Population Pharmacokinetic Approach to the Use of Low Dose Cyclosporine in Patients with Connective Tissue Diseases

Abstract

This study describes the population pharmacokinetics and dose personalization of cyclosporine in 36 patients with connective tissue diseases. A one-compartment open model with absorption was adopted as a pharmacokinetic model, and a nonlinear mixed effects model was used to analyze the population pharmacokinetic models. In the final model, age (AGE) and total body weight (TBW) were influential covariates on clearance (CL/F), which was expressed as CL/F (L/h)=17.8×(AGE/60)^−0.269×(TBW/46.9)^0.408, in addition to the volume of distribution (Vd/F), (L)=98.0 and absorption rate constant (K_a) (h⁻¹)=0.67 (fixed). The results of the present study provide novel insights into factors involved in determining the most suitable dose and dosing strategy for individual patients with connective tissue disease.

Connective tissue diseases are a group of heterogeneous disorders that are characterized by immunological reactions against self-antigens, autoantibody production, and the deposition of immune complexes onto sensitive tissues.¹⁾ These proteins are injured by inflammation associated with connective tissue diseases and may affect the skin, joints, bones, heart, blood vessels, lungs, eyes, and ears. Connective tissue diseases that directly affect the skin include systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), scleroderma, Sjogren’s syndrome, mixed connective tissue disease, systemic sclerosis, dermatomyositis, and polymyositis. The calcineurin inhibitor cyclosporine was previously shown to be crucial for maintaining immunosuppression in solid-organ transplantation.²⁾ In contrast, cyclosporine, as an immunosuppressant, has been used in the treatment of connective tissue diseases for more than 20 years.³⁾ Its efficacy has been demonstrated in many clinical trials.^4–7) In a clinical setting, the typical practice of cyclosporine dosage adjustments is frequently used to determine its concentration and then adjust the dosage by a trial-and-error approach. However, controversy still surrounds the dosage regimen, therapeutic concentration range, and method of therapeutic drug monitoring of cyclosporine in the treatment of collagen diseases^8–10); the pharmacokinetics of cyclosporine show large variations in blood levels following its oral or intravenous administration, not only interindividually, but also intraindividually,¹¹⁾ especially after oral dosing.¹²⁾ Recent studies reported low dose cyclosporine therapy, which was adjusted to gain of blood trough concentration approximate 80–150 ng/mL, was an effective and less toxic cyclosporine therapy for the regulation of connective tissue diseases with RA,¹³⁾ SLE,^14–16) and lupus nephritis.^17–19) Population pharmacokinetic methods, with the ability to incorporate covariates from sparse drug concentration data, offer an opportunity to generate a dosing regimen. Although a dosing method that takes into account the various factors that influence the pharmacokinetics of cyclosporine is needed, pharmacokinetic studies on cyclosporine in collagen disease patients are limited, as are population pharmacokinetic studies.

The pharmacokinetics of cyclosporine have not yet been estimated in collagen disease patients using the population pharmacokinetic approach. Difficulties have been associated with predicting a priori the optimal dosage regimen of cyclosporine for an individual patient with a collagen disease due to the large variability in its pharmacokinetics. Therefore, the pharmacokinetic behavior of this drug needs to be determined not only in individual collagen disease patients, but also in the collagen disease patient population. To clarify the observed variability in cyclosporine with collagen diseases, we performed a population pharmacokinetic analysis of low dose cyclosporine using a nonlinear mixed effects model approach.

PATIENTS AND METHODS

Patients and Data Sources

Routine clinical data, including 89 blood cyclosporine concentrations at a steady state, were collected from 36 patients (10 males and 26 females). A summary of the data for these patients is presented in Table 1. All patients received cyclosporine soft capsules (Mylan Inc., Tokyo, Japan) for the treatment of connective tissue diseases between April 2009 and March 2014 at Sasebo Chuo Hospital, Nagasaki, Japan. According to physicians’ decisions, the initial dosage of cyclosporine was a low dose orally administered at 2–3 mg/kg/d or 50 mg b.i.d. The daily dose was taken in two divided doses at approximately 12-h intervals. Patients receiving dialysis therapy and those with cancer were excluded. All blood samples were collected in tubes containing ethylene diamine tetraacetic acid as an anti-coagulant. These samples were stored at −20°C for later analysis. Whole blood cyclosporine levels were measured using a chemiluminescent immunoassay (CLIA) on an ARCHITECT analyzer (Abbott Japan Co., Ltd., Tokyo, Japan) using the ARCHITECT cyclosporine reagent kit. The lower limit of detection of this method was 25 ng/mL.

Table 1. Demographic and Clinical Data for the Study Population of Patients Receiving Cyclosporine

Number of patients (Male/Female)	10/26
Number of observations (Male/Female)	26/63
Age (year)	60 (15–95) a)
Total body weight (kg)	46.9 (34.0–84.4) a)
Diagnosis
Systemic lupus erythematosus	13
Rheumatoid arthritis	7
Adult on-set Still’s disease	5
Scleroderma	4
Dermatomyositis	2
Nephrotic syndrome	2
Vasculitis syndrome	2
Palmoplantar pustulosis	1
AST (IU/L)	22 (10–170) a)
ALT (IU/L)	20 (6–282) a)
Serum creatinine (mg/dL)	0.69 (0.32–4.42) a)
CCR (mL/min)	71.0 (20.3–165.5) a)
Cyclosporine (mg/kg/d)	2.2 (1.1–4.1) a)
Prednisolone (mg/kg/d)	0.6 (0–1.2) a)

a) Values are the median (range). AST, aspartate aminotransferase; ALT, alanine aminotransferase; CCR, creatinine clearance.

Population Pharmacokinetics of Cyclosporine

A population pharmacokinetic analysis was performed using the nonlinear mixed effect modeling software NONMEM^® version 7.3.0 (ICON Development Solutions, Ellicott City, MD, U.S.A.) with the first-order conditional estimation method. We used the one-compartment model following first-order absorption and elimination as an analysis model (ADVAN2, TRANS2). The typical value of the absorption rate constant (K_a) had to be assumed. Since the data set was predominantly constituted by the elimination phase, K_a was fixed at 0.67 h⁻¹, which has been reported previously for Japanese patients receiving cyclosporine capsules.²⁰⁾ Fundamental pharmacokinetic structural parameters were oral clearance (CL/F) and the volume of distribution (Vd/F).

Interindividual variability in pharmacokinetic parameters (CL/F and Vd/F) was modeled with log-normal distribution in the following equation.   

(1)

where P_i is the pharmacokinetic parameter for i-th individual, P_pop is the population mean value of the parameters, and η_i is a normally distributed random variable with mean zero and variance ω². The relationship between interindividual variability coefficients in CL/F and Vd/F in the OMEGA BLOCK option, ω_CL/F-Vd/F, was calculated as a negative value and the covariance step was unstable. Therefore, the OMEGA DIAGONAL option in the statistical model was used for this analysis.

The intraindividual residual variability was also modeled with exponential errors in the following equation.   

(2)

where C_ij is the j-th measured concentration in the i-th subject, Cpred_ij is the corresponding predicted concentration. ε_ij is the residual variability term, representing independent identically distributed normal statistical errors with mean zero and variance σ² for concentration.

Covariate Model

The covariates considered in CL/F were creatinine clearance (CCR) as an index of renal function; alanine amino transferase (ALT) as an index of hepatic function; the daily dose prednisolone (PSL) as an index of the influence of cytochrome P450 (CYP) 3A on cyclosporine metabolism; and gender (GEN), total body weight (TBW), and age (AGE) as indices of demographic factors. CCR was predicted using the Cockcroft–Gault equation.²¹⁾ The covariates considered in Vd/F were TBW and AGE as indices of demographic factors.

The covariates were modeled with power form and centered around the patient’s median expressed in the following equation.   

(3)

where P_pop is the population mean value of the parameters, θ_p is the typical pharmacokinetic parameter with the median covariate value, and θcov is the covariate scale factor. the corresponding medians of the covariate evaluation are shown in Table 1.

Model Evaluation and Validation

The parameter sensitivity of the final model was evaluated and validated via goodness of fit plots, the bootstrap method, and a visual predictive check (VPC). The goodness of fit of the final model was evaluated by checking the plots of observed, population predicted, and individually predicted concentrations and conditional weighted residuals.

The bootstrap resampling method was used to evaluate the stability of the final model. The final model was fit repeatedly to 200 additional bootstrap datasets. The entire procedure was performed in an automated fashion using DOS batch files (The Wings program for NONMEM ver.735),²²⁾ Microsoft Excel^®/Visual Basic for Application, in conjunction with NONMEM^®. The mean, standard error (S.E.), and 95% confidence interval obtained for bootstrap datasets were compared with final model estimates from the original dataset.

To perform a visual predictive check (VPC), 200 datasets were simulated using the parameter estimated by the final model with the SIMULATION option of NONMEM^®. VPC was performed by comparing the observed concentrations with 95% predicted intervals (PIs) simulated from the final population pharmacokinetic parameters. Exposure to cyclosporine increased in a dose-proportional manner, were showing an almost linear relationship over the entire dose range evaluated.²³⁾ Therefore, simulated cyclosporine concentrations were normalized to a dose of 50 mg. In addition, the median and 95% PIs were superimposed and compared with observations normalized to data of the 50-mg dose.

Simulation

A pharmacokinetic simulation was performed to support the cyclosporine initial dose strategy by using 1000 datasets with the SIMULATION option of NONMEM^® based on the final model in patients with connective tissue diseases. The probability (%) that the trough concentration of cyclosporine was maintained at 80–150 ng/mL^13–19) in a steady state was calculated as the ratio of the number of simulated patients to the total number of patients.

Statistical Analysis

To test the significance of various factors that influenced the pharmacokinetic parameters, the value of the objective function (OFV) determined in the NONMEM^® fitting routine was used. Covariates were searched by forward selection and confirmed in the final model by backward elimination. The difference in OFV (ΔOFV) obtained by comparing each model was asymptotically distributed according to the chi-squared test with the degree of freedom being equal to the difference in the number of parameters between the two models. The significance levels for the forward selection and backward elimination were set to p<0.05 (ΔOFV: 3.84) and p<0.01 (ΔOFV: 6.63), respectively. Individual estimates of the pharmacokinetic parameters for cyclosporine were generated by Bayesian feedback using the post hoc option in the NONMEM^® program after the population analysis.

Ethics

This study was performed in conformity with the Helsinki Declaration after approval by the Ethical Review Board of University of Toyama (approval number: clinical 26–33) and Sasebo Chuo Hospital (approval number: 2014-16). Patient privacy was fully protected and personal information was handled such that patients could not be identified.

RESULTS

Population Pharmacokinetics of Cyclosporine

Figure 1 shows whole blood cyclosporine concentrations after a steady state was used for the population pharmacokinetic analysis. A preliminary pharmacokinetic analysis was conducted to estimate the parameters of the basic model without any covariates. When the interindividual variations were included in the volume of distribution in model 0 and 1 in Table 2, OFV was significantly reduced. The basic models and parameter estimates were CL/F (L/h)=18.9, Vd/F (L)=98.6, and K_a (h⁻¹)=0.67 (fixed). The coefficients of variation (CV) of interindividual variabilities in CL/F and Vd/F were 16.7% and 19.0%, respectively, and CV of intraindividual residual variability was 15.4% (model 1 in Table 2). Each candidate covariate was added, in turn, to the base model and changes in the OFV were noted. AGE and TBW were incorporated into the clearance modeling in the forward selection step, significantly improving the estimates of cyclosporine clearance, but not those of ALT, GEN, PSL, and CCR, which was excluded in the forward selection step. TBW and AGE were not selected for Vd/F as significant covariates. Thus, we chose model 10 in Table 2 as the final model candidate. Backward elimination confirmed this analysis. The final models and parameter estimates were CL/F (L/h)=17.8×(AGE/60)^−0.269×(TBW/46.9)^0.408, Vd/F (L)=98.0, and K_a (h⁻¹)=0.67 (fixed). CV of interindividual variabilities in CL/F and Vd/F were 11.9% and 15.7%, respectively, and CV of intraindividual residual variability was 14.5%. Therefore, when comparing the basic and final models, CV of interindividual variabilities in CL/F and Vd/F were decreased by 4.8% and 3.3%, respectively, and CV of intraindividual residual variability was decreased by 0.9%.

Fig. 1. Concentration of Cyclosporine after Its Oral Administration to Patients with Connective Tissue Diseases (36 Patients, 89 Points)

Table 2. Summary of Covariate Model Building of Cyclosporine

	Model no.	Model	OFV	ΔOFV
	0	CL/F (L/h)=θ1, Vd/F (L)=θ2, Ka (h−1)=0.67 (fixed) without ω2 of Vd/F	711.50
Basic model	1	CL/F (L/h)=θ1, Vd/F (L)=θ2, Ka (h−1)=0.67 FIXED	705.66
Forward addition	2	Model 1+ALT on CL/F	705.63	0.03
	3	Model 1+GEN on CL/F	704.27	1.39
	4	Model 1+PSL on CL/F	703.94	1.72
	5	Model 1+CCR on CL/F	696.00	9.66*
	6	Model 1+TBW on CL/F	691.44	14.22*
	7	Model 1+AGE on CL/F	687.86	17.80*
	8	Model 6+CCR on CL/F	685.46	5.98*
	9	Model 7+CCR on CL/F	681.07	6.79*
Final model candidate	10	Model 7+TBW on CL/F	673.46	14.40*
	11	Model 10+CCR on CL/F	670.30	3.16
	12	Model 10+TBW on Vd/F	673.33	0.13
	13	Model 10+AGE on Vd/F	672.37	1.09
Backward elimination	14	Model 10–TBW on CL/F	691.44	−17.98**
	15	Model 10–AGE on CL/F	687.86	−14.40**

OFV, the minimum value of the objective function (－2 log likelihood) in each NONMEM run; ΔOFV, change in the value objective function for each equation; CL/F, oral clearance; Vd/F, volume of distribution; K_a, absorption rate constant; ALT, alanine aminotransferase; GEN, gender (0: male, 1: female); PSL, prednisolone; CCR, creatinine clearance; TBW, total body weight; AGE, age. * p<0.05, ** p<0.01.

Model Validation

Goodness of fit plots in the final model are shown in Fig. 2, and the observed concentrations were consistent with the population predicted concentrations and individual predicted concentrations (Figs. 2A, B). The goodness of fit was also evaluated graphically by the good distribution of points on conditional weighed residual versus population predicted concentrations and time after administration (Figs. 2C, D). These plots were distributed close to the trend line at y=x and its value ranged from ±3 unit range relative to the reference zero line. The results of parameter estimates of the final model and 99% successful computation from 200 bootstrap resampling are shown in Table 3. The mean value of the final model estimates were very similar to the estimates from bootstrapping, and fell within the 95% CI of these parameter estimates. The model validation using VPC also confirmed an acceptable agreement between the observed data and model-based simulated values. Most of the observed values lay within the 95% PIs at individual time points (Fig. 3).

Fig. 2. Goodness of Fit Plots for the Final Model

Observed concentrations versus population predicted concentrations A), observed concentrations versus individual predicted concentrations B), conditional weighted residuals versus population predicted concentrations C), and conditional weighted residuals versus the time after administration. The solid lines in A) and B) represent y=x. The dotted lines in C) and D) represent y=±3.

Table 3. Comparison of Parameter Estimates for the Final Model with Estimates from 200 Bootstrap Samples

Equation	Final model estimate	S.E.a)	Bootstrap validation (n=198)
			Mean	95% confidence interval
				Lower	Upper
CL/F (L/h)=θ1×(AGE/60)θ2×(TBW/46.9)θ3
θ1	17.80	0.700	17.89	16.80	19.02
θ2	−0.269	0.056	−0.271	−0.378	−0.181
θ3	0.408	0.106	0.404	0.223	0.569
Vd/F (L)=θ4, Ka (h−1)=0.67 FIXED
θ4	98.00	5.880	99.19	89.80	110.0
ω2CL/F	0.014	0.006	0.013	0.003	0.024
ω2Vd/F	0.025	0.020	0.027	0.000 b)	0.064
σ2	0.021	0.006	0.021	0.012	0.031

Successful ratio=99% (198/200). a) S.E. was estimated using bootstrap sample standard deviations. b) A finite value greater than zero. CL/F, oral clearance; Vd/F, volume of distribution; K_a, absorption rate constant; AGE, age; TBW, total body weight.

Fig. 3. Model Validation Using a Visual Predictive Check from 200 Data Sets of a Normalized 50-mg Dose of Cyclosporine (Number of Observations=89)

Plots contain observed concentrations (circle), the median (dotted line), and 95% prediction intervals (solid line) simulated based on the final model.

Simulation

A simulation was performed based on the final model to support the dosing strategy in patients with connective tissue diseases because AGE and TBW were identified as a covariates for CL/F. Cyclosporine concentrations were calculated by dividing the doses simulated for 1000 patients with AGE of 20, 40, 60, and 80, and TBW of 45, 60, and 75 kg, respectively. Simulations were performed for cyclosporine doses ranging from 25 to 150 mg (q12h). The probability (%) that the concentration of cyclosporine was maintained at 80–150 ng/mL^13–19) in a steady state is shown in Table 4.

Table 4. Achievement Rate of the Target Concentration Range 80–150 ng/mL for Various Initial Dosage and Cases of Demographic Profile in the Steady State

AGE	TBW (kg)	Maintenance dose every 12 h
		25 mg	50 mg	75 mg	100 mg	125 mg	150 mg
20	45	0	8.0	34.4	53.4	53.5	45.3
	60	0	2.1	14.5	35.2	49.6	50.7
	75	0	0.5	4.2	19.1	32.3	42.8
40	45	0.9	36.7	61.7	43.8	43.8	16.4
	60	0	16.4	49.0	57.0	47.9	33.8
	75	0	5.2	31.0	48.5	55.6	47.8
60	45	2.5	60.1	55.2	26.8	26.8	5.4
	60	0.3	35.1	59.9	47.1	28.4	16.5
	75	0.2	17.0	53.4	55.9	45.3	32.6
80	45	7.5	69.4	37.5	14.6	14.6	1.6
	60	1.5	53.3	58.4	34.1	18.4	8.1
	75	0.7	30.4	62.2	48.5	31.9	18.5

Each values is target achieving rate (%) from simulated for 1000 patients AGE, age; TBW, total body weight.

DISCUSSION

Interindividual variability in pharmacokinetics and drug responses area therapeutic premise, and the evaluation and management of this variability form the basis for individualized pharmacotherapy. This study was the first to apply a population pharmacokinetic modeling approach to assess the pharmacokinetics of an orally administered low dose of cyclosporine to patients with connective tissue diseases. Xiaoli’s group previously examined the population pharmacokinetics of cyclosporine in patients diagnosed with nephrosis and treated with cyclosporine.²⁴⁾ Nephrotic syndrome, consisting of massive proteinuria, hypoalbuminemia, edema, and hyperlipidemia, is a common complication of glomerular disease in children and adults.²⁵⁾ However, it is a very select group of connective tissue diseases. Furthermore, their data only consisted of trough blood cyclosporine concentrations. We collected data from the beginning of the distribution phase to the end of the elimination phase in the present study, which was designed to evaluate the population pharmacokinetics of an orally administered low dose of cyclosporine to patients with connective tissue diseases.

Large interindividual and intraindividual variabilities were detected in the pharmacokinetics of cyclosporine, and may be associated with various factors such as demographics, disease, physiological changes, and concurrent therapies. Therefore, candidates for covariates in the present study were composed of demographic data, and biochemical blood tests. In contrast, from the viewpoint of clinical versatility, we built a population pharmacokinetic model for cyclosporine without the genetic information of patients. However, previous studies showed that cyclosporine was a substrate for both P-glycoprotein and CYP3A,^26,27) and their well-recognized pharmacokinetic variabilities were explained, at least in part, by marked interindividual heterogeneity in the intestinal and hepatic multidrug resistance −1 gene and cytochrome P450 3A expression. Previous pharmacokinetic studies reported the pharmacokinetic of the interaction between cyclosporine and prednisolone in patients who underwent a kidney transplant.^28,29) The metabolisms of cyclosporine and prednisolone were mutually inhibited, increasing their blood concentrations. However, most patients with connective tissue diseases are concomitantly administered prednisolone, and 35 out of the 36 patients in the present study were administered prednisolone (data not shown). PSL was not selected for CL/F as a significant covariate in the present study. CL/F was influenced by AGE and TBW. A direct comparison of estimated final population pharmacokinetic parameters in this study with previous findings was not possible because an analysis has not yet been conducted for cyclosporine in patients with connective tissue diseases. The results of our analysis indicated that the CL/F of the median values (60-year-old patient with TBW of 46.9 kg) for patients treated with cyclosporine was 17.8 L/h. In a previous study on cyclosporine in patients with nephrotic syndrome,²⁴⁾ the pharmacokinetic parameters applied based on the median values of our data, showed that CL was 22.9 L/h. No significant differences were observed in CL/F between the present study and previous population pharmacokinetic analyses. The population pharmacokinetic parameters in the present study were very simple because the candidates of CL/F was only TBW and AGE as the demographic background, whereas those in the previous study required biochemical blood tests and combination therapies of cyclosporine.

In the validation data, goodness of fit plots of the concentrations predicted by the final model correlated strongly with the observed concentrations, which did not cause a marked deviation in the observed concentrations versus the predicted and individual concentrations. Furthermore, the conditional weighted residuals were nearly uniformly distributed within an acceptable range.^30,31) These results suggested that the optimal model was selected as the final model. The bootstrap method was used to evaluate the accuracy and robustness of the general model. The median values of the bootstrap procedure were similar to the parameter estimates from the original data set and 95% CI overlapped with those of the original data set. These results indicated that the accuracy and robustness of the general model were acceptable. The final population pharmacokinetic model was validated using the VPC method. VPC uses simulated responses from the posterior distribution of the parameters, which includes the uncertainty of the estimate in addition to the point estimates (θ, ω, and σ) to assess the ability of a model to represent the actual data. A one-compartment pharmacokinetic model with one-order absorption was appropriate to describe the concentration–time data of cyclosporine in this study. Cyclosporine concentrations were measured in the present study from blood samples collected more frequently in the elimination phase than in the distribution phase. The final model for cyclosporine provided a good prediction of concentration time profiles, particularly in the elimination phase, with most observed data lying within the 95% PIs closely reflecting data distribution at individual time points. An examination of individual model fits revealed reasonable correspondence between observed, population, and individual model predicted data.

This study has several limitations. First, data was very limited with only 89 samples. However, this was an important group of patients that had not been examined previously. There have been few pharmacokinetic data for cyclosporine in patients with connective tissue diseases. Furthermore, the standard error of ω²Vd/F was large. However, 95% CI from the bootstrap resampling method was used to evaluate the stability of the final model excluding zero. Second, standard treatment of connective tissue diseases go through long period of time administration of combined therapy. In addition, connective tissue diseases exhibit various symptoms of autoimmune diseases, and criteria of its effect are differently determined by type of diseases. Thus, it was difficult to investigate the exposure-efficacy and toxicity relationships of cyclosporine in the patient population. Therefore, in this study, it was adopted blood trough concentration of 80–150 ng/mL, as an effective and less toxic from the previous reports.^13–19) Administration plans for cyclosporine based on the area under the blood concentration–time curve (AUC) for transplantation patients have been extensively investigated and directions for use have been established. In recent years, Nagai et al. reported³²⁾ that the AUC was useful for monitoring the clinical and adverse effects of cyclosporine in dermatomyositis-associated progressive interstitial pneumonia patients. However, evidence for the directions for use of cyclosporine for connective tissue diseases is insufficient. In addition, several blood samples must be collected from patients to estimate the AUC. Although the AUC may be an index of a patient’s exposure to cyclosporine, it is seldom possible to determine this in routine outpatient clinic monitoring. Thus, in the present study, trough concentrations were used to adjust the cyclosporine doses for each individual.

The simulation suggests the achievement rate of the target concentration range of 80–150 ng/mL for various initial dosage and cases of demographic profiles in a steady state. The initial cyclosporine dose may be calculated from probability distribution on the basis of two markers in the final model. Dose adjustments to cyclosporine can then be performed following considerations of the condition of the patient and cyclosporine concentration. Table 4 is a 2-dimensional stochastic illustration, typically in a scale form, that has been designed for graphical estimations of a predetermined function. It enables the support of bedside judgments of cyclosporine initial dose by clinicians.

In conclusion, a population pharmacokinetic model for the administration of a low dose of cyclosporine to Japanese patients with connective tissue diseases was developed and evaluated. AGE and TBW were significant predictors of the CL/F of cyclosporine in these patients. From the viewpoint of pharmacokinetics, medicine, and physiology, the population pharmacokinetic parameters used in the present study were considered to be clinically valid. The results of the present study provide a novel insight into considering the most suitable dose and dosing regimen for individual patients with connective tissue diseases.

Acknowledgment

This work was partially supported by a Grant-in-Aid from the Research Foundation for Pharmaceutical Sciences of Japan.

Conflict of Interest

The authors declare no conflict of interest.

REFERENCES

• 1) Johnson AE, Gordon C, Bacon PA, Hobbs FDR. Undiagnosed systemic lupus erythematosus in the community. Lancet, 347, 367–369 (1996).
• 2) Kahan BD. Cyclosporine. N. Engl. J. Med., 321, 1725–1738 (1989).
• 3) Waldo FB, Kohaut EC. Therapy of focal segmental glomerulosclerosis with cyclosporine A. Pediatr. Nephrol., 1, 180–182 (1987).
• 4) Meyrier A. Treatment of idiopathic nephrotic syndrome with cyclosporine A. J. Nephrol., 10, 14–24 (1997).
• 5) Cattran DC, Appel GB, Hebert LA, Hunsicker LG, Pohl MA, Hoy WE, Maxwell DR, Kunis CL. A randomized trial of cyclosporine in patients with steroid-resistant focal segmental glomerulosclerosis. North America Nephrotic Syndrome Study Group. Kidney Int., 56, 2220–2226 (1999).
• 6) Masuda K, Nakajima A, Urayama A, Nakae K, Kogure M, Inaba G. Double-masked trial of cyclosporin versus colchicine and long-term open study of cyclosporin in Behcet’s disease. Lancet, 333, 1093–1096 (1989).
• 7) Callen JP, Krueger GG, Lebwohl M, McBurney EI, Mease P, Menter A, Paller AS, Pariser DM, Weinblatt M, Zimmerman G, AAD. AAD consensus statement on psoriasis therapies. J. Am. Acad. Dermatol., 49, 897–899 (2003).
• 8) Filler G. How should microemulsified Cyclosporine A (Neoral) therapy in patients with nephrotic syndrome be monitored? Nephrol. Dial. Transplant., 20, 1032–1034 (2005).
• 9) Rinaldi S, Sesto A, Barsotti P, Faraggiana T, Sera F, Rizzoni G. Cyclosporine therapy monitored with abbreviated area under curve in nephrotic syndrome. Pediatr. Nephrol., 20, 25–29 (2005).
• 10) Takeda A, Horike K, Onoda H, Ohtsuka Y, Yoshida A, Uchida K, Morozumi K. Benefits of cyclosporine absorption profiling in nephrotic syndrome: preprandial once-daily administration of cyclosporine microemulsion improves slow absorption and can standardize the absorption profile. Nephrology, 12, 197–204 (2007).
• 11) Ptachcinski RJ, Venkataramanan R, Burckart GJ. Clinical pharmacokinetics of cyclosporin. Clin. Pharmacokinet., 11, 107–132 (1986).
• 12) Lindholm A, Henricsson S, Lind M, Dahlqvist R. Intraindividual variability in the relative systemic availability of cyclosporin after oral dosing. Eur. J. Clin. Pharmacol., 34, 461–464 (1988).
• 13) Tugwell P, Bombardier C, Tugwell P, Gent M, Bennett KJ, Roberts RS, Ludwin D, Bensen WG, Carette S, Chalmers A, Klinkhoff AV, Esdaile JM, Kraag GR. Low-dose cyclosporin versus placebo in patients with rheumatoid arthritis. Lancet, 335, 1051–1055 (1990).
• 14) Tokuda M, Kurata N, Mizoguchi A, Inoh M, Seto K, Kinashi M, Takahara J. Effect of low-dose cyclosporin A on systemic lupus erythematosus disease activity. Arthritis Rheum., 37, 551–558 (1994).
• 15) Ogawa H, Kameda H, Amano K, Takeuchi T. Efficacy and safety of cyclosporine A in patients with refractory systemic lupus erythematosus in a daily clinical practice. Lupus, 19, 162–169 (2010).
• 16) Morton SJ, Powell RJ. An audit of cyclosporin for systemic lupus erythematosus and related overlap syndromes: limitations of its use. Ann. Rheum. Dis., 59, 487–489 (2000).
• 17) Fu LW, Yang LY, Chen WP, Lin CY. Clinical efficacy of cyclosporin a neoral in the treatment of paediatric lupus nephritis with heavy proteinuria. Br. J. Rheumatol., 37, 217–221 (1998).
• 18) Ogawa H, Kameda H, Nagasawa H, Sekiguchi N, Takei H, Tsuzaka K, Amano K, Takeuchi T. Prospective study of low-dose cyclosporine A in patients with refractory lupus nephritis. Mod. Rheumatol., 17, 92–97 (2007).
• 19) Moroni G, Doria A, Mosca M, Alberighi OD, Ferraccioli G, Todesco S, Manno C, Altieri P, Ferrara R, Greco S, Ponticelli C. A randomized pilot trial comparing cyclosporine and azathioprine for maintenance therapy in diffuse lupus nephritis over four years. Clin. J. Am. Soc. Nephrol., 1, 925–932 (2006).
• 20) Yoshida K, Kimura T, Hamada Y, Saito T, Endo T, Baba S, Shimada S. Comparative study of population pharmacokinetics upon switching of cyclosporine formulation from Sandimmune to Neoral in stable renal transplant patients. Transplant. Proc., 33, 3146–3147 (2001).
• 21) Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron, 16, 31–41 (1976).
• 22) Parke J, Holford NH, Charles BG. A procedure for generating bootstrap samples for the validation of nonlinear mixed-effects population models. Comput. Methods Programs Biomed., 59, 19–29 (1999).
• 23) Mueller EA, Kovarik JM, van Bree JB, Tetzloff W, Grevel J, Kutz K. Improved dose linearity of cyclosporine pharmacokinetics from a microemulsion formulation. Pharm. Res., 11, 301–304 (1994).
• 24) Xiaoli D, Qiang F. Population pharmacokinetic study of cyclosporine in patients with nephrotic syndrome. J. Clin. Pharmacol., 49, 782–788 (2009).
• 25) Tune BM, Mendoza SA. Treatment of the idiopathic nephrotic syndrome: regimens and outcomes in children and adults. J. Am. Soc. Nephrol., 8, 824–832 (1997).
• 26) Lown KS, Mayo RR, Leichtman AB, Hsiao HL, Turgeon DK, Schmiedlin-Ren P, Brown MB, Guo W, Rossi SJ, Benet LZ, Watkins PB. Role of intestinal P-glycoprotein (mdr1) in interpatient variation in the oral bioavailability of cyclosporine. Clin. Pharmacol. Ther., 62, 248–260 (1997).
• 27) Saeki T, Ueda K, Tanigawara Y, Hori R, Komano T. Human P-glycoprotein transports cyclosporin A and FK506. J. Biol. Chem., 268, 6077–6080 (1993).
• 28) Langhoff E, Madsen S, Olgaard K, Ladefoged J. Clinical results and cyclosporine effect on prednisolone metabolism of cadaver kidney transplanted patients. Proc. Eur. Dial. Transplant Assoc. Eur. Ren. Assoc., 21, 963–968 (1985).
• 29) Öst L. Impairment of prednisolone metabolism by cyclosporine treatment in renal graft recipients. Transplantation, 44, 533–535 (1987).
• 30) Beal SL, Sheiner LB. NONMEM User’s Guides. University of California, San Francisco (1999).
• 31) Hooker AC, Staatz CE, Karlsson MO. Conditional weighted residuals (CWRES): a model diagnostic for the FOCE method. Pharm. Res., 24, 2187–2197 (2007).
• 32) Nagai K, Takeuchi T, Kotani T, Hata K, Yoshida S, Isoda K, Fujiki Y, Shiba H, Makino S, Hanafusa T. Therapeutic drug monitoring of cyclosporine microemulsion in interstitial pneumonia with dermatomyositis. Mod. Rheumatol., 21, 32–36 (2011).