The possibilities of using HIV-1 molecular cluster analysis to study the epidemic process of HIV infection

Full Text

Introduction

Human immunodeficiency virus type 1 (HIV-1) is one of the most significant pathogens influencing the development of the global epidemic over the past decades. However, epidemiological surveillance of HIV infection has certain limitations associated with the long asymptomatic course of the disease, high levels of stigmatization of people living with HIV (PLHIV), as well as the limited informative value of classical epidemiological investigations.

Under these conditions, genomic epidemiological surveillance becomes crucial. It is based on monitoring the genetic properties of infectious disease pathogens — the driving force behind the development of the epidemic process — and allows for the collection of objective data on the structure and dynamics of the epidemic, independent of the completeness and reliability of information obtained during epidemiological investigations [1].

A distinctive feature of HIV-1 is its exceptionally high rate of evolution, which makes molecular genetic methods instrumental in studying the origin, spread, and dynamics of HIV infection, since the nucleotide sequence of HIV-1 becomes a unique and objective characteristic of PLHIV. It has been shown that in PLHIV with an epidemiological link, the viruses exhibit high genetic similarity (over 98%), and the degree of this similarity is inversely proportional to the length of the HIV-1 transmission chain [2].

This principle underlies the analysis of HIV-1 molecular clusters — one of the most widely used bioinformatics methods in epidemiological surveillance. HIV-1 molecular cluster analysis has been implemented in epidemiological surveillance practices at the national level in the United States [3], the United Kingdom [4], Canada [5], Germany [6] and China [7]. In Russia, HIV-1 molecular cluster analysis has been described only in relation to investigations of HIV-1 transmission during the provision of medical care , however, the scope of application of the method is much broader. At the same time, this field is new and raises a number of methodological and ethical questions.

In this regard, the aim of this review is to systematize the potential applications of HIV-1 molecular cluster analysis for studying the characteristics of the HIV epidemic, as well as to identify the methodological and ethical aspects of its application in the epidemiological surveillance system.

Methodological foundations for the identification of HIV-1 molecular clusters

HIV-1 molecular clusters (HIV-1 transmission clusters) are defined as groups of viral nucleotide sequences characterized by high genetic similarity and reflecting an epidemiological link between PLHIV from whom they were derived. The identification of an HIV-1 molecular cluster indicates that cases of HIV infection are linked through one or more epidemiological chains and represent a fragment of a broader HIV-1 transmission network. It has been shown that the rate of HIV-1 transmission within molecular clusters significantly exceeds the average rate of viral spread in the general population [8], allowing molecular clusters to be considered as indicators of active HIV-1 transmission networks .

It is extremely important to note that the composition and size of the identified HIV-1 molecular clusters critically depend on parameters such as the selected genetic distance threshold, the applied analytical approach, as well as the sequencing coverage.

The identification of viral molecular clusters in the study of HIV infection is carried out in two ways. The first, which is computationally simpler, is based on calculating the degree of genetic similarity among HIV-1 nucleotide sequences. The second method for identifying HIV-1 molecular clusters, which is computationally and methodologically more complex, is based on phylogenetic analysis. Phylogenetic analysis is most commonly performed using the maximum likelihood method with bootstrap support ranging from 100 to 1,000. However, more complex models, such as the Bayesian method, can also be used. Phylogenetic analysis not only allows for the identification of clusters based on genetic similarity but also takes into account the evolutionary history of the virus. The results of cluster analysis based on a phylogenetic approach are more reliable; therefore, at the individual level — for example, in investigations of nosocomial outbreaks [9] — as well as when studying samples of limited size, this approach should be applied. At the same time, for analysis at the population level, it is preferable to use an approach based on genetic distance analysis.

For molecular cluster analysis, it is acceptable to use both HIV-1 nucleotide sequences obtained through conventional population sequencing (Sanger sequencing) and those obtained through massively parallel sequencing (Next-Generation Sequencing, NGS). Most HIV-1 nucleotide sequences used for molecular cluster analysis were obtained for clinical purposes during routine HIV-1 drug resistance testing using Sanger sequencing of the pol gene regions which encode the virus protease and part of its reverse transcriptase, ranging in length from 500 to 1,000 nucleotides [10]. However, NGS is being used increasingly often, and currently, when analyzing large numbers of clinical samples, it is comparable to population sequencing in terms of labor intensity and cost. At the same time, NGS offers significant advantages in terms of informative value. In particular, NGS makes it much easier not only to sequence the most well-studied pol region, which contains the genes for protease, reverse transcriptase, and integrase, but also to obtain nucleotide sequences of nearly complete HIV-1 genomes [11]. The reliability of cluster analysis results is proportional to the length of the nucleotide sequences under study [12, 13]; therefore, an increasing number of studies are based on the analysis of complete viral genomes. The choice of HIV-1 sequencing method is determined by the study objectives, equipment availability, as well as the scale of molecular genetic surveillance.

An important parameter affecting the informative value and reliability of HIV-1 molecular cluster analysis results is sequencing coverage — the proportion of HIV-infected patients for whom the viral nucleotide sequence is known, relative to all PLHIV living in the area under study. To obtain reliable analysis results, the HIV-1 sequencing coverage must be at least 10% [14].

The choice of the genetic distance threshold is extremely important and fundamentally determines the results of the analysis of HIV-1 molecular clusters. Thresholds are selected based on the study objectives, as well as the characteristics of the sample under investigation (sequencing density, length of nucleotide sequences, duration of clinical sample collection, and geographic homogeneity). In particular, the U.S. Centers for Disease Control and Prevention (CDC) recommends using a 0.5% threshold when searching for epidemiologically linked cases of recent HIV transmission , and a 1.5% threshold when searching for all potentially linked cases of the disease [10]. Furthermore, a method for selecting the optimal threshold, at which the maximum number of clusters is identified in the sample under study, is widely used [2, 15].

When using a phylogenetic approach, in addition to the genetic distance threshold, a significance threshold for the identified cluster is also applied, which is determined by the branch support (bootstrap support). In this case, to identify clusters, the genetic distance threshold is set in the range from 0.5% to 4.5%, and the bootstrap support from 70% to 100%, with the most common values being 1.5% and 90% [10].

Identification of disease outbreaks and assessment of the epidemiological patterns of HIV infection

One of the most relevant applications of HIV-1 molecular cluster analysis is the identification of HIV infection foci, the determination of their boundaries, and their subsequent detailed characterization. Analysis of large (typically more than 4 people) and active (including HIV-infected individuals diagnosed within the last 12 months) molecular clusters allows for the identification of population groups and areas where transmission of the pathogen is currently most intense . Unlike historical clusters, active clusters indicate growing epidemic foci. Their identification can be effectively utilized in operational epidemiological analysis. The identification of rapidly growing clusters is of particular importance. Rapid growth may indicate a high viral load in HIV-infected individuals, active risky behavior, or a high density of social contacts. It has been shown that within rapidly growing clusters, the spread of infection can be more than 30 times higher than the regional average , therefore, such clusters represent priority targets for anti-epidemic measures. There is no single criterion for defining a cluster as rapidly growing; however, in certain studies, authors most often classify clusters as such if they have increased by 5 or more diagnosed cases over the past 12 months [16–18].

Comparing the characteristics of HIV-infected individuals within molecular clusters with those of patients outside the clusters makes it possible to identify factors associated with the accelerated spread of infection and to pinpoint at-risk groups. Analysis of viral molecular clusters makes it possible to identify a susceptible cohort—a group of individuals united by common social, behavioral, or geographical characteristics, within which conditions for sustained viral circulation persist. Identifying such cohorts allows us to understand not only current but also potential directions for the further spread of the infection. The results obtained can serve as the basis for developing the most effective preventive and anti-epidemic measures.

A specific application of viral molecular cluster analysis in identifying HIV infection foci is the application of phylogenetic analysis during investigations of outbreaks associated with the provision of medical care. The application of HIV-1 molecular cluster analysis in this field has a long history and such a substantial evidence base that it is currently used in forensic practice in many countries [9]. This application of molecular cluster analysis is also described in the regulatory and methodological documents of the Russian Federation .

Clarification of the structure of HIV-1 transmission routes

Molecular cluster analysis can be used to refine the structure of HIV-1 transmission routes. Most published studies in this field focus on two main areas. Firstly, cluster analysis is used to determine the likely route of infection among cases for which this information is missing. M. Horecki et al. described the likely structure of transmission routes by analyzing connections within clusters among patients for whom the route of infection was not determined during the epidemiological investigation [19]. The researchers calculated the number of connections that patients with an unknown route of infection form with patients infected through heterosexual and homosexual sexual contact, as well as with patients who contracted HIV through intravenous drug use. The ratio of the number of contacts between patients with an unknown route of infection and patients in other groups showed that 45.1% of unknown cases may be cases of HIV-1 infection through homosexual contact.

Another significant body of research focuses on identifying stigmatized routes of HIV transmission, such as sexual transmission through homosexual contact and parenteral transmission through injection drug use [20–22]. One of the earliest studies is that by S. Hué et al., which assessed the proportion of men who reported heterosexual transmission but who may actually have been infected through sexual contact with men [23]. The study was conducted on a large sample of PLHIV in the UK, including nucleotide sequences of HIV-1 subtype B. Analysis of HIV-1 molecular clusters was performed using a phylogenetic approach with additional validation of the clusters via Bayesian analysis. After identifying the clusters, the authors divided them into three types: heterosexual-only, homosexual-only, and mixed. In cases where a man reported infection through heterosexual contact but his virus was found in a homosexual cluster, the authors concluded that this was likely a hidden case of homosexual transmission.

Determination of HIV spread patterns and identification of imported cases of HIV infection

One of the key applications of HIV-1 molecular cluster analysis is the study of the spatial characteristics of the epidemic and the identification of imported cases of infection. Comparing viral nucleotide sequences obtained at the national or regional level with sequences from international databases allows for assessing the degree of containment of the epidemic, identifying the main routes of HIV-1 transmission, and tracking the further spread of the virus within the country.

Phylogenetic and phylogeographic analyses are widely used to reconstruct the spatial dynamics of the HIV-1 epidemic and determine the role of international and interregional virus spread. Within these approaches, viral nucleotide sequences are analyzed in conjunction with data on the geographic origin of samples, which allows for the reconstruction of probable virus transmission routes and the estimation of the time intervals of its spread into the area under study.

In this context, the application of the Bayesian method allows for determining the most likely geographic origin of the virus and calculating the likelihood of such an origin, which makes it possible to quantitatively assess the contribution of the virus spread to the formation of the epidemic.

One example of the application of such methods is the reconstruction of the spread of subtype A (formerly IDU-A) in the post-Soviet states. In a study by F. Díez-Fuertes et al., using Bayesian analysis, it was shown that the time to the most recent common ancestor for this variant of the virus dates back to the mid-1990s—approximately 10 years before its widespread transmission among people who inject drugs [24]. Odessa (Ukraine) was identified as the most likely site of the epidemic origin, after which the virus spread to Russia, Belarus, Kazakhstan, and Uzbekistan. The authors also demonstrated that migration flows between countries played a significant role in the early stages of the epidemic, whereas subsequently the spread of the virus was sustained primarily by internal transmission within individual countries.

A number of studies also demonstrate that analysis of viral molecular clusters allows for an assessment of the contribution of migration flows to the epidemic process of HIV infection in the studied region. For example, K. Serwin et al. showed that sub-subtype A6 entered Poland as a result of active migration flows from Ukraine and subsequently spread throughout the country [25].

Phylogeographic analysis makes it possible to determine whether an epidemic is predominantly domestic (with local clusters dominating) or sustained by ongoing international and interregional importations. If most clusters share a common ancestor located within the study area, this indicates a high degree of containment of the epidemic and the predominance of virus transmission within the country. Conversely, the identification of numerous molecular clusters of the virus originating from different geographic regions indicates a significant contribution from external sources of the virus.

Analysis of the genetic diversity of the virus can provide additional information about the spatial structure of the epidemic. The detection of HIV-1 genetic variants atypical for a given territory may indicate international importation of the infection. For example, molecular genetic studies of the HIV-1 epidemic in the countries of the former Soviet Union have shown that the widespread prevalence of subtype A is associated with several early cases of virus transmission, followed by rapid territorial spread of this variant to various regions of the country [26].

Thus, the analysis of HIV-1 molecular clusters not only helps identify local outbreaks within the country but also serves as a vital tool for strategic epidemic management, enabling evidence-based decision-making when planning prevention and epidemic control measures at the regional and national levels.

Evaluation of the effectiveness of preventive and anti-epidemic measures

By analyzing viral molecular clusters, not only is it possible to identify vulnerable groups and areas requiring additional preventive and anti-epidemic measures, but also to assess the effectiveness of these measures.

One such indicator is the change in the growth rates of molecular clusters. A comparative analysis of cluster growth dynamics before and after the implementation of preventive and anti-epidemic measures allows for an assessment of their impact on the rise in incidence. A decrease in the number of HIV-1 nucleotide sequences within large and growing clusters following the initiation of anti-epidemic measures indicates a reduction in the intensity of viral transmission.

P.J. Peters et al. used this approach to evaluate the effectiveness of epidemic control measures — including mass testing and a needle exchange program—in an outbreak that occurred among oxycodone users sharing needles in 2014–2015 in Indiana (USA) [27]. To this end, viral sequencing was performed on each newly identified HIV-infected patient, and an analysis of viral molecular clusters demonstrated that the growth of clusters associated with the identified outbreak ceased several months after the implementation of epidemic control measures. In a study by C. Alpren et al. describing an outbreak among opioid-using injection drug users in Massachusetts (USA) in 2015–2018, analysis of HIV-1 molecular clusters revealed ongoing viral transmission within the outbreak following the implementation of anti-epidemic measures, which served as the basis for their intensification [28].

Using molecular cluster analysis based on a phylogenetic approach, the effectiveness of antiretroviral therapy (ART) as a measure to prevent HIV-1 transmission was evaluated. In a number of large prospective studies of serodiscordant couples [29, 30] it was shown that in individuals receiving effective ART and having a consistently undetectable viral load, viral transmission does not occur, and the identified cases of infection among their sexual partners are epidemiologically unrelated and attributable to external sources of infection.

It is also possible to quantitatively assess the effectiveness of ongoing anti-epidemic and preventive measures by calculating the effective reproduction number (Re), which reflects the average number of secondary infections resulting from a single infected individual in the population. A Re value greater than one indicates the continued spread of the infection, whereas a value less than one indicates that the epidemic is subsiding.

Unlike classical epidemiological approaches based on case registration, phylodynamic methods allow for the estimation of Re based on the analysis of HIV-1 nucleotide sequences. The use of the Bayesian method makes it possible to reconstruct the dynamics of viral transmission over time and to assess the impact of implemented anti-epidemic and preventive measures on the intensity of the epidemic process.

Thus, in a study by Y.H. Chen et al., it was established through Re estimation that the greatest effect on reducing the intensity of the HIV epidemic in the United States at the population level was achieved by retaining patients on ART, as well as by measures targeting the vulnerable group of men who have sex with men, in particular pre-exposure prophylaxis [31].

It should be noted that in the practical work of public health authorities, the direct calculation of Re using phylodynamic models is used only to a limited extent, due to the complexity of bioinformatics analysis. Consequently, the epidemiological surveillance system more frequently uses an indirect indicator reflecting the intensity of transmission, specifically the growth dynamics of HIV-1 molecular clusters.

Thus, the application of these approaches in the epidemiological surveillance system allows for the detection of changes in the intensity of viral transmission earlier than it is possible through the analysis of incidence rates, and also provides a scientifically sound assessment of the effectiveness of ongoing interventions, enabling timely adjustments to HIV prevention strategies.

Limitations of molecular cluster analysis

Although molecular cluster analysis offers a wide range of possibilities for epidemiological surveillance, it has a number of limitations that must be taken into account. The most significant external limitation is the incompleteness and non-uniformity (bias) of the sample of HIV-infected individuals for whom viral nucleotide sequences are known. Indeed, only in a small number of studies has HIV-1 sequencing been performed for a significant proportion of patients [32]. For the most part, however, the cohorts analyzed constituted a small fraction of HIV-infected individuals residing in the study area [33, 34]. At the same time, it is known that HIV-1 drug resistance testing — which most often yields viral nucleotide sequences for subsequent use in molecular epidemiological studies — is conducted unevenly. However, even in cases where the HIV-1 nucleotide sequence is known for the majority of diagnosed HIV-infected individuals, it is necessary to take into account that there is unevenness in the provision of medical care and the identification of HIV-infected individuals. Hard-to-reach, stigmatized groups are less likely to be tested and engaged in the healthcare system, which also affects the results obtained. This limitation is not a flaw in the cluster analysis itself; however, it is the sequencing depth and the adequacy of the study sample that determine how accurately molecular clusters will reflect epidemiological links.

The most significant limitation of cluster analysis itself is the influence of the level of genetic similarity established by researchers on the identification of groups of related HIV sequences. Based on accumulated data, it is known that a threshold of 0.5% allows for the detection of recent cases, while increasing it to 3.0–4.5% should be used to identify outbreaks that have been developing over a long period of time [10].

It should also be noted that even when there is a high degree of genetic similarity among HIV-1 nucleotide sequences, molecular clusters do not reflect direct transmission of the virus; in particular, this is because PLHIV whose nucleotide sequences fall within the same molecular cluster may have been infected from a third source [8].

Furthermore, using existing methods for analyzing molecular clusters, it is not possible to determine the direction of viral transmission with high certainty. Recent studies show that this possibility potentially exists when using HIV-1 nucleotide sequences obtained through mass parallel sequencing [35]; however, the accuracy of the approach does not allow for its application at the individual level.

Ethical and legal aspects of molecular cluster analysis

The use of molecular cluster analysis should also be evaluated from ethical and legal perspectives. Despite the significant potential of this approach for improving epidemiological surveillance, its use carries risks of causing harm to individual PLWHIV and reducing the effectiveness of epidemic response at the population level. Experts are focusing on issues such as the criminalization of HIV infection, the stigmatization and discrimination of vulnerable population groups, as well as ensuring confidentiality and the protection of personal data.

The issue of the potential use of molecular genetic analysis results in legal proceedings is of particular importance. Given that many countries, including the Russian Federation, have established criminal liability for the intentional transmission of HIV to another person, there is a risk that cluster analysis data could be interpreted as evidence of transmission from one person to another. However, it must be noted that molecular cluster analysis can only identify probable epidemiological links and does not allow for the reliable determination of the direction of viral transmission or the fact of direct infection. Consequently, the application of the results of such studies as evidence in criminal cases is methodologically incorrect and may lead to miscarriages of justice. A permissible exception may be their use in strictly limited situations, such as when investigating nosocomial outbreaks, where it is possible to compare molecular and clinical-epidemiological data.

Another equally important aspect is the risk of increased stigmatization and discrimination against PLHIV, as well as certain social groups. Cluster analysis results may indicate a higher rate of transmission in certain populations, including people who inject drugs, men who have sex with men, or socially vulnerable individuals. Incorrect interpretation or public dissemination of such data may contribute to increased negative attitudes toward these groups, their social isolation, and reduced access to medical care. Ultimately, this may lead to the opposite effect—a decrease in HIV detection rates and a deterioration in control over the epidemic.

Significant risks are also associated with ensuring the confidentiality of genomic information about the virus and related data on PLHIV. Despite the use of depersonalized data, the demographic and epidemiological characteristics being analyzed could potentially lead to the indirect identification of PLHIV. In this regard, the conduct of molecular epidemiological studies must be accompanied by strict adherence to data protection requirements, including restricting access to information, applying anonymization procedures, and controlling data transfer between organizations.

The issue of informed consent warrants separate consideration. There is no consensus within the scientific and professional community regarding the necessity to obtain separate informed consent from patients for the use of genomic information about the virus in the context of epidemiological surveillance.

On the one hand, genomic data can be viewed as part of routine laboratory diagnostics and used in the interest of public health. On the other hand, given the potential risks to patients’ rights and interests, it seems appropriate to ensure that PLHIV are informed about the possible application of the data collected, including the analysis of molecular clusters. Increasing the transparency of such studies helps build trust in the healthcare system and reduces the risk of negative reactions from patients.

Thus, the implementation of HIV-1 molecular cluster analysis into epidemiological surveillance practice should be carried out in accordance with the “do no harm” principle and should be primarily focused on population-level applications. The application of analysis results at the individual level and their application in legal proceedings should be strictly limited to investigations of nosocomial outbreaks.

Conclusion

The analysis of HIV-1 molecular clusters is currently instrumental in genomic epidemiological surveillance of HIV infection, significantly expanding the capabilities of traditional epidemiological surveillance. Experience from countries where this method has been integrated into the HIV epidemiological surveillance system has demonstrated its high effectiveness for in-depth study of the epidemic process, providing objective information on its structure, intensity, and geographical characteristics. At the same time, the use of phylodynamic methods, including the determination of Re, expands the possibilities for quantitatively assessing the effectiveness of preventive and anti-epidemic measures, as well as identifying the characteristics of HIV-1 transmission.

It is important to note that the implementation of molecular cluster analysis in epidemiological surveillance must be carried out with due consideration of the methodological and ethical limitations identified in this review. Thus, the effectiveness and reliability of cluster analysis results directly depend on the quality and representativeness of the source data, including sequencing depth, the completeness of clinical and epidemiological information on PLHIV, as well as the correct selection of analytical parameters, primarily the genetic distance threshold. This necessitates a rigorous methodological approach to conducting research, upon which the interpretation of the results depends. An equally important point is taking into account ethical constraints related to the necessity to protect personal data, prevent stigmatization, and ensure that the results of the analysis are not used to the detriment of patients. A priority area is the application of results from the analysis of HIV-1 molecular clusters at the population level to improve epidemiological surveillance and enhance the effectiveness of measures aimed at reducing incidence.

Furthermore, the widespread implementation of this method in epidemiological surveillance of HIV infection in the Russian Federation requires a regulatory and methodological framework, which is currently limited to describing the application of the method in investigating cases of HIV-1 transmission during the provision of medical care.

The prospects for further development of the method are linked to increased access to HIV-1 sequencing technologies, which will enable the achievement of the sample depth necessary for obtaining reliable cluster analysis results, with the expanded use of NGS, which will increase the informative value of the analysis, and the development of platforms (databases) that ensure centralized storage of HIV-1 genomic data and clinical-epidemiological information on PLHIV and enable automated bioinformatics analysis. Another promising field for the development of the method is conducting cluster analysis in near real-time, which is only possible with rapid HIV-1 sequencing and timely database updates. Early detection of emerging clusters enables the timely planning of preventive and anti-epidemic measures aimed at preventing new infections.

Given the predicted increase in sequencing coverage among PLHIV and the ongoing improvement of bioinformatics tools, a qualitative leap in genomic epidemiological surveillance over the next decade can be expected [36]. This will create the conditions for a more accurate assessment of the epidemic situation and form the basis for evidence-based policy decisions and the development of appropriate measures aimed at reducing HIV incidence and preventing the emergence of epidemic outbreaks.

At the same time, the analysis of HIV-1 molecular clusters should not be viewed as an alternative to existing methods. Its most effective application is being part of a comprehensive approach in conjunction with classical epidemiological methods.

About the authors

Vasily G. Akimkin

Central Research Institute of Epidemiology

Email: crie@pcr.ru
ORCID iD: 0000-0003-4228-9044

Dr. Sci. (Med.), Professor, Full Member of RAS, Director

Russian Federation, Moscow

Alina A. Kirichenko

Central Research Institute of Epidemiology

Author for correspondence.
Email: kirichenko@cmd.su
ORCID iD: 0000-0002-7116-0138

Cand. Sci. (Med.), senior researcher, HIV diagnostic and molecular epidemiology laboratory

Russian Federation, Moscow

Dmitry E. Kireev

Central Research Institute of Epidemiology

Email: dmitkireev@yandex.ru
ORCID iD: 0000-0002-7896-2379

Dr. Sci. (Med.), Head, HIV diagnostic and molecular epidemiology laboratory

Russian Federation, Moscow

References

Supplementary files