São Paulo is the most populous State in Brazil and the driving force of the Brazilian economy, with an estimated population of 44,396,484 inhabitants in 2015 (21.6% of the Brazilian population and a third of the National grass product). The metropolitan area of São Paulo comprises 39 cities, and that of Campinas, 42 cities, corresponding to 47.8 and 9.6% of the state’s population, respectively.¹ In São Paulo State, 236,434 AIDS cases were reported from 1980 to 2014. In 2015, a total of 1,137 new HIV infections were reported in São Paulo city (1).

HIV commonly spreads more rapidly in specific groups, such as those men who have sex with men (MSM), sex workers and people who inject drugs. According to the national epidemiological report, since 2008, the HIV epidemic has shown a significant increase in the number of cases among MSM. In the 18–24 age group, the proportion increased from 30.2% in 2007 to 43.7% in 2012 (2).

The HIV-1 epidemic in São Paulo is dynamic with a genetic complexity driven by subtypes B, F1 and recombinant BF1 (3–6), however recent studies have shown a continuous increase of subtype C among São Paulo patients (7, 8). The Brazilian government has been sponsoring a large-scale implementation of highly active antiretroviral therapy (HAART) since December 1996 and until June 2013, approximately 400,000 patients were receiving HAART in Brazil (9). The HAART improves the quality of life of the people living with HIV (PLWH) globally, however, virological failure when using HAART can lead to the emergence of drug resistance mutations (DRM) that increase the chance of new infections with resistant strains (10). Most studies conducted in Brazil have shown a transmitted drug resistance (TDR) prevalence similar to that observed among developed countries (11) and São Paulo follows this trend with around 8% of TDR reported (12).

Although several studies have been conducted in São Paulo State, the dynamics and determinants of transmission clusters of HIV-1 subtypes prevalent in the state of São Paulo remain unknown and deserve further investigation. In the present study, we combined clinical and epidemiological data in order to evaluate the dynamics and determinants of transmission clusters of the prevalent HIV-1 subtypes in the state of São Paulo, Brazil.

To improve the knowledge of the HIV-1 transmission dynamics in the state of São Paulo, Brazil, this study considered a large dataset that examines the demographic and clinical characteristics of individuals infected by the subtypes B, C and F1. The observed prevalence of HIV-1 subtypes was similar to that observed in previous studies in the region being the epidemic was driven by subtype B, followed by subtype F and the less frequent, subtype C (21). However, when only patients with recent diagnoses (between January 2013 and February 2015) are considered, the frequency of subtype C is about 10%. This picture contrasts with that of the early 2000s when the prevalence of the subtype C in the region was approximately 3.8% (4). This proportion is consistent with more recent studies on HIV-1 diversity, which report an increase in the subtype C in São Paulo (8, 22). Another remarkable concern involving subtype C is the high proportion of ART-naïve patients compared to ART-experienced patients, supporting the hypothesis of a more recent introduction of subtype C in the state of São Paulo (23).

Our findings showed that subtype B had a lower median CD4 T-cell count compared to subtypes F1 and C, as also recorded in previous studies (24), however, that comparison revealed no differences in viral load levels between the HIV-1 subtypes. This study has not performed any survival analyses on the three subtypes, though other authors have suggested an association between the viral subtype and progression to AIDS (25–28).

This study includes samples from 53 different cities of São Paulo state which is responsible for 32% of the Brazilian AIDS epidemic. Surveillance studies on primary resistance performed in Brazil estimated TDR prevalence between 5 to 15%, depending on the geographic region (12, 29, 30). Another study conducted in 2015 detected a primary resistance rate of 9.2% in samples from patients with recent diagnoses in São Paulo state (31). The current study found 11.8% of TDR, with a high rate of resistance observed in subtypes F1 (18.2%), followed by subtypes B (12.2%) and C (3.8%). Previous studies conducted in the South of Brazil and involving a cohort of ART-naïve patients found a lower rate of DRM associated with resistance to subtype C compared to subtype B (32). Though many studies performed in different Brazilian states have reported rates of primary resistance, there is still a lack of information on this rate among different high-risk groups. It is known that, in Brazil, the MSM population is 29 times more susceptible to being infected compared to the whole population (33). In Latin America, MSM is the most commonly exposed to the risk of HIV-1 infection and represents the highest portion of new infections (34). The study by Bermúdez-Aza et al. (35), evaluated 299 HIV-infected MSM from 9 Brazilian cities and reported a rate of primary resistance of around 19.3% which was consistent with the data observed in the present study.

The concept of transmission clusters is controversial, and consensus has yet to be established. The definition of clusters adopted in this study was determined by bootstrap support ≥90% and a genetic distance ≤6.0%. Though many studies have adopted a genetic distance of up to 4.5%, we believe that the threshold adopted is necessarily more flexible, given the fact that our study considers 10 years of the epidemic, during which time patients studied have been in different phases of the disease and with different time of within-host viral evolution. Therefore, our study detected a putative low cluster prevalence of 18.3%. This low cluster prevalence could be explained in part because we were conservative when defining the cluster using two approaches, genetic distance and branch support. We found statistical differences in the clustering when the three circulating subtypes were evaluated. In subtype C, 40% of the patients were found to be transmission clusters, a value much higher proportion than that found among the B and F1 subtypes. One hypothesis for this difference is the fact that subtype C was introduced more recently and is affected by the distance from the founder virus (36, 37).

Studies using phylogenetics to describe transmission clusters are being increasingly used in developed countries. The study by Lewis et al. (38) describes the HIV-1 transmission cluster in London where was reported that approximately 25% of the MSM population was associated with various clusters. Meanwhile, a study involving a heterosexual cohort in the United Kingdom reported that only 5% of the population studied was associated with transmission clusters (39).

Based on our analysis of transmission clusters, it can be determined that, regardless of the viral subtype, clusters are most represented by the MSM and ART-naïve patients that are driving the epidemic in São Paulo state. The individual risk factors experienced by MSM have been well documented. They include high frequencies of different partners, unprotected sex, and high-index viral loads (40, 41). Another factor involved in MSM is the peculiar risk associated with anal sex, as the behaviors associated with it, increase the efficacy of viral transmission, resulting in a probability of HIV transmission via anal sex is 18 times higher than vaginal sex (42).

Another interesting finding was that patients with primary resistance to subtypes B and F1 had higher associations with transmission clusters. Indeed, many individuals carrying the virus with TDR, even after removing the TDR codons, were linked to other patients, a finding that suggests domestication or endogenic transmission events. Paradoxically, the mutations associated with resistance that prevail in these clusters do not seem to reduce viral fitness highlighting the importance of pre-treatment genotype testing among recently diagnosed individuals when NNRTI-based regimens are being considered. In this line, a study involving a Swiss cohort found that ART-naïve individuals are most commonly responsible for the transmission of resistant strains in the MSM (43).

We observed that of the HIV epidemic in the state of São Paulo, 81% of the clusters in subtype B were found to be endogenous, 11% originated in other states, and 8% were from other countries. This finding may be explained by the relatively early introduction of this subtype in Brazil, which occurred in the 1960s (44) as a result, the epidemic seems to be self-sustained. Among the subtypes studied, subtype C was found to be the most frequent in the cluster originating outside of Brazil; the clusters originating in southern Brazilian states were also found to have an important influence. This finding may also be explained by the recent introduction of subtype C into the state, as well as by the fact that this is the most prevalent subtype in the world (45). Other analyses also support the origin of the subtype C in São Paulo from cities in the south of Brazil (23). Subtype F1 was found to have an important portion of clusters originating in other Brazilian states and a low percentage of clusters from outside of Brazil, perhaps because of its low prevalence worldwide.

To further clarify the issue of transmission cluster origins within the state, we divided the state into four geopolitical regions. This division revealed that Regions Northeast and the Metropolitan area have very similar profiles. They reflected independent epidemics in which more than 50% of the subtype B clusters had local origins. These differences are likely caused by the fact that each of these regions possesses its independent economic centers and a relatively mature, even though concentrated epidemic. Subtype C was found to have an important component of endogenous clusters, particularly within the metropolitan area, and was also found to be associated with the MSM population. This finding is partially similar to reports in the South of Brazil, where the epidemic linked to this subtype is associated with both risk groups, the MSM and heterosexuals (46). However, the Northwest and South regions differed from the others described above in that only subtype B was found to have an important proportion of endogenous clusters. This difference may be due to the underrepresentation of samples from the regions and/or to the low prevalence of non-B subtypes in this geographic region. However, we evaluated a large dataset from most of the regions analyzed. We take into account the reported number of HIV infections by geographic location. Thus, the percentage of HIV-1 infected patients by location follows the distribution: area 1 (Northeast, NE; 0.78%), area 2 (Northwest, NW; 0.72%), area 3 (SOUTH, SO; 0.4%) and area 4 (Metropolitan, Metro; 1.4%). Highlighting that the area with the lowest representation (area 3), which has its representation ranging from 0.4 to 0.7% if we exclude the city of Santos, which for logistical issues was not represented in our sample cohort.

Within the state of São Paulo, we found independent micro-epidemics, the strongest of which involve subtype B. We further subdivided the São Paulo metropolitan area, with a population of over 14 million into 3 different sub-regions in an attempt to understand the role of each region in the local epidemic. One particularly noteworthy finding was the fact that there were only endogenous clusters of subtype C in Region 4a (the ABCD region) and in Region 4c (the São Paulo city). It is very likely that this subtype was introduced into these two sub-regions independently and then spread independently. Meanwhile, the subtype F1 epidemic is largely confined to the periphery of the metropolitan area, among individuals in the neighborhood of the region (region 4b) and is likely to have spread from there to other areas within the city or the opposite but maintained a local spread.

We recognize the limitations of our study and a potential bias in the results. The most important of which is the sample bias, the data sampling method used, and the disproportionate amount of samples from the different cities within the state. Our research utilized sequences from the partial pol region of the HIV genome, which inherently restricts the assessment of phylogenetic signals and particularly impacts the identification of recombination events in other genomic regions. We also realize that phylogeographic and migration analyses could give us further insight into the concern of intra-city transmission. On the other hand, this study was performed based on genotyping services open to all public clinics in the State. Albeit limitations in sample transportation may have diminished the amount of samples from areas farther away from the laboratory, it probably did not impact the subtype distribution and other potential inclusion bias.

Our data suggests that the epidemic in the state of São Paulo is driven by the MSM population with a higher level of TDR, regardless of the HIV-1 subtype, and that the epidemic is sustained by ART-naïve patients. The prevalence of subtype C in the epidemic is growing in the state, particularly in the greater São Paulo metropolitan area, where individuals with subtype C infection are twice as likely to be in transmission cluster than those infected with subtypes B and F1.

The authors acknowledge that the data presented in this study must be deposited and made publicly available in an acceptable repository, prior to publication. Frontiers cannot accept a manuscript that does not adhere to our open data policies.

The studies involving humans were approved by The study was approved by the institutional ethical committee (CAAE 02298012.6.1001.0059). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

VP: Data curation, Formal analysis, Investigation, Methodology, Supervision, Validation, Writing – original draft, Writing – review & editing. AP-P: Data curation, Formal analysis, Validation, Visualization, Writing – original draft, Writing – review & editing. CS: Data curation, Formal analysis, Validation, Visualization, Writing – original draft, Writing – review & editing. JP: Data curation, Methodology, Writing – original draft, Writing – review & editing. CA: Data curation, Methodology, Writing – original draft, Writing – review & editing. MP: Data curation, Methodology, Writing – original draft, Writing – review & editing. MM: Data curation, Writing – review & editing, Writing – original draft. LC: Data curation, Methodology, Writing – original draft, Writing – review & editing. EM: Data curation, Writing – review & editing, Writing – original draft. DA: Data curation, Methodology, Writing – review & editing, Writing – original draft. AA: Conceptualization, Data curation, Methodology, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing. LB: Conceptualization, Methodology, Project administration, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing.