Pregnancy and HIV Disease Progression during the Era of Highly Active Antiretroviral Therapy

Article (PDF Available)inThe Journal of Infectious Diseases 196(7):1044-52 · November 2007with11 Reads
DOI: 10.1086/520814 · Source: PubMed
Abstract
Before the availability of highly active antiretroviral therapy (HAART), there was no clear effect of pregnancy on human immunodeficiency virus (HIV) disease progression. This has not been assessed during the HAART era. We conducted an observational cohort study among HIV-infected women with >or=1 outpatient clinic visit between January 1997 and December 2004. HIV disease progression was defined as the occurrence of an AIDS-defining event or death. Of 759 women who met the inclusion criteria, 139 (18%) had had >1 pregnancy, and 540 (71%) had received HAART. There was no difference in HAART duration by pregnancy status. Eleven pregnant (8%) and 149 nonpregnant (24%) women progressed to AIDS or death. After controlling for age, baseline CD4(+) lymphocyte count, baseline HIV-1 RNA level, and durable virologic suppression in a Cox proportional hazards model that included propensity score for pregnancy, pregnancy was associated with a decreased risk of disease progression (hazard ratio [HR], 0.40 [95% confidence interval {CI}, 0.20-0.79]; P=.009]). In a matched-pair analysis of 81 pregnant women matched to 81 nonpregnant women according to age, baseline CD4(+) lymphocyte count, receipt of HAART, and date of cohort entry, pregnant women had a lower risk of disease progression both before (HR, 0.10 [95% CI, 0.01-0.89]; P=.04) and after (HR, 0.44 [95% CI, 0.19-1.00]; P=.05) the pregnancy event. Pregnancy was associated with a lower risk of HIV disease progression in this HAART-era study. This finding could be the result of the healthier immune status of women who become pregnant or could possibly be related to a beneficial interaction between pregnancy and HAART.
1044 JID 2007:196 (1 October) Ta i e t a l.
MAJOR ARTICLE
Pregnancy and HIV Disease Progression
during the Era of Highly Active Antiretroviral Therapy
Jennifer H. Tai,
5,a
Mercy A. Udoji,
5,a
Gema Barkanic,
1
Daniel W. Byrne,
4
Peter F. Rebeiro,
1
Beverly R. Byram,
6
Asghar Kheshti,
6
Justine D. Carter,
1
Cornelia R. Graves,
2
Stephen P. Raffanti,
1,6
and Timothy R. Sterling
1,3
1
Division of Infectious Diseases, Department of Medicine, and
2
Department of Obstetrics and Gynecology,
3
Center for Health Sciences Research,
and
4
Department of Biostatistics,
5
Vanderbilt University School of Medicine, and
6
Comprehensive Care Center, Nashville, Tennessee
(See the editorial commentary by Anastos, on pages 971–3.)
Background. Before the availability of highly active antiretroviral therapy (HAART), there was no clear effect
of pregnancy on human immunodeficiency virus (HIV) disease progression. This has not been assessed during
the HAART era.
Methods. We conducted an observational cohort study among HIV-infected women with 1 outpatient clinic
visit between January 1997 and December 2004. HIV disease progression was defined as the occurrence of an
AIDS-defining event or death.
Results. Of 759 women who met the inclusion criteria, 139 (18%) had had
11 pregnancy, and 540 (71%) had
received HAART. There was no difference in HAART duration by pregnancy status. Eleven pregnant (8%) and
149 nonpregnant (24%) women progressed to AIDS or death. After controlling for age, baseline CD4
+
lymphocyte
count, baseline HIV-1 RNA level, and durable virologic suppression in a Cox proportional hazards model that
included propensity score for pregnancy, pregnancy was associated with a decreased risk of disease progression
(hazard ratio [HR], 0.40 [95% confidence interval {CI}, 0.20–0.79]; ]). In a matched-pair analysis of 81P p .009
pregnant women matched to 81 nonpregnant women according to age, baseline CD4
+
lymphocyte count, receipt
of HAART, and date of cohort entry, pregnant women had a lower risk of disease progression both before (HR,
0.10 [95% CI, 0.01–0.89]; ) and after (HR, 0.44 [95% CI, 0.19–1.00]; ) the pregnancy event.P p .04 P p .05
Conclusion. Pregnancy was associated with a lower risk of HIV disease progression in this HAART-era study.
This finding could be the result of the healthier immune status of women who become pregnant or could possibly
be related to a beneficial interaction between pregnancy and HAART.
Studies conducted before the availability of highly active
antiretroviral therapy (HAART) showed that pregnancy
either slightly increased the risk of HIV disease pro-
Received 7 November 2006; accepted 23 February 2007; electronically published
29 August 2007.
Potential conflicts of interest: none reported.
Presented in part: 43rd Annual Meeting of the Infectious Diseases Society of
America, San Francisco, 6–9 October 2005 (abstract 689); 26th Annual Meeting
of the Society for Maternal Fetal Medicine, Miami Beach, 30 January–4 February
2006 (abstract 265); 13th Conference on Retroviruses and Opportunistic Infections,
Denver, 5–9 February 2006 (abstract 705).
Financial support: National Institutes of Health (grant P30 AI54999 to the
Vanderbilt-Meharry Center for AIDS Research and grant K24 AI065298 to T.R.S.);
Vanderbilt Medical Scholars Program (grant M01 RR-00095 to J.H.T.); Vanderbilt
General Clinical Research Center (grant M01 RR-00095 to D.W.B.).
a
J.H.T. and M.A.U. contributed equally to this study.
Reprints or correspondence: Dr. Timothy R. Sterling, 1161 21st Ave. South, A-
2209 Medical Center North, Nashville, TN (timothy.sterling@vanderbilt.edu).
The Journal of Infectious Diseases 2007;196:1044–52
2007 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2007/19607-0016$15.00
DOI: 10.1086/520814
gression or had no effect. Early studies noted a possible
association between pregnancy and accelerated disease
progression [1–3]. Observational studies conducted in
developing countries found that pregnancy was an in-
dependent predictor of increased HIV disease progres-
sion [4, 5]. Several studies conducted in the United
States and Europe, however, did not demonstrate an
increased risk of HIV disease progression in pregnant
women [6–11]. All of the above studies were conducted
in patient populations that received either no antiret-
roviral therapy (ART) or a single nucleoside reverse-
transcriptase inhibitor, in accordance with treatment
availability at that time. In a primarily pre–HAART-era
study of HIV-infected women in the United States with
1 versus 2 pregnancies, there was little difference in
CD4
+
lymphocyte counts, HIV-1 RNA levels, and time
to class C events, but women with 2 pregnancies had
a small survival advantage [12].
HAART provides greater virologic suppression and
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
Effect of Pregnancy during the HAART Era JID 2007:196 (1 October) 1045
immunologic recovery than pre-HAART regimens, and as a
result there have been dramatic improvements in HIV-asso-
ciated morbidity and mortality [13]. However, we are unaware
of studies that have fully assessed the relationship between preg-
nancy and HIV disease progression in women during the
HAART era.
SUBJECTS, MATERIALS, AND METHODS
Study population. The study included all HIV-infected
women who received HIV care (1 visit) between 1 January
1997 and 31 December 2004 at the Comprehensive Care Center
in Nashville, Tennessee. Follow-up started with the first clinic
visit after 1 January 1997. For inclusion, women must have
had either a CD4
+
lymphocyte count and/or HIV-1 RNA level
available at baseline. Baseline CD4
+
lymphocyte counts and
HIV-1 RNA levels were considered to be the earliest test results
obtained up to 120 days before, but no more than 365 days
after, the initial clinic visit during the study period.
The index pregnancy was the first pregnancy that occurred
after the initial clinic visit. Pregnant women lost to follow-up
before the pregnancy outcome and women pregnant at the end
of the study period were excluded. Before March 1999, most
pregnant women receiving HIV care at the Comprehensive Care
Center obtained obstetrical care at Vanderbilt University Med-
ical Center. After March 1999, most women received prenatal
care from an obstetrical clinic at the Comprehensive Care Cen-
ter, which included regularly scheduled visits with a health care
provider, nutritionist, and social worker.
Clinical data were entered into an electronic medical record
by medical providers at the time of the patient encounter, by
automated data upload (e.g., for laboratory results), or by clinic
personnel (e.g., for deaths). Laboratory data, ART (including
start and stop dates), dates of nutrition consults, use of nu-
tritional supplementation, and history of substance abuse were
validated by systematic chart review. If the chart noted (by
patient self-report and/or the provider) that the patient had
not been taking the prescribed ART, the patient was coded as
not taking ART during that period.
The study protocol was approved by the Institutional Review
Board of Vanderbilt University Medical Center. Clinical re-
search was conducted in accordance with the human experi-
mentation guidelines of the US Department of Health and
Human Services and Vanderbilt University.
Study definitions. HAART was defined as regimens of 7
days’ duration that contained 2 nucleoside reverse-transcriptase
inhibitors plus either a protease inhibitor, a nonnucleoside re-
verse-transcriptase inhibitor, or a third nucleoside reverse-tran-
scriptase inhibitor; 1 nucleoside reverse-transcriptase inhibitor,
1 protease inhibitor, and 1 nonnucleoside reverse-transcriptase
inhibitor; 2 protease inhibitors plus 1 nucleoside or nonnu-
cleoside reverse-transcriptase inhibitor; or any regimen con-
taining enfuvirtide. Non-HAART ART included mono- or
dual-nucleoside reverse-transcriptase inhibitor therapy. Dura-
ble virologic suppression was defined as having more unde-
tectable (
!400 copies/mL) than detectable (400 copies/mL)
HIV-1 RNA levels in persons with 2 levels available.
HIV disease progression was defined as any new AIDS-de-
fining event or death (combined end point). AIDS-defining
events were based on the 1993 US Centers for Disease Control
and Prevention classification criteria [14], excluding the cri-
terion of
!200 CD4
+
lymphocytes/mm
3
. All events were re-
viewed and confirmed by study investigators. Subjects were
censored at the time of their AIDS-defining event, the date of
last clinic encounter, or 31 December 2004 if the last clinic
encounter occurred after 31 December 2004.
Statistical analysis. To maximize power, all women who
met the inclusion criteria were included; we deemed power to
be sufficient to detect a clinically meaningful difference. Con-
tinuous variables were compared with the Wilcoxon rank-sum
test, and categorical variables were compared with the x
2
test
and Fisher’s exact test. The log-rank test was used for Kaplan-
Meier survival analyses.
Multivariate Cox proportional hazards models were used to
determine predictors of disease progression while adjusting for
confounding factors. Variables with in univariate anal-
P
! .10
yses were eligible for inclusion in the multivariate model. Var-
iables were retained in the final model only if significantly
associated with the outcome ( ) or deemed to be of
P .05
clinical importance.
Selection bias regarding factors associated with becoming
pregnant was addressed via a propensity-adjusted method [15].
The propensity score of the estimated probability of becoming
pregnant was derived from a logistic regression model that
predicted the occurrence of pregnancy on the basis of several
factors: age, race, marital status, baseline CD4
+
lymphocyte
count
1200 cells/mm
3
, baseline HIV-1 RNA level 110,000 cop-
ies/mL, durable virologic suppression, HAART duration, non-
HAART ART duration, mean number of patient encounters
per year, mean number of nutrition consults per year, receipt
of nutritional supplementation, and history of substance abuse.
Another method used to adjust for differences between preg-
nant and nonpregnant women was a matched-pair analysis in
which 1 nonpregnant woman was matched to 1 pregnant
woman (1:1 match). Subjects were matched according to date
of cohort entry (6 months), baseline CD4
+
lymphocyte count
(100 cells/mm
3
), receipt of HAART, and age at study entry
(5 years). A list of nonpregnant women who matched on all
criteria was generated for each pregnant woman, and the non-
pregnant woman with the closest date of cohort entry was
chosen. When a previously matched nonpregnant woman was
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
1046 JID 2007:196 (1 October) Ta i e t a l.
Table 1. Clinical and demographic characteristics of the study patients, by pregnancy status.
Characteristic
Pregnant
(n p 139)
Nonpregnant
(n p 620) P
Age, median (IQR), years 25 (23–31) 36 (30–42) !.001
a
Black race 68 (49) 322 (52) .57
b
Baseline CD4
+
lymphocyte count
Median (IQR), cells/mm
3
450 (312–660) 352 (177–560) !.001
a
Patients with 1200 cells/mm
3
126 (91) 441 (72) !.001
b
Baseline HIV-1 RNA level
Median (IQR), log
10
copies/mL 3.9 (3.1–4.6) 4.2 (3.2–4.9) .02
a
Patients with 110,000 copies/mL 67 (48) 348 (57) .06
b
Treatment during study period
HAART 119 (86) 421 (68)
!.001
b
Non-HAART ART 20 (14) 55 (9) .09
b
No ART 0 (0) 144 (23) !.001
b
HAART duration, median (IQR), months
c
29.0 (7.8–50.0) 27.2 (9.6–52.4) .99
a
Non-HAART ART duration, median (IQR), months
d
13.5 (4.8–26.2) 12.2 (5.0–30.7) .79
a
Durable virologic suppression
e
42 (31) 123 (22) .03
b
Follow-up, median (IQR), months 57.9 (33.4–82.1) 32.9 (10.8–64.4) !.001
a
Clinic attendance, median (IQR), visits/year 7.2 (4.7–12) 5 (3–7.3) !.001
a
Nutritional consultations, median (IQR), no./year 1 (0–3) 0 (0–1) !.001
a
Ever prescribed
Prenatal vitamin 109 (78) 156 (25)
!.001
b
Multivitamin 14 (10) 82 (13) .40
b
Ferrous sulfate 26 (19) 65 (10) .009
b
Folate 4 (3) 9 (1) .27
b
Any of the above 116 (84) 239 (39) !.001
b
Intravenous drug use as HIV risk factor 16 (12) 117 (19) .048
b
History of heavy alcohol use 13 (9) 96 (15) .08
b
NOTE. Data are no. (%) of patients, unless otherwise indicated. Baseline is the date of cohort entry. ART,
antiretroviral therapy; HAART, highly active ART; IQR, interquartile range.
a
Wilcoxon rank-sum test.
b
Fisher’s exact test.
c
Among patients who received HAART.
d
Among patients who received non-HAART ART.
e
Among 135 pregnant women and 557 nonpregnant women with 2 available HIV-1 RNA levels; suppression
was defined as having more undetectable (
!400 copies/mL) than detectable (400 copies/mL) HIV-1 RNA levels.
Table 2. Outcomes in study patients, by pregnancy status.
Outcome
Pregnant
(n p 139)
Nonpregnant
(n p 620)
AIDS-defining event only 7 (5) 42 (7)
Death only 1 (1) 67 (11)
Both AIDS-defining event and death 3 (2) 40 (7)
Total 11 (8) 149 (24)
NOTE. Data are no. (%) of patients.
the only match for a pregnant woman, the previously matched
pregnant woman was reassigned to the nonpregnant woman
with the next closest date of cohort entry. Multivariate Cox
proportional hazards models of HIV disease progression were
performed separately for time before the pregnancy event (de-
fined as live birth, abortion, stillbirth, or ectopic pregnancy)
and time after the pregnancy event. The date of the pregnancy
event in the pregnant woman was also used for the matched
nonpregnant pair.
(2-tailed) was considered to indicate statistical sig-P .05
nificance. Stata SE (version 8.2; Stata Corporation) and SPSS
(version 14; SPSS) were used for statistical analyses.
RESULTS
Patient characteristics. We identified 759 women who met
the inclusion criteria, of whom 139 (18%) had at least 1 preg-
nancy during follow-up. There were 174 pregnancies during
the study period; 30 women had 2 pregnancies and 5 women
had 3 pregnancies. Among the 139 index pregnancies, there
were 124 live births, 9 spontaneous abortions, 3 elective abor-
tions, 2 stillbirths, and 1 ectopic pregnancy.
Clinical and demographic characteristics of the cohort are
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
Effect of Pregnancy during the HAART Era JID 2007:196 (1 October) 1047
Figure 1. Kaplan-Meier survival curve of progression to a new AIDS-
defining event or death among the overall cohort, by pregnancy status.
The nos. of pregnant and nonpregnant women at risk during each interval
are given below the graph.
shown in table 1. Compared with nonpregnant women, preg-
nant women were younger and had higher median baseline
CD4
+
lymphocyte counts and lower median baseline HIV-1
RNA levels. HAART was received by 540 women (71%); 75
women (10%) received non-HAART ART only. All pregnant
women received ART. Pregnant women were more likely than
nonpregnant women to receive HAART, but, among women
who received HAART, treatment duration was comparable for
pregnant and nonpregnant women. Pregnant women were also
more likely to achieve durable virologic suppression. Duration
of follow-up was longer for pregnant women.
Subjects were assessed for other clinical factors potentially
associated with disease progression (table 1). Pregnant women
had higher rates of clinic attendance than nonpregnant women
and were more likely to receive nutritional consults and sup-
plements. Pregnant women were less likely to have intravenous
drug use as a risk factor for HIV acquisition.
Risk of disease progression. Outcomes by pregnancy status
are shown in table 2. The AIDS-defining events in pregnant
women were Candida esophagitis, cytomegalovirus retinitis,
disseminated cryptococcosis, lymphoma, wasting syndrome,
Pneumocystis jiroveci pneumonia (3), and recurrent bacterial
pneumonia (2).
In Kaplan-Meier analysis, pregnant women were less likely to
progress to an AIDS-defining event or death than nonpregnant
women ( , log-rank test) (figure 1). Results of univariate
P
! .001
Cox proportional hazards analysis of predictors of clinical disease
progression are displayed in table 3. Baseline CD4
+
lymphocyte
count
1200 cells/mm
3
, HAART duration, non-HAART ART du-
ration, durable virologic suppression, and pregnancy were all
associated with a lower risk of progression to an AIDS-defining
event or death; baseline HIV-1 RNA level
110,000 copies/mL
and increased age were associated with a higher risk of disease
progression. Although other clinical variables were evaluated,
only receipt of prenatal vitamins or any nutritional supplement
significantly affected disease progression.
Multivariate Cox proportional hazards models were then
constructed to assess the effect of clinical characteristics on HIV
disease progression after controlling for confounding variables
(table 4). After adjusting for baseline CD4
+
lymphocyte count
and HIV-1 RNA level, age, and durable virologic suppression,
pregnancy was associated with a decreased risk of disease pro-
gression (hazard ratio [HR], 0.40 [95% confidence interval
{CI}, 0.21–0.76]; ). The results did not differ when
P p .005
nadir CD4
+
lymphocyte count was used instead of baseline
count. Clinic visit rate and prescription of prenatal vitamins
or any other supplement assessed were not associated with
disease progression, and their inclusion in the model did not
affect the results. HAART duration and non-HAART ART du-
ration were included in other models that did not include
durable virologic suppression (all 3 variables could not be in-
cluded in a single model due to collinearity); pregnancy was
associated with a lower risk of disease progression in these
models as well (HR, 0.47 [95% CI, 0.24–0.89]; ). In aP p .02
multivariate model that included the propensity score for preg-
nancy in addition to other significant predictors, the risk of
disease progression was again significantly lower in pregnant
women than in nonpregnant women (HR, 0.40 [95% CI, 0.20–
0.79]; ). This association also held in a model con-P p .009
taining the propensity score as a single summary covariate (HR,
0.42 [95% CI, 0.21–0.83]; ).P p .01
An analysis was then performed of 81 pregnant women
matched to 81 nonpregnant women according to age, baseline
CD4
+
lymphocyte count, receipt of HAART, and date of cohort
entry. Clinical characteristics are presented in table 5. In mul-
tivariate models that adjusted for baseline CD4
+
lymphocyte