Impact of worker emigration on HIV epidemics in labour export areas: a molecular epidemiology investigation in Guangyuan, China

We aimed to investigate the molecular epidemic characteristics and viral transmission patterns of HIV-1 in a typical labor export area, Guangyuan city, China. Based on conducting phylogenetic trees and molecular transmission networks, a phylogenetic analysis was performed on HIV-1 pol sequences obtained from 211 migrant-history workers, 83 non-migrant-history individuals, and 21 migrant-history unknown individuals between January, 2012 and February, 2017 in Guangyuan city. Phylogenetic analysis revealed that CRF07_BC (48.3%, n = 152) and CRF01_AE (33.3%, n = 105) were the dominant strains in Guangyuan city, and circulated by multiple lineages with various epidemic characteristics. Geographic network analysis showed that Guangyuan city-related sequences with 20.3% CRF07_BC and 28.3% CRF01_AE were linked to that of other provinces, compared to that with 1.7% CRF07_BC and 5.0% CRF01_AE in cities of Sichuan. Molecular transmission network analysis further illustrated that migrant-history workers linked more sequences from other provinces than non-migrant-history individuals in both CRF07_BC (29.3% versus 0.0%, P = 0.013) and CRF01_AE (40.5% versus 10.0%, P = 0.001) networks. Our results highlighted that migrant-history workers in recent year played a vital role in fueling HIV-1 epidemic in Guangyuan city. Molecular transmission network analysis could be a useful approach for disclosing the transmission mechanism of HIV, which should be used in prevention and intervention efforts.

By the end of December 2017, approximately 758,610 people were living with HIV-1 in China, with an estimated 321,233 cases of AIDS. Among the 134,512 newly diagnosed HIV-1 infections in 2017, 69.4% was attributed to heterosexual and 25.3% to homosexual transmission 1 . One of the current challenges was that China faced the rapid increase of HIV-1 prevalence among floating population and lacking effective intervention strategies for this group 2,3 . With the rapid urbanization of China, outgoing migrant-workers had become the main component of the floating population in China, who worked outside the place where the household registrations were located 4 . The migrant-workers accounting for a large proportion of HIV-1 epidemic in China had caused the diversity of circulating HIV-1 and could further local fuel epidemics 5,6 . Although several molecular epidemiological investigations conducted in Shanghai, Beijing, Shenzhen, and others indicated that migrant-workers had a significant influence on the HIV-1 epidemics in coastal or metropolis 2,7,8 , barely researches focused on the condition of rural areas where export labors.
The newly diagnosed cases of HIV-1 in Sichuan province ranked first in China in 2017, and the prevalence of HIV-1 varied widely in regions 1 . The presence of relatively poor and developing towns and counties had made Sichuan become an important labor-exporting province, with a rapid increase in the prevalence of HIV-1. Guangyuan city, located in northern Sichuan province in western China (Fig. 1), was a mountainous agricultural city with 3.05 million registered populations currently. It was also renowned for its large output of labors in Sichuan. Every year, approximately one million migrant-workers were far away from their homes to seek better employment opportunities, including 21% inside and 54% outside their home province. Above 50% of them had become major migrant labors in Yangtze (mainly including Jiangsu and Shanghai), Zhujiang River delta areas (mainly located in the southeast of Guangdong), and the Bohai Rim (mainly including Beijing and Tianjin) 9,10 . Since the first HIV-1 case was identified in 1996 in Guangyuan city, the cumulative number of reported HIV-1-infected persons had reached 978 in 2017 (from a report of Guanyuan Municipal Center for Disease Control and Prevention[CDC]). In recent six years, the number of newly diagnosed HIV-1 infections increased from 81 in 2012 to 196 in 2017, showing a serious situation in HIV-1 prevalence. Therefore, we characterized HIV-1 genetic diversity and transmission network in this labor-exporting representative city, Guangyuan city, and tried to explore a possible mechanism for controlling the spread of HIV-1 in this kind of areas. Sequence data collection. A total of 10 mL whole blood samples were collected from the 434 studied participants. None of the individuals was exposed to antiretroviral treatment (ART) at the time of blood specimen collection. Plasma was separated from the whole blood within two hours after collection and stored at −80 °C until use. Blood plasma samples were subjected to viral RNA extraction and cDNA synthesis as previously described 11  Migration condition data collection. The 315 individuals with pol sequences were further conducted a questionnaire survey after sequence data collection. Migration condition data were collected through face-to-face or telephone interviews conducted by the staffs of CDC. These indicators included whether the subjects had gone out as migrant worker before diagnosed, age of the first going out as migrant worker, occupation of going out as migrant worker, and place of migrant working. 294 questionnaires were collected. Identification of HIV-1 subtypes and CRFs.

Identification of HIV-1 epidemic lineages.
For further analysis of HIV-1 epidemic lineages of predominant variants circulating in Guangyuan city, a BLAST-based search tool (https://www.hiv.lanl.gov/content/ sequence/BASIC_BLAST/basic_blast.html) was used to search against the closely related sequences from locals and nationwide in Sichuan province CDC drug resistance database and LANL database. Two sets of databases, database CRF07_BC with 664 CRF07_BC pol sequences and database CRF01_AE with 584 CRF01_AE pol sequences were built for both phylogenetic and molecular transmission network analyses. A monophyletic group in ML phylogenetic tree(built as mentioned above) with bootstrap support ≥0.9 was considered as an epidemic lineage.
Analysis of HIV-1 molecular transmission network. The flow chart of transmission network analysis included four steps: construction of phylogenetic tree, extraction of transmission cluster, identification of minimum patristic genetic distance and visualization of the network 3 . We used Cluster Picker 14 to extract transmission clusters from the phylogenetic tree, with the intra-cluster maximum pairwise distance <3.0% nucleotide substitutions per site and bootstrap support value ≥95%. The patristic genetic distances of all sequences within the available clusters were calculated in Patristic software (http://www.bioinformatics.org/patristic/manual.html) (Supplementary Material 2). Among all distances, one that minimizes the sum of edge weights (patristic genetic distances) was selected to define the linkages within a cluster 3 . Lastly, the network data were visualized and analyzed using the network software Cytoscape 3.5 15 .

Statistical analysis.
Chi-square test was used to calculate the distributions of demographic information between CRF01_AE and CRF07_BC, and compare the difference of linkages between migrant-history workers and non-migrant-history individuals in transmission networks. A p-value less than 0.05 was considered statistically significant. All statistical analyses were performed using SPSS v.20.0 software (IBM Company, New York, USA).

Ethics statement. The study protocol was reviewed and approved by the Institutional Review Board at
the Human Medical Research Ethics Committee of the Sichuan CDC. The objectives and the procedure of the study, and potential risks and benefits of participating in the study were given to potential participants during the recruitment of study subjects. Verbal and written consent procedures were given to the study participants and they had the right to discontinue the survey at any time. All research methods in this study were carried out by the approved guidelines.

Results
The mean age of the 315 participants was 39.4 ± 14.0 years (range 18-78) and male-to-female ratio was 3 High proportion of the migration in Guangyuan city. A total of 67.0% participants reported that they had the experience of being migrant workers before diagnosed, and the mean age of the first going out as migrant workers was 26.9 ± 11.0 years (rang 15.0-70.0). Most of the subjects who ever went out as migrant workers (migrant-history workers) were outside the home province (54.5%), whereas 15.7% and 2.8% stated "inside the home province but outside Guangyuan city" and "inside Guangyuan city but outside the domicile". The main occupation of these migrant-history workers was worker (37.0%), followed by waiter (15.6%), company employee (7.6%), salesman (6.6%), driver (1.9%), and others (1.9%) ( Table 1).

Discussion
As the uneven economic development in China, massive migrant-workers traveled from the rural areas to urban centers to seek better employment opportunities and to improve their living conditions. By the end of 2017, the size of outgoing migrant-workers had reached 171.9 million nationwide, and the majority of them were living in some international metropolis, such as Beijing, Shanghai, Guangzhou, et al. 10 . The strong association between HIV-1 infection and migrant-workers was well-established [16][17][18] . Engaging in high-risk behaviors 5 made them more vulnerable to  HIV-1 than the general population 19 . In addition, the high mobility of migrant-workers makes it difficult to monitor HIV-1 infection and manage care 20 . Guangyuan city in Sichuan was well-known as a labor-exporting city with approximately one million migrant-workers every year 21 . Overall, we elucidated the characteristics of HIV-1 genetic diversity, viral evolutional lineages, and transmission networks in this city. Two distinguishing features of HIV-1 epidemics in Guangyuan city included: 1) there were close transmission linkages between Guangyuan city and other provinces or municipalities, especially some international metropolis, for which people who ever went out as migrant-workers probably played a crucial role; 2) the vast majority of male heterosexuals were observed being shared links with other male heterosexuals and MSM, implying that some MSM might hide their real sexual orientation.
In this study, we found a remarkably high portion of a migrant-history HIV reported cases in Guangyuan city, and most of the migrant-history workers were outside the home province. The studies in some single towns (e.g., Unite Kingdom and Uganda), showed that only small amount of the transmission events occurred locally [22][23][24] . To further identify the reliable linkages between HIV reported cases in Guangyuan city and other areas, a wide range of reference sequences were included for both phylogenetic analysis and transmission network analysis. As expected, our results revealed that Guangyuan city sequences had a close link with sequences form other cities. Furthermore, a much higher proportion of links were found with sequences derived from the Yangtze, Zhujiang River delta areas and the Bohai Rim than that from other cities of Sichuan. Based on the statistical data, an approximately above 50% migrant-workers of Guangyuan city constantly flow to the cities of these areas, such as Guangdong (106.9 thousand), Zhejiang (86.3 thousand), Shanghai and Jiangsu (62.7 thousand), as well as Beijing and Tianjin (61.1 thousand) 9 . Thus we inferred that the migrant-workers flow between their work place and home with not only money but also HIV-1, for their higher levels of sexual risk including unprotected sex 25 . And they might serve as a bridge and facilitated viral transmission from other provinces/cities to home regions.
Although heterosexual contact remained the domination of acquiring the HIV-1 infection in Guangyuan city, the proportions of MSM in newly diagnosed HIV-1 infections posed a persistent increase in recent years. Usually, urban areas had more attraction to MSM, since large cities had relatively open culture and convenient sexual venues (bars, saunas, parks, sex clubs, et al.) 18 . These MSM migrants were more at highest risk for HIV-1 than other individuals in the migrant-workers [26][27][28] . More disturbing, a number of MSM might hide their real sexual orientation in China, because of the traditional cultural, historical reasons, as well as societal stigma and discrimination 29 , which could be wildly underestimating the role of the MSM in the HIV-1 transmission networks. Especially, the frequent high-risk bisexuality and unprotected behaviors of these MSM make their female partners more vulnerable to HIV-1 30,31 . This also suggested that the real infected routes of these male heterosexuals in the large cluster deserved further investigation. Importantly, the low mean genetic distance in two lager clusters with recently diagnosed individuals in both CRF07_BC and CRF01_AE networks illustrated that the dynamic networks would probably continue to expand in the future if not intervened.
However, the incomplete coverage of sequences might make our results subject to selection/sampling bias. We anticipated being able to make even stronger inferences about the transmission network characteristic through improving the completeness of molecular surveillance data in the future.

Conclusions
We, for the first time, revealed the current HIV-1 epidemic in a hotspot area of Sichuan province that exported labors through analysis of viral diversity and transmission networks, and further underscored the value of molecular sequences survey combined with epidemiologic and demographic data in guiding precautionary intervention at the local level.