Clinical–Epidemiological Characteristics and IFITM-3 (rs12252) Variant Involvement in HIV-1 Mother-to-Children Transmission Susceptibility in a Brazilian Population

Simple Summary Mother-to-children transmission (MTCT) is the main infection route of HIV-1, mainly occurring in pregnancy, delivery, and/or postpartum and it is a multifactorial phenomenon, where genetic variants play an important role. A case–control study was performed in HIV-1 infected mothers and their exposed infected and uninfected children from Pernambuco, Brazil. Our analysis shows that transmitter mothers have a significantly lower age at delivery, late diagnosis, deficiency in ART use (pregnancy and delivery), and detectable viral load in the third trimester of pregnancy compared with non-transmitter mothers. Infected children show late diagnosis, vaginal delivery frequency, and tend to breastfeed, differing significantly from uninfected children. Moreover, the genetic analysis reveals that a variant in the IFITM-3 gene (an important viral restriction factor) is significantly more frequent among infected than uninfected children. Abstract Mother-to-children transmission (MTCT) is the main infection route for HIV-1 in children, and may occur during pregnancy, delivery, and/or postpartum. It is a multifactorial phenomenon, where genetic variants play an important role. This study aims at analyzing the influence of clinical epidemiological characteristics and a variant (rs12252) in interferon-induced transmembrane protein 3 (IFITM-3), a gene encoding an important viral restriction factor, on the susceptibility to HIV-1 mother-to-children transmission (MTCT). A case–control study was performed on 209 HIV-1-infected mothers and their exposed infected (87) and uninfected (122) children from Pernambuco, Brazil. Clinical–epidemiological characteristics are significantly associated with MTCT susceptibility. Transmitter mothers have a significantly lower age at delivery, late diagnosis, deficiency in ART use (pregnancy and delivery), and detectable viral load in the third trimester of pregnancy compared with non-transmitter mothers. Infected children show late diagnosis, vaginal delivery frequency, and tend to breastfeed, differing significantly from uninfected children. The IFITM-3 rs12252-C allele and TC/CC genotypes (dominant model) are significantly more frequent among infected than uninfected children, but the statistical significance does not remain when adjusted for clinical factors. No significant differences are observed between transmitter and non-transmitter mothers in relation to the IFITM-3 variant.


Introduction
Mother-to-children transmission (MTCT) is the main infection route for HIV-1 in children under 13 years, occurring in pregnancy and delivery, or postpartum through breastfeeding. In 2019, it was responsible for about 1.8 million children living with HIV-1 around the world [1].
Even in the absence of preventive measures, 55% to 85% of exposed children will not be infected [2], revealing that MTCT has a complex and multifactorial nature. Several clinical-epidemiological characteristics, such as antiretroviral therapy (ART) use, viral load, and CD4+ T cell levels, may contribute to MTCT. Viral and host genetic factors can also be related to HIV-1 MTCT susceptibility, such as viral restriction factors gene variants [3][4][5][6].
In the context of HIV-1 infection, in vitro studies suggested that IFITM-3∆21 was associated with lower susceptibility to infection, by decreasing the infectivity of nascent viral particles, cell-virus fusion [27,28], and replication inhibition [11]. However, how IFITM-3 variants may influence HIV-1 infection susceptibility is still an open question. Indeed, it was suggested that the rs12252 C allele was associated with rapid disease progression, but not HIV-1 infection in a Chinese cohort [21].
Considering the IFITM-3 antiviral function and its expression in tissues such as placenta and oral mucosa, the present study analyzed the influence of IFITM-3 rs12252 polymorphism and clinical-epidemiological variables on HIV-1 MTCT susceptibility in a northeastern Brazilian population.

Study Design
A retrospective case-control study was performed with 209 HIV-1 infected mothers and their respective exposed children (87 infected and 122 uninfected). Mothers/children were recruited at the Institute of Integral Medicine of Pernambuco Professor Fernando Figueira (IMIP-PE) in Recife (northeast Brazil), from 2013 to 2016.
Inclusion criteria were HIV-1 infected mothers who had at least one detectable viral load (VL) (>40 copies/mL) during pregnancy, antiretroviral therapy (ART) use, and their respective children exposed to HIV-1 via MTCT under 13 years of age. The exclusion criteria were individuals without complete clinical parameters in medical records; mothers and their respective children who maintained VL undetectable throughout the gestational period; children infected by other transmission routes than MTCT. All individuals enrolled in the study were from the same geographical origin, the metropolitan area of Recife, Pernambuco, Brazil.
All individuals involved in the research had their medical records reviewed. The following information was collected from the children's medical records: sex, age at diagnosis, viral load, birth weight, type of delivery, and breastfeeding history. From the mothers, data were collected regarding childbirth, maternal diagnosis, use of ART (pregnancy and childbirth), and viral load during pregnancy.
The mothers were categorized as transmitters (case) and non-transmitters (controls). The exposed children were categorized as infected (at least two detectable VL, and/or positive HIV-1 serology) and uninfected (minimum of two undetectable VL, and/or negative HIV-1 serology). The exposed infected children were considered as the case group, while the exposed uninfected children were considered as controls.
All methodological procedures were assessed and approved by the IMIP-PE Human Research Ethics Committee (nº 2629-13).

DNA Extraction and Genotyping
Mothers' and children's genomic DNA was extracted from peripheral blood, using mini salting-out protocol [29].
Additionally, aiming to minimize possible confounding genetic factors well-known to be associated with susceptibility to HIV-1 infection, the CCR5∆32-rs333 (genetic factor associated with HIV-1 entry susceptibility protection) genotyping was performed by polymerase chain reaction (PCR), using the primer sequences: forward 5 -GTCTTCATTACACCTGCAGCTCT-3 and reverse 5 -CACAGCCCTGTGCCTCTT-3 . The amplicons were analyzed by 3% agarose gel electrophoresis using ethidium bromide as staining.

Statistical Analysis
Allelic and genotypic frequencies were estimated by direct counting using genotype transposer [30]. Hardy-Weinberg equilibrium (HWE) adherence and possible associations were verified through chi-square test (X 2 ) and Fisher's exact test, respectively.
Shapiro-Wilk test was used to assess if specific variables distribution agreed to a normal distribution before univariate statistical analyses; if they did not, Mann-Whitney tests were used for quantitative variables. Categorical variables were compared by Fisher's exact test. All tests were two-tailed with a significance level of α = 0.05 and performed using R program version 2.11.1 [31] or GraphPad Prism version 5.0 (GraphPad Software, La Jolla, CA, USA).
Additionally, we performed a multivariable logistic regression to assess any genetic association while controlling for all clinical-epidemiological data, also using R program. We included interaction terms to assess if there were any discernible effects of motherchild genotype interaction leading to susceptibility to or protection against MTCT.
The statistical power analysis of the study was performed using the G*power 3.1.9.7 software.

Results
The clinical-epidemiological characteristics of HIV-1 infected mothers and their respective children are displayed in Table 1. Table 1. Clinical and epidemiological characterization of children exposed to HIV-1 via vertical transmission and their respective mothers from a Pernambuco state population. Transmitter mothers show a lower median age at delivery (23.3 years), differing significantly from non-transmitter mothers (26.1 years, p = 0.011). Regarding the diagnosis, previous knowledge of HIV-1 serological status (before prenatal care) is significantly more frequent among non-transmitter mothers (41.8%) than transmitter mothers (4.6%; p = 0.009). The delayed acknowledgment (HIV-1 diagnostic at delivery and postpartum) is more common among transmitter mothers (21.8% and 51.7%, respectively) than in non-transmitter mothers (11.5%, p = 0.002; and 0.8%, p = 2.2e −16 , respectively).
The detectable VL in the third trimester of gestation is significantly more frequent in transmitter mothers (96.5%) than in non-transmitter mothers (66.4%; p = 7.4e −8 ). Significant differences are not observed for other pregnancy periods.
In relation to the children, most are female (>54.0%) with HIV-1 negative serology (58.4%) and with weight considered as normal (>48.0%). Univariate analysis shows a higher median of diagnosis age in infected children (0.96 years) than uninfected children (0.12 years, p = 0.0001). In addition, the frequency of uninfected children with insufficient weight (36.1%) is significantly higher than among infected children (13.8%, p = 0.0004).
Allelic and genotype distribution of CCR5 and IFITM-3 polymorphisms are shown in Table 2. The CCR5 and IFITM-3 genotype distribution in mothers and children are in accordance with HWE.
The rs12252-C allele is significantly more common in infected (20.2%) than uninfected children (11%, p = 0.014). Statistically significant differences are observed between infected and uninfected children by dominant model, with the TC/CC genotypes significantly more frequent among infected children (OR = 2.07, p = 0.034). Although no statistically significant differences are found in the codominant model, some trends are observed, resulting in levels of statistical significances closer to 0.05. Indeed, the TT genotype is the most frequent among infected (64.3%) and uninfected children (78.9%), and the TC and CC genotypes are more frequent in infected (30.9% and 4.8%, respectively) than uninfected children (20.2% and 0.9%, respectively). The analysis of IFITM-3 rs12252 is corrected by the presence of CCR5∆32 and the described association remains. For IFITM-3 SNP, the statistical power value is >0.99 among children exposed to HIV-1.
Additionally, we also performed an analysis of concordance and discordance of CCR5∆32 and IFITM-3 rs12252 genotypes between exposed children and their respective mothers; however, no significant differences are observed (Supplementary Table S1).

Discussion
HIV-1 MTCT susceptibility presents itself as a multifactorial and complex event, dependent of viral, maternal, and pediatric variables. In our study population, maternal characteristics such as low age at delivery, late diagnosis, no use of ART during pregnancy/delivery, and detectable VL in the last pregnancy trimester are associated with increased susceptibility to HIV-1 MTCT, corroborating with previous studies [32][33][34].
Younger mothers may lack experience to care for themselves and their infants, and possibly have a less favorable socioeconomic status than older women [31]. These facts can directly impact other preventive measures, leading to late diagnosis and inappropriate ART use, maximizing the transmission risks, as observed in our and other populations [33][34][35].
Undetectable VL maintenance in gestation proved to be of great importance for preventing viral transmission. The HIV-1 MTCT occurs mainly during the third trimester of gestation by reduction in the placental vascular integrity [2], as we observed in our group.
The lack of an early maternal diagnosis leads to a child's late diagnosis. In our population, infected children were born mostly through vaginal delivery, were breastfed, and had a higher age of diagnosis, corroborating with reports in the literature [36][37][38][39][40]. During vaginal delivery, infants are exposed to HIV-1 via fluids from the birth canal that penetrate their oropharyngeal cavity [36]. The elective cesarean delivery is an efficacious intervention for HIV-1 MTCT risk mitigation among infected mothers not taking ART [37].
In our sample, non-exclusive breastfeeding is predominant among infected children, which may be related to failures in health services. Most recruited mothers live far from large city centers, and, therefore, have precarious access to health services, leading to lack of precocious diagnosis/treatment. Napyo et al. [38] suggested that HIV-infected women with delivery not supervised by healthcare teams and without ART adherence were less likely to avoid exclusive breastfeeding, corroborating, at least in part, our results. In our sample, most of the deliveries were supervised by health workers, but deficiencies during care prevented the adoption of appropriate measures.
The insufficient weight in uninfected children was observed as a factor for lower susceptibility to infection, disagreeing with a report in a Zambian population [41], which did not find association between this variable and HIV-1 MTCT susceptibility. This result could be related to ART use in pregnancy, not observed in Zambia [41]. Thus, the ART use before or in the beginning of first trimester of pregnancy may increase the probability of the child being born with a lower weight [42].
Additionally, the rs12252-C allele of IFITM-3 is related to susceptibility to HIV-1 MTCT in our study group, differing from what is expected for this variant [11,22]. The presence of rs12252-C polymorphism in IFITM-3 can alter a splice-accepting site, encoding a protein lacking 21 amino acid residues [26] modifying the IFITM3∆21 protein solely in host cell membrane, thus, meaning a greater expected viral restriction. However, it is not completely clear whether the mutated protein, although present on the cell surface, maintains its functionality in viral restriction. In fact, Jia et al. [11] observed in vitro that IFITM3∆21 was more efficient in restricting the HIV-1 entry into the cell, being more abundantly incorporated into nascent viral particles than its wild form [22]. Therefore, the genetic association observed in our study needs to be confirmed in further functional studies.
Zhang et al. [21], investigating a Chinese cohort, observed that the rs12252-C allele is associated with a faster progression to AIDS, but not to HIV-1 susceptibility. The differences between the studies may lie in the studied groups, since Zhang et al. [21] compared HIV-1 infected adults against healthy men who have sex with men (MSM), while in our study, we analyzed exposed infected and uninfected children. In addition, Zhang et al. also observed that individuals with CC/CT risk genotypes exhibited higher viremia peaks, lower CD4+ T cells counts, and a greater risk for the rapid decline in the CD4+ T cells counts to levels below 350 cells/mL, when compared to control individuals [21].
Compton et al. [27], comparing different proteins of IFITM family, suggested that the IFITM-1 protein, which naturally does not have the N-terminal portion seen in IFITM-3, resembling IFITM3∆21, has a moderate restriction of HIV-1 infectivity compared to IFITM-3. In this sense, it is possible that, due to a still unknown mechanism, just being present in the plasma membrane is not sufficient to restrict the complete HIV-1 entry, the N-terminal portion being necessary for a more effective restriction, which could partially explain our findings.

Conclusions
Our results show the utmost importance of clinical-epidemiological characteristics in the susceptibility of HIV-1 MTCT, reinforcing the crucial role of a satisfactory prenatal screening provided by the health care system to mitigate HIV-1 transmission risks. In addition, being aware of the limitations (sample size, absence of some clinical variables, and functional validation), our study describes, for the first time, the association of the rs12252-C variant in IFITM-3 with susceptibility to HIV-1 MTCT, suggesting the importance of this gene in modulating susceptibility to viral entry.