https://orcid.org/0000-0002-1071-8337
Abstract
The human immunodeficiency virus (HIV) type 1 A6 variant is dominating in high-prevalence Eastern European countries, with increasing prevalence over the remaining regions of Europe. The recent war in Ukraine may contribute to further introductions of this A6 lineage. Our aim was to model the transmission dynamics of the HIV-1 A6 variant between Poland and Ukraine.
HIV-1 A6 partial pol sequences originating from Poland (n = 1185) and Ukraine (n = 653) were combined with publicly available sequences (n = 7675) from 37 other countries. We used maximum likelihood-based tree estimation followed by a bayesian inference strategy to characterize the putative transmission clades. Asymmetric discrete phylogeographic analysis was used to identify the best-supported virus migration events across administrative regions of Poland and Ukraine.
We identified 206 clades (n = 1362 sequences) circulating in Poland or Ukraine (63 binational clades, 79 exclusively Polish, and 64 exclusively Ukrainian). Cross-border migrations were almost exclusively unidirectional (from Ukraine to Poland, 99.4%), mainly from Eastern and Southern Ukraine (Donetsk, 49.7%; Odesa, 17.6% regions) to the Central (Masovian, 67.3%; Lodz, 18.2%) and West Pomeranian (10.1%) districts of Poland. The primary sources of viral dispersal were the Eastern regions of Ukraine, long affected by armed conflict, and large population centers in Poland.
The Polish outbreak of the A6 epidemic was fueled by complex viral migration patterns across the country, together with cross-border transmissions from Ukraine. There is an urgent need to include war-displaced people in the national HIV prevention and treatment programs to reduce the further spread of transmission networks.
(See the Editorial Commentary by Schapiro on pages 1725–6.)
By the end of 2020 in Ukraine, 309 971 people had a diagnosis of human immunodeficiency virus (HIV), with the second highest incidence (after Russia) of HIV-1 infections in Europe (37.5 cases per 100 000 people compared with 1.9 per 100 000 in Poland) [1]. In virtually every region of the former Soviet Union, including Ukraine, the A6 lineage represents the dominant HIV-1 sequence variant [2, 3]. Infections with this variant have been expanding over recent years and increasing prevalence of the lineage has also been observed in Western European countries—for example, Germany [4]. Of note, the most common variant of HIV-1 in Western and Central Europe remains subtype B [5], but in recent years in Poland, we have noted an increase in the frequency of A6, which has become the second most prevalent clade [6].
Importantly, the A6 variant has been associated with failure of the first approved long-acting injectable antiretroviral therapy (ART), cabotegravir-rilpivirine, and is now considered a risk factor for virologic breakthrough [7]. Furthermore, it is uncertain whether spread of this clade would also negatively affect the efficacy of cabotegravir as the HIV preexposure prophylactic option [8].
It should be emphasized that the scale and pattern of A6 circulation among geographic locations in Poland and neighboring countries remains largely unexplored; however, infection dynamics after the onset of war in the Donbas on the Ukrainian-Russian border in 2014 was previously investigated. Between 2012 and 2015, the primary source of the transmissions was internally displaced people from the Eastern to the Central and Southern regions of Ukraine [9]. In Poland, in the last decade, Ukrainian citizens were the predominant group of migrants. By February 2020, 1 390 978 Ukrainians lived in Poland, constituting 62.8% of all officially registered foreigners [10]. Since the beginning of the war in Ukraine in late February 2022, 6 431 533 civilians (mostly women and children) were forced to seek refuge in Poland [11]. The HIV disease burden in Ukraine is higher than in Poland, with approximately 0.9% versus 0.01% prevalence, respectively, and approximately 130 000 versus 15 000 people receiving ART as of March 2022 [12, 13]. Large-scale migration may affect HIV-1 epidemics within and among the regions of Poland and Ukraine, particularly when ART is disrupted by ongoing armed conflict in Ukraine [9] and the past 2 years of the coronavirus disease 2019 epidemic [14].
Molecular data from pathogen genomes have become an essential instrument in infectious disease research supporting public health interventions. Phylodynamic models combine virus genetic and epidemiological data with geospatial attributes, allowing investigation of the migration history of pathogens in the absence of detailed contact tracing data [15, 16]. In the current study, we used a discrete phylogeographic approach to reconstruct the historical spread of the HIV-1 within and between Poland and Ukraine. Our models expand the knowledge of the subtype A6 epidemic in Poland and uncover distinct patterns of spread of the A6 variant into an Eastern region of the European Union, providing a strong background for future analyses of the effect of war on HIV epidemics.
METHODS
Sequence Data Collation and Phylogenetic Inference
We collected all available HIV-1 subsubtype A6 partial pol sequences (HXB2 genome location positions from 2253 to 3554) with known sampling location (ie, regional—the first administrative level) from 1997 until the end of 2021 from Poland and Ukraine. For Poland, we gathered samples from Polish individuals, both ART naive and ART experienced, undergoing genotypic drug resistance testing in all 16 regions (details in the Supplementary Materials, Supplementary Table 1, and Supplementary Figure 1). All publicly available HIV-1 A6 pol sequences from Ukraine and other countries, along with sampling date and related geographic information, were retrieved from the Los Alamos National Laboratory HIV database [17]. The final data set comprised 9513 sequences, including 1185 from Poland, 653 from Ukraine, and 7675 background samples (Supplementary Table 2).
Identification of Polish and Ukrainian Transmission Networks and Phylogeographic Estimations
To identify transmission linkages that best resemble the epidemic dynamics, we applied validated analytical workflows described by Cuypers et al [18] and Dellicour et al [19] (see Supplementary Materials). We performed discrete phylogeographic analysis [20] to implement the diffusion model in BEAST (Bayesian Evolutionary Analysis Sampling Trees) software, version 1.10.4 [21]. Simultaneously with reconstructing the geographic migration history, the spread across population groups—defined as sequences (1) from Polish citizens, (2) from Ukrainian migrants sampled in Poland, or (3) of Ukrainian origin—was estimated. For inference of viral movement between location, we defined the originating location as the “from” location and the recipient location as the “to” location. The same approach was applied for migration events between populations (ie, local and migrant populations). The preceding phylogeographic inference identified all binational clades occurring across Poland and Ukraine that had ≥ 2 sequences with known origins. We incorporated the sampling region and population groups as discrete traits associated with the sampled genomes according to a continuous-time Markov chain, characterized by a matrix of asymmetrical transition rates among sampling traits [22]. A detailed description of methods is provided in the Supplementary Materials.
Ethical Considerations
The study protocol was approved by the Bioethical Committee of the Pomeranian Medical University, Szczecin, Poland (approval nos. KB-0012/26/17 and KB-0012/08/12). All patients gave informed consent for the sampling and clinical data processing, and all samples in the study were deidentified to maintain participants’ anonymity. The research was conducted in accordance with the Declaration of Helsinki.
RESULTS
Characteristics of the A6 Variant Sequence Data Set and Phylogenetic Analysis
The alignment of global HIV-1 subtype A6 genetic diversity (n = 9513) was based on sequences from 39 countries worldwide. Notably, most isolates originated from Eastern Europe, including the Polish sequences (n = 1185). Reference data were derived from the Los Alamos National Laboratory HIV Sequence Database [17] and grouped into 5 locations: the Ukrainian data set (n = 653); Russia (n = 5983); the former Soviet Union, except for Russia, Ukraine, and the Baltic countries (n = 1420); Europe, with Latvia, Lithuania, and Estonia (n = 233), and other (n = 39) (presented in Figure 1, with detailed description in the Supplementary Materials and Supplementary Table 1). Our data set contained the A6 variant sequences from each Polish region. Moreover, we obtained sequences from 24 of 27 administrative regions in Ukraine (Supplementary Figure 1 and Supplementary Table 1). Among the Polish sequences, isolates from people originating from Ukraine were distinguished (n = 293 [24.7%])
Human immunodeficiency virus type 1 A6 pol fragment sequence data set, showing collection years for the 9513 sequences analyzed in the current study. Sequences are divided by sampling location, as indicated. Inset provides an expanded view of the years between 1997 and 2005. Abbreviation: FSU, Former Soviet Union.
HIV-1 A6 Variant Dispersal in Eastern Europe
The maximum clade credibility summary tree (Figure 2) revealed that samples formed 2 primary distinct monophyletic clades, the first with 4906 sequences (clade1) and the second comprising 4607 branches (clade2). Clade 1 contained mostly Russian sequences (n = 3906 [79.6% sequences within the clade; 65.3% of all Russian samples]). The second clade also contained a considerable proportion of Russian sequences (n = 2077 [45.1% within the clade; 34.7% of all Russian samples]) while including almost all samples from Poland (n = 1105 [93.3%]) and Ukraine (n = 628 [96.2%]).
Global migration history of human immunodeficiency virus type 1 (HIV-1) subtype A6, estimated from the P and RT gene fragments, dated between 1997 and 2021. Colors of the branches (lines) and nodes (circles) in the HIV-1 A6 maximum clade credibility tree, constructed from the complete data set (n = 9513), indicate their most probable location. Abbreviation: FSU, Former Soviet Union.
The discrete phylogeographic analysis allowed to distinguish and estimate independent introduction events for the HIV-1 A6 variant into each of 6 investigated areas (Figure 1). In our data set, the spread of the virus involving Ukraine as the origin location accounted for 52.5% estimated number of total exportation events, making this country a major player in the dissemination of A6, followed by Russia (43.6%). Analysis of the A6 variant spread revealed that most of the migration events were toward countries in the former Soviet Union (40.3%), followed by Russia (34.8%) and Poland (15.4%). Based on the entire data set, the first introduction of the HIV-1 A6 subsubtype into Poland was timed around April 1991 (95% highest posterior density (HPD), May 1986 to January 1994), with Ukraine as the source location.
Transmission Dynamics Across Regions of Poland and/or Ukraine
To delineate the viral dynamics in the Poland and/or Ukrainian regions, we next analyzed the 206 A6 clades, including 1362 sequences (which reflected 74.1% of the entire Polish-Ukrainian data set) (Supplementary Figure 2 and Supplementary Table 4). All well-supported clades were downsampled to obtain informative data on location movements (see Methods). In addition, clades that did not provide inform on the movement between regions (ie, size <3 or sequences from a single location) were excluded from further analysis. After downsampling there were 89 Polish/Ukrainian clades (3–275 sequences) that included samples from ≥2 regions (in total, 773 sequences from 35 regions). Among those, 43 (48.3%) were binational, 20 (22.5%) contained only Ukrainian sequences, and 26 (29.2%) were exclusively of Polish origin (details in Supplementary Table 5).
The dynamics of HIV A6 transmission between Polish and Ukrainian regions indicated that, on average, 59.2% of migration events occurred within Poland and 25.0% within Ukraine. In comparison, 15.8% were binational transmissions, and a majority of these were directed toward Poland (99.4%) (Figure 3). From 2008, numerous HIV-1 A6 variant introductions from Ukraine fueled the Polish HIV-1 epidemic of this clade. The frequency of migration events from Ukraine to Poland during the last 20 years shows that so far it peaked around July 2013 (95% confidence interval, May 2008 to April 2019) (Figure 4). The continuing increase in transmissions within Poland has been more notable since 2010 (Figure 3).
Relative contributions of Poland and Ukraine to the total circulation of human immunodeficiency virus type 1 A6 between and within the 2 countries. The unbalanced dynamics of viral spread before 2006 may reflect low sampling density rather than the actual dispersal tendency.
Timing of import events from Ukraine toward Poland. Dates of binational migration events were estimated from virus genomes using a phylogeographic approach, with peak density at 11 July 2013 (95% confidence interval, 5 May 2008 to 15 April 2019).
Phylogeographic inference revealed a complex viral migration history across 2 adjacent countries, with multiple links between Polish regions and migration between Ukrainian and Polish areas (Figure 5).The highest number of cross-border migration events originated mainly from the Donetsk (49.7%) and Odesa regions (17.6%). The destination locations were primarily in the Masovian (67.3%), Lodz (18.2%), and West Pomeranian (10.1%) districts (Figure 6). The Central and Southern regions of Poland were the dominant source of the A6 variant for nationwide migration networks (Figure 6A ). The national transmission hub was the region of Lodz (23.5%), the migratory source for 5 states. Next, Lesser Poland (23.1%) linked with 5 centers, followed by the Masovian district (9.9%), where virus diffusion was observed only to Lower Silesia. The gravitational center of the Polish A6 variant epidemic was the Masovian district, with more than half of domestic migration events (51.3%) and entries from 8 regions. Lower Silesia was the second destination of virus flow (14.7%), followed by Silesia (10.2%).
Migration events between regions in Poland and Ukraine through time: before 2006 (A), between 2006 and 2010 (B), between 2010 and 2013 (C), between 2014 and 2016 (D), and after 2016 (E). Arrow thickness corresponds to the average number of inferred migration events. Inward movements between a particular group and location are depicted in the same color. Regions are colored according to the number of sequences included in the clades for A6 type. The underlying map is from the Database of Global Administrative Areas. We report only migration events associated with adjusted Bayes factor support ≥3. The city of Kiev and the Kiev region were considered a single area. Abbreviations: HIV, human immunodeficiency virus; PL, Poland/Polish; UA, Ukraine/Ukrainian.
Relative contributions of migration events within Poland (A), within Ukraine (B), and between the 2 countries (C) to the spread of the human immunodeficiency virus type 1 variant A6. Sankey plot shows the proportions of migration events from each source region (“From”) toward the sink region (“To”). We present only migration events associated with adjusted Bayes factor (BFadj) support ≥3 (all corresponding Bayes factors are displayed in Supplementary Table 3). Colors were chosen to clearly distinguish the different types of migration between regions.
Subsequently, we analyzed viral movement in relation to the origin of the source/destination across population groups. The domestic spread of the A6 lineage that we observed was not only between Polish citizens. In 3 directions (from Lesser Poland to Lodz, from Lower Silesia to Silesia, and from Silesia to the Masovian district; Figure 5D and 5E ), we noted the migration-related dispersal of the A6 variant between Polish citizens and migrants from Ukraine. Interestingly, a considerable interregional migration path from Lesser Poland to the Masovian district was identified among Ukrainian migrants with HIV diagnosed in Poland (Figure 5C ). The present study also revealed robust evidence of migration across Ukraine (Figure 6B ). The Donetsk region was the leading exporter, accounting for one-third of viral migration (33.3%) with A6 variant dispersal toward 4 centers.
Numerous A6 lineage dispersal events were observed from the Kiev region (including the Ukrainian capital), constituting more than a quarter (27.4%) of migrations, followed by the Southern region of Odesa (22.6% of migration events; 5 directions). Three centers concatenate Kiev as the predominant destination spot of migration events (29.4%). The second catchment of virus transmission was the Donetsk region (19.1%), especially for migrations from Odesa (16.7%) and the Poltava region (Figure 6B ). Notably, for 29.0% of migration events, sink destinations of A6 dispersal were unknown because of the lack of detailed sequence origin data (Figure 6B and 6C ). From a chronological perspective, our estimations showed that the density of migration events from Ukraine to Poland peaked between 2011 and 2016.
DISCUSSION
Our study presents the evolving transmission dynamics of HIV-1 A6 within and between Poland and Ukraine. We found that the increasing A6 epidemic in Poland has its source in Ukraine—both the earliest and the majority of the introduction events come from Ukraine. In addition, the maximum clade credibility summary tree (Figure 2) estimated from the global data set was in line with Ukraine being the origin of the current HIV-1 A6 pandemic [2]. Our calculations uncovered almost unidirectional viral migration from Ukraine to Poland (99.4%). The imbalance in the number of incoming and outgoing virus transmissions between Poland and Ukraine indicates that Poland may become a gateway for HIV-1 A6 influx into European Union countries, but this observation requires future studies and shows the need for international efforts.
HIV prevalence in the general population of Ukraine varied from 1.16% in 2008% to 0.9% in 2020 [23], indicating that until the end of 2021, >12 000 migrants to Poland might have been HIV infected. Annually, however, <9% (n < 140) of new registered infections in Poland were diagnosed among migrants with HIV [1, 24]. Our data set contained almost a quarter of the samples (24.7%; Supplementary Table 6) derived from people of Ukrainian citizenship; still, infrequent sampling among foreigners might result in missing links in identified Polish-Ukrainian clades. The Polish Ukrainian cross-border population migration patterns broadly match our A6 variant binational dispersal profile.
The most intense cross-border movement of the A6 variant occurred between 2011 and 2016 at the time of several powerful events in Ukraine: the Maidan demonstrations in Kyiv, the annexation of the Crimea, and the onset of the hybrid war in the East. The Donetsk region, where the armed conflict was long lasting, continuing since 2014, was the primary source of A6 dispersal. The Russian military offensive resulted in the ongoing internal and external migration away from conflict-bound territories [9]. The Donetsk region was also strongly affected by HIV, with the fifth highest prevalence in the country (1.3% population in 2019). The region with the highest prevalence (1.7% population in 2019) in Ukraine, the Odesa district [23], was the other central hub of variant dispersal.
The top destination spot was the Masovian region, where the Polish capital city, Warsaw, is located. With the highest population in Poland, it is a political, financial, and cultural center of the country. Because this Polish metropolitan area is a natural migrant destination, more than one-third of all foreigners in Poland live and work here [10]. Furthermore, migrants most often inhabit large cities that are regional capitals. Apart from Warsaw, many Ukrainians reside in Szczecin (the capital of West Pomerania) and Łódź, where the greatest growth in migrant flow was observed [10].
Our dual-trait analyses allowed us to explore simultaneously the national patterns of A6 variant movements. The reconstructed patterns of interprovincial spread of the HIV-1 variant confirmed our group’s previous findings [6] that the burden of the subtype (including A6) across Poland was unevenly dispersed geographically. The A6 variant transmissions were common in populous, urbanized, and well-connected regions in Central and Southwestern Poland, which might have been a key factor in the transmission dynamics between the main cities in the country. Previous molecular analyses on interregional transmission networks identified exclusively populations of men who have sex with men and subtype B across 9 Polish regions. Four regions, namely West Pomerania, Lesser Poland, Greater Poland, and Lower Silesia, were most involved in virus migrations. Of note, Central and Eastern Poland areas were missing in the cited study [25].
We also characterized the spatial distribution of A6 variant within Ukrainian regions. Because of the high HIV prevalence in Eastern and Southern Ukraine, each wave of internally displaced people likely includes many HIV-infected individuals. Moreover, internally displaced people with HIV infection more frequently resettled in the spots that imported more viral lineages [9]. Consistent with this trend, we found that viral dispersal of the A6 variant in Ukraine occurred between the Eastern, Central, and Southern regions. The Donetsk region was the leading hub for the within-Ukraine virus spread. Our data were in line with the observations that many internally displaced people moved away from the frontline area but many remained within the Donetsk region, just in other counties [9]. Next, we found that the Kiev region (with the city of Kiev) was the vital hub of A6 dispersal throughout the country. This region is an important industrial, scientific, and cultural center and accounts for 11.3% of the population of Ukraine [26]. The Odesa region, a historical epicenter of the A6 epidemic outbreak [2], was the third estimated transfer spot for ongoing virus circulation.
Our observations of viral movement from people with HIV of Ukrainian origin toward Poland and the large migration during the recent and ongoing war suggest that in coming years the intensity of migration events from Ukrainian people, including those with HIV diagnosed in Poland will likely increase, which will require notable surveillance efforts. Authorities should consider the size of the identified transmission networks and the high number of individuals as a challenge highlighting the urgent need for effective public health measures to inform virus prevention and elimination strategies. Poland may act as a hub of A6 dispersal to other European Union countries, especially with the expected further displacement of HIV-infected Ukrainian war refugees.
This may change the European HIV-1 subtype distribution pattern, especially with possible lower efficacy of long-acting injectable cabotegravie-rilpivirine ART and cabotegravir-based preexposure prophylaxis related to the A6 variant [7, 8]. Rilpivirine and cabotegravir resistance–associated mutations with the A6 variant are rare among treatment-naive and treatment-experienced HIV-1–infected patients from Poland. Nevertheless, subsubtype A6 becomes the second most prevalent lineage across real-life cohorts. Continued subtyping and screening for resistance-associated mutations will remain essential to the implementation of long-acting injectable treatment [27].
The principal limitation of our study was related to the availability of the HIV-1 pol sequences. The data set of Polish isolates may not fully reflect the prevalence of infections in the overall population because they originated from HIV-1 regional centers for primary transmitted drug resistance or treatment failure analysis; there is no national systematic drug resistance surveillance program. Moreover, our data set constituted only a fraction of all genotyped Ukrainian individuals, only those included in public databases, while there were no available sequences from 2017 onward. Most of the sequences lack metadata about the transmission mode, making dynamic analysis across risk groups impossible. To minimize drawbacks of the sequence data set, we used discrete trait analysis and changed model sensitivity to a biased sampling of subpopulations. We adopted a filter based on the relative abundance of samples from each location in the alignment, in a way that is very comparable to the tip-date randomization test for temporal signal, and we combined the geographic data from different partitions that describe distinct samples of the actual epidemic, in order to minimize the chance of false-positive migration linkages and connections with covariates [28, 29].
In conclusion, HIV-1 A6 subtype introductions to Poland originate most commonly from the Eastern regions of Ukraine that have been long affected by armed conflict. Poland's primary hubs of virus transmissions were located in large population centers. Geospatial A6 transmission networks were expanding in Poland, sustained by local viral dispersal and cross-border migration originating from Ukraine. More introductions may be expected since the initiation of war in 2022, which will require increased attention. There is an urgent need for expanded HIV diagnostics and ART initiation among war-displaced populations to reduce the transmission networks.
Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Author Contributions. K. Serwin was responsible for the design of the analyses, devised the study together with M. P., annotated viral genetic sequences with epidemiological information, developed the phylogenetic analysis and wrote the first draft of the article. A. C. performed the majority of phylogeographic analyses, assisted in writing the manuscript, and supported the analyses. K. Scheibe, A. U., and P. Z. performed all human immunodeficiency virus sequencing experiments. B. A. W., E. S., I. C., E. J., K. W. C., P. J., M. B. J., A. W., and B. S. provided samples from their clinical centers to be analyzed within this study and assisted in writing the manuscript. M. P. provided substantial support in every stage of analyses, assisted in phylogenetic analysis, and wrote the second draft of the manuscript. All authors reviewed and approved the final manuscript.
Financial support. This work was supported by the Polish National Science Centre (SONATA-BIS grant UMO-2018/30/E/NZ6/00696.).
References
Author notes
Potential conflicts of interest. B. A. W. reports payment or honoraria for lectures from Gilead and GSK. I. C. reports fees for lectures from Gilead. B. S. reports grants or contracts, paid to the organization, from Gilead and Janssen; consulting fees and support for attending meetings and/or travel, paid to the author, from ViiV and Gilead; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events, paid to the author, from ViiV, Gilead, and Merck; participation on a data safety monitoring or advisory board with ViiV, paid to the author; and receipt of equipment, materials, drugs, medical writing, gifts, or other services from AbbVie, provided to the organization. All other authors report no potential conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
