Safety and exposure of once-daily ritonavir-boosted atazanavir in HIV-infected pregnant women


Dr Donnie McGrath, Bristol-Myers Squibb, Research and Development, 5 Research Parkway, Wallingford, CT 06492, USA. Tel: +1 203 677 7346; fax: +1 203 677 6852; e-mail:



There are limited antiretroviral options for use in the treatment of HIV infection during pregnancy. The purpose of this study was to assess the safety, efficacy and appropriate dosing regimen for ritonavir (RTV)-boosted atazanavir in HIV-1-infected pregnant women.


In this nonrandomized, open-label study, HIV-infected pregnant women were dosed with either 300/100 mg (n=20) or 400/100 mg (n=21) atazanavir/RTV once-daily (qd) in combination with zidovudine (300 mg) and lamivudine (150 mg) twice daily in the third trimester. Pharmacokinetic parameters [maximum observed plasma concentration (Cmax), trough observed plasma concentration 24 hour post dose (Cmin) and area under concentration-time curve in one dosing interval (AUCτ)] were determined and compared with historical values (300/100 mg atazanavir/RTV) for HIV-infected nonpregnant adults (n=23).


At or before delivery, all mothers achieved HIV RNA <50 HIV-1 RNA copies/mL and all infants were HIV DNA negative at 6 months of age. The third trimester AUCτ for atazanavir/RTV 300/100 mg was 21% lower than historical data, but the Cmin values were comparable. The Cmin value for atazanavir/RTV 400/100 mg was 39% higher than the Cmin for atazanavir/RTV 300/100 mg in historical controls, but the AUCτ values were comparable. Twice as many patients in the 400/100 mg group (62%) had an increase in total bilirubin (>2.5 times the upper limit of normal) as in the 300/100 mg group (30%). Atazanavir (ATV) was well tolerated with no unanticipated adverse events.


In this study, use of atazanavir/RTV 300/100 mg qd produced Cmin comparable to historical data in nonpregnant HIV-infected adults. When used in combination with zidovudine/lamivudine, it suppressed HIV RNA in all mothers and prevented mother-to-child transmission of HIV-1 infection. During pregnancy, the pharmacokinetics, safety and efficacy demonstrated that a dose adjustment is not required for ATV.


Treatment guidelines for HIV-1 infection in pregnant women recommend highly active antiretroviral (ARV) therapy (HAART) with two nucleoside reverse transcriptase inhibitors (zidovudine and lamivudine) plus the nonnucleoside reverse transcriptase inhibitor nevirapine [1–3]. Some guidelines also recommend the ritonavir (RTV)-boosted protease inhibitor lopinavir as an optional third agent [1], although others recommend several boosted protease inhibitors as optional agents [2]. All other ARV drugs are alternative agents or for use in special circumstances [1,4]. However, there are questions and concerns regarding the two most frequently recommended third agents: treatment initiation with nevirapine is associated with an increased risk of symptomatic liver toxicity, often accompanied by a rash, which is potentially fatal [1,5]. Concerns with RTV-boosted lopinavir include uncertainty regarding whether an adjusted dose is necessary during pregnancy [6–8], and the common side effects of diarrhoea, nausea and vomiting and elevation of plasma lipids [9,10]. Therefore, an unmet medical need exists for additional recommended third agents for use during pregnancy.

Atazanavir (ATV) is a potent, well-tolerated, once-daily (qd) HIV protease inhibitor, with established efficacy in both treatment-naïve and treatment-experienced adult, nonpregnant HIV-infected patients [11,12] and is included as a preferred treatment option for nonpregnant HIV-infected patients [2]. HIV protease inhibitor drug levels are generally reduced during pregnancy [13–16], especially during the third trimester, because of metabolic and physiological changes associated with pregnancy [17]. In one study of lopinavir/RTV, compensation for the lower exposures required a dose increase to 533/133 mg twice daily (bid) from 400/100 mg bid in the third trimester to produce exposures similar to those in nonpregnant historical controls [7]. Conversely, Ripamonti et al. [18] reported that the standard dose of ATV/r (300/100 mg) resulted in ATV exposures in women in the third trimester that were similar to their postpartum exposures. The most frequently observed adverse event with ATV is elevations of indirect bilirubin as a result of uridine diphosphate glucuronosyl transferase (UGT) 1A1 inhibition, and this gives rise to the theoretical concern that using ATV in pregnant women may exacerbate physiological hyperbilirubinaemia in the neonates [1]. The purpose of this study was to determine the dosing regimen for ATV/r that produced adequate drug exposure during pregnancy compared with historical data in nonpregnant HIV-infected adults, and to assess the safety of ATV use in pregnancy.


Study design

In this multicentre, open-label, prospective, single-arm Phase I study, patients were enrolled in South Africa, Puerto Rico and the USA from 12 June 2006 to 12 September 2008. The primary objective was to determine the dosing regimen of ATV/r that produces adequate drug exposure during pregnancy when compared with historical data in nonpregnant HIV-infected adults. Secondary objectives included: (1) to measure the HIV RNA in mothers and the HIV DNA in infants born to women exposed to ATV/r during pregnancy; (2) to assess the safety of ATV/r in pregnant women and their infants; (3) to compare ATV/r drug concentrations in cord blood with those in maternal plasma at the time of delivery; and (4) to explore ATV/r drug exposure during the second trimester of pregnancy. The mothers were followed until 8–12 weeks postpartum and the infants were followed until 6 months of age. The laws and regulatory requirements of all participating countries were adhered to. This study was conducted in accordance with the ethical principles that have their origin in the Declaration of Helsinki, as defined by the International Conference on Harmonization and in accordance with the ethical principles underlying the European Union Directive 2001/20/EC and the United States Code of Federal Regulations, Title 21 Part 50 (21CFR50). The research protocol was approved by institutional review boards for each research site. Written informed consent was obtained from every patient or their legally acceptable representative prior to clinical trial participation, including informed consent for any screening procedure conducted to establish eligibility for the trial.

Participants and treatment

Patients who met the inclusion criteria were HIV-1-infected, pregnant women at ≥12 to ≤32 weeks of gestation with a CD4 cell count ≥200 cells/μL, with a singleton pregnancy, who agreed to formula-feed their infants throughout the study after delivery. Patients with the following ARV histories were included: (1) ARV-naïve patients with HIV RNA >400 copies/mL; (2) patients who were currently on HAART with HIV RNA <50 copies/mL and who switched to the study regimen for a reason other than virological failure of a protease inhibitor-based regimen; and (3) patients on HAART for ≤90 days with HIV RNA >50 copies/mL but ≥1 log10 copies/mL drop in HIV RNA within 90 days of screening. ATV-based HAART for ≥3 weeks was not allowed except for prior mother-to-child transmission prevention with documented HIV RNA <50 copies/mL at the time of discontinuation of ATV. Treatment was ATV/r 300/100 mg qd during the second trimester and 300/100 mg qd or 400/100 mg qd in the third trimester in combination with zidovudine/lamivudine 300/150 mg bid. The initial interim pharmacokinetic analysis occurred after the first 12 patients had received ATV/r 300/100 mg during the third trimester with pre-specified criteria for a dose increase to 400/100 mg for all subsequent mothers entering the third trimester. The pre-specified criteria included requirements for Cmin or AUCτ values. If more than two of 12 patients had an ATV Cmin<150 ng/mL but 10 of 12 patients had an ATV Cmin≥50 ng/mL, then ATV/r would be increased to 400/100 mg qd. The AUCτ criterion stated that, if the geometric mean of ATV AUCτ for these 12 patients was <30 000 ng h/mL but ≥15 000 ngh/mL, then ATV/r would be increased to 400/100 mg qd. The dose increase occurred if either criterion was met, and, if a dose escalation was required, all patients at ≥week 28 were given the higher dose. Prophylaxis for prevention of mother-to-child transmission of HIV infection with ARVs (zidovudine and lamivudine) and Pneumocystis jiroveci pneumonia prophylaxis were recommended for all infants.


Blood samples were collected after ≥2 weeks of adherent dosing. Adherence was assessed by pill count and was defined as taking all doses and the number of pills prescribed for each medication prescribed. ATV was sampled over one dosing interval (24 h post-dose) from the mother in the second trimester, the third trimester and postpartum (median 43 days; range 24–71 days). A single blood collection from the mother and the umbilical cord was performed at delivery. Samples were assayed by liquid chromatography and tandem mass spectrometry. For ATV and RTV, the standard curves were fitted by a 1/X2-weighted quadratic equation over the concentration ranges of 10.0–10 000 and 5.0–5000 ng/mL, respectively. Values for precision for the analytical quality control (QC) samples were a coefficient of variation (CV) no greater than 7.9% and 9.4% for ATV and RTV, respectively, with deviations from the nominal concentrations of no more than ± 9.4% for ATV and ± 7.6% for RTV. The historical reference data for the current study were pooled from nonpregnant HIV-infected women and men receiving ATV/r 300/100 mg with a nucleoside reverse transcriptase inhibitor (NRTI) backbone (that did not include tenofovir) in two previous clinical studies that had concluded nearest the start of this study [19,20]. These pooled pharmacokinetic data are also similar to the data in the product label for ATV/r 300/100 mg qd and thus were considered representative data for infected patients. Pharmacokinetic parameters (Cmax, Cmin and AUCτ) were derived by noncompartmental methods.


Descriptive statistics were calculated for the second trimester, third trimester (both 300/100 mg qd and 400/100 mg qd groups), postpartum and historical pharmacokinetic parameters and plasma concentrations, as well as for maternal and umbilical cord plasma concentration and the fetal:maternal ratio of plasma concentrations at delivery. Point estimates and 90% confidence intervals for the ratios of plasma concentrations of geometric means for ATV Cmax, AUCτ and Cmin in the third trimester for the 300/100 mg qd or 400/100 mg qd group relative to pooled historical data were calculated using historical data as a reference. Similar analyses were performed for the second trimester and postpartum data relative to the historical data. Efficacy analyses for treated mothers tabulated the proportion of subjects with HIV RNA <400 copies/mL and <50 copies/mL at the time of delivery, and summarized changes from baseline in log10 HIV RNA level and CD4 cell count over time. The proportion of infants with HIV-1 infection, as determined by DNA polymerase chain reaction (PCR), was tabulated for time-points from birth to 6 months of age.


A safety assessment occurred at each visit and was based on all treated patients, and included clinical examination and laboratory testing of the mothers and infants. All adverse events up to 30 days after the last dose of ATV/r were included. The infant's HIV DNA level was determined at delivery and at weeks 2, 6, 16 and 24. Bilirubin levels were assessed in infants on days 1, 3, 5 and 7 and at weeks 2 and 6. Incidences of adverse events were tabulated and reviewed for potential significance and clinical importance.


Patient characteristics

Sixty-nine women were screened and 41 were enrolled in this study. Twenty-eight patients were screen failures: 26 did not meet the study criteria; one was unable to comply with study procedures; and one was nonadherent. The baseline characteristics of mothers treated in the third trimester with ATV/r 300/100 mg were comparable to those of mothers treated with ATV/r 400/100 mg (Table 1). Thirty infants (75%) were born full term and 10 (25%) were born prematurely (one patient withdrew). The study design, interim analysis, pre-specified criteria and post interim analysis protocol are shown in Figure 1.

Table 1.   Demographics and baseline characteristics of mothers and infants*
 Mothers' third trimester regimen of atazanavir/RTV
300/100 mg400/100 mg
  • APGAR, appearance pulse grimace activity respiration; CDC, Centers for Disease Control and Prevention; RTV, ritonavir; SE, standard error.

  • *

    Percentages are based on patients with measurements.

  • The definition of premature is <37 weeks.

 Age (years) [mean (SE)]29 (1.1)28 (1.0)
 Gestational age at baseline (weeks) [mean (SE)]23 (1.2)23 (1.2)
Race [n (%)]
 Black/African American17 (85)19 (90)
 Caucasian3 (15)1 (5)
 Mixed race01 (5)
Region [n (%)]
 South Africa14 (70)15 (71)
 North America6 (30)6 (29)
 HIV RNA level (log10 copies/mL) [mean (SE)]3.52 (0.232)4.02 (0.171)
 CD4 cell count (cells/μL) [mean (SE)]435 (39.2)390 (25.0)
 Prior CDC class C events (n)00
 Prior antiretroviral treatment [n (%)]5 (25)3 (14)
 Current antiretroviral treatment [n (%)]4 (20)1 (5)
Gestational age at delivery (weeks)
 Mean (SE)38 (0.3)38 (0.3)
 Min, max35, 4035, 41
 Number of premature infants64
 Female [n (%)]8 (40)11 (55)
 Length (cm) [mean (SE)]48 (0.7)49 (0.5)
 Weight (kg) [mean (SE)]2.8 (0.10)2.9 (0.10)
 Number <2.5 kg34
 Number <1.5 kg00
APGAR score [mean (SE)]
 At 1 min8.55 (0.185)8.55 (0.135)
 At 5 min9.40 (0.134)9.20 (0.138)
Figure 1.

 Design of the original protocol, interim analysis, pre-specified criteria and post interim analysis protocol. RTV, ritonavir.

Interim analysis

Twenty women received ATV/r 300/100 mg in the third trimester. The interim analysis (Fig. 1) was performed on the first 12 of these 20 patients. The lowest Cmin observed in the first 12 patients was 196 ng/mL and the geometric mean of the Cmin was 514 ng/mL. Therefore, the Cmin analysis did not warrant a dose increase according to this pre-specified criterion. However, the geometric mean of the AUCτ (26 647 ng h/mL) fell inside the pre-specified range (<30 000 and ≥15 000 ng h/mL) for a dose increase; therefore, the dose was increased to 400/100 mg during the third trimester for an additional 21 patients. After the decision to increase the third trimester dose, patients who were in their second trimester underwent blood sampling for pharmacokinetic analysis of ATV/r 300/100 mg.


Of the 20 patients being treated with ATV/r 300/100 mg in the third trimester, one patient discontinued because of premature labour. The infant was born 12 days later. The maternal HIV RNA was <50 copies/mL at delivery. Two of 21 patients receiving ATV/r 400/100 mg in the third trimester discontinued before delivery. One patient withdrew consent to participate in the study after 3 weeks of therapy, and therefore maternal HIV RNA or infant HIV DNA results were not available at or after the delivery for this mother–infant pair. The second patient was diagnosed with pre-eclampsia with grade 3–4 transaminitis, and all ARVs were stopped. The infant was delivered by Caesarean section 4 days later. Maternal HIV RNA was <50 copies/mL prior to ARV discontinuation, and at delivery. Of the 41 mothers, 12 (29%) had vaginal births, 14 (34%) had scheduled Caesarean sections and 14 (34%) had unscheduled Caesarean sections.

Clinical efficacy

All maternal HIV RNA measurements were <400 copies/mL at the time of delivery for both treatment groups. At the time of delivery, maternal HIV RNA <50 copies/mL was achieved for all 19 patients on the 300/100 mg regimen and for 19 of 20 patients on the 400/100 mg regimen. One patient on ATV/r 400/100 mg had three consecutive HIV RNA measurements <50 copies/mL prior to delivery, an HIV RNA of 59 copies/mL at delivery, and subsequent re-suppression post-delivery to <50 copies/mL. All infants were HIV DNA negative at delivery and up to 6 months.

Pharmacokinetic analysis

Concentration–time curves for ATV/r 300/100 mg and 400/100 mg during pregnancy are shown in Figure 2. During the third trimester, the AUCτ and Cmax of ATV/r 300/100 mg were 21% and 27% lower, respectively, than historical data but the Cmin values were comparable (Table 2). During the same time period, the Cmin value for ATV/r 400/100 mg was 39% higher than the historical controls, but the AUCτ and the Cmax values were comparable (Table 2). All Cmin values observed were at least 10 times greater than the protein-binding adjusted EC90 values for ATV [effective concentration against 90% of viral isolates (EC90) equals 14ng/mL for wild-type virus], and the lowest Cmin observed was 199 ng/mL. The maternal and cord blood concentrations of ATV were similar between the two dosing regimens at the time of delivery (Table 2). The fetal:maternal ratios of plasma concentration (using cord concentration as surrogate for fetal) were 0.19 and 0.12 for the ATV/r 300/100 mg and 400/100 mg regimens, respectively.

Figure 2.

 Pharmacokinetic analysis of 300/100 mg and 400/100 mg atazanavir/ritonavir (RTV) during pregnancy, 300/100 mg atazanavir/RTV postpartum and 300/100 mg atazanavir/RTV in nonpregnant HIV-infected adults (historical data). Historical data were from women and men in studies AI424089 and AI424137.

Table 2.   Pharmacokinetic and safety results. (a) Atazanavir exposures during the antepartum and postpartum periods; (b) atazanavir exposures during the peripartum period; (c) adverse events and laboratory abnormalities in mothers and infants
 Second trimesterThird trimesterThird trimesterPostpartumHistorical*
300/100 mg (n=9)300/100 mg (n=20)400/100 mg (n=20)300/100 mg (n=36)300/100 mg (n=23)
 Geometric mean (ng/mL)37293291421156494485
 Geometric mean ratio0.830.730.941.26
 (90% CI)(0.65, 1.06)(0.56, 0.96)(0.76, 1.17)(1.09, 1.46) 
 Geometric mean (ng h/mL)34 39934 25246 60260 53343 388
 Geometric mean ratio0.790.791.071.40
 (90% CI)(0.61, 1.03)(0.63, 0.99)(0.84, 1.37)(1.19, 1.64) 
 Geometric mean (ng/mL)6646689171421662
 Geometric mean ratio1.001.011.392.15
 (90% CI)(0.65, 1.54)(0.73, 1.39)(0.96, 2.00)(1.65, 2.80) 
 Geometric mean (%CV)
 Third trimester300/100 mgThird trimester400/100 mg
Maternal concentration at delivery (ng/mL)(n=19)(n=17)
 1201 (63)1224 (54)
Cord blood concentration (ng/mL)(n=14)(n=15)
 224 (67)184 (68)
Fetal/maternal ratio(n=14)(n=15)
0.19 (41)0.12 (49)
 Mothers' third trimester regimen of atazanavir/RTV [n/total (%)]
300/100 mg400/100 mg
  • AUCτ, area under concentration-time curve in one dosing interval; CI, confidence interval; Cmax, maximum observed plasma concentration; Cmin, trough observed plasma concentration 24 hour post dose.

  • CV, coefficient of variation.

  • ALT, alanine aminotransferase; AST, aspartate aminotransferase; RDS, respiratory distress syndrome; RTV, ritonavir; ULN, upper limit of normal.

  • *

    Sources of historical values were studies AI424089 and AI424137.

  • Anaemia (n=1) was a related serious adverse event.

  • Anaemia (n=2) and hyperbilirubinaemia (n=1) were related serious adverse events.

  • §

    § Based on adverse event reports, not laboratory values.

  • Fasting lipids were measured at screening and postpartum week 4 only.

  • Infant born at 35.6 weeks following premature rupture of membranes; maternal bilirubin 2.3 g/dL on day of delivery; birth weight 1.8 kg; infant bilirubin 13 g/dL on day 3; received 2 days of phototherapy; discharged healthy.

  • **

    Threshold values for grade 3–4 abnormalities were not provided because they vary depending on the age of the newborn.

 Adverse events
  Serious adverse events
   Any serious adverse events7/20 (35)8/21 (38)
   All grade 2–4 treatment-related adverse events4/20 (20)6/21 (29)
  All grade 2–4 treatment-related adverse events of clinical interest
   Hyperbilirubinaemia§0/20 (0)1/21 (5)
   Anaemia1/20 (5)3/21 (14)
   Vomiting1/20 (5)0/21 (0)
   Diarrhoea1/20 (5)0/21 (0)
   Impaired glucose tolerance0/20 (0)1/21 (5)
 Grade 3–4 laboratory abnormalities
  Total bilirubin elevation (>2.5 × ULN)6/20 (30)13/21 (62)
  ALT (>5 × ULN)0/20 (0)1/21 (5)
  AST (>5 × ULN)0/20 (0)1/21 (5)
  Total cholesterol (≥240 mg/dL)1/14 (7)3/16 (19)
  Haematocrit (<24%)2/20 (10)1/21 (5)
  Hypocarbia (≤14 mEq/L)1/20 (5)8/21 (38)
 Serious adverse events
  Any serious adverse events10/20 (50)4/20 (20)
  Zidovudine overdose1/20 (5)0/20 (0)
  Anaemia1/20 (5)1/20 (5)
  Restrictive cardiomyopathy1/20 (5)0/20 (0)
  Hypoglycaemia0/20 (0)1/20 (5)
  Prematurity1/20 (5)1/20 (5)
  Respiratory distress/neonatal RDS2/20 (10)0/20 (0)
  Hyperkalaemia1/20 (5)0/20 (0)
  Infections and infestations4/20 (20)2/20 (10)
  Hyperbilirubinaemia0/20 (0)1/20 (5)
  Jaundice1/20 (5)0/20 (0)
  Cerebral ischaemia1/20 (5)0/20 (0)
  Vomiting1/20 (5)0/20 (0)
 Bilirubin-related adverse events and laboratory abnormalities
  Any grade jaundice10/20 (50)10/20 (50)
  Any grade total bilirubin elevation5/20 (25)9/20 (45)
  Grade 3–4 total bilirubin elevation3/20 (15)4/20 (20)
  Number of infants who received phototherapy33
 Grade 3–4 laboratory abnormalities**
  Hypoglycaemia0/20 (0)1/19 (5)
  Haemoglobin0/6 (0)1/6 (17)
  Hyperkalaemia4/20 (20)6/18 (33)

Maternal adverse events and laboratory abnormalities

Both ATV/r 300/100 mg and 400/100 mg were well tolerated, with no unanticipated adverse events (Table 2). Similar numbers of serious adverse events were observed for the two regimens: seven of 20 (35%) and eight of 21 (38%) for ATV/r 300/100 mg and 400/100 mg, respectively. Grade 3–4 laboratory abnormalities included elevation of total bilirubin (>2.5 times the upper limit of normal), which occurred at twice the rate with 400/100 mg (62%) than with 300/100 mg (30%) (Table 2).

Infant adverse events and laboratory abnormalities

Ten of 20 and four of 20 infants born to mothers who received ATV/r 300/100 mg and 400/100 mg during the third trimester, respectively, experienced serious adverse events (Table 2). One infant was exposed to an overdose of zidovudine. Safety events related to bilirubin elevation and grade 3–4 laboratory abnormalities for infants are listed in Table 2.

Bilirubin elevation of any grade occurred in 14 of 40 infants (35%). No difference in the time course of bilirubin elevation occurred between infants whose mothers received the 300/100 mg and 400/100 mg doses of ATV/r in the third trimester. Among infants who had elevated bilirubin in the first 14 days, none was elevated to grade 1 because thresholds for grade 1 bilirubins are higher in the first 2 weeks of life to account for normal, physiological bilirubin elevations [21]. Six infants underwent phototherapy (at ages ranging from 3 to 6 days); four infants had 2 days of phototherapy, one had 3 days, and one had 4 days; bilirubin levels associated with these events ranged from 8.1 to 13 mg/dL.

The total bilirubin level was <4 mg/dL in all infants by week 12. All grade 3–4 bilirubins occurred after day 14; however, levels were decreasing by day 14 in all cases. The infants' total bilirubin levels on delivery day correlated weakly with the mothers' total bilirubin levels at delivery (Fig. 3a). When the infants' total bilirubin levels were compared with the mothers' bilirubin levels over the last 4 weeks of pregnancy, an even weaker correlation was observed (Fig. 3b).

Figure 3.

 Analysis of bilirubin levels in mothers and infants. (a) The relationship between infants' total bilirubin levels on delivery day and mothers' total bilirubin levels at delivery. (b) The relationship between infants' total bilirubin levels on delivery day and mothers' mean total bilirubin levels over the last 4 weeks of pregnancy.


ATV/r 300/100 mg qd dosing had a lower AUCτ and Cmax in the third trimester of pregnancy compared with the AUCτ and Cmax in nonpregnant adults; however, Cmin values were similar. Increasing the dose of ATV/r to 400/100 mg achieved AUCτ and Cmax values similar to those in nonpregnant adults and higher Cmin values. ATV concentrations were elevated in the postpartum period. This observation has been made for other protease inhibitors [22–24]. ATV concentrations appear to normalize by week 7 postpartum [18].

The ratio of maternal to cord blood ATV indicates that, as with other protease inhibitors [25], ATV does not freely cross the placenta; however, in this study, plasma protein binding in cord blood was lower than in maternal blood, indicating that free drug concentrations in the fetus were approximately twice as high as in the mothers at a similar total (bound and unbound or ‘free’) ATV plasma concentration [26], suggesting that the levels achieved in the cord blood may provide some antiviral protection to the fetus [18].

A theoretical concern with ATV use during pregnancy is an increase in bilirubin in newborns [1]. This could occur either by passive back-diffusion of infant bilirubin across the placenta, as a result of saturation of protein binding of bilirubin in mothers with elevated maternal bilirubin, or theoretically through the effect of UGT1A1 inhibition in the fetus as a result of ATV crossing the placenta. The placenta normally only allows unidirectional flow of bilirubin from the fetus to the mother, and not from the mother to the fetus. Unbound fetal bilirubin crosses the placenta as a consequence of the greater albumin-binding capacity of maternal serum, and is then bound to maternal albumin and carried to the maternal liver for conjugation and elimination. Theoretically, a persistence of very high maternal bilirubin levels might disrupt the normal transplacental flow of fetal bilirubin, leading to intrauterine hyperbilirubinaemia. The actual threshold of maternal bilirubin level and the duration of elevation required to disrupt normal transplacental flow are unknown, and data are limited to case studies with conflicting results [27–30]. This study had a conservative rule whereby a maternal bilirubin level of 10 mg/dL at any time or a level of 7.5 mg/dL persisting for 2 weeks mandated discontinuation of the study drug. However, this rule was not invoked during the study. Overall the rate of grade 3–4 hyperbilirubinaemia observed in this study was, as expected because of the reduced ATV exposures in pregnancy, lower than observed in studies of ATV/r 300/100 mg in nonpregnant adults; for example, in study AI424089, grade 3–4 bilirubinaemia was 59% [31], in contrast to the 30% observed in the current study.

This study found a weak correlation between maternal bilirubin, both on the day of delivery and over the 4 weeks prior to delivery, and infant bilirubin. Although cord blood concentrations of ATV were <20% of the plasma concentrations on average, the free drug concentrations in the fetus were, as noted, higher than in the mothers at similar total (bound+free) ATV concentrations [26]. While the ATV that crossed the placenta may have inhibited fetal UGT1A1, the placental transport system and maternal elimination of fetal bilirubin appeared to be adequate to deal with any elevated fetal bilirubin. The observed pattern of infant bilirubin was generally consistent with the neonatal physiological elevations of bilirubin. Six infants (15%) did undergo phototherapy; however, infant jaundice and phototherapy are not rare. In fact, about 60% of otherwise healthy term infants will experience jaundice and about 10% of them will require some form of treatment (phototherapy or exchange transfusions) [32,33].

Regarding safety overall for the infants, only three serious adverse events were reported as related to drugs used in the study, with the drug implicated being zidovudine, and only two serious adverse events were hepatobiliary (hyperbilirubinaemia and jaundice). The majority of serious adverse events (12 of 14) experienced by infants whose mother received ATV/r 300/100 mg were unlikely to be, or were not, related to the study medication.

Regarding efficacy, the selection of a suitable threshold can be controversial; maintaining a plasma concentration of protease inhibitors above a certain threshold appears to be correlated with positive outcome. The US Department of Health and Human Services Treatment guidelines suggest a minimum ATV Cmin of 150 ng/mL if therapeutic drug monitoring is to be used [34]. Recognizing the variability in the Cmin value, and that a reduction in the third trimester might be expected, the exposure criterion for Cmin was set to allow up to two patients to fall below 150 ng/mL and the given regimen still to be considered viable. For AUC, the criterion was set at a geometric mean of 30 000 ng h/mL based on our rationale that a reduction of up to 30% in ATV AUC would not compromise outcome. The criteria for a dose increase within the current study were based on the assumption that, although exposures were likely to be lower in pregnant patients, the relationship between these AUC and Cmin values would be largely consistent with that in nonpregnant patients. Reductions of 20–30% in ATV AUC and Cmin were observed when ATV was given in combination with tenofovir, with no apparent loss of antiviral effect [35]. Indeed, the recent CASTLE study (AI424138) indicated that, even though tenofovir lowered ATV exposures, the antiviral efficacy was very good and comparable to that for twice-daily lopinavir/RTV to 96 weeks [36]. In this study, the lowest observed AUC fell below the range of historical reference values, but the relationship between AUC and Cmin differed in this population, where Cmin values were higher than in nonpregnant patients at similar AUC values. At ATV/r 300/100 mg qd, the range of observed Cmin values in the third trimester was very comparable to the historical reference [interquartile range 455.5–986.0 ng/mL (current study) vs. 370–1035.3 ng/mL (historical)]. Furthermore, with data from 20 patients, the geometric mean AUC for 300/100 mg qd meets the predefined criterion for AUC. Although this result appears to conflict the interim analysis with 12 patients, considering the known variability in ATV pharmacokinetics, these two estimates of the population mean are not incompatible.

On the basis of the pharmacokinetic data in this study, particularly Cmin, a dose adjustment does not appear to be necessary during pregnancy. The seeming disconnect between the decision to study a second cohort at 400/100 mg qd and the recommendation of 300/100 mg qd is based in large part on the differing relationship between AUC and Cmin in this population. After reviewing the pharmacokinetic data as a whole, the dosing recommendation is rational despite this apparent contradiction within the study. Any consideration of a dose increase should also take relative safety profiles and ease of compliance with a new dosing regimen into account. For the latter consideration, switching from one 300 mg capsule to two 200 mg capsules of ATV at the beginning of the third trimester may lead to dosing errors and compliance problems. In this regard, not having to dose-adjust during pregnancy and complicate the ATV/r 300/100 mg treatment regimen could be viewed as a potential benefit. Regarding safety considerations, both ATV/r 300/100 mg and 400/100 mg were well tolerated with no unexpected, related adverse events; however, maternal grade 3–4 hyperbilirubinaemia occurred more frequently at the higher dose. It should be noted that ATV/r 300/100 mg dosing with tenofovir may not be adequate in pregnant women, as tenofovir has been shown to reduce ATV exposures when the two drugs are coadministered. A study is ongoing to assess ATV/r 400/100 mg dosing with tenofovir during pregnancy [37].

A limitation of this study is that the historical controls comprised both men and women who were primarily Caucasians from the Americas and Europe, and this study included primarily women from South Africa. However, previous studies of ATV/r 300/100 mg have showed no significant pharmacokinetic differences by gender [38] or differences in clinical outcome by race [39].

The clinical outcomes from this Phase I study suggested that treatment of pregnant mothers with ATV/r 300/100 mg qd and zidovudine/lamivudine bid was efficacious in the suppression of HIV RNA in these patients, and, together with 6 weeks of prophylaxis in the infants, it prevented mother-to-child HIV-1 infection. The pharmacokinetics, safety and efficacy data obtained in this study suggest that, when ATV/r is used during pregnancy, a dose adjustment is not required for ATV. This indicates that ATV/r 300/100 mg in combination with a zidovudine/lamivudine 300/150 mg bid backbone may be a good treatment option for HIV-infected pregnant women.


The study team would like to acknowledge the mothers and their families for their participation and commitment during the study. We thank Bristol-Myers Squibb employees Moegsina Gomez, Marina Mathew, Kristy Grimm, Awny Farajallah and Sophia Hilaly for their support and contributions to the successful completion of the study and Yonghua Wang for her help with the statistical analysis. This Bristol-Myers Squibb-supported study is also known as Study AI424182 and is registered with, number NCT00326716. Professional medical writing and editorial assistance was provided by Carolyn Carroll and funded by Bristol-Myers Squibb.

Conflicts of interest: F.C. reports receiving research support from Bristol-Myers Squibb, GlaxoSmithKline, Tibotec, Schering Plough, Gilead Sciences and Abbott Laboratories; C.Z. reports receiving grant support from Tibotec, Pfizer, Bristol-Myers Squibb, Advent and the NIH institutes: NIAID, NCRR and NIMH. C.Z. also reports being a member of the Tibotec Presidents Council (advisory group). M.B. reports receiving research support from Pfizer, Boehringer-Ingelheim and Bristol-Myers Squibb, receiving lecture fees from Bristol-Myers Squibb, Roche and Aspen, and receiving financial support for conference attendance from Roche. O.O. reports receiving research support from Johnson & Johnson, Tibotec, Bristol-Myers Squibb, ViiV, Pfizer, Merck and Clinlogix, and consulting fees from Gilead Sciences. O.O. also reports being on the speaker bureau for Gilead Sciences and Abbott Laboratories. E.V., T.E., M.C., R.B., W.H., V.W. and D.M. report being employees and shareholders of Bristol-Myers Squibb.