Zidovudine triphosphate and lamivudine triphosphate concentration-response relationships in HIV-infected persons

Objective: To quantitate intracellular concentrations of zidovudine and lamivudine triphosphate and explore relationships with virologic and immunologic responses to antiretroviral therapy.

Cited Here...: Eight antiretroviral-naive, HIV-infected persons with CD4 T cell counts > 100 × 10⁶ cells/l, and HIV RNA in plasma > 5000 copies/ml participating in a prospective, randomized, open-label study of standard dose versus concentration-controlled therapy with zidovudine, lamivudine, and indinavir.

Cited Here...: Peripheral blood mononuclear cells and plasma were collected frequently throughout the study for quantitation of intracellular zidovudine triphosphate and lamivudine triphosphate concentrations, and zidovudine and lamivudine concentrations in plasma. CD4 T cells and HIV RNA in plasma (Roche Amplicor Ultrasensitive Assay) were measured at baseline and every 4 weeks throughout the study. Relationships among intracellular and plasma concentrations, and CD4 T cells and HIV RNA in plasma were investigated with regression analyses.

Cited Here...: Significant relationships were observed between the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate and the baseline level of CD4 cells. Lamivudine triphosphate concentrations were related in a linear manner to the apparent oral clearance of lamivudine from plasma. A direct linear relationship was found between the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate. The percent change in CD4 cells during therapy and the rate of decline in HIV RNA in plasma were related to the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate.

Conclusion: These studies into the intracellular clinical pharmacology of nucleoside reverse transcriptase inhibitors illustrate potential clinical implications as determinants of therapeutic success. Moreover, these findings provide several leads and a strong impetus for future investigations with nucleoside reverse transcriptase inhibitors particularly when given in combination and sequentially.

The active moiety of nucleoside reverse transcriptase inhibitors (NRTI) is not the parent compound, but rather the intracellular triphosphate, which blocks viral DNA synthesis by reverse transcriptase inhibition and chain termination. Formation of the intracellular triphosphate of NRTI has been shown to be dependent upon the activation state of the target cell, and this has lead to the classification of these agents as cell activation-dependent, or -independent ^[1,2]. Cell activation-dependent nucleosides (zidovudine and stavudine) are preferentially phosphorylated in activated cells and yield higher ratios of intracellular triphosphate to endogenous deoxynucleoside triphosphate than in resting cells. Cell activation-independent agents (didanosine, lamivudine, and zalcitabine) are the converse. This classification scheme has been useful clinically: it predicted that a combination of cell activation-dependent and independent agents might have additive-to-synergistic activity, as was shown for zidovudine and lamivudine ^[3]. However, the amount of NRTI triphosphate formed is actually the sum of several processes influencing formation and degradation. Other important characteristics include the ability of NRTI to enter cells, and the affinity for nucleoside kinases and phosphatases ^[4]. The complex nature of NRTI phosphorylation allows the possibility of interplay among these processes between cell activation-dependent and -independent agents. The objective of this study was to quantitate zidovudine triphosphate and lamivudine triphosphate concentrations, and to explore relationships among these concentrations and markers of anti-HIV response in HIV-infected subjects.

Methods

Human subjects and study design

Study participants were enrolled in an ongoing randomized, open-label trial of standard dose versus concentration-controlled therapy with zidovudine, lami vudine, and indinavir. Antiretroviral-naive HIV- infected persons age 18-60 years, with plasma HIV RNA ≥ 5000 copies/ml and CD4 T cell count ≥ 100 × 10⁶ cells/l were eligible for the study. Exclusion criteria included the presence of an active opportunistic infection that would require interruption of antiretroviral therapy, and persons with a known history of non-adherence with medications or scheduled physician and clinic visits. This investigation was approved by the Human Subjects Committee of the University of Minnesota, and was conducted at the Outpatient Clinic of the General Clinical Research Center at the University of Minnesota. Subjects were informed about the study and gave written consent prior to participation.

All participants initially received lamivudine 150 mg twice daily, and indinavir 800 mg every 8 h for the first 2 weeks; zidovudine was initiated at a dose of 100 mg twice daily for the first week and then increased to 200 mg twice daily for the second week to minimize gastrointestinal side effects. At week 2, patients were randomized to either standard therapy consisting of zidovudine 300 mg twice daily, lamivudine 150 mg twice daily, and indinavir 800 mg every 8 h or concentration-controlled therapy. Study participants randomized to concentration-controlled therapy received an individualized regimen developed to maintain targeted antiretroviral drug concentrations in plasma. These target concentrations were a steady-state average plasma concentration (C_ss) of 0.19 mg/l for zidovudine and 0.44 mg/l for lamivudine, and a trough concentration (C_min) of ≥ 0.15 mg/l for indinavir ^[5]. The initial duration of the study was 24 weeks. Participants maintaining a plasma HIV RNA < 200 copies/ml were eligible to continue their assigned therapy for an additional 12 months.

Laboratory evaluations

Peripheral blood mononuclear cells (PBMC) were obtained from 15 ml whole blood collected at baseline prior to the start of therapy. PBMC were obtained in all subjects at week 2, 2 h following the observed simultaneous administration of zidovudine (200 mg) and lamivudine (150 mg); this sample collection approach was repeated at weeks 30 and 56 in eligible participants. PBMC were obtained between 2 and 8 h post doses of zidovudine and lamivudine at weeks 8, 16, and 24 during the initial 24 week study, and then at weeks 28, 36, 44, 52, 72, and 80 in the subjects with plasma HIV RNA < 200 copies/ml that continued the study. The strategy for collection of PBMC samples was chosen based upon pharmacokinetic considerations for an acceptable degree of fluctuation within the sampling window, and a timeframe that was practical for subjects attending an outpatient clinic. Available clinical data indicated the half-life of zidovudine triphosphate was approximately 7 h while that of lamivudine triphosphate was approximately 15 h ^[6,7]. Thus, the expected variability within a 2-8 h window for zidovudine triphosphate was 45%, and for lamivudine triphosphate it was 34%. Plasma samples for measurement of zidovudine and lamivudine concentrations were obtained at each time point that a PBMC sample was obtained. Additionally, plasma samples for zidovudine and lamivudine were obtained over an 8 h period at week 2; this sample collection strategy was repeated at weeks 30 and 56 in eligible participants.

The concentrations of both zidovudine triphosphate and lamivudine triphosphate were quantified in each PBMC sample. Quantification of zidovudine triphosphate in PBMC was performed with a combined cartridge-radioimmunoassay methodology similar to that described by Robbins ^[8]. Intracellular nucleotides from PBMC extracts were separated using solid phase anion exchange (SPE) cartridges. Extracts containing 10 × 10⁶ cells were loaded onto the anion exchange cartridges (Accell Plus QMA, 3cc, Waters Corporation, Milford, Massachusetts, USA). The zidovudine mono- and diphosphates were eluted using 74.5 mM KCl washes. The triphosphate fraction, eluted with 1 M KCl, was collected and subjected to enzymatic dephosphorylation with an acid phosphatase [sweet potato, type XA (Sigma, St. Louis, Missouri, USA)]/acetate buffer. After phosphatase treatment, this fraction containing free zidovudine was applied to a Waters Oasis SPE cartridge and washed with water and 10% methanol to remove any endogenous thymidine. Zidovudine was eluted with methanol, the methanol was evaporated to dryness under nitrogen, and the sample extract was reconstituted in radioimmunoassay (RIA) buffer. A 10% methanol wash was used in our procedure instead of the 5% utilized in the method of Robbins. This was necessary due to the higher affinity of thymidine for the Oasis cartridge. Quantification of zidovudine was accomplished with a modified commercially available RIA (ZDV-Trac; DiaSorin, Stillwater, Minnesota, USA). Zidovudine standard solutions ranged from 100 to 4000 fmol per assay tube (10-400 fmol/1×10⁶ cells added). Quality controls were prepared using PBMC extracts (10 × 10⁶ cells/ml) from normal volunteers. These were spiked with zidovudine triphosphate (Moravek Biochemicals, Inc., Brea, California, USA) at levels of 300, 1000, and 3000 fmol/10 × 10⁶ PBMC and run with patient unknowns. The limit of quantitation for this assay was 10 fmol/1 × 10⁶ PBMC.

Quantification of lamivudine triphosphate in PBMC was performed by separation of lamivudine monophosphate, diphosphate, and triphosphate on anionexchange SPE cartridges, followed by removal of the phosphate group from lamivudine triphosphate with acid phosphatase, and HPLC quantitation. Sample extracts equivalent to 2 × 10⁶ PBMC were loaded onto SPE cartridges (Waters Accell Plus QMA, 1cc). The cartridges were washed with water and 60 mM and 95 mM KCl to remove lamivudine mono- and diphosphate, respectively. The lamivudine triphosphate was eluted with 400 mM KCl, and BHET was added as the internal standard. This fraction was subjected to enzymatic dephosphorylation with acid phosphatase (sweet potato, type XA)/acetate buffer, and following dephosphorylation the solution was loaded onto a Waters Oasis cartridge. Lamivudine was eluted with methanol, the extract dried, and the residue reconstituted in 200 μl water. Fifty μl of this reconstituted solution was quantified on a Hewlett-Packard LC/MS system with UV spectroscopy at 274 nm. This approach is similar to that described by Solas, with the following modification: the concentration of KCl used to remove the lamivudine mono- and diphosphate was reduced; the SPE cartridges were not washed with 20 mM KCl; and acid phosphatase was used for enzymatic dephosphorylation ^[9]. The standard curve for lamivudine ranged from 550 to 55 000 fmol/10⁶ PBMC. Quality control samples, prepared using PBMC extracts from normal volunteers, were spiked with lamivudine triphosphate (Moravek Biochemicals, Inc.) at nominal concentrations of 1500, 3000, and 8000 fmol/10⁶ PBMC, and run with patient unknowns. The limit of quantitation was 550 fmol/1 × 10⁶ PBMC.

Plasma concentrations of zidovudine and lamivudine were quantified by a validated simultaneous HPLC procedure. The lower limit of quantitation for both drugs was 25 ng/ml using a 200 μl sample volume. The between-day coefficients of variation ranged from 1.0% to 3.8% with an overall accuracy of 99.2% for zidovudine, and from 5.5% to 9.4% with an overall accuracy of 93.7% for lamivudine. Apparent oral plasma clearance of zidovudine and lamivudine was determined by fitting a one-compartment model with first order absorption, an absorption lag phase, and first order elimination to the concentration-time data using maximum a posteriori probability-Bayesian estimation (ADAPT II, version 4.0) as described previously ^[5,10]. Zidovudine and lamivudine steady-state concentrations in plasma were then determined with standard pharmacokinetic equations.

A clinical assessment and measurement of hematologic parameters and clinical chemistries were performed with every clinic visit. Urinalysis, cholesterol, and triglycerides were performed every 3-4 months. Adverse reactions were graded and managed using the approach developed by the AIDS Clinical Trials Group ^[11]. CD4 cells and plasma HIV RNA (Roche Amplicor Ultrasensitive Assay) were measured at baseline and every 4 weeks during the study. Relationships among plasma HIV RNA, CD4 cell counts, and pharmacologic characteristics were explored with regression analysis.

Results

Zidovudine triphosphate and lamivudine triphosphate data were obtained in eight HIV-infected persons. The median HIV RNA in plasma at baseline was 24 011 copies/ml (range, 6890-333 531 copies/ml); the median baseline CD4 T cell count was 385 × 10⁶ cells/l (range, 115-634× 10⁶ cells/l). Two participants met the criteria for virologic failure (plasma HIV RNA > 200 copies/ml) at the end of the first 6 months of therapy. The remaining six participants all had plasma HIV RNA < 20 copies/ml after 6 months and continued therapy. Four of the six participants completed the entire 18 month study. One individual completed 7 months of therapy before moving out of the geographic area, and one completed 12 months and then voluntarily ceased study participation. Both had plasma HIV RNA < 20 copies/ml at the time they left the study.

A total of 77 PBMC samples were obtained from these eight subjects, and concentrations of both zidovudine triphosphate and lamivudine triphosphate were quantified. The average time post dose that these samples were collected was 4.6 h. The median zidovudine triphosphate concentration for all 77 samples was 50 fmol/1 × 10⁶ PBMC (range, < 10-177 fmol/1 × 10⁶ PBMC), and the overall median lamivudine triphosphate concentration was 8687 fmol/1 × 10⁶ PBMC (range, < 550-16 577 fmol/1 × 10⁶ PBMC). A strong relationship was present between these 77 intracellular zidovudine triphosphate and lamivudine triphosphate concentrations (Fig. 1). The median number of PBMC samples obtained in each of the eight participants was 10 (range, 5-13). The individual median triphosphate concentrations for each of the eight participants ranged from 13.8 to 96.4 fmol/1 × 10⁶ PBMC for zidovudine triphosphate, and from 2352 to 13 024 fmol/1 × 10⁶ PBMC for lamivudine triphosphate. These individual median triphosphate concentrations were related to baseline CD4 cell counts as shown in Fig. 2; there was no relationship with baseline plasma HIV RNA.

Fig. 1

Fig. 2

Zidovudine and lamivudine median steady-state concentrations in plasma were 0.24 mg/l (range, 0.18-0.32 mg/l) and 0.52 mg/l (range, 0.43-0.69 mg/l), respectively, for the eight subjects. The median apparent clearance from plasma was 1.57 l/h per kg (range, 1.16-2.53 l/h per kg) for zidovudine, and 0.40 l/h per kg (range, 0.30-0.53 l/h per kg) for lamivudine. No relationship was found between the plasma concentrations of zidovudine and intracellular zidovudine triphosphate concentrations. A significant relationship was observed between the weight-adjusted apparent oral clearance of lamivudine in plasma and intracellular concentrations of lamivudine triphosphate (Fig. 3); lamivudine triphosphate concentrations tended to be higher with lower lamivudine trough plasma concentrations, but this relationship was not statistically significant (r² = 0.35;P = 0.1). There was no relationship between the plasma oral clearance values of zidovudine and lamivudine.

Fig. 3

The percent change in absolute CD4 T cell count from baseline to the end of therapy for each participant was strongly related to their median intracellular concentration of zidovudine triphosphate; this relationship was also apparent for lamivudine triphosphate but was marginally significant (Fig. 4). There were no significant relationships between the percent change in CD4 T cells and either the steady-state concentrations in plasma of zidovudine or lamivudine. Similarly, the rate of decline in HIV RNA in plasma from baseline to nadir was not related to the steady-state plasma concentrations of zidovudine or lamivudine. However, the rate of decline in HIV RNA in plasma from baseline to nadir was related significantly to the intracellular concentrations of both zidovudine triphosphate and lamivudine triphosphate (Fig. 5). Stepwise regression analyses that included the concentrations of zidovudine triphosphate, lamivudine triphosphate, and the trough concentration of indinavir as independent variables, and the percent change in CD4 T cells or the rate of decline in HIV RNA in plasma as dependent variables were performed. The final model for the percent change in CD4 T cells included only the intracellular concentrations of zidovudine triphosphate (r² = 0.82;P = 0.0013), while the final model for the rate of decline in HIV RNA in plasma included only the intracellular concentrations of lamivudine triphosphate (r² = 0.75;P = 0.003).

Fig. 4

Fig. 5

Discussion

In this study, a strong relationship was observed between the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate. The intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate were related to the baseline level of CD4 T cells. Lamivudine triphosphate concentrations were related in a direct linear manner to the apparent oral weight-adjusted plasma clearance of lamivudine. Importantly, the percent change in CD4 cells during therapy and the rate of decline in HIV RNA in plasma were related to the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate.

The relationship between the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate is a new finding, and may illustrate an interplay among three processes involved in the intracellular anabolism and catabolism of these NRTI. First, zidovudine has been shown in vitro to reduce the amount of lamivudine triphosphate formed in phytohemagglutinin-stimulated PBMC and U937 cells ^[12,13]. We hypothesize that the probable mechanism of this inhibition is the ability of zidovudine to increase dCTP pools in stimulated and resting cells, which results in feedback inhibition of deoxycytidine kinase activity. Deoxycytidine kinase is the enzyme responsible for the initial phosphorylation of lamivudine to lamivudine monophosphate. Zidovudine has also been shown to decrease dTTP pools in resting cells; an increase in dTTP pools enhances deoxycytidine kinase ^[2,14]. Thus, zidovudine appears to have the ability to exert a rate-limiting step on the formation of lamivudine triphosphate. Second and third, these data may illustrate the contribution of (perhaps competition for) common pathways involved in the intracellular anabolism and catabolism of both of these NRTI. Presumably common pathways include the affinity for nucleoside diphosphate kinase for conversion of the diphosphate to triphosphate ^[15], and the susceptibility of the triphosphate moieties to catabolism by phosphatases.

The biology of higher intracellular zidovudine triphosphate and lamivudine triphosphate concentrations in subjects with lower CD4 cell counts at baseline is not obvious. Increases in nucleoside kinase activity or reduced phosphatase activity are plausible hypotheses. A progressive reduction in 5′-phosphatase activity in patients with advanced HIV disease and lower CD4 cell counts has been described ^[16,17]. Two prior investigations have shown that the intracellular concentrations of zidovudine monophosphate were related in an inverse manner to CD4 cell counts ^[16,18]. In one of these studies, total zidovudine phosphate concentrations were also related in an inverse manner to the CD4 cell count ^[16], while in the other a direct linear relationship was found between zidovudine triphosphate and the CD4 cell count ^[18]. Clearly, additional studies are warranted to understand the reasons for, and any clinical implications of, the correlation between the number of CD4 cells and the formation of intracellular nucleoside anabolites.

The intracellular concentrations of lamivudine triphosphate were related to the weight-adjusted oral clearance of lamivudine from plasma. This relationship indicated that as systemic clearance increased the amount of lamivudine triphosphate formed increased. As has been suggested for a similar relationship between zidovudine monophosphate and zidovudine plasma concentrations, this finding may represent the lymphocyte compartment as a route of systemic clearance for lamivudine ^[18]. No relationship was found between the oral clearance of zidovudine or zidovudine plasma concentrations and zidovudine triphosphate concentrations. This has been reported by others ^[18,19]. The lack of an obvious correlation for zidovudine here, however, does not mean that none exists, particularly in light of published data that has shown patients receiving a higher zidovudine dose to implement a plasma concentration-controlled regimen had higher zidovudine triphosphate concentration compared with those receiving the lower standard dose ^[20]. No relationships were found between plasma clearances of zidovudine and lamivudine. This is as expected given the different elimination pathways of these two drugs: zidovudine is eliminated primarily by hepatic metabolism while lamivudine is excreted primarily as unchanged drug via the kidney ^[21,22].

The results of this study, which demonstrate a relationship between the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate, may contribute to the understanding of two clinical observations. In a study of lamivudine in combination with either zidovudine or stavudine in antiretroviral-naive persons, a greater decrease (P = 0.052) in plasma HIV RNA was found in the recipients of stavudine/lamivudine compared with zidovudine/lamivudine after 20-24 weeks of therapy; there was no difference in virologic response after 44-48 weeks of therapy, however ^[23]. In a comparison of the kinetics of virologic response to zidovudine/lamivudine or stavudine/lamivudine, the maximum reduction in plasma HIV RNA and the reduction sustained over 24 weeks of therapy appeared to be greater with the stavudine/lamivudine combination ^[24,25]. Whereas zidovudine has been shown to reduce the in vitro phosphorylation of lamivudine, stavudine was not shown to have an effect on endogenous dCTP pools in resting cells nor on the amount of lamivudine triphosphate formed ^[2,12]. Thus, a greater virologic response to the combination of stavudine/lamivudine compared with zidovudine/lamivudine may arise early because of differences in the amount of lamivudine triphosphate formed.

A diminished response to subsequent NRTI-containing regimens has been observed in subjects who have had prior NRTI therapy, although this has not been a universal finding ^[25-27]. In one study, the blunted response to combination stavudine/lamivudine therapy could not be explained on the basis of virologic genotypic studies ^[25]. A small companion pharmacologic study suggested that long-term prior therapy with zidovudine might have impaired the phosphorylation of stavudine and lamivudine ^[28]. A molecular basis for this suggestion has not been clearly elucidated, and other investigators have shown that concentrations of stavudine triphosphate were not different in subjects with and without previous zidovudine therapy ^[29]. The present study, which demonstrated a correlation in the ability of subjects to phosphorylate two NRTI (Fig. 1) with different activation state classifications, allows a related, but different hypothesis to explain these findings: those persons that phosphorylate one NRTI poorly may phosphorylate a subsequent NRTI poorly.

Finally, relationships were found between the intracellular concentrations of zidovudine triphosphate and lamivudine triphosphate and the percent change in CD4 cells during therapy and the rate of decline in HIV RNA in plasma. These findings confirm previous work that has shown a greater increase in CD4 cells in subjects who had higher intracellular concentrations of zidovudine triphosphate and extends this observation to lamivudine triphosphate ^[20]. In addition, quantification of the pharmacologically active moieties of zidovudine and lamivudine has demonstrated, for the first time, that the rate of fall in HIV RNA in plasma is dependent upon these intracellular triphosphate concentrations. No relationships were found between plasma zidovudine or lamivudine concentrations and the percent change in CD4 cells or the rate of decline of HIV RNA in plasma. The elusiveness of quantitative relationships among plasma concentrations of NRTI, intracellular triphosphate concentrations, and anti-HIV effect, should not be taken as conclusive evidence that quantitation of NRTI plasma concentrations has no clinical utility. The relationships between the concentrations of the parent NRTI and the formation of the intracellular triphosphate are complex and multifactorial. Thus, efforts to develop and investigate more elaborate models, which incorporate additional data on patient and pharmacologic characteristics, remain warranted. Quantitation of intracellular triphosphate concentrations is technically and analytically challenging, and the ability to extrapolate these techniques to routine patient care is doubtful at this time. Nevertheless, the quantification of intracellular triphosphate concentrations of zidovudine and lamivudine has provided new insights into the clinical pharmacology of these agents, and revealed potential clinical implications as determinants of therapeutic success with antiretroviral therapy. Together these findings provide several leads and a strong impetus for future investigations with NRTI and their phosphorylated anabolites, particularly when given in combination and sequentially.

Acknowledgements

The authors thank K. Henry and T. Schacker for their referral of persons to this study, the staff of the General Clinical Research Center for their patient-care assistance, and the individuals who generously volunteered to participate in this study.