Correlation between Plasma, Intracellular, and Cervical Tissue Levels of Raltegravir at Steady-State Dosing in Healthy Women

Abstract

Raltegravir is an antiretroviral with potential value for preexposure prophylaxis (PrEP) against HIV, but the intracellular pharmacokinetics in genital tissue have not been described. In this study, healthy, HIV-uninfected nonpregnant women took 400 mg of raltegravir twice daily for 22 days. On day 8, 15, and 22, blood was collected 0, 4, 6, 8, and 12 h and cervical biopsy specimens taken 0, 6, and 12 h after raltegravir dosing. Plasma and intracellular raltegravir concentrations in peripheral blood mononuclear cells (PBMC) and cervical tissue were measured by tandem mass spectrometry. Linear mixed effects models evaluated correlations between different sample types, as well as differences in concentration between phases of the menstrual cycle. Ten women were enrolled: 9 completed all three visits and 1 completed two visits. The age (mean ± standard deviation) of participants was 30 ± 8 years. Trough plasma concentrations of raltegravir 12 h after a directly observed dose were above the HIV 95% inhibitory concentration (IC₉₅) of 33 nM (14.6 ng/ml) in 96% of measurements, compared to 67% of PBMC and 89% of cervical tissue trough values. Across all measurements, only 2% (3/135) of plasma values fell below the IC₉₅, compared to 10% (13/135) for PBMC and 6% (5/81) for cervical tissue. There was no impact of menstrual phase on raltegravir concentrations. In conclusion, cervical tissue raltegravir concentrations were no greater than plasma concentrations, and ∼10% of all cervical tissue trough values were below the IC₉₅, making the current twice-daily formulation of raltegravir impractical for PrEP.

INTRODUCTION

Preexposure prophylaxis (PrEP) using antiretroviral medication has proven effective for infants and adults to prevent HIV acquisition. Studies have shown significant reduction in risk of HIV acquisition in high-risk infants born to HIV-infected women (1, 2), men who have sex with men (3), and HIV-1 serodiscordant heterosexual partners (4). The ideal agent for PrEP should have a long half-life and high concentrations in the genital tract at the site of viral exposure, should be nontoxic, and should not be part of first-line HIV regimens. Tenofovir was chosen as a potential PrEP agent in part because it fulfills almost all of these criteria (5). However, tenofovir is part of the first-line HIV treatment regimen in many parts of the world (6), raising concerns about the impact of development of drug resistance if HIV transmission occurs while taking PrEP.

Raltegravir is an HIV integrase inhibitor which blocks strand transfer of viral DNA and integration into the host's genome and has minimal adverse effects (7). Genital fluid concentrations of raltegravir are higher than plasma concentrations, an attractive quality for a PrEP agent (8, 9). However, no studies have examined intracellular tissue concentrations of raltegravir in the genital tract, where the level of the drug would be critical to ensuring efficacy (10). Raltegravir is a substrate for P glycoprotein transporter, which is expressed in cervical tissue (11) and is partially regulated by progesterone (12, 13), and thus intracellular concentrations could vary throughout the menstrual cycle.

In this pharmacokinetic (PK) study, we aimed to quantify genital tissue concentrations of raltegravir in comparison to concentrations in plasma and peripheral blood mononuclear cells (PBMC) and to evaluate whether concentrations and/or the proportion of values below the 95% inhibitory concentration (IC₉₅) change over the course of a menstrual cycle. We hypothesized that intracellular genital trough concentrations would exceed plasma trough values and that trough concentrations would be lower in the proliferative phase of the menstrual cycle due to higher concentrations of progesterone.

MATERIALS AND METHODS

Study procedures.

This was an open-label pharmacokinetic study to assess plasma and intracellular concentrations of raltegravir in white blood cells and genital tract tissue over the course of a menstrual cycle. Participants were recruited at the University of Washington in Seattle, WA, by online and print advertisements. Women were eligible for the study if they were healthy, between 18 and 50 years of age, menstruating regularly, and not using hormonal contraception. Participants were screened for pregnancy, HIV, and liver and kidney dysfunction prior to randomization. Exclusion criteria included irregular menstrual bleeding, diagnosed liver and/or kidney disorder, confirmed HIV infection, use of hormonal contraception, use of prescribed or over-the-counter proton pump inhibitors or H2 blockers, breast-feeding, and allergy to benzocaine. Participants were asked to refrain from using vaginal medications, creams, or lubricants in the 48 h prior to clinic visits and were advised to refrain from sexual intercourse for the duration of the study. This study received institutional review board approval from the University of Washington. All participants provided written informed consent prior to participation in the study.

Participants received 400 mg of raltegravir twice daily for 22 days. Participants were instructed to take doses with food and asked to either keep a medication diary to record dosage times or text message the study coordinator when taking each scheduled dose. Study visits were scheduled on day 8, 15, and 22 of medication use. At each visit, a follow-up questionnaire was administered to assess medication adherence, last menstrual period (LMP), use of vaginal medications, any drug- and biopsy-related side effects or clinical adverse events, and abstinence from sexual intercourse. After confirmation of no adverse events and a negative pregnancy test result, a peripheral intravenous line was placed for blood collection. Serial blood samples were collected prior to a directly observed dose of medication (0 h) and then 4, 6, 8, and 12 h after. Midway through the study, we added a single additional blood sample at the 0-h time point for measurement of serum progesterone. Cervical biopsy samples were collected at 0, 6, and 12 h. Biopsies were performed using Tischler forceps after numbing of the cervix with 20% benzocaine anesthetic spray (Beutlich Pharmaceuticals, Waukegan, IL). Bleeding was stopped using direct pressure and silver nitrate. One week after the final dose of study medication, participants received a follow-up phone call to determine the date of the last menstrual period and to check for possible adverse events.

Blood and tissue sample processing.

Blood was drawn into an 8-ml cell preparation tube (CPT; BD Biosciences, Franklin Lakes, NJ) with sodium heparin anticoagulant for PBMC isolation and a 6-ml tube for plasma collection. All samples were processed within 15 min of collection. Plasma was separated by centrifugation, and aliquots were stored at −80C. For isolation of PBMC, the CPT was initially spun at 1,800 × g for 30 min at room temperature, and then the plasma and PBMC layers above the gel gradient were mixed and transferred to a 50-ml conical tube. An aliquot was removed for counting cells, and PBMC count was determined using a hemocytometer. The PBMC suspension was spun at 400 × g for 10 min at 4°C, the supernatant was removed, and the pellet was resuspended in 1 ml of cold phosphate-buffered saline (PBS). The cell suspension was added to a 1.5-ml microcentrifuge tube with 150 μl of Nyosil (Nye Lubricants, Inc., Fairhaven, MA) and centrifuged at 17,000 × g for 1 min at 4°C. The PBS supernatant above the oil layer was removed, and 1 ml of liquid chromatography-mass spectrometry (LCMS)-grade sterile water was used to wash the tube above the oil layer. The oil layer was then removed and the cell pellet resuspended in cold 70% methanol and stored at −80°C. Cervical tissue samples were transported on ice, weighed, homogenized using a disposable pestle, and then suspended in 70% methanol and stored at −80°C.

Measurement of raltegravir plasma, PBMC, and tissue concentrations.

As previously described (14), plasma samples (10 μl) were combined with C-13 internal standard and ammonium acetate and extracted with methanol. Once all protein had precipitated, water was added (to decrease the percentage of organic being injected onto the separation system) and the samples were centrifuged. A volume of 20 μl was injected and separated using reverse-phase chromatography with a gradient from 75:25 (vol/vol) 1% formic acid in H₂O–1% formic acid in acetonitrile (ACN) to 10:90 (vol/vol) 1% formic acid in H₂O–1%formic acid in ACN. The analyte is detected using positive-mode electrospray ionization liquid chromatography-tandem mass spectrometry (ESI LC-MS/MS). Using this assay, raltegravir can be quantitated over a range from 1 ng/ml (lower limit of quantitation [LLOQ]) to 7500 ng/ml (upper limit of quantitation [ULOQ]) with precision and accuracy error less than 15%, except at the LLOQ, at which 20% is acceptable.

Intracellular PBMC and tissue methanol extracts were also determined by LC-MS/MS, as described previously (14). A sample volume of 200 μl was combined with C-13 internal standard and ammonium acetate, and the raltegravir was detected by LC-MS/MS. Using this assay, raltegravir can be quantitated over a range from 0.0025 ng/ml (LLOQ) to 10 ng/ml (ULOQ) with interassay and intra-assay precision and accuracy error less than 15%, except at the LLOQ, at which, according to FDA guidelines, 20% is acceptable. Both the plasma and intracellular assays have been validated as consistent with FDA guidelines.

Conversion of units of measurement between different sample types.

To compare concentrations between the different sample types, we converted all intracellular and tissue concentrations to ng/ml. Per Simiele et al., we assumed that one cell had a volume of 282.9 fl (15), which gives a conversion of 1 fmol/million cells equal to 1.6 ng/ml. We assumed that 1 g of tissue was equivalent to 0.96 ml (16), which gives the conversion of 1 fmol/mg equal to 0.46 ng/ml.

Statistical analysis.

The following parameters were used in our analyses: trough concentrations immediately prior to each observed dose of medication on study visit days (i.e., time zero [C_{0 h}] and 12 h [C_{12 h}]), mean concentration of all values for a given visit day (C_all), and area under the curve of drug concentrations (AUC_0–12). Pharmacokinetic analysis was performed utilizing Phoenix WinNonlin (Certara, St. Louis, MO). AUC_0–12 was calculated with the linear/log trapezoidal method. Other pharmacokinetic parameters (C_{0 h} and C_{12 h}) were determined with Phoenix WinNonlin. Menstrual phase was determined by progesterone level (luteal phase when progesterone > 5 ng/ml and proliferative when progesterone < 2 ng/ml) when available, or by day since LMP (luteal phase when >15 days since LMP and proliferative phase when <15 days since LMP) when progesterone levels were not available or were indeterminate.

Linear mixed-effect models were used to test for differences between C_{0 h} and C_{12 h} for raltegravir. For these models, raltegravir concentrations were log transformed to stabilize variance. Linear mixed effects account for intrasubject correlation due to repeated measures (each subject had up to three separate visits, with 3 to 5 measures at each visit). An intracluster correlation coefficient was calculated to measure the proportion of variation in raltegravir concentrations due to differences between people versus between visits for the same person.

We estimated correlations between plasma, PBMC, and cervical concentrations (all measures were log transformed) using a linear mixed-effect modeling approach (17) that accounts for repeated measures. Finally, linear mixed models were used to evaluate the effect of menstrual cycle on log-transformed C_{12 h}. Effects of phase (luteal versus proliferative) and progesterone levels on raltegravir concentrations were each tested. Differences in AUC_0–12 (log transformed) between luteal and proliferative phase were also tested with mixed models.

RESULTS

Participant characteristics.

Ten women were enrolled, 9 of whom completed all three visits and 1 of whom completed two visits. Samples from two visits for one participant were unable to be analyzed due to mislabeling. The age (mean ± standard deviation) of participants was 30 ± 8 years. Two participants reported one missed dose of medication several days prior to their study visits, while one participant reported two separate missed doses, one of which was the night prior to her first study visit. One participant developed a severe headache and stopped study participation after the day 14 visit. Four women reported mild side effects of the study medication at one or more study visits: 2 reported nausea, 1 reported vomiting, 1 developed leg aches and knee pain, 1 reported congestion, and 2 reported dry mouth. Five participants reported vaginal blood spotting after at least one study visit, which resolved with no additional treatment in all cases. Although reporting of normal menstrual cycles was an inclusion criterion, one participant had a cycle longer than 35 days during our study.

Plasma, PBMC, and tissue pharmacokinetics of raltegravir.

The trough plasma concentrations of raltegravir 12 h after the directly observed dose (C_{12 h}) in the clinical research center were above the IC₉₅ of 33 nM (14.6 ng/ml) in 96% of measurements. Only 67% of PBMC trough values were above the IC₉₅ of 14.6 ng/ml, but cervical tissue trough concentrations were above this value 89% of the time. In addition to minimum or trough values, overall mean concentrations have been associated with virologic failure in raltegravir-containing regimens (18). Among our participants, plasma C_alls were higher than PBMC or cervical tissue concentrations (Table 1).

TABLE 1.

Summary of raltegravir pharmacokinetic values at steady state in each of three sample types: plasma, PBMC, and cervical tissue

Parameter	Value for:
	Plasma	PBMC	Cervix
AUC0–12, ng · h/ml
Mean + SD	5,089 ± 4,029	3,050 ± 2,908	1,803 ± 1788a
Median (interquartile range [IQR])	3,696 (1,931, 6,744)	1,491 (976, 4,593)	1,158 (813, 2,155)a
Call, ng/ml
Mean ± SD	431 ± 341	263 ± 248	149 ± 148
Median (IQR)	286 (186, 548)	127 (85, 390)	106 (69, 183)
Values below IC95	3/135 (2%)	13/135 (10%)	5/81 (6%)
C0 h, ng/ml
Mean ± SD	446 ± 509	258 ± 327	144 ± 145
Median (IQR)	301 (170, 547)	143 (99, 253)	99 (44, 194)
Values below IC95	1/27 (4%)	1/27 (4%)	1/27 (4%)
C12 h, ng/ml
Mean ± SD, ng/ml	160 ± 394	75 ± 134	117 ± 229
Median (IQR)	47 (28, 105)	23 (10, 45)	56 (21, 90)
Values below IC95	1/27 (4%)	9/27 (33%)	3/27 (11%)
P value for comparison of C0 h and C12 hb	<0.0001	<0.0001	0.0248
Intracluster correlation coefficientc	0.080	0.235	0.345

^aAUC calculated using only 0-, 6-, and 12-h values.

^bFor linear mixed model comparing 0- and 12-h values.

^cProportion of variation in concentrations due to variation between individuals.

At presentation for a clinic visit, mean concentrations of raltegravir drawn immediately before the directly observed dose (C_{0 h}) were significantly higher than the trough concentrations 12 h after directly observed therapy (C_{12 h}) (Table 1). In medication diaries, all participants reported taking the medication 12 h before presentation, but in some participants variation in C_{0 h} levels from visit to visit was significantly greater than the 95% confidence interval (95% CI) of assay variation (Fig. 1), suggesting that adherence to dosing schedule was not optimal. The intracluster correlation coefficients (ICC) suggest that variation between participants contributed very little to overall variation in raltegravir concentrations: the ICC for plasma was 0.080, suggesting that 92% of variation in plasma raltegravir concentrations is due to variation within individuals (Table 1).

FIG 1.

Median raltegravir concentrations (minimum to maximum) at each time point in each sample type on day 8, day 15, and day 22 of medication administration.

Cervical biopsy sample size varied considerably, ranging from 6 to 129 mg. Using mixed modeling, the effect of biopsy sample size on raltegravir concentrations was estimated, and we found that for each milligram increase in tissue weight, the concentration of raltegravir decreased 2% (95% CI, 1 to 3%; P = 0.0005).

Relationship between plasma concentration and PBMC or tissue concentration.

Plasma concentrations of raltegravir were highly correlated with those in PBMC (r = 0.82) and cervical tissue (r = 0.81). The correlation between PBMC and cervical tissue concentrations was slightly weaker (r = 0.69). The median ratio of PBMC concentration to plasma concentration was 0.49 (range, 0.07 to 9.53), while the median ratio of cervical concentrations to plasma was 0.56 (range, 0.02 to 5.12).

Variation in plasma, PBMC, and tissue concentrations across the menstrual cycle.

The effect of the menstrual cycle on trough raltegravir concentrations in plasma, PBMC, and the cervix did not reveal consistent patterns (Fig. 2). Linear mixed models revealed no statistically significant association between phase of menstrual cycle or progesterone levels and raltegravir concentrations in either plasma, PBMC, or the cervix (data not shown).

FIG 2.

Trough concentrations of raltegravir 12 h after a directly observed dose at visits in the proliferative and luteal phases for the nine participants with more than one visit analyzed. Median value, 95% confidence interval.

DISCUSSION

Our results show that cervical tissue and PBMC concentrations of raltegravir were lower than plasma concentrations and more likely to fall below the IC₉₅. Both tissue and PBMC concentrations of raltegravir correlate with plasma and show little sign of accumulation, making plasma measurements of raltegravir a reasonable proxy for intracellular concentrations. The lack of raltegravir accumulation in the cervical tissue was unexpected, as prior work had shown significantly higher levels of raltegravir in the cervicovaginal fluid than in plasma (8).

There is significant variation among antiretrovirals in the ratio of genital tract concentrations to blood plasma. A Microbicide Trials Network study (MTN-001) compared the pharmacokinetics of steady-state tenofovir administered either orally or as an intravaginal gel and found that the extracellular cervicovaginal fluid concentrations of tenofovir were significantly higher than plasma concentrations with oral dosing. However, vaginal tissue concentrations of tenofovir diphosphate were lower than those in PBMC with oral administration (19). A separate study of a single oral dose of tenofovir reported higher vaginal tissue concentrations of tenofovir diphosphate than found in PBMC (20). Zidovudine, lamivudine, and emtricitabine have also been found to have higher concentrations in cervicovaginal fluid than in blood plasma, while lopinavir, atazanavir, and efavirenz all have lower cervicovaginal fluid concentrations than blood plasma concentrations (21). In this study, we measured raltegravir concentrations within the cervical tissue and in PBMC, as these seem to be most relevant to preventing HIV acquisition via the genital tract of women.

Plasma concentrations of raltegravir have significant inter- and intrasubject variability across studies (8, 18, 22, 23). In initial dose-finding studies, there was no dose response seen in effectiveness against HIV, nor was there any pharmacodynamic relationship between minimum concentrations, AUC, or maximum concentrations and the medication's effectiveness in reducing plasma viral concentrations (7). More recent work has shown that both trough plasma concentrations and the mean of all measured drug concentrations are related to effective suppression of HIV replication (18), but the relationship is not as direct or linear as it is for other medications. One possible explanation for this could be that raltegravir is concentrated within cells, so lower extracellular (plasma) concentrations are not correlated with high intracellular concentrations. However, in our and other studies, raltegravir concentrations in PBMCs are strongly correlated with plasma concentrations and do not show any accumulation of the drug inside the cells that could explain the lack of correlation between plasma concentrations and efficacy (22, 24, 25). It is likely that 100% of measured drug concentrations do not need to be above the IC₉₅ to achieve PrEP efficacy. In a substudy of participants in the iPrex study of daily tenofovir and emtricitabine for PrEP, only 20 to 40% of time points for nonseroconverters had tenofovir diphosphate concentrations above the calculated 90% effective concentration (EC₉₀) of 16 fmol/10⁶ cells (26), compared to 0 to 20% of those who became infected (3). The IC₉₅ for raltegravir was calculated using in vitro infectivity assays, and it is unclear how well this correlates to protective efficacy in vivo.

Polymorphisms in the multidrug resistance gene (MDR1) that encodes the P glycoprotein transporter can influence intracellular concentrations of raltegravir and other agents and therefore potentially antiretroviral (ARV) efficacy (27). P glycoprotein concentrations increase with the luteal phase of the menstrual cycle (12); thus, we hypothesized that P glycoprotein may efflux more raltegravir across the cell membrane in the luteal phase (a time of higher progesterone) rather than proliferative phase of the menstrual cycle (a time of lower progesterone) and decrease intracellular raltegravir concentrations. We did not find an association with menstrual phase, but given the small sample size, a small effect cannot be ruled out.

Cervical biopsy sample size was associated with tissue raltegravir concentrations. As raltegravir concentrations in tissue are initially reported as fmol/mg, it is not surprising that tissue weight and concentrations are related. However, an increase in biopsy specimen size is usually related to increased depth of excision and more stromal tissue in a biopsy sample (as opposed to more epithelial surface area). The decrease in raltegravir concentrations with increasing biopsy sample size could suggest that raltegravir is not evenly distributed throughout the tissue and has lower concentrations in the stroma or that extraction methods work less well on thicker stromal tissue than the surface epithelium.

Our study was small but collected several samples from participants, giving a large number of individual measurements of raltegravir concentrations. In a pharmacokinetic study, patient adherence is a challenge. The high trough C_{0 h}s may suggest poor compliance with dosing at a regular time, despite reporting of uniform times in their medication record; however, these may also represent the known high levels of variation in raltegravir concentrations (8, 18, 22, 23). We only collected three cervical biopsy samples at each visit; thus, the values for cervical tissue AUC are based on a 3-point curve, as opposed to the 5-point curve for plasma and PBMC, making comparison difficult.

Our results suggest that genital tissue concentrations of raltegravir are not significantly higher than plasma concentrations, in contrast to concentrations in genital secretions. Additionally, tissue concentrations and intracellular concentrations in white blood cells are more likely to fall below the IC₉₅ than plasma concentrations. Given that twice-daily dosing is needed to maintain genital tissue and PBMC concentrations, these evaluations of raltegravir, in its current formulation, suggest that the pharmacokinetic characteristics may not be suitable for an agent for PrEP.