Utility of therapeutic drug monitoring in the management of HIV-infected pregnant women in receipt of lopinavir

Summary

The pharmacokinetics of antiretroviral drugs in pregnancy is poorly understood. We reviewed the use of therapeutic drug monitoring (TDM) in clinical settings to document plasma concentrations of lopinavir during pregnancy and investigated how clinicians acted upon TDM results. A retrospective review was carried out of all HIV-infected pregnant women taking boosted lopinavir-based highly active antiretroviral therapy (HAART) at five National Health Service (NHS) centres in the UK between May 2004 and March 2007. Seventy-three women in receipt of lopinavir were identified, of whom 89% had plasma lopinavir concentrations above the suggested minimum recommended for wild-type HIV. Initial TDM results prompted dosage change in 10% and assessment of adherence and/or pharmacist review in 11%. TDM was repeated in 29%. TDM can play an important role in the clinical management of HIV-positive pregnant women, allowing informed dose modification and an alternative measure of adherence.

INTRODUCTION

With the exception of zidovudine (ZDV) in the second and third trimesters, antiretroviral therapy is not licensed for use in pregnancy, but is widely prescribed to reduce mother-to-child transmission and to improve maternal health.¹ Pregnancy remains a contraindication to entry to studies of antiretroviral therapies. Consequently, therapies are prescribed in this setting without pharmacokinetic, toxicity and efficacy data to inform of their safe and most effective use. Physiological changes in pregnancy are known to alter volume of distribution, protein binding, body fat to water ratio and renal function, which may increase the interpatient variability in drug concentrations.² As these compounds are increasingly prescribed in pregnant women, there is urgent need for data on many antiretroviral therapies in pregnancy, especially those more recently introduced.

There is now increasing use of ritonavir-boosted protease inhibitors as part of highly active antiretroviral therapy (HAART) in pregnancy. In the UK, saquinavir and lopinavir are widely prescribed as first-line therapy in pregnancy, while atazanavir is used as second-line therapy, often when a once-daily regimen is deemed imperative. When HAART is initiated during pregnancy, there is an extra time constraint to fully suppress HIV replication not only prior to term, but also in anticipation of preterm delivery, which occurs in 20–25%.³ Furthermore, there is concern that subtherapeutic concentrations of the protease inhibitor, particularly in the third trimester of gestation, might lead to the development of treatment failure and resistance to other components of the regimen.

While it has been suggested that no dose adjustment is required for ritonavir-boosted saquinavir,⁴ there have been conflicting reports in relation to ritonavir-boosted lopinavir (LPV/r).^5,6 Data from a formal pharmacokinetic study showed lower lopinavir concentrations during the third trimester compared with paired postpartum concentrations leading to the suggestion that all pregnant women taking lopinavir should receive a higher dose.⁵ However, data from observational studies suggest considerable interpatient variability and these authors suggest a more targeted approach to dose escalation.⁶ Thus to ensure adequate drug concentrations during pregnancy, clinicians have begun to utilize therapeutic drug monitoring (TDM). However, there is no consensus as to when best to conduct TDM or of its overall clinical use. We have therefore reviewed the use of TDM performed on HIV-infected women taking LPV/r during pregnancy at five UK centres. We aimed to evaluate the effectiveness of TDM, document plasma concentrations of lopinavir during pregnancy and assess how clinicians act upon TDM results.

METHODS

A retrospective review was carried out of all HIV-infected pregnant women taking LPV/r-based HAART between May 2004 and March 2007 at five National Health Service (NHS) centres in the UK: King’s College Hospital, London; the Royal Liverpool University Hospital, Liverpool; Mayday University Hospital, Croydon; Northampton General Hospital, Northampton; and St Mary’s Hospital, London.

The inclusion criteria included pregnancy and having at least one TDM result sampled while receiving LPV/r at steady state (greater than 2 weeks). The formulation of LPV/r given in the majority was soft-gel capsules, but seven (10%) received the melt-extruded tablet formulation. The demographic details collected for each patient were as follows: ethnicity, gestation age at TDM and maternal age. The clinical/pharmacological details collected were as follows: CD4+ lymphocyte cell count, HIV plasma viral load (VL), lopinavir plasma concentration, lopinavir formulation and dose, concurrent medications, liver enzymes, infant HIV DNA polymerase chain reaction (PCR) results at age three months, adherence, mode of delivery and clinicians’ response to abnormal TDM results.

Drug concentrations were measured using validated high-performance liquid chromatography-tandem mass spectrometry, with a lower limit of quantification of 250 ng/mL.⁷ The suggested minimum concentration required for the full suppression of wild-type HIV replication is 1000 ng/mL.⁸ Lopinavir plasma concentrations obtained in individual pregnant subjects receiving LPV/r twice daily were plotted on population pharmacokinetic curves. Population pharmacokinetic profiles were obtained from 23 non-pregnant subjects receiving LPV/r 400/100 mg twice daily (University of Liverpool, unpublished data).

All statistical analyses were performed using SPSS version 15.0. Univariate associations between variables and virological outcome were explored using the t-test for continuous variables and the χ² or Fisher’s exact test for categorical variables. Logistic regression was used for multivariate analyses. In some cases, variables were transformed into dichotomous variables prior to analysis, e.g. adherence results were divided into complete (excellent) or incomplete (good/poor); viral load into virologically suppressed (<50 copies/mL) or not suppressed (>50 copies/mL) and TDM results into sufficient (>1000 ng/mL) or insufficient (<1000 ng/mL).

All tests were two sided and P values <0.05 were considered significant.

RESULTS

Seventy-three women took LPV/r during their pregnancy. Fifty-six (77%) were black African, 12 (16%) Caucasian, one (1%) Afro-Caribbean, one (1%) south east Asian and three (4%) were of mixed race. Their median age was 29 years (range 15–44).

Of the 73 pregnant women taking LPV/r, 46 (63%) took a nucleoside analogue reverse transcriptase inhibitor backbone of ZDV plus lamivudine (3TC). The other backbones prescribed are outlined in Table 1.

Table 1.

Details of ART backbones prescribed with LPV/r

Other antiretroviral agents	No. of women	%
ZDV/3TC (Combivir®)	46	63.0
TDF/3TC	3	4.1
TDF/ZDV	3	4.1
ABC/ZDV/3TC (Trizivir®)	3	4.1
DDI/ZDV	3	4.1
ABC/TDF	2	2.7
Unknown	2	2.7
DDI/3TC	1	1.4
DDI/ABC	1	1.4
DDI/TDF	1	1.4
ABC/3TC	1	1.4
ABC/NVP	1	1.4
ABC/ZDV	1	1.4
TDF/3TC/D4T	1	1.4
SQV/TDF/ABC	1	1.4
TDF/SQV	1	1.4
DDI/ABC/3TC/SQV	1	1.4
TDF/ZDV/3TC	1	1.4

ART = antiretroviral therapy; LPV/r = ritonavir-boosted lopinavir; ZDV = zidovudine; 3TC = lamivudine; TDF = tenofovir; ABC = abacavir; ddI = didanosine; NVP = nevirapine; D4T = stavudine; SQV = saquinavir

Twenty-nine women (40%) conceived while on HAART or had taken antiretroviral therapy previously, of which eight (11%) were receiving lopinavir preconception. The remainder commenced HAART at a median gestational age of 24 weeks (range 3–35).

Overall pretreatment median CD4+ lymphocyte (CD4) count was 326 × 10⁶ cells/μL (range 13–869). The median pretreatment CD4 of those who commenced therapy during pregnancy (n = 64) was 331 × 10⁶ cells/μL (range 34–869). Those already on treatment (n = 9) tended towards a lower CD4 with a median of 52 × 10⁶ cells/μL (P = 0.009; range 13–38 in 4 women, in 3 no data were made available and 2 had a CD4 of >300).

The median pretreatment plasma HIV RNA was 12,580 copies/mL (range <50–422,000); three women (4%) were virologically suppressed (<50 copies/mL). At the time of first TDM, the median VL was 182 copies/mL (range <40–252,000), where the women had been taking LPV/r for an average of 19 days (Table 2); 22 (30%) women had a VL of <50 copies/mL; VL was again measured in 65/73 (89%) women at 36 weeks gestation. Fifty-two (80%) of these women had a plasma HIV RNA <50 copies/mL (range ,50–412,000).

Table 2.

Details of TDM samples (n = 73)

Details of TDM samples	Median (range)
Days on LPV/r at sampling	19 (7–180)
Hours since last dose	12 (1–17)
VL at trough sampling (copies/mL)	182 (40–252,000)
Trough concentration on LPV/r  400/100 mg (ng/mL), n = 66	4334 (<250–17,486)
Trough concentration on LPV/r  533/133 mg (ng/mL), n = 7	5197 (<250–7718)

TDM = therapeutic drug monitoring; LPv/r = ritonavir-boosted lopinavir; VL = viral load

LPV/r was administered every 12 hours and levels were taken as close to 12 hours post dose as possible; however, only 78% of the TDMs were performed within two hours of the 12-hour ideal (range 1–17 hours, median 12 hours). The median gestational age at first TDM was 29 weeks (range 9–38). The trimester of initial TDM was third in 44/73 (60%), second in 21/73 (29%) and not clearly documented in 8/73 (11%). Only 7/21 (33%) patients who’s initial TDM was taken in second trimester had a repeat TDM ensuring adequate LPV/r concentrations in the third trimester.

Adherence, as defined by individual physicians, was reported as excellent in 55 (75%), good in six (8%), poor in 10 (14%) and not reported in two (3%). Medication error in one patient resulted in ingestion of half-dose LPV/r for two weeks.

In 65/73 (89%), the plasma concentration of lopinavir in the initial TDM sample was above the suggested minimum concentration for wild-type HIV (Figure 1). One of 73 (1%) had a concentration above the 90th centile for non-pregnant adults. In 8/73 (11%) samples, the lopinavir concentration was <1000 ng/mL.⁸ Six of these eight samples had concentrations below the limit of quantification (<250 ng/mL). There was no correlation between TDM results and VL (r² = 0.0075).

Figure 1.

Lopinavir plasma concentrations obtained in individual pregnant subjects receiving lopinavir/ritonavir twice daily. Each marker represents an individual concentration. Lines represent population centile data (P50 = solid, P25 and P75 = dashed, P10 and P90 = dotted) from population pharmacokinetic profiles obtained from 23 male HIV-positive subjects receiving lopinavir/ritonavir 400/100 mg twice daily (University of Liverpool, unpublished data)

As a result of their initial TDM, LPV/r dose was adjusted in seven patients (10%) (5; 7% increased; 2; 3% decreased). Other interventions as a result of the initial TDM included those with low levels (8; 11%) having adherence reviews and re-assessment by the pharmacist.

Repeat TDM was performed in 21 (29%) patients: five had a low lopinavir concentration at the initial TDM; four to monitor a change in dosage, of which three had increased lopinavir dose for the third trimester and one had decreased the dose; in two there were concerns regarding toxicity; one had detectable VL despite a normal initial TDM; two after switching formulation; one as part of formal pharmacokinetic study; one had the initial TDM in the first trimester, so this was repeated in the third trimester; five had no reason given or it was not clear from the data provided, although one was stated to be a ‘routine’ TDM.

Univariate analysis of the variables revealed that length of time on the current regimen (>3 weeks) was significantly associated with viral suppression (<50 HIV RNA copies/mL) at time of first TDM (P = 0.01). Previous treatment with HAART was borderline significant at this time also (P = 0.05). However, in multivariate analysis only length of time on current regimen remained a significant predictor of viral suppression, P = 0.02, odds ratio (OR) = 6.3 (95% confidence interval [CI], 1.9–20.6). Baseline VL, baseline CD4 cell count, TDM result and adherence were not associated with virological outcome.

Similar analysis of viral suppression at 36 weeks gestation revealed that only adherence was a significant predictor of viral suppression both in univariate and multivariate analysis (P = 0.007; P = 0.003, OR = 9 [95% CI, 2–38.7]). No association was found between baseline VL, baseline CD4 cell count, TDM result, length of time of HAART or previous exposure to HAART.

Although not adequately powered to detect birth outcomes, there was one (1/73, 1.4%) HIV-positive infant (tested at 3 months) in the cohort. In this case, the mother seroconverted during pregnancy. In addition there were three (3/73, 4%) antenatal/perinatal infant deaths (1 trisomy, and twins born prematurely at 23/40). No other infant abnormalities were detected at the time of the study.

DISCUSSION

In this large retrospective case series, a wide variability in lopinavir concentrations in the third trimester of pregnancy was observed, confirming earlier reports.^5,6

It has been suggested that concentrations of lopinavir below 1000 ng/mL are not adequate for suppression of wild-type virus,⁸ and although studies have not provided a cut-off value, there have been positive correlations between lipid profiles⁹ and liver enzyme elevation¹⁰ and lopinavir pharmacokinetics. There is currently a lack of pharmacokinetic data in HIV-infected pregnant women and as such ‘therapeutic’ levels of lopinavir are based on data obtained from a Caucasian male cohort (University of Liverpool, unpublished data).

The current guidelines on the management of HIV-positive pregnant women (BHIVA/CHIVA, 2008) point out that LPV/r appears to be well tolerated and clinically effective, but recommend TDM for protease inhibitors (PIs) during pregnancy and suggest this should take place at steady state (2 weeks or more into therapy) and be repeated in the third trimester.¹¹

Some clinicians in this study monitored lopinavir concentrations twice during pregnancy as ‘routine’ with the argument physiological changes between second and third trimester may otherwise be missed. Others feel this unnecessary and unsupported by current evidence and opt for concentration to be measured during the third trimester. These differences in clinical practise have been fuelled by conflicting reports of lopinavir levels in pregnant women during late pregnancy. Stek et al.⁵ showed lower concentrations of lopinavir during the third trimester at standard dosing than at six weeks postpartum in the same women. Based on their prospective trial of 21 HIV-infected pregnant women on LPV/r/FTC, the PACTG 1026s team argue the higher dose of LPV/r (3 tablets twice daily) provides appropriate exposure during third trimester and should be considered during second trimester. They showed that by using three lopinavir tablets twice daily from the start of third trimester, women achieved lopinavir concentrations that equated to those measured postpartum in the same women dosed with two lopinavir tablets twice daily.¹² They argue that this is the correct approach rather than looking at whether effective concentrations are achieved and the virological outcomes. Manavi et al.¹³ have also published their experience with 26 HIV-infected pregnant women in receipt of lopinavir/r, where four (15.4%) were found to have subtherapeutic concentrations.

However according to Bouillon-Pichault et al.¹⁴ LPV/r 400 mg/100 mg twice daily regimen was adequate in reaching the minimum concentration required to suppress wild-type virus replication in PI-naïve pregnant women, but may not be suitable for PI-experienced pregnant women. Lyons et al.⁶ showed adequate drug levels in their cohort of third-trimester patients and argued that no increase of lopinavir is routinely warranted for the third trimester.

Lopinavir levels vary widely between non-pregnant patients taking LPV/r. The additional metabolic and physiological changes known to occur during pregnancy make a strong case for performing lopinavir TDM antenatally. However, interpretation of the results and their use as a guide to further management is not so clear.¹⁵ It is a complicated situation with others finding significant intraindividual variability in antiretroviral concentrations suggesting dosing decisions made on a single TDM may be limited.^16,17

Our study revealed that adherence as assessed by the clinician was a significant predictor of viral outcome, whereas TDM result was not. There is still debate, however, over whether adherence can be ‘tested’ or verified using repeated TDM or whether VL response to adherence counselling is more appropriate as well as cost-effective. Of the six instances of lopinavir being undetectable in plasmas, all clinicians tackled poor adherence issues with time spent educating the patient including a review with the pharmacist. One clinician felt this satisfactorily tackled the issue of detectable VL and low TDM and that a repeat TDM was not warranted. He felt the initial TDM did not add to his management, as he was aware of adherence issues that were also evidenced by a detectable VL. The other clinician repeated two patients’ TDM in combination with addressing adherence issues.

TDM directly altered management in the cases where adherence was good, but levels still subtherapeutic (n = 4, 5%). The lopinavir dose in these individuals was increased or increases were planned. It would appear the initial TDM did alter clinical management in these cases; however, another study would be required to see if it altered outcome.

There is an absence of large randomized prospective studies showing that TDM improves clinical outcome. This study, generated from the University of Liverpool TDM database, is the largest of its kind in published literature to date. As well as reporting usage of TDM for lopinavir in pregnancy, it highlights the lack of consensus between clinicians on the exact role of TDM in clinical management. Further analyses of this and other databases should be performed enhance our understanding of this subject.