Young Pregnant Women Living with HIV/AIDS in Criciuma,Southern Brazil,Are Infected Almost Exclusively with HIV Type 1 Clade C

D ifferent studies had shown that in South Brazil the AIDS epidemic is distinct from all other areas in the Americas. HIV-1 clade B prevails in the Americas, but the occasional identification of non-B variants in the continent had become almost a norm. ^{1 –3} In South America HIV-1 F is identified as a minority variant in most areas. Toward the south of the continent, clade F and BF recombinants become increasingly more important ^{4 –6} with BF recombinants being the most prevalent variant in the city of Buenos Aires. ^{7, 8} On the other hand, a distinct situation is found in southern Brazil, with an important presence of HIV-1 C. This variant is observed from rates of 30%, at the most septentrional metropolitan area of the southern states, ⁹ to rates of 45% in the most meridional harbor city, Rio Grande. ¹⁰ The highest rate of HIV-1 C identified in the region (70%) is at the harbor city of Itajai in Santa Catarina, one of the cities with the highest AIDS prevalence rate. ¹¹ Although it is conceivable that small towns might reflect the situation observed in the larger metropolitan areas, available studies have accessed only the major metropolitan areas or harbor cities in the region. To improve our understanding of HIV molecular epidemiology in the region, we assessed a cohort of pregnant and childbearing women living with HIV/AIDS in three small towns in the coastal area of State of Santa Catarina attending the specialized clinic at Criciuma from January to December 2007.

The study was open to all pregnant women in this period and to 20 mothers of children living with HIV/AIDS with previous obstetric follow-up. These women agreed to participate and signed an informed consent form. TCD4 count and HIV RNA viral load were obtained from the local Reference Laboratory. Biological material (plasma and/or PBMCs cell) was collected and preserved at −20°C until transport in dry ice to the Retrovirus Laboratory of Adolfo Lutz Institute (State of São Paulo) for genetic sequencing. HIV RNA was extracted using the Quiamp viral RNA mini kit (Quiagen, Germany) according to the manufacturer's instructions. Complementary DNA synthesis was performed with the SuperScript III enzyme (Life Technologies, USA). HIV polymerase fragments spanning the entire protease (PR) gene and partial reverse transcriptase (RT, codons 1–235, relative to HXB2) were obtained by nested PCR, ¹² and gag and envelope ^13,14 were obtained from some cases with subtype C at the pol gene. Purified products were sequenced in an automatic ABI 3100 Genetic Analyzer (Applied Biosystems, USA). Chromatograms were assembled with Sequencher software and manually edited to identify nucleotide mixtures. Screening for HIV-1 subtypes was done with NCBI and REGA HIV-1 web-based subtyping tools and further characterized by phylogenetic analysis. Multiple alignments were performed using Clustal W with a reference set available in the Los Alamos HIV-1 database, using outgroup HIV-1 OSEMP1300. Neighbor joining (NJ) and maximum likelihood (ML) trees were constructed based on nucleotide substitution models (Modeltest v3.7), tested with a hierarchical likelihood ratio test using PAUP (v4b10). Zero length and bootstrapping were performed using the stepwise addition algorithm for 1000 replicates.

A representative NJ tree (GTR + I + G model) is depicted in Fig. 1. Sequences showing discordant genomic subtyping, outlier behavior on NJ trees, or any non-B signal in NCBI were selected for bootscanning analysis using SIMPLOT 2.5 to evaluate recombination breakpoints and inspection of signatures. The mutation profile was analyzed at the Stanford HIV Drug Resistance Database and tropism prediction by the 11/25 rule and at two websites, Geno2Pheno (Max Planck Institute, http://coreceptor.bioinf.mpi-inf.mpg.de/index.php) and PSSM (http://indra.mullins.microbiol.washington.edu/pssm). CCR5 (R5) prediction was assigned if results were concordant in the three methods. All the sequences generated in this study were submitted to GenBank with accession numbers GQ401279 to GQ401338. A mid-p 95% confidence interval (95% CI) was used to estimate the proportions confidence interval using Epi Info (CDC, USA).

OPEN IN VIEWER

FIG. 1.

Representative neighbor-joining phylogenetic tree of partial pol region sequences of the study (in bold), generated with the GTR + I + G model, using Paup 4.0. Reference HIV-1 subtypes were obtained from the Los Alamos HIV Sequence Database. Bootstrap values above 70% in key branches are depicted, with branches with zero length test significant values of *p < 0.05 and **p < 0.01. A shadow box highlights the clustering of the query sequence (BR07SC338, underlined) with HIV CRF_31 BC references.

HIV sequences were analyzed from 42 individuals: 24 woman pregnant at collection and 18 previous clinic clients, 14 woman and 4 unrelated children. Children's material was processed and amplified whenever the child's mother presented an inadequate sample for sequencing. The median age was 27 years old, the median TCD4 was 504 cell/mm³, and the median viral load was 3.57 log₁₀. Table 1 describes demographic and laboratory data of the study population. Among the women, all referred to sexual intercourse as the source of infection, with a median age of 15 years at the first sexual relationship. Poor adherence to antiretroviral therapy (ARV) medication was reported by 33% of cases. Additional sociobehavioral information may be found elsewhere (S.A. Manenti et al., unpublished observations) HIV-1 subtype C at pol was observed in most cases, 33/42, 78.6% (95% CI: 64–89%). Mosaic genomes were observed in 2/42, 4.8% (95% CI: 0.8–19%) with one CB and one CF. The CB mosaic at pol showed a small B segment at RT with breakpoints compatible to that of CRF_31BC (Figs. 1 and 2). Subtype F at pol was observed in 4/42, 9.5% (95% CI: 3–21%) and B pol in 3/42, 7.1% (95% CI: 1.8–18%). HIV gag and env from a subset of subtype C at pol were additionally sequenced.

OPEN IN VIEWER

FIG. 2.

(a) HIV-1 subtype composition and breakpoint position of mosaics at the pol region and reference HXB2 (K03455); (b–d) SIMPLOT images of mosaics identified at pol. (b) mosaic CF (BR07SC339), (c) HIV CRF_31 BC reference, and (d) study sample BC mosaic (BR07SC338) analyzed with the same parameters (window 250 bp; step 10, Kimura two-parameter).

All env sequences had a predicted R5 tropism (Table 1). Discordant gag/pol/env regions were observed in 1/9, 11% of cases with more than one region evaluated. Figure 3 depicts the partial pol amino acid alignment of international consensus B and consensus C (NCBI), our dataset ^9,11,12 of southern Brazil consensus C and study C sequences from nontransmitting pregnant women and transmitting mothers (nontransmitting pregnant women for study participants pregnant at the time of the study with an HIV-negative child on follow-up evaluation and transmitting mother for women participating in the study with a previous HIV-positive child). Typical clade C amino acid polymorphisms at protease (Fig. 3a) were observed, but our sequence set had amino acid signatures present in the south Brazil subtype C consensus but not in the international NCBI subtype C consensus (S12T, I19L, N37K, and K41N), the two first present in clade B consensus. L63P are observed in the majority of pregnant women, all with use of protease inhibitors. At reverse transcriptase (RT) (Fig. 3b) the southern Brazil data set and study sequences showed differences from the international consensus (E39D and G123D) and polymorphisms at A173T, V317L,and I326V were observed among pregnant women.

OPEN IN VIEWER

FIG. 3.

Partial pol amino acids alignments. (a) HIV PR and (b) HIV RT including B and C international consensus obtained at NCBI, Brazilian south C consensus (published ^9,11,12 datasets) along with HIV-1 subtype C consensus from study females as Women, for nonpregnant women, and Pwomen, for women pregnant at the time of collection.

All pregnant women received either ARV treatment or prophylaxis for prevention of vertical transmission. Follow-up information has revealed that no cases of HIV transmission were observed among these pregnant women. As for the mothers enrolled, all had one or more HIV-positive offspring. In most cases the ARV medication constituted an NRTI backbone (zidovudine + lamivudine) with nevirapine or one PI, lopinavir or nelfinavir (Table 1). Although HAART has been used (either for treatment or prophylaxis) at the time of collection, only 15/41, 37% of cases, had a plasma viremia below 500 copies/ml. The most common mutation observed to the NRTI class was M184V in 11/40, 27.5% of cases, and to NNRTI the codon K103N (5/40, 12.5%). At PR, the major resistance mutation, L90M, was observed in 3/41, 7.3% of cases. Drug resistance mutations (DRMs) to one class were observed in 4/42, 9.5%, to two classes 13/42, 31%, and triple class resistance in 3/42, 7.1%. Table 1 depicts the DRM and predicted ARV susceptibility observed in this study.

This is the first report of a molecular epidemiological evaluation of young women living with HIV/AIDS identified at an obstetric clinic serving small communities in south Brazil. A low proportion of subtype B (7%) was observed. Even when considering the confidence interval (95% CI: 1.8–18%), large due to the small sample size, such a low estimate for subtype B pol genomes has not been reported in any study in the Americas. Some polymorphisms are observed, most common to other HIV-1 C sequences from south Brazil. Our small sample size is a major drawback for the interpretation of these findings. However, as we evaluated a small community, in fact that sampling comprises almost all available pregnant women at the time when study participants were recruited.

With a total population of about 220,000 inhabitants in the major city and nearby villages, and considering the national estimates for HIV infection (0.61), the estimation for the number of individuals living with HIV/AIDS in this area would be of 1342 individuals. With a prevalence estimate for females of 0.42% (0.8% among males), and assuming that females correspond to about half of the population, we could estimate that about 462 women would be HIV infected in the area. In this setting, the small sample not only represents most women living with HIV/AIDS visiting the obstetric clinic at the time of the study but almost 10% of the HIV-infected female population at the region. Another limitation is the limited access to quality samples for analysis. In some cases where it was not possible to obtain an HIV genome sequence from the women, material from their offspring was studied instead. All these cases were confirmed as mother-to-child transmission (data not shown).

Although the preponderance of clade C in this population might somehow be modulated by an increase in vertical transmissibility of clade C, studies in the region ¹⁵ and abroad ¹⁶ do not support a differential transmissibility of HIV clades so we can assume that the data from their children represent the HIV clades from their mother. In fact, the proportion of clade B among these children is higher than that observed among adult females. Therefore, our data might document the actual predominance of clade C among HIV-infected young females in this area.

Resistant mutations were present in most treated women. NNRTIs and NRTIs were more common in this population. This is in accordance with the literature ¹⁷ in which cases failing therapy based on lopinavir, resistance mutations to this drug are less often observed. It is noteworthy that only 37% of cases had a viral load below 500 copies/ml at the time of collection, highlighting the fact that although Brazil has free access to HIV medication for treatment or prophylaxis, and extensive effects have been deployed to foster mother-to-child transmission control, additional efforts in medication adherence are needed.

To improve the comparability with other studies, we have applied a methodology similar to that used in other cities in South and Southwest Brazil. As in other towns, recombinants have been frequently observed. In this population two samples showed recombination at pol and one discordant gag/pol/env genome. It is noteworthy that one pol recombinant structure shows a breakpoint compatible to the CRF-31, suggesting the presence of this variant in the region. A CRF_31-like recombinant pattern is common in some areas of the largest metropolitan area in the region, Porto Alegre. ¹² Moreover, it has been detected in other areas of the States of Rio Grande do Sul ¹⁸ and Santa Catarina. ¹¹ This may suggest that either this breakpoint constitutes a recombinant hotspot or the variant may have some fitness advantage that favors its spread in the region. Another finding of potential interest is the fact that in this small sample, frequent mosaics (8.6%) were observed among clade C genomes.

The fact that the gene regions studied comprised only part of the HIV genome suggests that the actual figure would be higher. Frequent B/F mosaics, reflecting the decades of B and F coexistence in areas of Brazil such as São Paulo and Rio de Janeiro, have been documented by different groups. ^{4, 5} Although not proof of the advantage of the HIV-1 C clade over the others, the historical data of a mostly B epidemic in the area in the 1990s, along with the presence of CF and CB mosaics, might suggest that the C clade prevails in the area after a period of a few decades of B/C/F coexistence. Monitoring the expansion of the HIV-1 C and its mosaics to other regions is an important issue in the surveillance of the dynamic of the HIV epidemic in Brazil.