Epidemiological and phylogenetic analysis for non-B subtypes of human immunodeficiency virus type 1 in Busan, Korea

Recent data on non-B subtypes’ epidemiology among patients infected with human immunodeficiency virus-1 (HIV-1) in Korea are lacking. We aimed to assess the changing trends in the epidemiology of non-B subtypes of HIV-1 in Korea using phyloepidemiological analyses. We analyzed the demographic records and sequencing data obtained from genotypic drug resistance tests between 2005 and 2019 from 517 patients infected with HIV attending a tertiary care hospital in Busan, Korea. Subtyping and phylogenetic analyses with reference sequences were performed. Additionally, transmission clusters were identified via maximum-likelihood trees. Non-B subtypes accounted for 21.3% of the 517 sequences. CRF01_AE (52.7%) was the most common non-B subtype, followed by CRF02_AG (16.4%), A1 (11.8%), and C (5.5%). The prevalence of non-B subtypes decreased from 36.4 to 13.4% by 2009, while it increased to 27.4% between 2015 and 2019. Among patients with non-B subtypes, the proportion of overseas sailors decreased from 66.7 to 7.5%; contrarily, the proportion of men-who-have-sex-with-men (MSM) increased from 0 to 46.9% over the study period. We identified 8 transmission clusters involving non-B subtypes, with sizes ranging from 2 to 4 patients, including 3 clusters containing MSM. Our results highlight the changes in the epidemiological trends of non-B subtypes of HIV-1 in Korea.

www.nature.com/scientificreports/ decreasing 3 . According to the HIV surveillance data from the Korean Centers for Disease Control and Prevention (KCDC) in 2000, heterosexual contact was a major mode of HIV transmission and MSM only accounted for 28% 8 . In a recent nationwide HIV cohort study of Korea that included 1442 patients between 2006 and 2016, MSM accounted for 60.4% 9 . The proportion of MSM was higher in younger age groups, 71.5% in 18-29 years and 92.9% in 18-19 years, respectively 3 . Currently, more than 80% of the HIV epidemic in Korea is attributed to subtype B 6,10 . A recent nationwide study demonstrated that subtype B was identified in 93.1% of 927 samples collected between 1999 and 2012 6 . Furthermore, several previous studies reported that the Korean HIV epidemic was composed of a distinct monophyletic clade, Korean clade B (B k ), rather than the pandemic form of subtype B 6,10-12 . B k was estimated to have emerged less than 10 years after the introduction of subtype B in Korea. The B k epidemic rapidly increased until the early 1980s, followed by modest growth until the 1990s, after which there was a plateau 10,13 . MSM are increasingly identified as the major high-risk groups in the B k epidemic in Korea 10 . These findings showed that the HIV-1 subtype distribution in association with changes in the socio-demographic determinants of the HIV infection has dynamically changed over time in Korea. While the epidemiology of HIV-1 subtype B in Korea is well established, little is known about HIV-1 non-B subtypes. Therefore, to gain a deeper understanding of HIV-1 non-B subtype epidemics in Korea, we conducted a molecular epidemiological study using available sequence data collected from genotypic drug resistance tests between 2005 and 2019. We aimed to evaluate the changing trends in the epidemiology of HIV-1 non-B subtypes, identify transmission clusters, and characterize the people in those clusters in Korea 14 .

Results
Study population. The demographic and clinical characteristics of the 517 patients are summarized in  (Fig. 2). The subtype distribution within non-B subtypes had also changed over time. Before 2000, CRF02_AG was the most commonly identified subtype, followed by A1 and CRF01_AE. Since then, the CRF01_AE diagnoses continuously increased from 25.0% in 1985-1999 to over 57.5% in 2015-2019 (Fig. 2). Demographic changes were also observed over time (Fig. 3). Among the 75 men infected with non-B subtypes, the proportion of MSM significantly increased in recent years, from 12.5% in 2010-2014 to 46.9% in 2015-2019 (P = 0.007) (Fig. 3a).

Discussion
The southeastern region of Korea, including Busan, the largest international port city in Korea, was one of the most affected areas of the HIV-1-related early epidemic in Korea. Many HIV infections have been identified in OSs and their spouses 3 . Of the 103 HIV infections newly diagnosed in this region between 1988 and 1997, 81% of cases were transmitted through heterosexual contacts, and 80% of them were OSs or their spouses 15 . In a previous phylogenetic analysis of nef gene from 102 OSs and 14 of their spouses enrolled between 1985 and 1998, non-B subtypes accounted for 80% of the cases, and among them, CRF02_AG was the most common, followed by CRF01_AE and A1 5,6 . With the change in the epidemic pattern in Korea, the proportion of MSM in this region also increased over time, from 12.6% in 1990-1997 to 41.2% in 2002-2011 15 . In this study, approximately 80% of HIV-1 sequences were of subtype B, and MSM accounted for 67% of them, which is broadly in line with previous epidemiological studies in Korea 3,9,15 . www.nature.com/scientificreports/ However, we found that the epidemiological profile of HIV-1 non-B subtype infection in this region has been slowly changing over the decades. Trends showed that the proportion of non-B subtypes among the newly diagnosed HIV-1 patients initially decreased from 36.4% in 1988-1999 to 13.4% in 2005-2009 as the number of OSs and their partners gradually decreased, and the number of MSM relatively increased over time. The recent upward trend in the proportion of non-B subtypes may be partially explained due to the increasing numbers of OWs and immigrant workers from endemic countries with non-B subtypes. However, we also found that the proportion of MSM significantly increased in recent years, from 12.5% in 2011-2014 to 46.9% in 2014-2019.
In the phylogenetic analysis, MSM were more commonly included in clusters than heterosexual men and women. MSM in clusters are more likely to be younger and recently diagnosed with HIV than heterosexual men and women. Of the 18 MSM with non-B subtypes, 7 (38.9%) were included in 3 small clusters, with sizes ranging from 2 to 3 patients. Cluster 4 (CRF01_AE) included 3 young MSM and 2 of them reported homosexual contacts in Korea. Overall, 75% (9/12) of MSM with CRF01_AE were recently diagnosed with HIV between 2015 and 2019 and the majority of them reported no history of sexual activity overseas. These findings suggest that CRF01_AE is recently increasing in the MSM population in this region. Recent studies in Asian countries have shown that CRF01_AE strains have increased in their MSM population 16 . Recent studies in China have shown a dramatic shift in genotype distribution from subtype B to CRF01_AE among MSM 17,18 . Therefore, our findings underscore the importance of strengthening epidemiological surveillance in this region.
We also identified a small transmission pair harboring CRF56_cpx (cluster 7) that involved 2 young MSM with sequences sampled in 2019. This subtype was circulated in MSM in France and was identified in the Philippines 19,20 . Only one case (0.1%) was detected in 927 samples collected between 1999 and 2012 in a nationwide study in Korea 6 . Both clustered individuals were recently diagnosed with HIV and had no history of sexual activity overseas. Therefore, we speculate that this subtype may have sporadic local transmission, possibly in MSM, in Korea.
We also found a mixed cluster harboring CRF07_BC (cluster 8) that involved 2 MSM and a heterosexual man [21][22][23][24] . Two non-clustered patients with CRF07_BC were Chinese MSM whose sequences were sampled in 2016 and 2018, respectively. CRF07_BC mainly circulated among IDUs in China in early 2000, spread from IDUs to their sexual partners and further to the general population in China, particularly after a dramatic increase in the MSM population since 2009 25 . In 2003, CRF07_BC was identified in Taiwan, with likely migration between 1998 and 2001, and has also become the predominant strain among IDU in Taiwan 26,27 . CRF07_BC has also spread to some countries in South East Asia that share borders with China, possibly through drug trafficking routes [27][28][29] . This subtype was also rarely reported in Korea, with a prevalence of 0.1% (1/927) in 1999-2012 6 . Considering that our sequences in the CRF07_BC clusters were sampled between 2016 and 2019, our results suggest that the CRF07_BC is circulating in local transmission chains, possibly between different risk groups in Korea. Therefore, a further nationwide study is needed.
The proportion of women among patients with non-B subtypes have fluctuated over time. The majority of women reported that they were infected through heterosexual contact with HIV-positive husbands or male sexual partners when they were not aware of their partner's infection status. Their subtype distribution was mostly dependent on their partner's subtypes. In the phylogenetic analysis, 28% (7/25) of women were included in 4 small clusters that belonged to CRF01_AE, with sizes ranging from 2 to 4 patients. Two clusters included only women and the remaining 2 were heterosexual male/female clusters. A male/female cluster included a married couple, an OW, and his partner, but epidemiological connections could not be identified in the other 3 clusters.
In OSs and their partners, CRF02_AG and A1, mainly circulating among African and non-Chinese Asian countries, were most commonly found 5,30 . In the 1980-1990s, many Korean OSs were workers employed at global shipping companies and were sent abroad as skilled workers; some of them were infected with HIV, particularly in Africa 31 . We found an A1 pair (cluster 6) containing 2 OSs. However, there was no epidemiological link between the 2 OSs. On the other hand, CRF_01AE accounted for more than 80% of OWs and their partners, which may be associated with the increasing numbers of OWs to Southeast Asia and China 30 .
Our findings should be considered in the context of several limitations. First, this was a hospital-based, retrospective, observational study. The collection of epidemiological information might have been limited because of the retrospective nature of our study. Therefore, we cannot rule out the presence of unmeasured confounding factors. Second, our study was conducted at a single center, and a small number of HIV cases were included. Therefore, the results should be generalized to other regions of the country with caution. Although the proportion of OS and their spouse were higher in this region at the beginning of epidemic, recent trends of increasing of MSM and decreasing age at HIV diagnosis are similar to those of nationwide study. During the study period, approximately 8.6% of the Korean population living with HIV infection visited the study hospital 2 . The southeastern region of Korea accounts for about 15% of all HIV infected patients in Korea 32 . Therefore, approximately 58% of HIV-infected patients in this region attended the study hospital and 43.2% of them were included for subtype analysis. Third, since the HIV-1 genotype resistance assay was routinely performed at the time of entry to care in our institution since 2011, some of the subjects treated before this time have not been included in our dataset, which might have resulted in selection bias. Many OSs were not included in the study due to death, transfer, or loss of follow-up before enrollment, or suppressed viral load due to ART at enrollment. In addition, selection bias was also possible because the sequences were only from persons engaged in clinical care who had an available genotype. Fourth, about 20% of the subjects were treatment-experienced patients, and their genotypic resistance tests were mostly performed when they returned to care after lost to follow-up, or some were performed at the time of treatment failure. Although there was no statistically significant difference in their proportion between subtype B group and non-B group, the patients who were diagnosed with HIV in the earlier years were more likely to be included in the treatment-experienced group, which might have resulted in selection bias. However, we included only treatment naïve patients in the phylogenetic cluster analysis. Fifth, Viroseq HIV-1 genotyping system used in this study is generally optimized for HIV-1 subtype B and has been www.nature.com/scientificreports/ used to sequence non-B subtypes with different genotyping sensitivities. It can also be affected by sensitivity of detection of different versions of the system 33,34 .

Conclusions
In this study, despite the dominance of HIV-1 subtype B infection in the Korean HIV population, we found that the proportion of non-B infections has recently increased. Changes in epidemiological profiles have also influenced the distribution of HIV-1 subtypes in Korea. HIV-1 non-B subtypes are no longer limited to infections acquired abroad, such as those in OSs, and are spreading to multiple transmission groups, including MSM. Our results also highlight the continuous expansion of HIV-1 non-B subtype diversity, suggesting introducing new subtypes in certain risk groups or sexual linkages between different subgroups. Implementing continued HIV-1 molecular surveillance is essential for understanding and monitoring transmission cluster dynamics to improve HIV prevention efforts in Korea.

Methods
Study population. Among 1,197 36 , and jpHMM-HIV (http:// jphmm. gobics. de/ submi ssion_ hiv, Institute of Microbiology and Genetics, University of Gottingen, Niedersachsen Germany) 37 . Additionally, we used HIV BLAST (https:// www. hiv. lanl. gov/ conte nt/ seque nce/ BASIC_ BLAST/ basic_ blast. html, Los Alamos National Laboratory, USA) to identify the nearest HIV-1 sequences from different geographical proximities. In the case of discordance between the various systems, we relied on the manual molecular phylogenetic analysis. The pol sequences were aligned with reference sequences of respective subtypes via ClustalW in BioEdit (version 7.2, Biological Sequence Alignment Editor, Raleigh, NC, USA). The 104 reference gene sequences of the subtypes were obtained from the HIV database (https:// www. hiv. lanl. gov). A phylogenetic tree was constructed using the neighbor-joining method based on the Kimura 2-parameter model with 1000 bootstrap replicates in MEGA version 10.1.8 (https:// www. megas oftwa re. net).

Phylogenetic analysis of non-B subtypes.
A maximum-likelihood phylogenetic tree that includes 83 sequences of ART-naive patients with non-B subtypes and 104 HIV-1 reference sequences were constructed via MEGA (version 10.1.8). Using MEGA, we selected the nucleotide substitution model most appropriate for analyzing our data set: the general-time reversible model with the proportion of invariable sites and gamma plus invariant sites distributed rate heterogeneity (GTR + G + I model). The reliability of the branching orders was assessed via a bootstrap analysis of 1000 replicates. We defined clusters as clades with high branch support values (> 90%) and short branch lengths (genetic distance < 0.015) 38 . All phylogenetic trees were visualized via FigTree version 1.4.4 (https:// tree. bio. ed. ad. uk).