Adverse effects of antiretroviral therapy in pregnant women infected with HIV in Brazil from 2000 to 2015: a cohort study

The significant progress in antiretroviral treatment caused a striking decline in morbidity and mortality associated with human immunodeficiency virus (HIV) infection and a dramatic reduction in mother-to-child transmission (MTCT) [1–3]. Antiretroviral therapy for pregnant women is proven to be the most efficient intervention by reducing the rate of MTCT to lower than 2% and decreasing maternal and child mortality [4, 5].

The World Health Organization’s (WHO) Option B+ recommends the use of antiretroviral combined therapy (ART) to all pregnant and lactating women infected with HIV, independently from the CD4 count or the disease status. It also recommends maintaining its use after birth to control maternal disease and to prevent MTCT and sexual transmission of HIV [6]. A global treatment trend followed this recommendation not only in the developed countries but also in the ones with restricted economic resources. The Brazilian protocol follows the same guidelines to prevent MTCT of HIV [7].

Nevertheless, the increased use of potent and complex antiretroviral regimens during pregnancy can cause adverse effects in pregnant women and their newborns [8, 9]. The most frequent maternal adverse effects are hematologic, hepatic and dermatologic alterations, metabolic disturbances, pre-eclampsia, and viral resistance [10–14].

The Women’s Hospital at University of Campinas School of Medical Sciences (CAISM/UNICAMP) has been running a program for pregnant women infected with HIV since 1988. The objective of this study was to evaluate the ART maternal adverse effects in a large cohort of pregnant women infected with HIV followed at the Obstetric Clinic in this Brazilian public university hospital between 2000 and 2015.

Observational analytic study based on the evaluation of a historic cohort in a population of pregnant women infected with HIV between 2000 and 2015 followed at the Obstetric Clinic at the University of Campinas School of Medical Sciences (CAISM/UNICAMP). CAISM is the reference hospital for high risk pregnancy in Campinas, the second biggest city in the State of São Paulo with 1.2 inhabitants and an Human Development Index (HDI) of 0.8. Mostly, CAISM serves pregnant women without health insurance and from low socioeconomic status. Follow-up of pregnant women infected with HIV at CAISM started in 1988. ART drugs started in 1994. The ART drug schemes varied in different periods: from 1994 to 1999 monotherapy with AZT; from 1999 to 2000 double AZT scheme with 3TC; from 2000 on, an association of AZT/3TC with a NNRTI (nevirapine) or a PI (nelfinavir). In 2006, nelfinavir was substituted for lopinavir/ritonavir (LPV/r). The AZT/3TC/LPV/r scheme used as a first line treatment for infected pregnant women since 2006 was substituted for TDF/3TC/EFV in 2015. By the end of 2017 efavirenz (EFV) was substituted for raltegravir (RAL) according to the new Brazilian guidelines for treatment. AZT was the drug of choice to compose the ART scheme during pregnancy until 2015 although tenofovir had been used as a substitute for AZT mainly in patients with anemia since 2010.

The women were selected from the clinical records in the medical files and from the epidemiologic Health Surveillance Agency data available in the clinic. A specific form was developed to collect all the information. The following variables were analyzed: pregnant women’s epidemiologic and clinical characteristics, antiretroviral treatment administrated, delivery characteristics, MTCT of HIV, adverse hematologic and hepatic effects, and glucose and lipids metabolic disorders. The time of pregnancy when antiretroviral treatment was introduced was described in terms of gestational weeks; women receiving ART in their last menstrual period were considered in treatment at conception. Peripartum viral load was measured at least 6 weeks prior to delivery. Substance use (cocaine/crack), alcohol use or smoking practice were considered previous or current in pregnancy. For the global analysis, the combination of antiretroviral drugs was classified as: monotherapy with nucleoside reverse transcriptase inhibitor (NRTI) zidovudine (AZT); double therapy with NRTI (AZT and lamivudine-3TC) or combined therapy (ART) defined as a combination of at least three drugs with no less than one protease inhibitor (PI) or one non-nucleoside reverse transcriptase inhibitor (NNRTI). Women using the PI darunavir (DRV), lopinavir (LPV) and atazanavir (ATV) also received booster of ritonavir (r). To evaluate the adverse effects, the main assessed outcomes were: laboratory abnormalities: anemia (hemoglobin < 11 g/dl), low platelets count (platelets < 150,000/ml), liver function test abnormalities (at least one: alanine aminotransferase-ALT > 34 U/L, aspartate aminotransferase-AST > 45 U/l, alkaline phosphatase-ALCPH > 104 U/l, gamma-glutamyl transferase-GGT > 42 U/l, bilirubin > 1,0 mg/dl); diabetes (glycemic curve with at least one abnormal value: fasting ≥92 mg/dl, after the first hour ≥180 mg/dl, after the second hour ≥152 mg/dl) and dyslipidemia (at least one abnormality: total cholesterol ≥200 mg/dl, triglyceride 150 ≥ mg/dl). Dyslipidemia criteria were used for adult population, not specified for pregnant women. Hemoglobin (Hb) level less than 11 g/dL was defined according to the Brazilian Health Ministry guidelines: mild (Hb 10–10,9 g/dL), moderate (Hb 8–9,9 g/dL) and severe (Hb ≤ 8 g/dL) [15].

The DAIDS Adverse Effect Scale (DAIDS AE) was used to quantify the severity of adverse effects. This grading scale provides an AE severity grading scale ranging from grades 1 to 5 with descriptions for each AE based on the following general guidelines: Grade 1 indicates a mild event, Grade 2 indicates a moderate event, Grade 3 indicates a severe event, Grade 4 indicates a potentially life-threatening event, Grade 5 indicates death [16].

All newborns exposed to HIV infection were followed by the Pediatric Immunodeficiency Service at the same University Hospital. The cases with no final HIV infection diagnosis due to loss of follow-up were contacted by phone or telegram. This study was approved by the Institution’s Ethics in Research Committee (Protocol #351/2006).

Descriptive analysis was shown in terms of absolute (n) and relative (%) frequencies, mean, median and standard deviation. Chi-square or Fisher exact test (n < 5) was used to analyze the association between the categorical variables. For the continuous variables, the Student t-test (parametric data) or Mann-Whitney (non-parametric data) were used. The specific effects of the different antiretroviral regimens were analyzed using relative risk (RR) with its respective p values. A multivariate Cox Logistic Regression was done. A 95% confidence interval (CI) and a significant level of 0.05 were used. Statistical analysis was performed using SAS version 9.4.

Between 2000 and 2015, there were 47,841 births at the university hospital. From these, 801 were pregnant women infected with HIV, with a 1.67% prevalence rate. Data referring to 14 multiple births were not replicated for the maternal analysis. Figure 1 shows all the eligible cases, the follow-up losses and the final number of 793 pregnancies and 787 children. The MTCT rate in this cohort was 2.3%. In the last 5 years of this study, of the 350 infected pregnant women, there were only three cases of HIV transmission, corresponding to a MTCT rate of 0.8%. For the few women who did deliver HIV positive infants three were in use of monotherapy with AZT, one was not taking any ART, one was in use of an unknown ART. The other ones were receiving three-drugs regimens.

Maternal adverse effects were: dyslipidemia (82%), anemia (56%), liver function test abnormalities (54.5%), including hyperbilirubinemia (11.6%), fasting glycemia alteration (19.2%), thrombocytopenia (14.1%), and allergic skin reaction (2.7%).

In our study pregnant women infected with HIV presented significant rates of adverse effects, most of low severity. Anemia and hepatic alteration occurred in more than half of the cases and lower platelets count in almost 15%. Less than 3% presented ART-related allergic reaction.

The Obstetric Clinic in CAISM/UNICAMP is a reference service for HIV infected women in the Campinas region and serves a low-income population with less schooling. The great majority of the cases are highly complex including users of psychoactive substances (mainly crack). In this updated cohort of 15 years of medical care, the great majority of women were exposed to ART during pregnancy, delivered by C-section, received intravenous zidovudine during labor, did not breastfeed and their newborns received oral AZT for 6 weeks.

The antiretroviral drugs used in pregnant women during the study years varied according to the Brazilian Ministry of Health’s recommendations. Since 2000, an ART regimen with three combined drugs has been in use [7].

One unquestionable benefit of using ART during pregnancy is the viral load control which directly impacted MTCT. In our study, analyzing 793 pregnancies of women infected with HIV, the MTCT rate was 2.3%, close to the World Health Organization (WHO) Global Plan’s goal [17, 18]. Nevertheless, it is important to stress that among the 350 infected pregnant women in the cohort’s last 5 years, there were only three cases of MTCT, all from women with lower adherence to treatment and in abuse of substances. Previous studies from the same clinic showed MTCT rates of 2.9% in 2000 and 3.7% in 2009, the last one strongly related to no adherence to antiretroviral treatment during pregnancy [19, 20].

It is known that using complex and potent antiretroviral regimens during pregnancy can cause adverse effects for the mother and the newborn, including the organogenesis period [8, 9]. The physiologic changes in pregnancy affect the kinetics of absorption, distribution, biotransformation and elimination of medication, potentially altering pregnant woman’s susceptibility to the toxicity of different drugs [7]. These changes cause pharmacokinetic alterations with consequent reduction of exposure to most of the drugs during the gestational period. However, it is not always necessary to adjust their dosages until the end of pregnancy [21].

The most frequent and significant maternal adverse effects studied are hematologic, hepatic and dermatologic alterations, metabolic disturbances like diabetes mellitus and dyslipidemia, pre-eclampsia and viral resistance [8, 22].

Our study demonstrated a global anemia rate of more than 50% in pregnant women exposed to ART, which is in accordance with the international and national literature. Recent study showed high anemia occurrence in Brazilian pregnant women [23]. This adverse effect was not associated with any specific antiretroviral drug in our multivariate data analysis but was associated with the starting of ART during pregnancy, with the occurrence of infectious complications and with peripartum CD4 count lower than 200 cells/mm³.

It is known that anemia in pregnant women infected with HIV can be associated with other causes besides ART, such as advanced maternal disease, iron and folic acid deficiency, malnutrition, intestinal parasitosis, and increased intravascular volume (hemodilution); all significantly associated with increased risk for maternal mortality [24, 25].

The use of zidovudine is one of the causes of cytopenia in pregnant women infected with HIV. In general anemia occurs at the beginning of treatment, in the first 12 weeks of use, and it is frequently associated with low CD4 count and advanced disease [26–28], as demonstrated in our data.

Even when it is known that AZT is not the only cause of anemia in pregnancy, and that ART is linked to improved health, particularly in lower income countries, anemia is the most common adverse effect associated with AZT use in pregnancy, independently of the regimen [29–31]. A study with 1070 pregnant women in sub-Saharan Africa between 2005 and 2008 reported the same results [32]. Many studies have been demonstrating high rates of anemia independently from the treatment regimen, as the one in South Africa with 408 pregnant women and 64.2% rate of anemia [33]. Similarly, our multivariate analysis showed that low CD4 count is a risk factor associated with anemia.

Our data did not show a significant difference in the occurrence of anemia between pregnant women exposed to AZT or TDF regimens. We did not compare the use of monotherapy with AZT and combined ART because of the small number of cases using only AZT (23 cases). A clinical trial (PROMISE Trial) in six sub-Saharan African countries and India compared monotherapy with combined regimens (AZT/3TC/LPV/r or TDF/Emtricitabine/LPV/r) and demonstrated that maternal adverse effects were more frequent with combined regimens, although the difference was not significant [34].

A study with 117 pregnant women in 2011 demonstrated adverse hematologic and hepatic effects and gestational cholestasis particularly associated with nevirapine, zidovudine and nelfinavir use. The higher rates of discontinuation or substitution due to adverse effects or intolerance occurred with these same drugs, while only 1% of patients in use of lopinavir/ritonavir had the drugs changed [35]. Our data also evidenced an increased risk of elevated hepatic enzymes with the use of combined ART. In the multivariate analysis, the use of nevirapine was associated with increased risk for hepatic alteration during pregnancy, although NFV and ATV were also associated with these adverse effects.

In our cohort, nevirapine was the most utilized NNRTI. It was associated with increased hepatic alterations and hypersensitivity skin reactions, in agreement with recently published researches. Using NVP in pregnant women, especially with CD4 count higher or equal to 250 cells/mm³ is debatable, due to elevated risk for hepatic toxicity and skin allergic reaction already demonstrated [36, 37]. However, an elevated risk for hepatic alteration and allergic reaction was not demonstrated in pregnant women exposed to nevirapine with CD4 > = 250 cells/mm³ in our cohort. This drug is considered an alternative, seldom recommended for pregnant women in Brazil [7].

Data from a 2006 Brazilian study evaluating NVP adverse effects in 197 pregnant women reported a 5.5% toxicity rate, 4.5% of exanthema with only one case of Stevens-Johnson Syndrome, and 1.5% of hepatotoxicity including one severe case [38]. That study also described hepatitis C virus as the only risk factor for NVP toxicity, which was also demonstrated in recent studies [39, 40]. In our cohort, pregnant women co-infected with hepatitis C virus did not present higher hepatotoxicity due to ART use.

In our study, despite the low rates of maternal skin allergic reaction associated with NVP, there were three cases of Stevens-Johnson Syndrome with total recovery.

A clinical trial in Uganda compared LPV/r and efavirenz (EFV) use in 389 women infected with HIV during pregnancy and puerperium and found no difference in the adverse effects such as anemia and neutropenia related to the two drugs. On the other hand, the pregnant women exposed to EFV presented higher viral suppression at delivery compared to the ones in use of LPV/r, although this difference was extinguished one year after birth [41]. In our study only four pregnant women were using EFV by pregnancy’s conclusion and only one of them used it continuously from the beginning to the end. Although for years EFV has been largely used in adults’ antiretroviral regimens due mainly to its highly potent effects and its simpler posology, only recently and after its security was demonstrate, was this drug suggested as the preferable regimen for pregnant women [6, 42]. For this reason, in 2015 the Brazilian Health Ministry recommended EFV as the chosen drug for initial treatment [43]. Nevertheless, recent studies discussed its significant collateral effects, in special the neuropsychiatric one and the increased risk for suicide [44, 45], besides its lower potentiality when compared to the promising integrase strand inhibitors [46]. Because of this, the most recent Brazilian recommendation is to use raltegravir as first choice to complete ART regimen in pregnancy [7].

Until 2015 the main recommended drug for pregnant women in Brazil was lopinavir/ritonavir (LPV/r). Atazanavir/ritonavir (ATV/r), which was highly recommended in international guidelines, was restricted in Brazil to first line PI intolerance cases and was associated with higher risk for elevated bilirubin, without clinical implications to the newborn. Bilirrubin elevation is a known adverse effect of atazanavir during pregnancy or anytime of use. It is known that protease inhibitors can cause lipid metabolic changes, metabolic syndrome, lipodystrophy, changes in the carbohydrate metabolism and increased cardiovascular risk in every HIV infected patient [47–49]. Some studies had demonstrated that during pregnancy women in use of PI also presented higher rates of dyslipidemia, metabolic syndrome, lipodystrophy, diabetes and pre-eclampsia [50–52], as some of our data also demonstrated.

When evaluating the different ART classes, PI use was associated with elevated risk for dyslipidemia and fasting glycemia alterations when compared to NVP use. In the multivariate analysis, LPV use was associated with elevated risk for alteration in fasting glycemia, demonstrating the need for its early evaluation and diagnosis in the pregnant women exposed to PI, and the need to revise the criteria for indicating PI to women with known risk factors for diabetes, such as obesity.

On carbohydrates metabolism, a study showed that some antiretroviral drugs increased the TNFα, IL-6 e IL-1β pro-inflammatory cytokines involved in changes of adipocytes function and in the decrease of adiponectin, a positive modulator for insulin sensitivity [53]. A multi-center randomized clinical trial with 1407 women observed a general rate of gestational diabetes in 2.1%, higher with PI use (4.6%) [54].

Regarding birth outcomes, the authors used the same pregnant women’s cohort and their newborns to analyze neonatal adverse effects such as prematurity, low birth weight, small for gestational age week, neonatal anemia and hepatic alterations. The results on birth outcomes are described in another article which has just been published [55].

The high Caesarian section rate in this cohort occurred because from 2005 to 2015 most of pregnant women infected with HIV delivered by C-section following changes in the service’s protocol after the European Collaborative Group published their study [56]. On the other hand, prematurity was not considered iatrogenic and was not associated with C-section as shown in the recently published article by the same authors [55].

Our study focus was on the evaluation of adverse effects to which we produced consistent data on a clinical cohort in use of ART following the Brazilian Health Ministry’s recommendations, using guidelines very similar to the international ones. Data collection based on information registered in the women’s and children’s clinical records might have contributed to a possible register bias. Another limitation was having no HIV negative controls. The authors did not compare adverse effects in HIV negative control women because some tests to evaluate toxicity (hepatic alterations and dyslipidemia) are done only when the pregnant woman is in use of specific medication such as ART.

Our data showed more than 700 evaluations of pregnant women infected with HIV followed in a specialized clinic showing that the ART choices recommended by the Brazilian Health Ministry were adequate, leading to low gravity adverse effects and reaching excellent rates of MTCT.

Despite the growing evidence on the unquestionable benefits of antiretroviral treatment during pregnancy, the clinical monitoring on drug safety and efficacy is of extreme importance to optimize the treatment recommendations.

The evolution of the national guidelines of antiretroviral therapy for pregnant women improved adherence to the treatment and resulted in a significant reduction of mother-to-child transmission of HIV throughout the years. Despite the high frequency of maternal adverse effects, they are mostly of low severity. Parallelally, recommending newer ART medications with improved efficacy and significantly more favorable tolerability profiles (introduction of tenofovir/lamivudine plus raltegravir for first line) should reduce the incidence of maternal ART-related adverse effects, which will reduce maternal as well as infant morbidity and potentially mortality.