- Research article
- Open Access
- Open Peer Review
Risk Factors for HIV-1 seroconversion among Taiwanese men visiting gay saunas who have sex with men
BMC Infectious Diseasesvolume 11, Article number: 334 (2011)
Men having sex with men (MSM) accounts for 33.6% of all reported cases of HIV-1 infection in Taiwan. The aim of this study was to investigate the epidemiology of HIV-1 infection among MSM in gay saunas in Taiwan.
Patrons of 5 gay saunas were recruited for a weekly volunteer counseling and testing program from 2001 to 2005. Questionnaires were collected for a risk factor analysis. HIV-1 subtypes were determined using DNA sequencing and phylogenetic analyses.
HIV-1 prevalence rates among MSM in gay saunas in 2001 through 2005 were 3.4%, 5.1%, 8.9%, 8.5%, and 8.3%, respectively. In total, 81 of 1, 093 (7.4%) MSM had HIV-1 infection. Fifty-two HIV-1 strains were genotyped, and all of them were subtype B. HIV-seropositive men were significantly younger than the seronegatives. Only 37.1% used condoms every time during sexual intercourse. A multivariate logistic regression analysis showed that the risk factors for HIV-1 were being uncircumcised (odds ratio (OR) = 2.19; 95% confidence interval (CI), 1.08~4.45); having sexual intercourse with at least 2 partners during each sauna visit (≥ 2 vs. ≤ 1, OR = 1.71; 95% CI, 1.02~2.89); and the role played during anal intercourse (versatile vs. an exclusively insertive role, OR = 2.76; 95% CI, 1.42~5.36).
Overall, 7.4% Taiwanese MSM participating in this study had HIV-1 subtype B infection. Uncircumcised, being versatile role during anal intercourse, and having sex with more than one person during each sauna visit were main risk factors for HIV-1 infection.
Since the global outbreak of HIV/AIDS, HIV has been transmitted among different vulnerable populations. According to a recent review, the global trend of HIV-1 infection among men having sex with men (MSM) has continued to increase, especially in East Asia, Africa, and Russia . As of September 2011, there were 22, 335 known cases of HIV infection in Taiwan, and Taiwanese nationals accounted for 96.5% of those cases. Of the Taiwanese nationals, 92.5% were male. Populations at risk for HIV infection include injection drug users (30.4%), homosexual men (38.3%), heterosexual men (21.6%), and bisexual men (8.1%). Therefore, MSM are a high risk group for HIV infection in Taiwan .
Since 1996~1997, the use of highly active antiretroviral therapy (HAART) has significantly decreased HIV/AIDS-related mortality . It has been reported that unprotected sex has increased among MSM after the introduction of HAART, which has resulted in dramatic decreases in the morbidity and mortality from HIV-1 infection . Taiwan began to provide free HAART to all HIV-1/AIDS patients in April 1997 [5, 6]. Although successful HAART can reduce the contagiousness of HIV-1, the fact that HIV-1 transmission still occurs in the ART-era indicates that transmission occurs from undiagnosed, untreated, or unsuccessfully treated patients .
Risks factors for HIV-1 infection in MSM include unprotected anal intercourse [8–10], multiple sexual partners , a history of drug use [12–14], a history of commercial sexual transactions [11, 15], and a history of other sexually transmitted diseases (STDs) . In Taiwan, homosexuals usually gather in parks, gay bars, and gay saunas, and because gay saunas provide the opportunity for sexual intercourse, individuals who frequent saunas are at a high risk of contracting STDs. Therefore, gay saunas are an ideal place for voluntary counseling and testing for HIV-1 infection . The aims of this study were (1) to determine the prevalence and trends of HIV-1 infection among MSM frequenting gay saunas from 2001 to 2005; (2) to identify the risk factors associated with HIV-1 infection; and (3) to determine the viral subtypes of HIV-1 strains found in MSM in Taiwan.
Before beginning the study, we contacted gay saunas in Taipei, Hsinchu, and Taichung to inquire as to their willingness to cooperate in participant recruitment. Anonymous HIV testing and education were provided during peak hours on weekends. Blood samples were obtained by venipuncture in private rooms. A single participant was interviewed at a time.
Gay saunas in Taipei were visited every other weekend, while those in Hsinchu and Taichung were visited once a month. All recruited participants provided blood samples, and they could choose whether or not to complete a questionnaire. This research study was approved by the Institutional Review Board of National Yang-Ming University. Before testing for HIV-1/2 antibodies, researchers provided the necessary counseling. Test results were available 2 weeks after the blood samples were obtained. Participants with negative results were encouraged to undergo follow-up screening every 6 months, while participants with positive results were initially counseled by a professional social worker and referred to a hospital for further counseling. Positive blood samples then underwent HIV-1 subtyping.
HIV-1 Screening, Subtyping, and Phylogenetic Analysis
Participants provided 5~10 ml of peripheral blood for HIV-1 antibody testing and subtyping. Recombinant HIV-1/HIV-2 EIA (Murex Diagnostics Limited, Dartford, UK) was used for HIV screening. Positive samples were reconfirmed by HIV Western blot 2.2 (Genelabs, Singapore). A nested polymerase chain reaction (PCR) was used to determine the gag and env subtypes . Sequence alignments with different subtype reference sequences from the Los Alamos HIV database (available at: http://www.hiv.lanl.gov) were conducted using the BioEdit program (vers. 7.0.9) . Phylogenetic trees were constructed using the Neighbor-joining (NJ) and maximum-likelihood (ML) methods using the MEGA program (vers. 5.0) . The General Time Reversible (GTR)+G+I substitution model was the best fitting model as tested using MEGA5 Bayesian information criterion (BIC) scores, the corrected Akaike information criterion (AIC) value, and the maximum-likelihood value (lnL). The gamma distribution shape parameter (G) was 1.0342. The robustness of the phylogenetic trees was statistically evaluated by a bootstrap analysis with 1000 bootstrap samples.
Pearson's χ2 test and Fisher's exact test were performed in a univariate analysis to investigate differences in demographic data and consumer behavior patterns among MSM. Student's t-test or an analysis of variance (ANOVA) was conducted to examine age differences. A multivariate logistic regression analysis was performed to identify risk factors associated with HIV-1 infection.
Nucleotide Sequence Accession Numbers
HIV-1 env sequences obtained in this study were deposited in GenBank with the accession numbers JN631752~JN631791.
Between 2001 and 2005, 1, 093 participants (81 HIV-1 positive and 1, 012 HIV-1 negative) participated in this study. The positive rates of HIV-1 and syphilis were 7.4% and 7.6%, respectively. A trend analysis showed that HIV-1 positive rates significantly increased from 3.4% in 2001 to 6.1% in 2002, and to 8.9% in 2003 (p = 0.031). Then, it reached a plateau at 8.5% in 2004 and 8.4% in 2005. In contrast, 2 peaks syphilis infection rates were seen in 2002 and 2005 (Table 1).
The mean age of participants was 32.4 (SD, 8.1) years, with a range of 17~81 years. Among participants, 54.87% had a high school degree, and 91.7% were single (Additional file 1). Significant differences were found between participants who brought their own condoms to the sauna every visit (79.5% HIV-1 positive vs. 66.9% HIV-1 negative, p < 0.05). Among those who did not bring their own condoms, more HIV-1-positive participants than their counterparts purchased condoms from the counter (53.3% vs. 39.1%, p < 0.05) (Additional file 2).
HIV-1-positive participants had sexual intercourse with more people in a 3-month period on average than their counterparts (4.8 ± 5.1 vs. 4.0 ± 6.4, p < 0.05). More HIV-1-negative than HIV-1-positive participants had insertive roles (35.9% vs. 19.5%). Overall, 27% of study participants were circumcised, including 16.5% HIV-1-positive and 26.8% HIV-1-negative participants (p = 0.045) (Additional file 3). The major reasons for testing or not were shown in Additional file 4. Nineteen percent of the study participants had an STD history (Additional file 5).
According to univariate logistic regression, several factors were associated with HIV-1 infection, they included: those with versatile roles (OR = 2.55, 95% CI = 1.41~4.61), those performing anal intercourse (OR = 2.73, 95% CI = 1.08~6.86), those performing anal rimming (OR = 2.03, 95% CI = 1.24~3.34), those having sex with at least 2 partners per visit (OR = 1.73, 95% CI = 1.08~2.80), being uncircumcised (OR = 1.85, 95% CI = 1.01~3.42), and those aged under 35 years (OR = 2.70, 95% CI = 1.07~6.79), as shown in Table 2.
Results from stepwise multivariate logistic regression analysis indicated that those with versatile roles (OR = 2.76, 95% CI = 1.42~5.36), those had sexual intercourse with at least 2 men during each visit (OR = 1.71, 95% CI = 1.02~2.89), being uncircumcised (OR = 2.19, 95% CI = 1.08~4.45) were related to HIV-1 infection (Table 3).
Consistent tree topologies were observed in both the NJ and ML trees. Analysis of the env subtype in 52 HIV-1-positive participants showed that all participants had subtype B (Additional file 6). Results showed that some HIV-1 strains from gay saunas and Venereal Disease Control Institute (VDCI) clustered with bootstrap values of > 85%.
According to the Centers for Disease Control, in 2009, the HIV-positive rate based on screening at major hospitals, blood centers, military bases, and prisons was 54.96/100, 000 (1651/3, 003, 735) . In this study, we screened 1, 093 participants and found a syphilis-positive rate of 7.6% and a HIV-1-positive rate of 7.4%, which were significantly higher than those reported by the Centers for Disease Control.
From a physiological point of view, the strong diffusion capabilities of the rectum facilitate HIV infection . Analysis by the number of sexual partners showed a direct correlation between the number of sexual partners and the probability of HIV infection . In addition, condom use can significantly decrease propagation of the HIV virus. In this study, over 80% of study participants had anal intercourse, which was comparable to the 71.4% reported by Ko et al. in 1997 . The percentage of participants wore a condom every time and for the entire duration of sexual intercourse (Additional file 3) was similar to that reported by Ko et al. , suggesting that through the years, the proportion of homosexuals performing anal intercourse has remained unchanged and the proportion of condom use has remained low.
Frequent visits to saunas increase the probability of contracting STDs . Participants in our study had a higher rate of STDs, and participants with STDs are more susceptible to HIV infection (Additional file 5). The route of STD transmission is a risk factor for HIV infection and that STDs increase the HIV infection rate 2~5-fold . A stepwise multiple regression analysis revealed that role played during anal intercourse, the number of sexual partners during each sauna visit, and the circumcision status were risk factors associated with HIV-1 infection in MSM frequenting gay saunas. These were also found to be risk factors in other countries [16, 22].
Male circumcision is a religious commandment in Judaism and Islam. The world prevalence of men with circumcision is 12.5%~33%, and is especially high in the USA, Canada, and Africa and among Islamic people . Approximately 80% of US men are circumcised , in which white males had the highest prevalence (> 70%) and Hispanics had the lowest prevalence (≤ 40%) . Ko et al. reported that the prevalence of circumcision in Taiwan was 7.2%~8.7% . In this study, our findings indicated that circumcision of MSM provides a certain degree of protection against HIV-1 infection. Previous studies showed that male circumcision reduced the likelihood of female-to-male transmission of HIV-1 infection by 50%~60% [27–29]. Concerning the protective effect of circumcision for MSM, several studies reported that circumcision had a significant protective effect [30–33]. In contrast, other studies reported that circumcised MSM had greater chances for contracting HIV infection although they were statistically nonsignificant [34, 35]. Since Buddhism and Taoism are the two main religions in Taiwan (44.5% and 24.1%, respectively), while Christianity and Islam populations are relatively small (10% and 1%, respectively) , religion might not have been a confounder in this study. Recently, an emerging epidemic among Taiwanese MSM was revealed when the number of new HIV-infected homosexuals increased from 571 (17% of all new HIV-infected) in 2005 to 1, 275 (71%) in 2010 . Circumcision may be one of the approaches that should be adapted to control the epidemic of HIV-1 infection among Taiwanese MSM in the future.
HIV screening among high-risk group is cost-effective. Unlinked anonymous testing can provide high-risk individuals with a secure means of testing and can prevent a selection bias . In this study, we performed purposive sampling and snowballing in order to increase the sample size . However, because the sampling was not random, results from this study cannot be used to represent the prevalence of HIV infection in gay sauna patrons or of the general MSM population across Taiwan. In addition, because the screening was anonymous, most participants used aliases, and only a few participants agreed to be followed-up. This hindered our follow-up efforts and impeded us from estimating the incidence in a cohort study. The seropositive rate in this study was high. We therefore suggest that future studies expand the sampling area and survey other places where MSM congregate so as to estimate the size of the MSM population and thereby fully monitor trends of the HIV virus in this high-risk population.
Of MSM participating in this study, 7.4% (81/1093) were infected with HIV-1, and all of them had subtype B. The circumcision status, the role played during anal intercourse, and the number of sexual partners during each sauna visit were the three main risk factors for HIV-1 infection.
This study accessed gay saunas to obtain a deeper understanding of the sexual behavior of MSM in Taiwan. Results from this study can serve as a reference for researchers interested in these behavioral patterns and for local authorities promoting health education among MSM.
Van Griensven F, de Lind van Wijngaarden JW, Baral S, Grulich A: The global epidemic of HIV infection among men who have sex with men. Curr Opinion HIV AIDS. 2009, 4: 300-307. 10.1097/COH.0b013e32832c3bb3.
Center for Disease Control, R.O.C. (Taiwan): HIV/AIDS Data. [http://www.cdc.gov.tw/np.asp?ctNode=2231&mp=5]
Wolitski RJ, Valdiserri RO, Denning PH, Levine WC: Are we headed for a resurgence of the HIV epidemic among men who have sex with men?. Am J of Public Health. 2001, 91: 883-888.
Law MG, Prestage G, Grulich A, Van de Ven P, Kippax S: Modelling the effect of combination antiretroviral treatments on HIV incidence. AIDS. 2001, 15: 1287-1294. 10.1097/00002030-200107060-00011.
Fang CT, Hsu HM, Twu SJ, Chen MY, Chang YY, Hwang JS, Wang JD, Chuang CY, Division of AIDS and STD, Center for Disease Control, Department of Health, Executive Yuan: Decreased HIV transmission after a policy of providing free access to highly active antiretroviral therapy in Taiwan. J Infect Dis. 2004, 190: 879-885. 10.1086/422601.
Hung CC, Chen MY, Hsieh SM, Sheng WH, Chang SC: Clinical spectrum, morbidity, and mortality of acquired immunodeficiency syndrome in Taiwan: a 5-year prospective study. J Acquir Immune Defic Syndr. 2000, 24: 378-385.
Cohen MS, Gay CL: Treatment to prevent transmission of HIV-1. Clin Infect Dis. 2010, 50: S85-95. 10.1086/651478.
Ekstrand ML, Stall RD, Paul JP, Osmond DH, Coates TJ: Gay men report high rates of unprotected anal sex with partners of unknown or discordant HIV status. AIDS. 1999, 13: 1525-1533. 10.1097/00002030-199908200-00013.
Elford J, Bolding G, Maguire M, Sherr L: Sexual risk behaviour among gay men in a relationship. AIDS. 1999, 13: 1407-1411. 10.1097/00002030-199907300-00019.
Piaseczna MA, Craib KJ, Li K, Chan K, Weber AE, Strathdee SA, Martindale S, Schechter MT, Hogg RS: Longitudinal patterns of sexual behavior and condom use in a cohort of HIV-negative gay and bisexual men in Vancouver, British Columbia, Canada, 1995-2000. J Acquir Immune Defic Syndr. 2001, 28: 187-193.
Weber AE, Chan K, George C, Hogg RS, Remis RS, Martindale S, Otis J, Miller ML, Vincelette J, Craib KJ, Mâsse B, Schechter MT, LeClerc R, Lavoie R, Turmel B, Parent R, Alary M: Risk factors associated with HIV infection among young gay and bisexual men in Canada. J Acquir Immune Defic Syndr. 2001, 28: 81-88.
Rhodes F, Corby NH, Wolitski RJ, Tashima N, Crain C, Yankovich DR, Smith PK: Risk behaviors and perceptions of AIDS among street injection drug users. J Drug Educ. 1990, 20: 271-288. 10.2190/3G2P-3VGG-ACY4-7D65.
Rietmeijer CA, Wolitski RJ, Fishbein M, Corby NH, Cohn DL: Sex hustling, injection drug use, and non-gay identification by men who have sex with men. Associations with high-risk sexual behaviors and condom use. Sex Transm Dis. 1998, 25: 353-360. 10.1097/00007435-199808000-00006.
Wolitski RJ, Fishbein M, Johnson WD, Schnell DJ, Esacove A: Sources of HIV information among injecting drug users: association with gender, ethnicity, and risk behaviour. AIDS Community Demonstration Projects. AIDS Care. 1996, 8: 541-555. 10.1080/09540129650125506.
Weber AE, Craib KJ, Chan K, Martindale S, Miller ML, Schechter MT, Hogg RS: Sex trade involvement and rates of human immunodeficiency virus positivity among young gay and bisexual men. Int J Epidemiol. 2001, 30: 1449-1454. 10.1093/ije/30.6.1449.
Van Beneden CA, O'Brien K, Modesitt S, Yusem S, Rose A, Fleming D: Sexual behaviors in an urban bathhouse 15 years into the HIV epidemic. J Acquir Immune Defic Syndr. 2002, 30: 522-526. 10.1097/00126334-200208150-00008.
Lin YT, Lan YC, Chen YJ, Huang YH, Lee CM, Liu TT, Wong WW, Yang JY, Wang CT, Chen YM: Molecular epidemiology of HIV-1 infection and full-length genomic analysis of circulating recombinant form 07_BC strains from injection drug users in Taiwan. J Infect Dis. 2007, 195: 1283-1293. 10.1086/513437.
Hall T: BioEdit: a user-friendly biological sequence alignment editor and analysis program for Window 95/98/NT. Nucleic Acid Symp Ser. 1999, 41: 95-98.
Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S: MEGA5: Molecular Evolutionary Genetics Analysis using Maximum Likelihood, Evolutionary Distance, and Maximum Parsimony Methods. Mol Biol Evol. 2011, 28: 2731-2739. 10.1093/molbev/msr121.
Ko YC, Chang SJ: Sexual patterns and human immunodeficiency virus infection among homosexuals in Taiwan. Sex Transm Dis. 1992, 19: 335-338. 10.1097/00007435-199211000-00007.
Ko NY, Shiau C, Sheu SL: Informational needs and information-seeking behaviors of recently diagnosed HIV seropositive gay men. Hu Li Za Zhi. 1997, 44: 32-40.
Wasserheit JN: Epidemiological synergy. Interrelationships between human immunodeficiency virus infection and other sexually transmitted diseases. Sex Transm Dis. 1992, 19: 61-77.
Totaro A, Volpe A, Racioppi M, Pinto F, Sacco E, Bassi PF: Circumcision: history, religion and law. Urologia. 2011, 78: 1-9. 10.1016/j.urology.2010.10.043.
Xu F, Markowitz LE, Sternberg MR, Aral SO: Prevalence of circumcision and herpes simplex virus type 2 infection in men in the United States: the National Health and Nutrition Examination Survey (NHANES), 1999-2004. Sex Transm Dis. 2007, 34: 479-484.
McKinney CM, Klingler EJ, Paneth-Pollak R, Schillinger JA, Gwynn RC, Frieden TR: Prevalence of adult male circumcision in the general population and a population at increased risk for HIV/AIDS in New York City. Sex Transm Dis. 2008, 35: 814-817. 10.1097/OLQ.0b013e318175d899.
Ko MC, Liu CK, Lee WK, Jeng HS, Chiang HS, Li CY: Age-specific prevalence rates of phimosis and circumcision in Taiwanese boys. J Formos Med Assoc. 2007, 106: 302-307. 10.1016/S0929-6646(09)60256-4.
Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, Puren A: Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: the ANRS 1265 Trial. PLoS Med. 2005, 2: e298-10.1371/journal.pmed.0020298.
Bailey RC, Moses S, Parker CB, Agot K, Maclean I, Krieger JN, Williams CF, Campbell RT, Ndinya-Achola JO: Male circumcision for HIV prevention in young men in Kisumu, Kenya: a randomised controlled trial. Lancet. 2007, 369: 643-656. 10.1016/S0140-6736(07)60312-2.
Gray RH, Kigozi G, Serwadda D, Makumbi F, Watya S, Nalugoda F, Kiwanuka N, Moulton LH, Chaudhary MA, Chen MZ, Sewankambo NK, Wabwire-Mangen F, Bacon MC, Williams CF, Opendi P, Reynolds SJ, Laeyendecker O, Quinn TC, Wawer MJ: Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial. Lancet. 2007, 369: 657-666. 10.1016/S0140-6736(07)60313-4.
Kreiss JK, Hopkins SG: The association between circumcision status and human immunodeficiency virus infection among homosexual men. J Infect Dis. 1993, 168: 1404-1408. 10.1093/infdis/168.6.1404.
Buchbinder SP, Vittinghoff E, Heagerty PJ, Celum CL, Seage GR, Judson FN, McKirnan D, Mayer KH, Koblin BA: Sexual risk, nitrite inhalant use, and lack of circumcision associated with HIV seroconversion in men who have sex with men in the United States. J Acquir Immune Defic Syndr. 2005, 39: 82-89. 10.1097/01.qai.0000134740.41585.f4.
Tabet S, Sanchez J, Lama J, Goicochea P, Campos P, Rouillon M, Cairo JL, Ueda L, Watts D, Celum C, Holmes KK: HIV, syphilis and heterosexual bridging among Peruvian men who have sex with men. AIDS. 2002, 16: 1271-1277. 10.1097/00002030-200206140-00010.
Templeton DJ, Jin F, Mao L, Prestage GP, Donovan B, Imrie J, Kippax S, Kaldor JM, Grulich AE: Circumcision and risk of HIV infection in Australian homosexual men. AIDS. 2009, 23: 2347-2351. 10.1097/QAD.0b013e32833202b8.
Millett GA, Ding H, Lauby J, Flores S, Stueve A, Bingham T, Carballo-Dieguez A, Murrill C, Liu KL, Wheeler D, Liau A, Marks G: Circumcision status and HIV infection among Black and Latino men who have sex with men in 3 US cities. J Acquir Immune Defic Syndr. 2007, 46: 643-650. 10.1097/QAI.0b013e31815b834d.
Bartholow BN, Goli V, Ackers M, McLellan E, Gurwith M, Durham M, Greenberg AE: Demographic and behavioral contextual risk groups among men who have sex with men participating in a phase 3 HIV vaccine efficacy trial: implications for HIV prevention and behavioral/biomedical intervention trials. J Acquir Immune Defic Syndr. 2006, 43: 594-602.
Tzeng HM, Yin CY: Demands for religious care in the Taiwanese health system. Nurs Ethics. 2006, 13: 163-179. 10.1191/0969733006ne835oa.
Lyu SY, Su LW, Chen YM: Effects of education on harm-reduction programmes. Lancet. 2011
McFarland W, Caceres CF: HIV surveillance among men who have sex with men. AIDS. 2001, 15: S23-32.
Kerr-Pontes LR, Gondim R, Mota RS, Martins TA, Wypij D: Self-reported sexual behaviour and HIV risk taking among men who have sex with men in Fortaleza, Brazil. AIDS. 1999, 3: 709-717.
The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2334/11/334/prepub
We thank the MSM who participated in this study, and peer educators and social workers from the Living with Hope Organization for their help in collecting the questionnaires, and staff of the AIDS Prevention and Research Center of National Yang-Ming University for their administrative support and technical assistance. The authors also acknowledge the technical support provided by the Sequencing Core Facility of National Yang-Ming University's Genome Research Center (YMGC). The Sequencing Core Facility is supported by the National Research Program for Genomic Medicine (NRPGM) of the National Science Council. This study was supported in part by the Centers for Disease Control of Taiwan (DOH92-DC-1034, DOH93-DC-1017, and DOH94-DC-1032). Parts of this paper were presented at the XV International Conference on AIDS held in Thailand and the 7th TPE AIDS Conference in 2004.
The authors declare that they have no competing interests.
SFL, YTL, and YJC collected samples and data, performed the statistical analysis, and drafted the manuscript. MC helped draft the final manuscript. SWH and HFL carried out the sequence alignment and phylogenetic analyses. WWW, HCT, and YHL participated in the design of the study. YMC and SYL conceived of the study, participated in its design and coordination, and helped draft and revised the manuscript. All authors read and approved the final manuscript.
Yen-Ju Chen, Yu-Ting Lin contributed equally to this work.