|Home | About | Journals | Submit | Contact Us | Français|
China is facing a rapid upsurge in cases of human immunodeficiency virus type 1 (HIV-1) and hepatitis C virus (HCV) infection due to large numbers of paid blood donors (PBD), injection drug users (IDU), and sexual partners of infected individuals. In this report, a total of 236 HIV-1-positive blood samples were collected from PBD, IDU, and their sexual partners in the most severely affected provinces, such as Henan, Yunnan, Guangxi, and Xinjiang. PCR was used to amplify the p17 region of gag and the C2-V3 region of env of HIV-1 and the 5′ noncoding region and a region of E1/E2 of HCV. Genetic characterization of viral sequences indicated that there are two major epidemics of HIV-1 and multiple HCV epidemics in China. The PBD and transfusion recipients in Henan harbored HIV-1 subtype B′, which is similar to the virus found in Thailand, and HCV genotypes 1b and 2a, whereas the IDU in Yunnan, Guangxi, and Xinjiang carried HIV-1 circulating recombinant forms 07 and 08, which resemble those in India, and HCV genotypes 1b, 3a, and 3b. Our findings show that the epidemics of HIV-1 and HCV infection in China are the consequences of multiple introductions. The distinct distribution patterns of both the HIV-1 and HCV genotypes in the different high-risk groups are tightly linked to the mode of transmission rather than geographic proximity. These findings provide information relevant to antiviral therapy and vaccine development in China and should assist public health workers in implementing measures to reduce the further dissemination of these viruses in the world's most populous nation.
China appears to be on the brink of a tremendous upsurge in cases of human immunodeficiency virus type 1 (HIV-1) infection (12). The most recent statistics released by the Chinese government in November 2003 show that the estimated number of infected individuals has increased to 840,000 (1). However, many health experts working in China believe that this number is a significant underestimate and that the epidemic is even more widely spread (2, 12). Although this prevalence (0.06%) of infection seems remarkably low relative to China's population of 1.3 billion, the Joint United Nations Program on HIV/AIDS projects that if the epidemic were left unchecked, the number of HIV-1-infected individuals could reach 10 million or more by 2010 (3). Similar statistics for hepatitis C virus (HCV) infection in China are lacking at this time, but it is feared that the upsurge in HCV infection could be as severe as that of HIV-1.
Among those at high risk for HIV-1 infection in China, intravenous drug users (IDU) appear to have been the major group affected first. They now account for 68% of reported infections in China (2). Many of these IDU reside in Yunnan Province, in close proximity to the Golden Triangle, where a high prevalence of HIV-1 infection was found due to needle sharing (Fig. (Fig.1A)1A) (19, 33, 42, 45, 46). Since the first reports of HIV-1 infection in 1989, multiple viral genotypes have been detected in Yunnan. In 1990, subtype B strains were found in 90% of IDU, but this subtype was gradually replaced by subtypes B′, C, and E (now called circulating recombinant form 01 [CRF01]) (6, 9, 10, 16, 17, 31). By 1996, multiple genotypes coexisted in this population, providing an ideal scenario for HIV-1 to recombine (6, 9, 10, 16, 17, 30). It is therefore not surprising that recombinant forms are becoming predominant in some parts of Yunnan (10, 30, 35, 36). Subsequently, HIV-1 has spread through drug trafficking not just to the neighboring province of Guangxi (Fig. (Fig.1A)1A) (8, 20, 38) but also to the northwest province of Xinjiang (Fig. (Fig.1A)1A) (2, 3, 30).
HIV-1 infection of PBD in the central province of Henan (Fig. (Fig.1A)1A) and adjacent areas constitutes the second major epidemic in China (29a, 32, 34, 43, 44). This outbreak has generated a tremendous amount of media attention (5, 12, 23-25). However, few scientific studies have been focused on this particular group of patients. Unsanitary blood collection in this region began in the middle of 1980s and peaked between 1992 and 1996, when the government banned paid blood donations. The unsanitary collection practices often involved the pooling of blood from several donors, followed by the removal of plasma and the reinfusion of cells into donors (12). These pooled samples became increasingly contaminated with HIV-1 or HCV (5, 12, 25, 32). Such practices, in addition to the reuse of needles and unsterilized equipment, offered an efficient means of viral transmission among the local population. It is thus not surprising that the prevalence of HIV-1 infection in certain villages in Henan can be as high as 17 to 60% (12, 32, 34, 43, 44, 47). The problem of PBD was not confined to Henan but was also common in the neighboring provinces of Shanxi, Anhui, and Hubei (Fig. (Fig.1A)1A) (12). Although paid blood donors (PBD) accounted for only 9.7% of the officially reported HIV-1 infections in China (2, 12), it is widely believed that their numbers may well exceed half a million (2).
Despite the alarming spread of HIV-1 and HCV across China, there has been no large-scale study of the patterns of distribution of these viruses among various high-risk groups. The severity of HIV-1 and HCV coinfection has also not been assessed. In this report, we have systematically characterized both HIV-1 and HCV strains circulating among PBD and transfusion recipients in Henan and among IDU from Yunnan, Guangxi, and Xinjiang. Our findings show that in China there are two major epidemics of HIV-1 and multiple epidemics of HCV. The distinct distribution patterns of HIV-1 and HCV genotypes in the different high-risk groups are tightly linked to the mode of transmission rather than geographic proximity.
A total of 236 HIV-1-positive blood samples were collected between 1998 and 2001 from individuals seeking treatment at the local hospitals in severely affected provinces. The study was conducted according to guidelines set by the local ethical review committees. The study sites included Number 6 People's Hospital in Zhengzhou, Henan; Number 3 People's Hospital in Kunming, Yunnan; the Center for Disease Control Hospital in Nanning, Guangxi; Bayi Steel Corporation Hospital in Urumqi; and the Yining Prevention and Disease Control Station, Xinjiang. A total of 121 samples came from 15 cities in Henan (Fig. (Fig.1B);1B); 38 came from Kunming, Yunnan; 21 came from Nanning, Guangxi; and 56 came from Urumqi or Yining, Xinjiang (Table (Table1).1). The blood samples collected in Henan were from PBD and their sexual partners, whereas those in Yunnan, Guangxi, and Xinjiang were from IDU and their sexual partners (Table (Table1).1). A small number of PBD in Henan started their blood donation as early as 1985, but the majority (>90%) was actively involved between 1992 and 1996. Most PBD engaged in blood donation multiple times, some as much as 100 times or more. At the time of sampling, approximately 50% of HIV-1-positive PBD showed symptoms of AIDS or opportunistic infections. Many of the IDU from Yunnan and Guangxi were also symptomatic at the time of study. IDU from Xinjiang, however, were largely asymptomatic. Blood was separated into plasma and peripheral blood mononuclear cells by centrifugation through a Ficoll-Hypaque gradient. HIV-1 or HCV infection status was determined by an enzyme immunoassay and confirmed by a Western blot assay. Since only a limited number of peripheral blood mononuclear cell samples were available, plasma was used in most cases to obtain HIV-1 and HCV RNA for subsequent analysis. A further 13 HIV-1-positive blood samples were collected in Beijing (Fig. (Fig.1A)1A) from individuals who contracted HIV-1 sexually. An additional 50 HCV-positive blood samples were also collected from Queen Mary Hospital in Hong Kong, along with 4 HCV-positive samples from the neighboring province of Guangdong (Fig. (Fig.1A).1A). The risk factors for these individuals were blood transfusion and injection drug use.
The procedures for proviral DNA and viral RNA extraction and for cDNA synthesis have been previously described (39, 40). PCR amplification, nucleotide sequencing, and phylogenetic analyses using the program CLUSTAL W and the PHYLIP package were carried out as described previously (39, 40). Genetic distances were calculated with the Kimura two-parameter model, which allows different rates of transition and transversion and different frequencies of the four nucleotides (13). All RNA and DNA extractions and PCR amplifications were carried out with appropriate negative controls to detect possible contamination during the procedure. To study the genetic relationships and the potential for recombination among viral strains, we focused on two separate regions of both HIV-1 and HCV. For HIV-1, the p17 region of gag and the C2-V3 region of env were selected, both of which have been commonly used to determine the genetic relationships among HIV-1 strains (18, 22). For HCV, the 5′ noncoding region (NCR) and a region of E1/E2 were chosen because of the availability of a large number of such sequences in the database for comparative studies. The PCR primer-binding sites for both HIV-1 and HCV were based on the published sequences of geographical variants and were as highly conserved as possible. The primer sequences used for the p17 and C2-V3 regions of HIV-1 have been published (11, 39, 40). The primer pair used to amplify the HCV 5′NCR was previously described by Chan et al. (7), while those for the E1/E2 region are Lqz187 (antisense primer, 5′-CCYACBACMACDGGGCTNGGDGTGAARCARTA, used in both the first- and second-round amplifications [ambiguous nucleotides are expressed in International Union of Applied and Pure Chemistry nomenclature]), Lqz188 (first-round sense primer, 5′-CAYCGBATGGCHTGGGAYATGATGATGAA), and Lqz189 (second-round sense primer, 5′-TGGGAYATGATGATGAAYTGGTC). To check for potential contamination, prior to analysis the sequences obtained were compared to all known sequences in the HIV database by a BLAST search (http://hiv-web.lanl.gov/content/index) (15).
Nucleotide sequence accession numbers. The sequences have been submitted to GenBank. The accession numbers are AY731828 to AY731864, AY731866 to AY731946, and AY731948 to AY732099.
Of the 236 HIV-1-infected individuals, 157 were male and 79 were female; the age range was 23 to 43 years, with an average of 33 years (Table (Table1).1). Thirty-eight percent (89 of 236) were PBD or transfusion recipients from Henan, while 25% (60 of 236) were IDU from Yunnan, Guangxi, or Xinjiang (Table (Table1).1). The remaining 37% (87 of 236) were infected through sexual contact with PBD, transfusion recipients, or IDU. Overall, 57% of the total were coinfected with HCV. However, almost all of the coinfected individuals were PBD, transfusion recipients, or IDU, suggesting efficient transmission of both HIV-1 and HCV through these high-risk activities. Coinfection with HCV was relatively rare (4 to 6%) in persons who acquired HIV-1 via sexual contact (Table (Table11).
To study the pattern of distribution of HIV-1 genotypes, we obtained 122 p17 sequences and 116 C2-V3 sequences from PBD and transfusion recipients in 15 cities in Henan; from IDU in the cities of Kunming (Yunnan), Nanning (Guangxi), and Urumqi and Yining (Xinjiang); and from those who acquired infection through sexual contact in Beijing. We first estimated the evolutionary divergence for sequences obtained within and between each geographic location by analyzing the genetic distances using the Kimura model, which allows for different rates of transition and transversion. As shown in Table Table2,2, the mean nucleotide distances in p17 and C2-V3 varied substantially. Within Yunnan, Guangxi, Xinijiang, and Henan, the genetic distances were rather small, ranging from 1.5 to 5.9% for p17 and from 2.4 to 11.1% for C2-V3. In contrast, Beijing had the most divergent HIV-1 population, with mean genetic distances of 11.8% for p17 and 22.9% for C2-V3 (Table (Table2),2), levels of diversity normally seen between various HIV-1 genotypes (4, 11, 14, 18, 27-29). Finally, the mean genetic distances between different geographic regions was higher than that found within regions. This was particularly true when sequences from Henan and Beijing were compared with those from Yunnan, Guangxi, and Xinjiang (Table (Table2).2). On average, Henan sequences differed by 17.1% in p17 and 24.4% in C2-V3 from individuals in Yunnan, Guangxi, and Xinjiang, suggesting that IDU carry a very different strain of HIV-1 from those in PBD in Hunan (Table (Table2).2). Similar levels of genetic diversity were found when we compared sequences from Beijing to those from Yunnan, Guangxi, and Xinjiang. However, the most noteworthy finding was that the genetic distances between viruses from Yunnan, Guangxi, and Xinjiang were relatively small, ranging from 4.2 to 5.1% in p17 and 4.9 to 8.9% in C2-V3 (Table (Table2).2). Such low levels of genetic diversity are commonly found within an infected individual or between known transmission pairs (4, 11, 27-29). These findings suggest that HIV-1 strains in IDU and their sexual partners in these three provinces are closely linked, possibly originating from a single founder virus.
A phylogenetic approach was also used to describe the relationships between HIV-1 strains from these regions. Figure Figure22 depicts neighbor-joining trees for p17 and C2-V3 sequences obtained from Yunnan, Guangxi, Xinjiang, Henan, and Beijing, each of which is color coded. For clarity, only a few representative sequences from each region are shown, together with reference sequences from different HIV-1 subtypes. The most obvious and consistent result from the phylogenetic analysis was that sequences from Yunnan, Guangxi, and Xinjiang, despite being geographically separated, formed a single cluster for both p17 and C2-V3 (Fig. (Fig.2).2). This conclusion is in complete agreement with that inferred from an inspection of the genetic distances (Table. (Table.2),2), again suggesting that these strains circulating in IDU and their sexual partners may have a single origin. By comparison with reference subtypes, we conclude that the majority of the sequences from these three provinces resemble subtype C, with a particularly close relationship to CRF07 and -08. CRF07 and -08 are recombinants between subtype B′ and C, first described in Yunnan and Guangxi (20, 30). These viruses are more closely related to subtype C strains found in India than to those in African countries such as Zambia, Ethiopia, Uganda, or Brazil, suggesting the possibility of Indian origin. A few sequences from Yunnan and Guangxi clustered with subtype A, B, or B′ in p17 or with subtype E (CRF01) in C2-V3 (Fig. (Fig.22).
All HIV-1 sequences from Henan formed a completely different genetic group from those obtained from Yunnan, Guangxi, and Xinjiang, regardless of the region analyzed (p17 or C2-V3). Sequences from Henan grouped with subtype B′ sequences from Thailand and with the B′ strain (YN.RL42) first identified in Yunnan in 1991 (10). Furthermore, despite being derived from 15 different cities within Henan (Fig. (Fig.1B),1B), these sequences formed a tightly linked population, with an average genetic distance similar to that seen within a single infected person. There was no evidence of geographic clustering of sequences from a particular city. Such a pattern of distribution of viral sequences leads us to conclude that the HIV-1 strains present in PBD, transfusion recipients, and their sexual partners in Henan are likely to have a single origin, possibly arriving from Yunnan, where the B′ sequence was predominant in the early 1990s (10).
Not surprisingly, sequences from Beijing were the most divergent for both p17 and C2-V3. Of the 17 samples collected, three subtypes (A, B, and B′) in p17 and six subtypes (A, G, E, C, B, and B′) in C2-V3 were found (Fig. (Fig.2),2), suggesting multiple introductions of HIV-1 into Beijing. Many of the sequences were closely related to the B′ sequences from PBD and transfusion recipients or to CRF07 and CRF08 from IDU, indicating that these strains from distant provinces have already spread to the capital city. Although many sequences grouped within subtype B, they were quite distinct from the typical North American subtype B sequences (Fig. (Fig.2).2). The consensus Beijing subtype B amino acid sequence had a unique motif at the tip of the V3 loop, IHVGWGRS. In the entire HIV sequence database (6,980 sequences) at Los Alamos National Laboratory, this sequence occurred only once. It is impossible to predict whether this unusual signature sequence will become more prevalent among the patients infected in Beijing in the future. However, this unique sequence may serve as a convenient marker for future molecular epidemiologic studies. By comparing the clusters of sequences from the same patient in both p17 and C2-V3 phylogenetic trees, we identified four sequences (BJ-3, −4, −11, and −16) from Beijing as intersubtype recombinants (Fig. (Fig.2).2). The four individuals from which these sequences were derived reported sexual relationships either with infected individuals from Africa (BJ-16, A-G recombinant) or with individuals frequently traveling to Yunnan (BJ-11, B-C recombinant), to Thailand (BJ-4, A-E recombinant), or abroad (BJ-3, B′-B recombinant) (Fig. (Fig.2).2). Two further recombinant strains were found in Yunnan (YN-23, A-E recombinant) and Guangxi (GX-7, B′-C recombinant) (Fig. (Fig.22).
To study the pattern of distribution of HCVs, we obtained 103 5′NCR sequences and 89 E1/E2 sequences from IDU in Yunnan, Guangxi, and Xinjiang and from PBD and transfusion recipients in Henan. An additional 27 5′NCR sequences and 18 E1/E2 sequences were obtained from Hong Kong and the neighboring province of Guangdong for comparison. Figure Figure33 depicts the neighbor-joining tree for E1/E2 sequences; the samples are color coded in the same way as for the HIV-1 analyses. Among the 12 sequences from Yunnan and Guangxi, several genotypes, including 1a, 1b, 3a, and 3b, with one strain from Guangxi grouped loosely with genotype 6, were identified (Fig. (Fig.3).3). Sequences from Xinjiang grouped with three distinct genotypes, 3a, 3b, and 1b, characterized by a long branch on the tree (Fig. (Fig.3).3). The intragenotype genetic distance was 5.7% for genotype 3a, 4.5% for 3b, and 3.7% for 1b. Sequences in genotypes 3a and 3b have a close genetic relationship with those from Yunnan and Guangxi. Sequences in genotype 1b, however, grouped together with one sequence from Yunnan and many sequences from PBD in Henan (Fig. (Fig.3).3). These results suggest that despite the considerable geographic distance, HCV strains in Xinjiang are related to those in Yunnan and Guangxi. This clustering pattern suggests that HCV transmission must have occurred on at least three separate occasions, giving rise to genotypes 1b, 3a, and 3b, seen among IDU in Xinjiang. Sequence analysis of the 5′NCR gave the same pattern of relationships (data not shown).
Sequences from PBD and transfusion recipients in Henan clustered with HCV genotypes 1b and 2a. In particular, the sequences of genotype 2a formed a tightly related group, with an intragenotype diversity of 7.9%. The genotype 1b sequences could be further subdivided into four distinct clusters, each characterized by minimal diversity and distinguishable from other clusters by high bootstrap values (Fig. (Fig.3).3). This finding implies that there were at least five independent introductions of HCV into the PBD population. Examination of the geographic origins of these sequences showed that genotype 2a was largely restricted to the northeastern Henan cities of Xinxiang, Zhengzhou, Kaifeng, Shangqu, and Zhouko and that genotype 1b was found in almost all the cities studied (Fig. (Fig.1B).1B). Within the subgroups of genotype 1b, no geographic clustering was found. At this stage, it is not possible to determine the origins of these viruses among PBD and transfusion recipients. They may have been introduced from Yunnan at the same time when the HIV-1 B′ strain was introduced, or they may simply be the result of the expansion of preexisting strains during the blood collection process. Further studies of patients infected with HCV alone may allow us to test these two hypotheses. Last, sequences from Hong Kong (Fig. (Fig.3)3) were the most divergent, with three distinct genotypes, 1b, 2a, and 6, found among 14 E1/E2 sequences. Four samples obtained from the province of Guangdong (Fig. (Fig.3)3) were found to contain three genotype 6 sequences and one genotype 1b sequence, showing that genotype 6 is not confined to Hong Kong (21, 26, 41).
We have performed the first large-scale molecular epidemiologic study of HIV-1 and HCV infection among IDU, PBD, transfusion recipients, and their sexual partners in China. We found that both HIV-1 and HCV have already spread far beyond the southwest provinces of Yunnan and Guangxi into the distant northwest province of Xinjiang and the central province of Henan. We have shown that PBD, transfusion recipients, and their sexual partners in Henan harbor predominantly HIV-1 subtype B′ and HCV genotypes 1b and 2a. IDU and their sexual partners in Kunming, Nanning, Yining, and Urumqi carry HIV-1 CRF07 and -08 and HCV genotypes 1b, 3a, and 3b. This distribution pattern of HIV-1 and HCV strains is more due to the mode of transmission than to geographic proximity. Viruses within each high-risk group were genetically very similar and phylogenetically linked, suggesting recent separate outbreaks of HIV-1 and HCV infection. A limited number of sequences were available from individuals infected by homo- or heterosexual transmission in the capital city, Beijing. In this group, multiple subtypes of HIV-1 were found, some of which were related to strains seen in IDU, PBD, and transfusion recipients, suggesting that these strains of HIV-1 have spread into the general population. Taken together, our findings show that the epidemics of HIV-1 and HCV infection in China are the consequences of multiple introductions.
The current distribution pattern of HIV-1 among IDU, PBD, transfusion recipients, and their sexual partners is a reflection of the highly dynamic turnover of viral strains over the past 15 years in Yunnan. The initial HIV-1 epidemic in China was identified among the IDU in Yunnan in 1989 and was caused by a mixture of viruses closely resembling North American subtype B and Thai subtype B (B′) (10). Subsequently, the B′ subtype began to dominate, increasing from 20% of all infections detected in 1990 to 90% in 1996 (10, 31). During the same period, a second epidemic began among IDU in Yunnan, with strains genetically related to subtype C viruses from India (16, 17). In 1994, China began to experience an exponential increase in the number of infected individuals, an expansion in the complexity of virus genotypes, and the detection of infections far beyond Yunnan province (2, 3, 12). Multiple recombinant forms then emerged. A circulating recombinant form between subtypes A and E (CRF01) from a commercial sex worker in Yunnan who had worked in Thailand for a prolonged period was described (9). CRF07 and -08 were found along the drug trafficking route in Yunnan (30). By 1996 and 1997, an HIV-1 outbreak was reported among IDU in the neighboring province of Guangxi, where viruses from one city were found to be closely related to those in Yunnan (20, 37, 38). We believe that observed low levels of divergence in Yunnan in this study may be due to multiple causes. First, it may be reflective of intrinsic slow intragenotypic divergence of the specific CRF (CRF07 and -08) during the course of transmission through IDU. Second, it may also be due to the fact that samples were collected in only one hospital in the capital city, Kunming, despite the fact that patients came to this hospital from different and distant places. Nevertheless, the level of divergence in Yunnan (0.0592) is still significantly higher than that in Guangxi (0.0257) and Xinjiang (0.0145). However, compared to what has been found in Beijing, it is clear that Beijing is the most divergent place as far as HIV-1 is concerned since multiple genotypes were identified within a limited number of samples collected. By 1998, HIV-1 infection had been reported from all 31 provinces in China and from numerous autonomous regions and municipalities (12). It is impossible to describe the chain of transmission events that led to epidemics in Guangxi, Xinjiang, and Henan. However, the close relationship between viruses found in Yunnan, Guangxi, and Xinjiang in our study strongly argues that one major outbreak of HIV-1 among Chinese IDU is likely to have a common origin.
Furthermore, it seems that CRF07 and -08 strains and the B′ strain were transmitted to those inland provinces through individuals with different risk behaviors in different time periods. As subtypes B and B′ in Yunnan were replaced by other subtypes during the mid-1990s (10), it is likely that the dissemination of subtype B′ into Henan occurred in the early 1990s, earlier than the transmission of CRF07 and -08 into Guangxi and Xinjiang. This hypothesis is supported by the relatively higher degree of genetic diversity in both p17 and C2-V3 observed for viruses from Henan (4.4 and 6.7%, respectively) than for those from Guangxi (2.6 and 2.4%) or Xinjiang (1.5 and 3.8%) (Table (Table2).2). Moreover, this period of infection is consistent with the time when unsanitary blood collection practices became common in 1992.
The molecular epidemiology of HCV among IDU, PBD, and transfusion recipients, by contrast, is more complicated than that of HIV-1. This fact may be due in part to the history of the HCV epidemic in China. The observed distribution patterns of HCV genotypes among different high-risk groups in our study may be the result of expansion of preexisting strains or novel introductions, just as with HIV-1. For IDU, our results favor the latter possibility, as evidenced by the almost superimposable distribution patterns of these two viruses (Fig. (Fig.22 and and3).3). However, in the cases of PBD and transfusion recipients, the presence of at least five distinct HCV clusters and the lack of an immediate connection with the HCV seen in Yunnan support the hypothesis of an expansion of preexisting HCV strains. Further studies of Henan patients infected with HCV alone may allow us to clarify the situation.
In conclusion, our study helps to highlight the severity of HIV-1 and HCV infections in China. Although we have focused on IDU, PBD, transfusion recipients, and their sexual partners, our observations with regard to Beijing samples suggest that it is likely that HIV-1 and HCV are beginning to spread into the general population, most probably through sexual transmission. Even though paid blood donations have been outlawed in China since 1996, the tragedy of the HIV-1 epidemic in Henan and adjacent provinces has only recently been revealed. Furthermore, the complications that are likely to result from chronic HCV infection have yet to be discussed. There is no doubt that the Chinese Ministry of Health will endeavor to make sure that “blood heads” will not return to practice their illegal blood collection businesses. We also hope that our findings will be useful to public health workers who are implementing measures to reduce the further dissemination of these viruses in the world's most populous nation.
We are indebted to the patients for their participation and to Jerry Zaharatos for critical reading of the manuscript.