Home | About | Journals | Submit | Contact Us | Français |

**|**Antimicrob Agents Chemother**|**v.54(1); 2010 January**|**PMC2798545

Formats

Article sections

Authors

Related links

Antimicrob Agents Chemother. 2010 January; 54(1): 207–212.

Published online 2009 October 26. doi: 10.1128/AAC.01027-09

PMCID: PMC2798545

V. Kohl,^{1} C. Müller,^{1} O. A. Cornely,^{2,}^{3,}^{*} K. Abduljalil,^{1,}^{4} U. Fuhr,^{1} J. J. Vehreschild,^{2} C. Scheid,^{2,}^{5} M. Hallek,^{2} and M. J. G. T. Rüping^{2}

Department of Pharmacology, University of Cologne, Cologne, Germany,^{1} Department I of Internal Medicine, University of Cologne, Cologne, Germany,^{2} Clinical Trials Center Cologne, ZKS Köln, BMBF 01KN0706, University of Cologne, Cologne, Germany,^{3} Simcyp Limited, Sheffield, United Kingdom,^{4} Stem Cell Transplantation Program, Department I of Internal Medicine, University of Cologne, Cologne, Germany^{5}

Received 2009 July 22; Revised 2009 September 30; Accepted 2009 October 19.

Copyright © 2010, American Society for Microbiology

This article has been cited by other articles in PMC.

The objectives of the present study were to elucidate the factors influencing the pharmacokinetics of prophylactically administered posaconazole in allogeneic hematopoietic stem cell transplant (SCT) recipients. Between May 2007 and November 2008, clinical data were obtained from all SCT recipients at the University Hospital of Cologne undergoing therapeutic drug monitoring (TDM) of serum prophylactic posaconazole concentrations. The posaconazole concentrations were determined by high-performance liquid chromatography. We developed a population pharmacokinetic model using nonlinear mixed-effect modeling (NONMEM). The list of covariates tested included age; body weight; body height; gender; posaconazole dose; race; coadministration of antineoplastic chemotherapy; day of stem cell transplantation; concomitant ranitidine, pantoprazole, cyclosporine, or tacrolimus administration; coincident fever; diarrhea; and plasma gamma-glutamyltransferase activity. A total of 149 serum posaconazole concentrations from 32 patients were obtained. A one-compartment model with first-order absorption and elimination as the basic structural model appropriately described the data, with the apparent clearance being 75.8 liters/h (95% confidence interval [CI], 65.2 to 86.4 liters/h) and the apparent volume being distribution of 835 liters (95% CI, 559 to 1,111 liters). Among the covariates tested, significant effects were found for age (decrease in the volume of distribution of 123 liters per year of age) and the presence of diarrhea (59% loss of bioavailability). A basis for prediction of the mean posaconazole concentrations in allogeneic SCT recipients with hematological malignancies is provided for a given dose. Corresponding adjustments of the starting dose according to the presence of diarrhea and according to age appear to be justified before TDM results are available.

Posaconazole, a new triazole, offers broad-spectrum antifungal activity against *Candida*, *Aspergillus*, and *Fusarium* species, as well as the zygomycetes. The safety and tolerability profile is favorable (6). Its use for antifungal prophylaxis in patients with hematological malignancies is supported by two large phase III clinical trials (2, 15). In the first trial, patients receiving induction chemotherapy for acute myelogenous leukemia or myelodysplastic syndrome were randomized to receive either oral suspensions of posaconazole 200 mg three times a day (t.i.d.), fluconazole 400 mg once a day (q.d.), or itraconazole oral solution 200 mg twice a day. The incidence rates of proven and probable invasive fungal disease (IFD) as well as the all-cause mortality rates were significantly decreased (2). The second trial assessed prophylactic posaconazole 200 mg t.i.d. administered orally in allogeneic stem cell transplantation (SCT) recipients treated with immunosuppressive drugs for severe graft-versus-host disease (GVHD). The reference prophylaxis in that trial was fluconazole 400 mg q.d. The incidence of IFD was comparable between the two groups (incidence, 9.0% and 5.3%, respectively), but posaconazole was associated with a reduced incidence of invasive aspergillosis (*P* = 0.006) and improved attributable mortality (*P* = 0.046) (15).

On the basis of these results, posaconazole prophylaxis was implemented in the SCT unit of the University of Cologne, one of the largest referral centers in Germany providing hematology and infectious diseases services to a population of approximately 2.5 million and carrying out about 50 allogeneic SCTs per year. Posaconazole prophylaxis is initiated at the beginning of the conditioning regimen before allogeneic SCT and is administered until the discontinuation of immunosuppressive therapy, which usually occurs at day 100 after SCT.

Serum posaconazole concentrations are influenced by several known factors: diarrhea, vomiting, nutritional state, ethnicity, drug-drug interactions, and increased gamma-glutamyltransferase (γGT) concentrations (9, 10). Therapeutic drug monitoring (TDM) of concentrations in serum is regularly carried out in our institution to monitor the impacts of these already known factors and to identify other possible factors affecting the concentrations in serum. In the study described in this report, we estimated the values of the pharmacokinetic parameters for prophylactic posaconazole in allogeneic SCT recipients by population pharmacokinetic methods from clinical TDM data.

Clinical data for prophylactic posaconazole exposure were obtained from all SCT recipients at the University of Cologne Hospital who underwent TDM as a standard of care between May 2007 and November 2008. Demographic data, comedications, concomitant chemotherapies, laboratory parameters, and febrile and diarrheal episodes were obtained from the prospective neutropenic cohort study in Cologne.

Trough serum posaconazole concentrations were measured by high-performance liquid chromatography (HPLC), as described previously (13). The lower limit of quantification was 50 ng/ml. The intraday and interday accuracies were within ±10.9% and ±5.4%, respectively. All concentrations measured from the beginning of treatment were used; having achieved steady state was not a prerequisite for inclusion of the data in the analysis.

The pharmacokinetic properties of posaconazole were analyzed by a population approach by using the nonlinear mixed-effect software NONMEM (version VI; Globomax, LLC, Hanover, MD). All data collected were fitted to a one-compartment model by using the NONMEM subroutine ADVAN2. The first-order conditional estimation method with interactions (FOCE INTER) was used. The corresponding pharmacokinetic parameters, clearance (CL), volume of distribution (*V*), and the absorption rate constant (*k _{a}*), were integrated into the population pharmacokinetic model. The following assumptions were made: (i) posaconazole was administered as an oral suspension. Therefore, the bioavailable fraction of the drug could not be estimated. Accordingly, the model estimates are the values of apparent clearance (CL/

Population modeling was used to predict the posaconazole serum concentrations. The predicted concentrations were represented as population and individual a priori estimates (population predicted values and individual predicted values, respectively).

In the first step of the population pharmacokinetic modeling, a basic one-compartment model with a first-order elimination pathway was found to be appropriate as a structural population model, i.e., without covariates, including an interindividual variability term on CL/*F*.

The individual patient pharmacokinetic parameters of the basic model were estimated by using the following equations:

(1)

(2)

where CL_{j} represents the estimate of CL/*F* in the *j*th individual, as predicted by the regression model; *V _{j}* represents the estimate of

The 95% confidence intervals (CIs) were estimated as follows for all model parameters:

(3)

where θ_{PP} represents the estimate of the population parameter, SE represents its associated standard error, and *Z* shows the interval coefficient for a standard normal distribution (*Z* = 1.96 for a 95% CI). The most appropriate population model had to meet the following evaluation criteria: (i) the minimal objective function value (OFV) provided by NONMEM was used for selection of the most suitable population model and for statistical analysis. This value of the best model should be significantly lower than those of alternative models, on the basis of the likelihood ratio (LR) test. A reduction in OFV of >3.84 was considered statistically significant (*P* < 0.05) for the inclusion of one additional parameter (3, 7). (ii) The observed serum posaconazole concentrations and the respective predicted values were inspected visually and for the most appropriate model should show a more random distribution across the line of unity in comparison to the distributions of alternative models. (iii) The 95% confidence intervals for the parameter estimates should not include zero or unity (whichever one is applicable). (iv) The values should be physiologically plausible.

Potential covariates were tested on the basis of the structural model. The most favorable covariates were selected and incorporated into the model, so that a final model could be constructed.

The list of individual covariates included age (years); body weight (kg); body height (cm); gender; daily dose of posaconazole (mg, absolute values); ethnicity (Caucasian/other); stem cell transplantation (yes/no); coadministration of chemotherapy, ranitidine, pantoprazole, cyclosporine, or tacrolimus (yes/no for each); fever (yes/no); diarrhea (yes/no); and γGT levels (U/liter). These candidate covariates were incorporated into the population model in a stepwise manner to determine if they had a significant impact.

For instance, the coadministration of chemotherapy (Chem) as a covariate was added to screen its potential influence on the apparent total clearance, as follows:

(4)

Another example demonstrates the addition of age (Age) as a covariate to test its potential influence on the apparent volume of distribution:

(5)

To assess the validity of the model, a modified jackknife evaluation of the potential final model was performed (12) by repeating the estimation of population pharmacokinetic parameters with reduced data sets from which three cases were randomly excluded. A total of 100 runs were carried out. Covariates with a significant influence mediated by individual cases only were excluded from the final covariate model.

A total of 149 serum posaconazole concentrations were measured by HPLC during routine therapeutic drug monitoring of 32 allogeneic SCT recipients with hematological malignancies (range of number of samples per patient, 1 to 12; median number, 5). The demographic data and the clinical data for the covariates assessed are shown in Table Table1.1. The overall mean of all posaconazole concentration measurements was 411 μg/liter ± 333 μg/liter (range, 25 to 1,871). The mean maximum concentration of drug in plasma was 654 μg/liter ± 443μg/liter (range, 90 to 1,871). Patients unable to take oral medication, e.g., due to severe oral mucositis or vomiting, were switched to intravenous antifungals. Therefore, data for such patients are not part of this analysis.

Population pharmacokinetic analysis by use of a one-compartment model with first-order absorption and elimination as the basic structural model, including an interindividual variability term on CL/*F*, as shown in equation 6 (see also equation 1), was able to describe the collected data appropriately.

(6)

where η_{CL} represents the random variable with a mean value of zero and variance ω^{2}, which distinguishes the apparent clearance for the *j*th individual from the mean clearance value for the population predicted by the regression model.

Dose changes during therapy enabled the separate estimation of the apparent total clearance and the apparent total volume of distribution. As only trough values were available, a more sophisticated structural model (e.g., a two-compartment model) could not be developed.

When each covariate was tested separately, only body weight showed a significant impact on the apparent clearance of posaconazole. A significant impact on the apparent volume of distribution was observed for age and body height. No chosen covariate had a significant impact on CL/*F* and *V*/*F* simultaneously, as shown in Table Table2.2. These screening results were considered for the construction and the selection of a final population model. The process of modeling activities and the results of covariate screening with the NONMEM program are given in Table Table3.3. The simultaneous incorporation of age (on *V*/*F*) and the presence of diarrhea (on CL/*F* and *V*/*F*) as covariates in the basic model were associated with a significant and pronounced drop in the OFV (the change in the OFV [ΔOFV] = 24.0). Addition of further covariates to that model which had significant effects when they were added as single covariates to the basic model gave equivocal results. While the separate inclusion of ethnicity, comedication (with pantoprazole, ranitidine, or cyclosporine), time of SCT, γGT levels, and fever showed no significant reduction in OFV, the inclusion of comedication with tacrolimus resulted in a significant reduction in OFV (ΔOFV = 9.5); however, the 95% CIs included unity. The simultaneous addition of comedication with pantoprazole and comedication with tacrolimus led to a further slight decrease in the objective function value in the model tested (ΔOFV = 2.4).

These two models with three and four individual covariates, i.e., diarrhea, age, and comedication with tacrolimus (plus comedication with pantoprazole) were examined by use of the modified jackknife evaluation. Comedication with tacrolimus as a covariate lost its significant effect when the data for specific individuals were excluded from analyses, and the 95% CIs for the respective parameters occasionally included unity. The remaining three covariates, i.e., diarrhea, age, and comedication with pantoprazole, showed stable and statistically significant outcomes in the modified jackknife evaluation. However, the simultaneous incorporation of these three cofactors (model 3, as given in Table Table3)3) was not associated with a further decrease in OFV compared to the OFV achieved with the model with only two covariates.

For the reasons described above, a model with the simultaneous inclusion of the covariates diarrhea and age (model 2, as given in Table Table3)3) was selected as the final model for the current population pharmacokinetic analysis. An overview of goodness-of-fit-plots for the basic model and the final covariate model is given in Fig. Fig.11.

Overview of the goodness-of-fit plots and their improvement in the two models constructed. Model 1 (left column) is the basic model. Model 2 (right column) represents the final with age and diarrhea as covariates. The solid diagonals in the upper panels **...**

Parameter estimates and their respective 95% confidence intervals are presented in Table Table4.4. Estimates of the interindividual and the residual variabilities are expressed as coefficients of variation (CV). In the final model, the occurrence of diarrhea was related to a significantly higher apparent clearance as well as to a higher apparent volume of distribution of posaconazole. This is equivalent to the loss in bioavailability of 59% compared to the value obtained in the absence of diarrhea. Increasing age was significantly associated with a reduction in the apparent volume of distribution by 123 liters per year of age. These changes correspond to a respective decrease in the level of exposure as a result of diarrhea and an increase with age for a given dose. The remaining interindividual variability in CL/*F* not explained by the covariates was characterized by a coefficient of variation of 26.9%. The CV for residual variability was 42%.

In this study, the values of the pharmacokinetic parameters for posaconazole administered for antifungal prophylaxis to allogeneic SCT recipients were estimated by population pharmacokinetic methods performed with data obtained from TDM. The population estimate for CL/*F* in individuals with the median age (i.e., 49 years) and without diarrhea was 67 liters/h, and that for *V*/*F* was 2,250 liters in such patients. Among a broad range of covariates tested, significant effects were found for age (123-liter reduction in *V*/*F* for each year in patients) and diarrhea, which apparently caused a 1.7-fold reduction in bioavailability. Thus, this population approach made routine clinical and biochemical data usable for evaluation of the pharmacokinetics of posaconazole in the population studied. The effects of several covariates on the pharmacokinetic parameters identified in this study are supported by similar findings in some classical pharmacokinetic studies. The effect of diarrhea on the pharmacokinetics of posaconazole is not unexpected. This disturbance is likely to cause a shorter residence time of this antifungal in the gastrointestinal tract and is highly prevalent in chemotherapy and SCT recipients (9, 10). In 194 patients with acute myelogenous leukemia or myelodysplastic syndrome with documented diarrhea, the average plasma posaconazole concentration was 37% lower than that in patients without diarrhea (*P* < 0.0001) (9). According to the pharmacokinetic analysis with data for 246 allogeneic SCT recipients with GVHD, the same effect of this disturbance on the average plasma posaconazole concentration was observed, i.e., a 38% reduction in the average plasma posaconazole concentration in patients with diarrhea compared with that in patients without diarrhea (10).

Previous studies with healthy volunteers (11), allogeneic SCT recipients with GVHD (10), and patients with refractory IFD and/or persistent neutropenia (4) have not shown a clinically relevant effect of demographic variables such as gender, ethnicity, body weight, and age on the pharmacokinetics of posaconazole. However, a significant impact of ethnicity was described by a classical pharmacokinetic study with neutropenic patients receiving remission induction chemotherapy for hematologic malignancies. In that study, Caucasian patients had 21% higher average plasma posaconazole concentrations at steady state in comparison to those of black patients and 13% higher concentrations in comparison to those of Asian patients (9).

In the current study, older age was significantly related to a reduction in the apparent volume of distribution. This plausible relationship, which is observed for many drugs (8, 14), was, however, not found in several classical studies either in patients with hematological malignancies (9, 10) or in healthy volunteers (1). Of note, when allogeneic SCT recipients with GVHD were stratified into three groups by age (<18, 18 to 45, and >45 years), 106 patients >45 years of age had an 11% higher average plasma posaconazole concentration than 133 patients aged 18 to 45 years (10).

Comedication with ranitidine as an individual cofactor did not show any significant effect on the pharmacokinetic properties of posaconazole in our study. These results are in accordance with data available in the literature (1).

Increased serum γGT activity is a well-known adverse event during the administration of azoles (15). A significant impact of an increase in γGT levels as an individual cofactor on the pharmacokinetic behavior of posaconazole has been described by a pharmacokinetic analysis with 194 neutropenic patients (9). In that study, patients with γGT levels greater than or equal to two times the upper limit of normal had significantly lower average plasma posaconazole concentrations at steady state than patients with γGT levels less than two times the upper limit of normal (*P* = 0.018). These results were not considered to be clinically significant and did not necessitate posaconazole dosage adjustments (9). Our study did not confirm any significant effect of this cofactor.

It is not surprising that the demonstrable effects of covariates in the various studies are not identical, as there have been considerable differences between the populations studied so far, including the method of evaluation, the sample size, the frequency of the occurrence of individual covariate values, underlying hematological malignancy, etc. Most of the patients in the population studied had a satisfying general condition and had an average age of 49 years. Thus, patients with severe concomitant illnesses, advanced age, or other contraindications to SCT were not included. Furthermore, the patients received standard posaconazole doses.

For these reasons, the findings from the assessment of the effect of the concomitant administration of important drugs in SCT patients remain equivocal. For instance, we observed a plausible and clinically relevant negative impact of comedication with tacrolimus and comedication with cyclosporine as covariates on apparent clearance, if they were added separately to the basic model. However, a stepwise inclusion of the same two covariates into the final model, i.e., with age and diarrhea as cofactors (for example, model 4, as given in Table Table3),3), led to an inverse effect of comedication with tacrolimus as well as comedication with cyclosporine. The significant negative effect of comedication with pantoprazole on the apparent clearance, when it was included separately or in combination with age and diarrhea as cofactors (model 3, as given in Table Table3),3), contradicts the findings obtained in studies with healthy volunteers (1) as well as neutropenic patients undergoing conventional chemotherapy (1). According to a classical pharmacokinetic analysis with patients receiving remission induction chemotherapy for acute myelogenous leukemia or myelodysplastic syndrome, the average plasma posaconazole values were significantly lower (26%, *P* = 0.001) in patients concomitantly taking proton pump inhibitors, an effect that appears to be mediated by pH (13). The administration of posaconazole to 12 healthy volunteers under conditions of increased pH by use of concomitant esomeprazole administration decreased the mean maximal plasma posaconazole concentration and the area under the concentration-time curve by 46% and 32%, respectively (11). To safeguard the effectiveness of antifungal prophylaxis with posaconazole, it would be important to revisit the effects of individual comedications in larger prospective studies.

In summary, the outcomes of the population analysis presented here provide the basis for prediction of the mean concentrations of posaconazole in patients with hematological malignancies undergoing allogeneic SCT for a given dose and, thus, for initial dose selection. While a dose adjustment according to age and the presence of diarrhea appears to be justified in the absence of further information, this needs to be validated by routine therapeutic drug monitoring, as the model explained only a minor fraction of the overall variability in posaconazole exposure.

V.K., K.A., M.H., U.F., and C.M. have no conflicts of interest.

O.A.C. has received research grants from Astellas, Bayer, Basilea, Genzyme, Gilead, Pfizer, Merck, Optimer, Schering-Plough, and Vicuron, has been a consultant to Astellas, Basilea, f2G, Gilead, Pfizer, Merck, Mölnlycke, Nektar, Schering-Plough, and Zeneus, and has served at the speakers’ bureau of Astellas, Gilead, Merck, Pfizer, Schering-Plough, SpePharm, and United Medical. M.J.G.T.R. has served at the speakers’ bureau of Schering-Plough, MSD, and Gilead Sciences. J.J.V. has served at the speakers’ bureau of Schering-Plough, Merck, and Pfizer and received research grants from Astellas, Infectopharm, and Schering-Plough.

^{}Published ahead of print on 26 October 2009.

1. **Aktories, K., U. Förstermann, F. Hofmann, and K. Starke.** 2005. Allgemeine und spezielle Pharmakologie und Toxikologie, vol. 9. Auflage. Urban & Fischer, Munich, Germany.

2. **Cornely, O. A., J. Maertens, D. J. Winston, J. Perfect, A. J. Ullmann, T. J. Walsh, D. Helfgott, J. Holowiecki, D. Stockelberg, Y. T. Goh, M. Petrini, C. Hardalo, R. Suresh, and D. Angulo-Gonzalez.** 2007. Posaconazole vs. fluconazole or itraconazole prophylaxis in patients with neutropenia. N. Engl. J. Med. 356**:**348-359. [PubMed]

3. **Davidian, M., and D. Giltinan.** 1995. Nonlinear models for repeated measurement data. Chapman and Hall, London, United Kingdom.

4. **Ezzet, F., D. Wexler, R. Countney, M. Martinho, A. Cellary, C. Huber, R. Hachem, D. Kontoyiannis, K. Topelt, G. Corcoran, O. Cornely, I. Raad, and A. J. Ullmann.** 2003. Abstr. 43rd Intersci. Conf Antimicrob. Agents Chemother., abstr. A-1565.

5. **Ezzet, F., D. Wexler, R. Courtney, G. Krishna, J. Lim, and M. Laughlin.** 2005. Oral bioavailability of posaconazole in fasted healthy subjects: comparison between three regimens and basis for clinical dosage recommendations. Clin. Pharmacokinet. 44**:**211-220. [PubMed]

6. **Farowski, F., J. J. Vehreschild, and O. A. Cornely.** 2007. Posaconazole: a next-generation triazole antifungal. Future Microbiol. 2**:**231-243. [PubMed]

7. **Hung, T. Y., T. M. Davis, K. F. Ilett, H. Karunajeewa, S. Hewitt, M. B. Denis, C. Lim, and D. Socheat.** 2004. Population pharmacokinetics of piperaquine in adults and children with uncomplicated falciparum or vivax malaria. Br. J. Clin. Pharmacol. 57**:**253-262. [PMC free article] [PubMed]

8. **Klotz, U.** 2009. Pharmacokinetics and drug metabolism in the elderly. Drug Metab. Rev. 41**:**67-76. [PubMed]

9. **Krishna, G., M. AbuTarif, F. Xuan, M. Martinho, D. Angulo, and O. A. Cornely.** 2008. Pharmacokinetics of oral posaconazole in neutropenic patients receiving chemotherapy for acute myelogenous leukemia or myelodysplastic syndrome. Pharmacotherapy 28**:**1223-1232. [PubMed]

10. **Krishna, G., M. Martinho, P. Chandrasekar, A. J. Ullmann, and H. Patino.** 2007. Pharmacokinetics of oral posaconazole in allogeneic hematopoietic stem cell transplant recipients with graft-versus-host disease. Pharmacotherapy 27**:**1627-1636. [PubMed]

11. **Krishna, G., A. Moton, L. Ma, M. M. Medlock, and J. McLeod.** 2009. Pharmacokinetics and absorption of posaconazole oral suspension under various gastric conditions in healthy volunteers. Antimicrob. Agents Chemother. 53**:**958-966. [PMC free article] [PubMed]

12. **Lindbom, L., P. Pihlgren, and E. N. Jonsson.** 2005. PsN-Toolkit—a collection of computer intensive statistical methods for non-linear mixed effect modeling using NONMEM. Comput. Methods Programs Biomed. 79**:**241-257. [PubMed]

13. **Muller, C., M. Arndt, C. Queckenberg, O. A. Cornely, and M. Theisohn.** 2006. HPLC analysis of the antifungal agent posaconazole in patients with haematological diseases. Mycoses 49(Suppl. 1)**:**17-22. [PubMed]

14. **Sansone-Parsons, A., G. Krishna, J. Simon, P. Soni, B. Kantesaria, J. Herron, and R. Stoltz.** 2007. Effects of age, gender, and race/ethnicity on the pharmacokinetics of posaconazole in healthy volunteers. Antimicrob. Agents Chemother. 51**:**495-502. [PMC free article] [PubMed]

15. **Ullmann, A. J., J. H. Lipton, D. H. Vesole, P. Chandrasekar, A. Langston, S. R. Tarantolo, H. Greinix, W. Morais de Azevedo, V. Reddy, N. Boparai, L. Pedicone, H. Patino, and S. Durrant.** 2007. Posaconazole or fluconazole for prophylaxis in severe graft-versus-host disease. N. Engl. J. Med. 356**:**335-347. [PubMed]

Articles from Antimicrobial Agents and Chemotherapy are provided here courtesy of **American Society for Microbiology (ASM)**

PubMed Central Canada is a service of the Canadian Institutes of Health Research (CIHR) working in partnership with the National Research Council's national science library in cooperation with the National Center for Biotechnology Information at the U.S. National Library of Medicine(NCBI/NLM). It includes content provided to the PubMed Central International archive by participating publishers. |