In many countries, the increased incidence of HIV among MSM has become one of the major challenges in HIV prevention. Israel is a mixed and pluralistic country with nearly 8.2 million people in 2017 , of those 78.1% were Jews and 21.9% were Arabs: Muslims (N = 1 500 000 million, 83.0%), Christians (N = 168 300, 9.3%) and Druze (N = 139 300, 7.7%).
The Israeli population has a relatively low overall HIV-burden. In 2016, there were 7448 individuals recorded with HIV infection . We have previously reported that between 1986 and 2010, 96.3% of the 3349 HIV-infected men were Jews and 3.7% were Arabs: 48.1% of the Arabs (and 51.2% of the Jews) were MSM . This high rate of HIV-1 positive Arab MSM (AMSM) was unexpected, as these individuals experience unique difficulties in constructing their gay identity .
This study aims, for the first time, to retrospectively assess the spread of HIV-1 among the AMSM and Jewish MSM (JMSM) diagnosed in Israel between 2005 and 2016 in order to better understand the interethnic sexual connections between the two diverse population groups.
The National HIV Reference Laboratory database [5,6] was mined for all Israeli-born, HIV-positive men, MSM and non-MSM, identified between 2005 and 2016. The ethnicity (Jews/Arabs) and religion (Jewish/Christians/Muslims/Druze) was cross-matched with the National Civil Registry to verify the ethnicity and religion of each patient.
Partial protease (codons 4–99) and reverse transcriptase (codons 38–247) Sanger-based sequences  from the first available sample collected less than 6 months following diagnosis from treatment-naive MSM were retrospectively assessed. All AMSM patients for whom samples were available were included. Using a stratified random selection design, samples from ∼40% of all JMSM in each of the studied years were selected. Transmitted drug-resistance mutations (TDRM) were identified using https://hivdb.stanford.edu/page/who-sdrm-list/. Subtypes were defined by the REGA HIV-1 subtyping tool version 3.0.
Phylogenetic analysis included the major subtypes identified in Israel, A, B and C. Partial protease and reverse transcriptase nucleotide sequences and 14 early reference sequences from Los-Alamos database (Supplementary Table 1, http://links.lww.com/QAD/B381) were aligned with Clustal Omega . Drug resistance codons were stripped and the resultant partial protease-reverse transcriptase were combined (Supplementary Table 2, http://links.lww.com/QAD/B381) using custom R script. The general time reversible model, with proportion of invariable sites and gamma plus invariant sites-distributed rate heterogeneity (GTR+G+I model) was used . Bayesian Markov chain Monte Carlo (MCMC) method was run for 10 million iterations, with a 10% burn-in period and samples saved every 10 000 iterations using BEAST version 1.8 . Lineage trees were visualized in FigTree version 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). Clusters of epidemiologic relevance (with posterior probability >0.9) were studied. To validate the identified clusters (in subtype A and B trees) a representative sequence was chosen from each of the clusters and, using Blastn , the 10 most closely related sequences, were added as control sequences (Supplementary Table 3, http://links.lww.com/QAD/B381). Subsequently, phylogenetic analysis was repeated as described above (Supplementary Figures 4a and b, http://links.lww.com/QAD/B381).
Statistical analysis and ethical approval
Comparison of age distribution of JMSM and AMSM was performed with Mann–Whitney test (SPSS version 25.0). Fisher exact test was applied for categorical variables. The study was approved by the institutional review board of the Ministry of Health (MOH 2017–005). De-identified data (after cross- reference with the National Civil Registry) was used in the study, waiving the need for informed consent.
Demographic characteristics of the study population
During the study period (2005–2016), 1143 Israeli-born men who self-identified as MSM were diagnosed with HIV infection. Their overall number gradually increased between 2005 and 2013 (highest between 2011 and 2013), and then decreased in later years. Of those, 6.4% (73/1143) were AMSM: 57 Muslims and 16 Christians, while 93.6% (1070/1143) were JMSM (Supplementary Figure 1, http://links.lww.com/QAD/B381). The median age of AMSM was 33 years [interquartile range (IQR) 26.1–40.4] and JMSM was 33.3 years (IQR 27.1–40.4, P = 0.3). Diagnosis during seroconversion was identified in 8.7% (99/1143): 9.0% (96/1070) JMSM and 4.1% (3/73) AMSM (P = 0.2). Additionally, 247 Israeli-born, non-MSM men were diagnosed, with higher proportion of Arabs compared to the AMSM in the male MSM population (19%, 46/247 non-MSM versus 6.4%, 73/1143 MSM, P < 0.01). Overall, 84% of Jews and only 61% of Arabs self-identified as MSM (1070/1271 Jews and 73/119 Arabs, P < 0.01).
HIV-1 subtypes and transmitted resistance mutations
HIV-1 partial protease-reverse transcriptase sequences were analyzed for 62 AMSM (52 Muslims and 10 Christians) and 440 JMSM (Supplementary Figure 1, http://links.lww.com/QAD/B381). The majority (82.1%, 412/502) were infected with subtype B: 80.6% (50/62) and 82.3% (362/440) in AMSM and JMSM, respectively, (P = 0.8); 9.4% (47/502) had subtype A: 8.1% (5/62) AMSM and 9.5% (42/440) JMSM (P = 0.8); 4.2% (21/502) carried subtype C: 4.8% (3/62) AMSM and 4.1% (18/440) JMSM (P = 0.7). CRF02-AG and CRF01-AE were detected in 4.4% (22/502): 6.5% (4/62) of Muslim MSM and 4.1% (18/440) of JMSM (P = 0.5).
The overall prevalence of patients diagnosed with any TDRM was 13.5% (68/502, Table 1). The TDRM rate was higher in JMSM compared to AMSM: 14.1% (62/440) versus 9.7% (6/62), respectively, but the difference was not statistically significant. No significant temporal differences in the prevalence, distribution across HIV subtypes, or years of diagnosis of TDRM between AMSM and JMSM were observed (data not shown). Within the AMSM, TDRM were identified in Muslims only; Christians AMSM were TDRM-free. However, due to the low number of HIV-infected Christian MSM, statistical significance was not achieved. The most common TDRM identified were the nonnucleoside reverse transcriptase inhibitor (NNRTI) mutation K103N/S (9%, 45/502), the protease inhibitor mutation L90M (3.2%, 16/502) and the nucleotide reverse transcriptase inhibitor (NRTI) mutations M184V (1.2%, 6/502) and T215S (1.4%, 7/502). K103S was identified in subtype A only (1.1%, 5/440 JMSM, none in AMSM, P < 0.01) and K103N was more common in subtype B (9.2%, 38/412 of subtype B versus 2.2%, 2/90 of any other subtype, P < 0.01).
Clustering of HIV-1 sequences from Arab and Jewish MSM
Phylogenetic analysis of HIV-1 sequences of subtypes A and C (Supplementary Figures 2 and 3, respectively, http://links.lww.com/QAD/B381) revealed similar results: sequences of AMSM were mixed within clusters of JMSM in all three subtype trees. In subtype-A tree, three of the AMSM were in cluster 2 and two in cluster 3. The most common TDRM, forming a single transmission cluster among subtype-A tree, was the K103S/N mutation. Interestingly, patients harboring the K103S, which has a one- base pair change from K103N, were diagnosed in later years (2013 and 2014).
The subtype B tree showed 31 small clusters of less than 4 individuals and five large clusters (including 15, 16, 16, 30 and 35 individuals each) with no detectable common ancestor (with posterior probability >0.9) and with AMSM sequences intermingled within the different JMSM clusters (Fig. 1a). Two clusters were defined by a specific TDRM: K103N (8/15 sequences in cluster 1, Fig. 1b) and L90 M (6/16 sequences in cluster 2, Fig. 1c). While the most recent common ancestor in cluster 2 originated from ∼1990, the K103N cluster originated more than 10 years later. Three AMSM clustered within the L90M cluster and three in the K103N cluster, none of which had TDRM. Notably, while sequences from Muslim AMSM grouped with the L90M JMSM sequences, only Christian AMSM HIV-1 sequences intermingled with the more recent K103N cluster.
This report presents the first molecular analysis of HIV-1 infection among AMSM in Israel. While 1070 JMSM were diagnosed with HIV-1 between 2005 and 2016, only 6.4% (73) were AMSM, lower relative to the proportion of Arabs (20%) in the Israeli population . This lower prevalence could reflect the relatively more liberal Jewish society regarding sex and sexuality while AMSM reside within a more traditional society where MSM activities are not encouraged . Misreporting of MSM activities among AMSM could account for their relatively lower proportion, as previously suggested (1–11% of heterosexual men) in United Kingdom .
Proportionally more JMSMs were diagnosed during seroconversion. This may be related to the greater awareness to infection in JMSM or to the conservative nature of the Arab society, resulting in delayed referral to medical care. Consequently, future interventions for the AMSM community should include increasing the awareness for the possible clinical signs of the infection, better access to HIV tests and to pre and postexposure prophylaxis.
Subtype B was the most common among Israeli-born, HIV-1-infected MSM, as already reported . However, the prevalence of the various HIV-1 subtypes infecting AMSM and JMSM was nearly identical, suggesting shared transmission networks between the two populations. Indeed, it was previously speculated that AMSM preferred JMSM as their sex partners so that their sexual orientation and practices would not be brought to the attention of their families and friends in the Arab community .
The overall 13.5% prevalence of TDRM identified in MSM was higher than the overall 8.3% identified in a weighted analysis performed with samples collected between 2008 and 2010 from over 25 countries in Europe and Israel, reporting 5% NRTI TDRM and 3.1% NNRTI TDRM in the MSM population . The high rate of TDRM we identified in MSM in Israel calls for resistance testing, early treatment and for constant clinical surveillance, especially for the JMSM population.
Phylogenetic analysis revealed joined clusters of JMSM and AMSM, some of which shared a common TDRM which is highly suggestive for secondary transmissions among drug-naive HIV carriers [17,18]. Generally, viruses harboring a specific TDRM were grouped together. Several JMSM and one AMSM had the thymidine analogue T215S mutation related to previously used AZT  which was maintained due to its relatively weak effect on viral fitness [20,21]. The partial protease L90M TDRM related to the MSM population in Tel-Aviv  characterized subtype B sequences and was observed in clusters with AMSM. As both nelfinavir and saquinavir/r, which select for L90M, are rarely prescribed in Israel, the ongoing transmission of this TDRM has limited clinical impact.
The most prevalent TDRM were the K103N and K103S conferring resistance to efavirenz and nevirapine. In Europe K103N/S was also the most prevalent NNRTI ( detected in recently infected MSM). NNRTI mutations are still considered to have the largest impact of drug susceptibility and merit continuous monitoring of NNRTI resistance in treatment-naive patients.
This study is subject to several limitations. First, a reporting bias may be present, as sexual behavior and identity may be underreported. Second, we precluded comparison between Christian and Muslim MSM due to their small number. Third, this study included only a random sample of JMSM, making the generality of the results questionable. Nevertheless, this study clearly demonstrated that AMSM and JMSM share similar HIV-1 sequences, reflecting sexual intermingling between the two distinct ethnic groups and demonstrates the high prevalence of TDRM within these populations. Future interventions should address both ethnic groups using adequate messages directed for all MSM.
The authors would like to thank Zehava Yossefi, for data-base management and Yehudit Posen for critical review of the manuscript
Authors’ contributions: Z.M. and O.M. designed the study. E.B., M.W., O.M. and N.S.Z. preformed the laboratory analysis. E.M. critically assessed the study design. N.S.Z. and O.M. summarized the results. Z.M., O.M. and N.S.Z. wrote the paper. All authors have read and approved the final manuscript
Conflicts of interest
The authors have no conflicts of interest.
1. Central Bureau of Statistics J, Israel. Statistical Abstract of Israel 2017- No 68. 2017. http://wwwcbsgovil/reader/shnatonenew_sitehtm
. [Accessed 1 February 2017]
2. Chemtob D. HIV/AIDS in Israel: Periodic Epidemiological Report 1981-2015. 2016. http://wwwhealthgovil/PublicationsFiles/AIDS_PeriodicReport2016pdf
. [Accessed 1 February 2017]
3. Mor Z, Weinstein R, Grotto I, Levin Y, Chemtob D. Thirty years of HIV in Israel: current epidemiology and future challenges
. BMJ Open
2013; 3:pii: e003078.
4. Jaspal R, Cinnirella M. Identity processes, threat, and interpersonal relations: accounts from British Muslim gay men
. J Homosex
5. Michaeli M, Wax M, Gozlan Y, Rakovsky A, Mendelson E, Mor O. Evaluation of Xpert HIV-1 Qual assay for resolution of HIV-1 infection in samples with negative or indeterminate Geenius HIV-1/2 results
. J Clin Virol
6. Rakovsky A, Gozlan Y, Bassal R, Wax M, Shirazi R, Bakhanashvili M, et al. Diagnosis of HIV-1 infection: performance of Xpert Qual and Geenius supplemental assays in fourth generation ELISA-reactive samples
. J Clin Virol
7. Moscona R, Ram D, Wax M, Bucris E, Levy I, Mendelson E, et al. Comparison between next-generation and Sanger-based sequencing for the detection of transmitted drug-resistance mutations among recently infected HIV-1 patients in Israel, 2000-2014
. J Int AIDS Soc
8. Bennett DE, Camacho RJ, Otelea D, Kuritzkes DR, Fleury H, Kiuchi M, et al. Drug resistance mutations for surveillance of transmitted HIV-1 drug-resistance: 2009 update
. PLoS One
9. Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, Li W, et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega
. Mol Syst Biol
10. Posada D. jModelTest: phylogenetic model averaging
. Mol Biol Evol
11. Drummond AJ, Rambaut A. BEAST: Bayesian evolutionary analysis by sampling trees
. BMC Evol Biol
12. Boratyn GM, Camacho C, Cooper PS, Coulouris G, Fong A, Ma N, et al. BLAST: a more efficient report with usability improvements
. Nucleic Acids Res
2013; 41 (Web Server issue):W29–33.
13. Mor Z, Grayeb E, Beany A. Arab men who have sex with men in Israel: knowledge, attitudes and sexual practices
. HIV Med
14. Hue S, Brown AE, Ragonnet-Cronin M, Lycett SJ, Dunn DT, Fearnhill E, et al. Phylogenetic analyses reveal HIV-1 infections between men misclassified as heterosexual transmissions
15. Levy I, Mor Z, Anis E, Maayan S, Leshem E, Pollack S, et al. Men who have sex with men, risk behavior, and HIV infection: integrative analysis of clinical, epidemiological, and laboratory databases
. Clin Infect Dis
16. Hofstra LM, Sauvageot N, Albert J, Alexiev I, Garcia F, Struck D, et al. Transmission of HIV drug resistance and the predicted effect on current first-line regimens in Europe
. Clin Infect Dis
17. Price MA, Wallis CL, Lakhi S, Karita E, Kamali A, Anzala O, et al. Transmitted HIV type 1 drug resistance among individuals with recent HIV infection in East and Southern Africa
. AIDS Res Hum Retroviruses
18. Garcia-Lerma JG, MacInnes H, Bennett D, Weinstock H, Heneine W. Transmitted human immunodeficiency virus type 1 carrying the D67N or K219Q/E mutation evolves rapidly to zidovudine resistance in vitro and shows a high replicative fitness in the presence of zidovudine
. J Virol
19. Sallam M, Sahin GO, Indriethason H, Esbjornsson J, Love A, Widell A, et al. Decreasing prevalence of transmitted drug resistance among ART-naive HIV-1-infected patients in Iceland, 1996-2012
. Infect Ecol Epidemiol
20. de Ronde A, van Dooren M, van Der Hoek L, Bouwhuis D, de Rooij E, van Gemen B, et al. Establishment of new transmissible and drug-sensitive human immunodeficiency virus type 1 wild types due to transmission of nucleoside analogue-resistant virus
. J Virol
21. Cong ME, Heneine W, Garcia-Lerma JG. The fitness cost of mutations associated with human immunodeficiency virus type 1 drug resistance is modulated by mutational interactions
. J Virol
22. Turner D, Amit S, Chalom S, Penn O, Pupko T, Katchman E, et al. Emergence of an HIV-1 cluster harbouring the major protease L90 M mutation among treatment-naive patients in Tel Aviv, Israel
. HIV Med
* Drs Zuckerman and Mor contributed equally to this study.