications were carried out in an MJ Research PTC‐100 (Waltham, MA, USA) machine. The cycling parameters for 13 µl reactions in rapid‐cycling PCR machines were: 1 min at 96 °C followed by five cycles of 25 s at 96 °C, 45 s at 70 °C, 45 s at 72 °C, followed by 21 cycles of 25 s at 96 °C, 50 s at 65 °C, 45 s at 72 °C, followed by four cycles of 25 s at 96 °C, 60 s at 55 °C and 20 s at 72 °C.

Statistical analysis

To assess the association of HLA variations with SARS infection, the frequencies of HLA class I and II alleles were estimated by direct counting in both the patient and the control groups. The frequencies of HLA alleles (HLA antigens) in patients with SARS and healthy controls were each tested for deviation from Hardy–Weinberg equilibrium, and compared using a χ2 test (continuity correction for the χ2 test was used where a small number of alleles was observed (E < 5) and a two‐tail Fisher's exact test was used to calculate the P‐value when E < 1. Corrected P‐values (Pc) were obtained by multiplying the number of antigens or alleles (Edward, 1974: reference not listed) at each locus: A (n = 15), B (n = 32) and DRB1 (n = 13). A Pc value < 0.05 was considered statistically significant. Odds ratios (OR) rather than relative risks (RR) were quoted since ORs are most commonly used in HLA association studies. The values were obtained from standard contingency table analysis using Haldane's modification of Woolf's method (Haldane, 1956). An OR value > 1.0 implies that subjects carrying the antigen are at an increased risk of a disease (a positive association), while an OR value < 1.0 implies that subjects carrying the antigen are at a lower risk of a disease than those not carrying the antigen (a negative association).

To assess the association of HLA alleles with the severity of our SARS cases, SARS patients were divided into assisted ventilation (mask and intubation) (n = 30) and non‐assisted ventilation groups (n = 65). The patients were also grouped according to the occurrence of lymphocytopenia (< 1 × 109 L−1) (n = 63). Four grades were scored to indicate the area of lung infiltration as indicated by the chest radiographs: grade 1 indicated 1/4 (n = 25), 2 for 2/4 (n = 58), 3 for 3/4 (n = 10) and 4 for 4/4 (n = 2). The association between lymphocytopenia and HLA variation was analysed using a one‐way anova using the SNK test for multiple comparisons. Mordridit score analysis was applied when comparing associations between lung infiltrates and HLA. All analyses were performed using spss 11.0. All probabilities were two tailed with a P‐value < 0.05 considered statistically significant.

Results

The allelic frequencies of HLA‐A, ‐B and ‐DRB1 variants in the 95 SARS patients and 403 healthy control subjects are listed in 1, 2, 3, respectively. The Fisher's exact test showed that in SARS patients the frequencies of HLA‐A23 [2.1%, P = 0.0361, OR, 21.6 (95%CI, 0.39–36.90)], ‐A34 [2.1%, P = 0.0361, OR, 21.6 (95%CI, 0.39–36.90)], ‐B60 [37.9%, P = 0.0061, OR, 1.9 (95%CI, 0.23–2.13)] and HLA‐DRB1*12 [32.6%, P = 0.046, OR, 1.6 (95%CI, 0.20–2.01)] were higher than the control group, whereas the frequencies of HLA‐A33 [12.6%, P = 0.02, OR, 0.5 (95%CI, 0.39–1.37)], HLA‐B58 [11.6%, P = 0.027, OR, 0.5 (95%CI, 0.07–1.41)] and HLA‐B61 [1.05%, P = 0.0176, OR, 0.2 (95%CI, 0.01–2.05)] were lower. However, after correction of the P‐value by multiplying the number of alleles at each locus, all statistical significance was lost. With regard to the association of HLA polymorphism with severity of SARS, the auxiliary ventilation, areas of lung infiltrate and lymphocytopenia all seemed to occur randomly with HLA‐A, ‐B, and ‐DRB1 alleles, since none of the associations tested were significant (data not shown).

Table 1

Allelic frequencies of human leucocyte antigen (HLA)‐A in SARS patients and healthy controls

HLA‐A Allele	SARS		Normal		OR (95%CI)	P	Pc
	n	%	n	%
1	5	5.26	11	2.73	2.07 (0.50–3.75)	0.349
2	49	51.58	207	51.36	1.01 (0.64–1.56)	0.97
3	2	2.11	17	4.22	0.59 (0.23–2.76)	0.503
11	54	56.84	207	51.36	1.24 (0.70–1.72)	0.336
23	2	2.11	0	0	21.58 (0.39–36.90)	0.0361	ns
24	25	26.32	103	25.56	1.05 (0.62–1.69)	0.879
26	4	4.21	13	3.23	1.42 (0.41–3.32)	0.872
29	2	2.11	7	1.74	1.41 (0.30–4.46)	1
30	7	7.37	24	5.96	1.31 (0.49–2.58)	0.608
31	4	4.21	14	3.47	1.32 (0.40–4.21)	0.968
32	0	0	7	1.74	0.28 (0.07–4.63)	0.418
33	12	12.63	95	23.57	0.48 (0.39–1.37)	0.02
34	2	2.11	0	0	21.58 (0.39–36.90)	0.0361	ns
68	1	1.05	7	1.74	0.83 (0.19–4.43)	0.981
74	0	0	2	0.50	0.84 (0.10–9.01)	1

n, number of individuals carrying a given allele; P, P‐value; Pc, corrected P‐value; OR, odds ratio; ns, not significant.

Table 2

Allelic frequencies of human leucocyte antigen (HLA)‐B in SARS patients and healthy controls

HLA‐B Allele	SARS		Normal		OR (95%CI)	P	Pc
	n	%	n	%
7	1	1.05	13	3.23	0.46 (0.16–3.19)	0.419
8	1	1.05	7	1.74	0.84 (0.19–4.43)	0.981
13	20	21.05	86	21.34	1.00 (0.58–1.71)	0.951
14	1	1.05	1	0.25	4.26 (0.26–13.49)	0.345
16	9	9.47	38	9.43	1.04 (0.49–2.12)	0.989
18	0	0	4	0.99	0.46 (0.08–6.21)	1
27	2	2.11	16	3.97	0.63 (0.23–2.85)	0.568
35	7	7.34	18	4.47	1.77 (0.54–3.02)	0.366
37	3	3.16	6	1.49	2.31 (0.41–5.02)	0.503
44	1	1.05	18	4.47	0.33 (0.14–2.70)	0.206
45	0	0	3	0.74	0.60 (0.09–7.24)	1
46	18	18.95	99	24.57	0.73 (0.50–1.51)	0.245
47	0	0	2	0.50	0.84 (0.10–9.01)	1
48	5	5.26	16	3.97	1.43 (0.45–3.04)	1
50	0	0	4	0.99	0.46 (0.08–6.21)	1
51	11	11.58	47	11.66	1.02 (0.51–1.98)	0.982
52	3	3.16	17	4.22	0.84 (0.31–2.80)	0.855
53	1	1.05	0	0	12.81 (0.21–33.73)	0.191
54	5	5.26	16	3.97	1.43 (0.45–3.04)	0.779
55	4	4.21	15	3.72	1.23 (0.39–3.07)	1
56	0	0	11	2.73	0.18 (0.06–3.69)	0.215
57	0	0	5	1.24	0.38 (0.08–5.52)	0.589
58	11	11.58	87	21.59	0.49 (0.38–1.41)	0.027	ns
60	36	37.89	97	24.07	1.93 (0.83–2.13)	0.0061	ns
61	1	1.05	31	7.69	0.19 (0.11–2.05)	0.0176	ns
62	17	17.89	48	11.91	1.63 (0.68–2.24)	0.119
63	0	0	1	0.25	1.40 (0.10–12.92)	1
67	1	1.05	5	1.24	1.15 (0.21–5.35)	1
71	1	1.05	7	1.74	0.84 (0.19–4.43)	0.981
72	1	1.05	0	0	12.81 (0.27–33.73)	0.191
75	10	10.53	49	12.16	0.88 (0.47–1.89)	0.658
76	0	0	1	0.25	1.40 (0.10–12.92)	1

n, number of individuals carrying a given allele; P, P‐value; Pc, corrected P‐value; OR, odds ratio; ns, not significant.

Table 3

Allelic frequencies of human leucocyte antigen (HLA)‐DRB1 in SARS patients and healthy controls

HLA‐DRB1 Allele	SARS		Normal		OR (95%CI)	P	Pc
	n	%	n	%
1	2	2.11	5	1.24	1.94 (0.33–5.43)	0.873
3	13	13.68	74	18.36	0.72 (0.47–1.61)	0.28
4	20	21.05	89	22.08	0.95 (0.57–1.68)	0.827
7	11	11.58	50	12.41	0.95 (0.50–1.92)	0.825
8	14	14.74	48	11.91	1.30 (0.60–2.10)	0.453
9	22	23.16	104	25.81	0.88 (0.56–1.59)	0.593
10	5	5.26	11	2.73	2.07 (0.50–3.75)	0.349
11	6	6.32	45	11.17	0.57 (0.34–1.80)	0.161
12	31	32.63	92	22.83	1.64 (0.77–2.01)	0.046	ns
13	6	6.32	45	11.17	0.57 (0.34–1.80)	0.161
14	12	12.63	63	15.63	0.80 (0.48–1.73)	0.462
15	19	20.00	114	28.29	0.64 (0.48–1.41)	0.1
16	12	12.63	35	8.68	1.55 (0.62–2.38)	0.237

n, number of individuals carrying a given allele; P, P‐value; Pc, corrected P‐value; OR, odds ratio; ns, not significant.

No deviation from Hardy–Weinberg equilibrium was observed when the frequencies of HLA alleles were tested in the SARS patients or the healthy controls (P > 0.05).

Discussion

Classical HLA classes I and II molecules are essential to present antigenic peptide (including viral peptide) for T‐cell recognition and thus initiate an adaptive immune response. To assist the survival of a population, the genetic cluster of HLA loci manifests a striking level of polymorphism to protect the host from any pathogen mutation eluding host immune surveillance. Antigenic peptides binding to different alleles of HLA molecules vary in their affinity, with the HLA allotype thus affecting the host immune response to a given antigen.

Studies have indicated that HLA variations are associated with susceptibility or resistance to a wide range of infections, including malaria, tuberculosis, leprosy, HIV and hepatitis virus persistence (De Vries ; Hill, 1998; Liu ). SARS has been shown to be caused by infection with SARS‐CoV, with major lung involvement and consequent high infectivity and mortality. Although many questions remain to be answered, immunological mechanisms are believed to be involved in the development of the disease. As previously noted, in the 2002–03 SARS epidemic, the disease was mostly confined to South‐East and East Asian populations, suggestive of greater susceptibility to SARS‐CoV.

Guangdong province, where the SARS epidemic started, has a population of 85.2 million. The first case of SARS that fulfilled the WHO case definition was reported in Foshan, a city about 20 km from Guangzhou (formerly Canton), the provincial capital (Zhong ; Xu ). Phylogenetic studies have indicated that the populations of Guangdong, Hong Kong and Taiwan populations largely share common origins (Chu ). For example, over 98% of the Hong Kong population are Chinese, and the majority have ancestral origins in neighbouring Guangdong province and are of Han Chinese ethnicity (Chang & Hawkins, 1997). The present‐day population of Taiwan comprises indigenous Taiwanese groups, and people of Minnan and Hakka ethnicity who are the descendants of settlers from the provinces of Fujian and Guangdong during the last few centuries (Lin ). It is of interest that the SARS cases in Taiwan mainly consisted of southern Chinese origin Taiwanese rather than the indigenous peoples (Lin ).

Similarities in the genetic profiles of these three populations, Guangdong, Hong Kong and Taiwan offers the opportunity of identifying possible associations between HLA alleles and SARS infection. Hypothetical SARS‐associated HLA alleles among the three populations are listed in Table 4, and it is obvious that the SARS‐associated HLA alleles that have been reported in any single population could not be verified in either of the other two populations. It also seems that the Guangzhou population was more susceptible to the SARS epidemic irrespective of their HLA profiles.

Table 4

Human leucocyte antigen (HLA) alleles reported suspect being associated with SARS in three populations

HLA allele Population	SARS		Control		OR (95%CI)	P	Pc
	n (N)	%	n (N)	%
B*4601/B46
Taiwan	15 (66)	22.72	52 (380)	13.68	1.86 (1.02–3.54)	0.06
(A)
(B)	15 (66)	22.72	25 (202)	12.38	2.08 (1.04–4.24)	0.04	ns
(C)	6 (10)		25 (202)	12.38	10.62 (2.80–40.26)	0.0008	0.0279
(D)	6 (10)		9 (56)	16.07	7.83 (1.83–33.47)	0.007	ns
Hong Kong	27 (83)	32.5	5125 (18 774)	27.3	1.28 (0.81–2.04)	0.3229
Guangzhou	18 (95)	18.95	99 (403)	24.57	0.73 (0.50–1.51)	0.245.
B*1301/B13
Taiwan	1 (66)	1.515	34 (380)	8.947	0.16 (0.02–0.97)	0.02	ns
Hong Kong	20 (83)	24.1	3298 (18 774)	17.6	1.49 (0.90–2.47)	0.1466
Guangzhou	20 (95)	21.05	86 (403)	21.34	1.00 (0.58–1.71)	0.951
B*0703/B7
Taiwan	NA		NA
Hong Kong	9 (83)	10.8	544 (18 774)	2.9	4.08 (2.03–8.18)	0.00072
Guangzhou	1 (95)	1.05	13 (403)	3.23	0.46 (0.16–3.19)	0.419
DRB1*0301
Taiwan	NA		NA
Hong Kong	1 (79)	1.3	42 (250)	16.8	0.06 (0.01–0.47)	0.00008
Guangzhou	13 (95)	13.68	74 (403)	18.36	0.72 (0.47–1.61)	0.28

In Taiwan study, HLA allele frequencies were estimated by direct counting. For HLA–B46 association: (A) 33 cases of probable SARS patients compared with 190 healthy unrelated controls; (B) 33 cases of probable SARS patients with 101 high risk non‐infected health care workers; (C) 5 cases of ‘severe cases’ (deceased or intubated) probable SARS patients with 101 high risk non‐infected health care workers; (D) 5 cases of ‘severe cases’ (deceased or intubated) probable SARS patients with 28 ‘excluded fever patients’. For the other alleles listed, 33 cases of probable SARS patients compared with 190 healthy unrelated controls.
n, number of individuals carrying a given allele; N, number of cases tested; P, P‐value; Pc, corrected P‐value; OR, odds ratio; ns, not significant; NA, not available.

The mutation of SARS‐CoV could give rise to variations and discrepancies in the results of the HLA association studies in the three populations, and large‐scale studies are essential for the demonstration of an association of HLA variation with susceptibility to a disease. On the other hand, mutation of SARS‐CoV might result in changes of its T‐cell receptor epitope, which would require a specific HLA allele(s) for presentation. Because of the poor fidelity of RNA‐dependent RNA polymerase, genetic variation typically gives rise to a heterogeneous virus pool in RNA populations, including coronaviruses. This feature makes viruses highly adaptable and contributes to difficulties in preventing and controlling viral diseases. The virus isolated from the Guangzhou patients is the prototype of the SARS‐CoV, with a reported moderate mutation rate, that subsequently was found in Hong Kong and countries around the world (Zhao ). The Guangzhou SARS‐CoV has the highest homology (> 99%) with the sequences of the SARS‐CoV that were isolated from SARS cases in Hong Kong, Beijing, Singapore, Canada and the USA, suggesting that all of these SARS‐CoV are closely related (Zhong ), therefore mutation of SARS‐CoV could not be considered as a major contributory factor in the lack of HLA association observed in the present study.

During the SARS episode in Guangdong, six of the 95 SARS patients died of the disease. No samples from these deceased patients were available for our study. However, a history of lymphocytopenia, lung involvement and auxiliary ventilation were considered as surrogate measures of the severity of the disease, and no significant association of HLA variations with these parameters was observed among the patients, who instead displayed a random distribution of HLA‐A, ‐B and ‐DRB1 alleles.

In conclusion, the confined regional spread of the 2002–03 SARS epidemic appears to have been mostly dependent on factors such as relative population density, the frequency and extent of travel of affected individuals, and the rapid introduction and uptake of protective measures, rather than to HLA variation in the population.