Association of Low CD4 Cell Count and Intrauterine Growth Retardation in Thailand

Abstract

Objective: 

Each year, intrauterine growth retardation (IUGR) affects 20-30 million neonates worldwide, mostly in resource-limited settings. Increased perinatal and infant mortality has been associated with IUGR. Some studies have suggested that HIV infection could increase the risk of IUGR. To confirm this hypothesis, we examined the association between HIV-related factors and the risk of IUGR in Thailand.

Patients and Methods: 

Data from a cohort of 1436 HIV-infected pregnant women enrolled in the “Perinatal HIV Prevention Trial-1”, a clinical trial conducted from 1997 to 1999 in Thailand, were analyzed using a logistic regression, adjusting for risk factors usually associated with IUGR.

Results: 

The rate of IUGR was 7.6%. Adjusting for a short maternal height, low body mass index, small weight gain during pregnancy, and infant female sex, a low maternal CD4 percentage was independently associated with IUGR (odds ratio 0.96, per 1% increment, 95% confidence interval 0.93 to 0.99, P = 0.03).

Conclusions: 

The current World Health Organization recommendation to initiate combination antiretroviral therapy for immunocompromised women as early as possible during pregnancy for their own health and for the prevention of HIV mother-to-child transmission is likely to also decrease the incidence of IUGR. Encouraging immunocompromised HIV-infected women who plan to become pregnant to wait until immune restoration has been achieved may help to reduce the risk of IUGR.

INTRODUCTION

Intrauterine growth retardation (IUGR) is defined as a birth weight below the 10th percentile of weight for the corresponding gestational age. After prematurity, IUGR is the second cause of perinatal mortality in developing countries.¹ Neonates affected by IUGR have a higher susceptibility to various diseases such as respiratory distress, ventricular hemorrhage, neonatal sepsis, and have overall a higher morbidity than non-IUGR neonates.² Adults born with IUGR have been shown to be at higher risk of diabetes mellitus, obesity, and hypertension.³

In resource-limited settings, the most frequent factors of IUGR are nutritional.⁴ Where malaria is endemic, it also significantly contributes to IUGR.¹ In industrialized countries, smoking is the main risk factor for IUGR,⁴ followed by gravidic hypertension and preeclampsia.⁵ Other factors such as female sex of the child, primiparity, and ethnicity have also been associated with the risk of IUGR.⁴

In 1987, the World Health Organization (WHO) estimated the rate of IUGR at 23.8% (9.4-54.2) in resource-limited countries, affecting approximately 30 million infants per year.¹ In Thailand, the rate of IUGR was estimated to be 8.5% in 1983.⁶ Because the definition of IUGR relies on the gestational age at delivery, for which an accurate measurement is often unavailable in resource-limited settings, a low birth weight (LBW), defined as below 2500 g, is often used as a proxy for IUGR. In HIV-infected women, studies have reported conflicting results with regards to IUGR due to the use of different definitions (LBW or IUGR) or the lack of adjustment for confounding factors.^7-13 Using the stringent definition of IUGR, we analyzed the HIV-specific determinants of IUGR in a cohort of HIV-infected pregnant women in Thailand.¹

Patients and Methods

The Perinatal HIV Prevention Trial-1 was a randomized, double-blind, multicenter, clinical trial comparing the efficacy of zidovudine regimens of different durations in mothers and children to prevent HIV perinatal transmission in Thailand.¹⁴ Between June 1997 and December 1999, consenting HIV-positive women meeting the selection criteria were enrolled. Women with contraindication to zidovudine or hydramnios were excluded. At baseline, sociodemographic characteristics were recorded, medical and obstetrical history taken, physical, obstetrical, and ultrasound examinations performed, and blood drawn for viral load and CD4 count. At 28-week pregnancy, women were randomized to receive either a long course of zidovudine (300 mg, twice a day), starting at 28 weeks of gestation, or a short course, starting at 35 weeks of gestation. The physical and biological examinations were repeated at 32, 35, and 40 weeks of pregnancy and immediately after delivery. Gestational age was determined using an algorithm validated in this population, combining ultrasound measurement, date of last menstruation period, uterine height, and Ballard score.¹⁵

Immediately after birth, neonates were examined by a pediatrician who determined the Ballard score of maturation, and peripheral blood was collected within 24 hours for the diagnosis of HIV transmission by DNA polymerase chain reaction, using Roche Amplicor version 1.5. Infants found to be HIV DNA polymerase chain reaction positive within 7 days of birth were considered infected in utero and infants found positive thereafter were considered infected intrapartum.¹⁶

The study was approved by the ethical committees of the Harvard School of Public Health and the Ministry of Public Health in Thailand.

The primary outcome was IUGR. The criterion for IUGR was a birth weight below the 10th percentile of local reference curves of birth weight for gestational age.¹⁵

A univariate analysis was first performed to identify risk factors for IUGR among non-HIV-specific variables (obstetrical, nutritional, sociodemographic, and physical examination findings collected at preentry, during pregnancy, and at birth). We then examined the role of HIV-related factors (CD4 percentage, HIV RNA viral load at preentry, HIV infection status of the newborn, duration of maternal zidovudine prophylaxis) on the risk of IUGR, using a logistic regression, adjusting for all known confounding factors and factors found to be associated with IUGR in the univariate analysis at the P level of 0.20. The duration of zidovudine prophylaxis and viral load were categorized into quartiles, and the other continuous variables were included as continuous variables after assessment of their linearity. We used the CD4 percentage of the total lymphocyte count shown to be less variable than the absolute count.¹⁷ To compute its fraction of attributable risk of IUGR, we categorized the CD4 percentage (above or below the median). The WHO HIV Classification was not included in the multivariate analysis because its stages are known to be highly correlated with CD4 percentage. Other models were built using CD4 absolute count instead of percentage and LBW as the outcome instead of IUGR, to confirm our findings. Interactions were taken into account when statistically significant (P < 0.05) and medically relevant, using dummy variables. Model fitness was assessed with the Hosmer and Lemeshow test. The STATA function aflogit was used to calculate the fraction of adjusted risk of IUGR attributable to HIV-related factors. STATA software (version 8.0) was used for statistical analyses.

RESULTS

Of the 1436 HIV-infected women enrolled in the study, 26 were lost to follow-up before delivery, and gestational age could not be reliably determined for 100. The characteristics of these 126 excluded mothers were similar to the others, except for the hepatitis C virus-positive serology (9.9% of the women excluded were hepatitis C virus positive versus 4.4% of those analyzed, P = 0.01). We excluded 10 sets of twins who had lower birth weight than singletons as twin-specific weight curves were not available, 12 cases of neonates with congenital malformations and 17 mothers who had experienced gravidic hypertension, because these both groups have known causes of IUGR.

The characteristics of the 1271 mothers and infants included in this analysis are presented in Table 1. More than half of the women were primiparous (58.9%). The mean baseline percentage of CD4 was 19.8% at 26 weeks of gestation, and 92.0% were in stages 1 or 2 of the WHO Classification. The mean viral load was 3.9 log₁₀ copies per milliliter. The ultrasound examination was performed at a median gestational age of 21 weeks (interquartile range 18-24). The rate of in utero HIV transmission was 2.7% [95% confidence interval (CI) 1.8 to 3.6].

TABLE 1: Descriptive Analysis of the Study Population (n = 1 271)

The IUGR rate was 7.6% (95% CI 6.2 to 9.1), and the mean birth weight was 2979 (2955-3003) g. The mean gestational age at delivery, 39 weeks, did not differ between IUGR and non-IUGR neonates (P = 0.50).

Table 2 presents the unadjusted and adjusted odds ratios (ORs) for the characteristics associated with the risk of IUGR (P < 0.20) in the univariate analysis. Two HIV-related factors, CD4 percentage and in utero HIV transmission, were significantly associated with the risk of IUGR in the univariate analysis and the known risk factors-maternal weight gain, body mass index, height, and infant sex. However, smoking (25 women) and intravenous drug use (IVDU) (7 women) were not found significantly associated with IUGR. None of the 7 infants born from IVDU mothers had an IUGR. There was a significant interaction between parity and mean weekly weight gain. A high CD4 percentage was associated with a lower risk of IUGR (OR 0.96, per 1% CD4 percentage increment, 95% CI 0.93 to 0.99, P = 0.01). The association between in utero transmission and IUGR was no longer statistically significant after adjusting for the other factors (OR = 2.32, 95% CI 0.79 to 6.80, P = 0.12). The results were unchanged when in utero infected neonates were excluded (data not shown).

TABLE 2: Risk Factors of IUGR: Univariate Analysis (Factors Associated With P <0.2 are Shown) and Multivariate Analysis

A CD4 percentage below the median (20%) was associated with a higher risk of IUGR (OR of 1.9, 95% CI 1.1 to 3.2, P = 0.02) and contributed to 28% (95% CI 4.8 to 50) of the risk of IUGR in this population. In contrast, the CD4 percentage was not significantly associated with the risk of LBW when used as the outcome (data not shown). The test of Hosmer and Lemeshow showed that the model was adequate (P = 0.22).

DISCUSSION

The CD4 percentage was the main risk factor of IUGR in this population of HIV-infected pregnant women in Thailand. A lower maternal CD4 percentage was independently associated with a higher risk of IUGR (OR 0.96, per 1% increment, P = 0.01). Previous studies investigating the role of HIV on the risk of IUGR have not found this association. This difference may be explained by the use of LBW instead of the more stringent definition of IUGR as the outcome.^8-11 LBW can result from prematurity or IUGR, and preterm neonates who have an appropriate birth weight for their gestational age have a LBW but are not affected by IUGR. Indeed, when we used LBW in our analysis, no association with CD4 percentage was observed.

Two studies, 1 on 487 neonates in the United States and Puerto Rico¹² and 1 on 718 neonates in Tanzania,¹⁰ failed to demonstrate an association between IUGR and CD4 cell count. However, these studies used the absolute CD4 count, which has a wider variability than the percentage. In a study on 634 neonates by Stratton et al,¹² also conducted in the United States and Puerto Rico, there was an association between LBW and a CD4 percentage below 14%.¹³ However, when the IUGR was used as the outcome, the analysis did not show such an association and the only risk factor was the use of intravenous drugs during pregnancy. Because the method for estimating the gestational age was not reported, it is not possible to determine if this evaluation was accurate, especially because most women were enrolled late during pregnancy.

The underlying pathogenesis of the association between low CD4 percentage and IUGR observed in our study is unknown. The maternal production of some cytokines, induced by chronic activation of the immune system and advanced disease stage, may have a negative impact on the fetal growth. It has been reported that placentas of HIV-infected women produce more cytokines (tumor necrosis factor-alpha, interleukin-6 and interleukin-1-béta) than noninfected women.¹⁸ In addition, a high rate of tumor necrosis factor-alpha has been associated with the occurrence of LBW in malaria-infected pregnant women,¹⁹ and a similar mechanism could be hypothesized. Yet, additional studies are needed to understand the mechanisms of the negative effect of low CD4 percentage on fetal growth. However, considering the relatively high risk of IUGR attributable to a poor immune status (28%), immune restoration with antiretroviral treatment during pregnancy could be beneficial.

Consistent with other studies, neither zidovudine prophylaxis duration nor HIV viral load was associated with IUGR.^12,13 In our study, as in the study conducted in Tanzania,¹⁰ the relationship observed between in utero transmission and IUGR was not significant. This could be related to the small number of infants infected in utero. However, the timing of HIV transmission, either intrapartum or in utero, is difficult to ascertain, and it is not possible to determine if IUGR is a consequence or a cause of early HIV transmission.

We have to acknowledge some possible biases in our study. The rate of IUGR, 7.6%, estimated in our study population is slightly lower than the rate of 8.5% previously reported in the general Thai population.²⁰ Our rate may have been underestimated as 39 cases at risk of IUGR-twin pregnancies, gravidic hypertension or congenital malformations, and factors which are not HIV related-were excluded from our analysis. Also malaria, a known risk factor for IUGR in endemic settings, was not taken into account because only 1 site among the 27 was located in an endemic area. Finally, smoking and IVDU, which are also known risk factors of IUGR, are very uncommon among Thai pregnant women and involved only 7 and 25 women, respectively, in our study. However, the HIV-unrelated risk factors of IUGR observed in our study were consistent with those previously reported, supporting the validity of our definition of IUGR.²¹ Also the fact that the mean gestational age of 39 weeks observed in our study was similar to other cohorts of HIV-infected pregnant women confirms the accuracy of our algorithm to determine the gestational age.

We have shown that a poor immune status is independently associated with a greater risk of IUGR in our cohort of Thai HIV-infected women. This finding is important because all infants born to HIV-infected pregnant women with IUGR, infected and not infected, are at higher risk of mortality and morbidity in the neonatal period.

The current WHO guidelines for the prevention of mother-to-child transmission of HIV in limited resource settings recommend providing pregnant women with highly active antiretroviral therapy (HAART) during pregnancy if they need it for their own health, that is, if their CD4 count is less than 200 cells per cubic millimeter or they present HIV-related symptoms.²² The results of our study demonstrate that this strategy may also reduce the risk of IUGR. These results have to be weighed against those of studies reporting an increased risk of prematurity among women receiving HAART during pregnancy,^23-25 in particular, a protease inhibitor-containing regimen,²⁶ although recent studies seem to be reassuring.^27-29 Further research is needed to ascertain the positive effect of HAART during pregnancy on the rate of IUGR in women with low and high CD4 cell counts at the beginning of pregnancy. Several perinatal HIV prevention trials about to start, comparing third trimester HAART to zidovudine plus intrapartum nevirapine, may provide important insights. However, because restoration of the immune function in HIV-infected patients treated with HAART requires more than few months, immunocompromised HIV-infected women should be advised to initiate pregnancy only after a satisfactory immune restoration.

ACKNOWLEDGMENTS

We thank all the women and their infants who participated in the study and the clinical coinvestigators who cared for the patients and collected the data and would like to express our gratitude to Tim R. Cressey for the editorial review of the article.