The eGFR parameter model helps predict blood drug concentration of amisulpride in patients with schizophrenia

doi:10.21203/rs.3.rs-4676992/v1

Download PDF

Article

The eGFR parameter model helps predict blood drug concentration of amisulpride in patients with schizophrenia

https://doi.org/10.21203/rs.3.rs-4676992/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Objective

This study aimed to examine inter-patient variability in amisulpride levels, using a multiple regression model with a dataset acquired from Chinese patients with schizophrenia.

Methods

Plasma concentrations of amisulpride (153 data from 128 patients) were analyzed by four-covariate model and eGFR model and were validated using goodness-of-fit and external prediction ability analyses. The amisulpride levels were detected by 2D-LC-UV system.

Results

The eGFR model adequately characterized amisulpride levels with adjusted R² 0.457. The dose and eGFR values significantly affected concentrations, as follows: amisulpride concentration = 838.561–9.169*eGFR + 1.273*dose. To remain below the laboratory alert level, recommended dose should remain below 300 mg/d, 500 mg/d, and 700 mg/d for patients with eGFR of 60 ml/min/1.73 m², 90 ml/min/1.73 m², and 120 ml/min/1.73 m², respectively. Once and twice daily treatment regiments yielded comparable trough concentration values.

Conclusion

The eGFR model is simpler and more accurate in predicting the concentration of AMI, and provides an evaluation method for safe and individualized use of AMI.

Health sciences/Medical research/Drug development

Biological sciences/Drug discovery/Drug safety

Biological sciences/Drug discovery/Pharmaceutics

Biological sciences/Drug discovery/Pharmacology

amisulpride

pharmacotherapy

pharmacokinetics

schizophrenia

estimated glomerular filtration rate

concentration

Amisulpride (AMI) is the second-generation drug for schizophrenia treatment and a potent antagonist of dopamine D₂ and D₃ receptors [1]. AMI is almost entirely eliminated by the kidneys; the 24-hour excretion rate in urine is 90%, and the elimination half-life is between12-20 hours. Because of its marginal metabolic elimination, drug interactions involving drug metabolizing enzymes are relatively rare. High doses of AMI block postsynaptic dopamine receptors, inhibiting dopaminergic hyperactivity; meanwhile, low doses of AMI selectively block presynaptic D₂/D₃-dopamine auto-receptors, thereby enhancing dopaminergic transmission. In clinical practice, the reference dose range is wide, starting with 200–400 mg/d for patients with schizophrenia positive symptoms and ending at 800 mg/d. It is recommended that patients with mainly negative symptoms take low doses within the range of 50-300mg/d [2]. Treatment dose should also fully consider the risk of adverse reactions, such as extrapyramidal symptoms, prolonged corrected QT interval, increased serum prolactin levels, and weight gain [3]. Taking fully consideration of the risk of elevated prolactin levels, The Arbeitsgemeinschaft für Neuropsychopharmakologie und Pharmakopsychiatrie consensus guidelines strongly recommend AMI therapeutic drug monitoring with in therapeutic range of 100–320 ng/mL, while defining the laboratory alert level as 640 ng/mL [4].

AMI plasma levels differ among patients, including many patients with drug levels outside the reference range [5]. In most patients, clinical efficacy corresponds to high AMI levels [6], AMI plasma concentration is closely correlated with drug dose (r² = 0.96, P < 0.0001) [7] and increases along with daily oral dose administered (r = 0.50, p < 0.001) [8]. However, the relationships among age, sex, concomitant medications, body weight, and plasma concentration remain unclear [6–11]. For example, Müller and Bergemann reported higher AMI plasma levels in females than in males; in contrast, Fengli reported no such sex-based differences. Müller suggested that adjusting for sex-based differences in renal clearance rates (10–15% lower in females than in males) may eliminate these differences. Nevertheless, differences in analytical methodologies among studies may yield conflicting results.

Huang et al. reported on the use of a population pharmacokinetics model to quantify any association between age and AMI clearance and to help predict the AMI concentration of schizophrenia in China [12]. AMI clearance is reduced 2.5-3-fold in renal impaired and older adult patients, which is likely related to an age-related reduction in renal function. Renal function should be considered at the time of prescribing, as patients with mild and moderate renal insufficiency present with an AUC (area under curve) of AMI that is two- to ten-times higher, respectively.

This retrospective study aimed to examine the relationship between age, sex, renal clearance, dose, co-medication and AMI concentration using real-world data and focusing on renal function, which is associated with less liver metabolism but mainly renal clearance. Renal function, expressed as estimated glomerular filtration rate (eGFR), was included in a regression model, replacing age, sex, and comedication. This decision was based on the fact that eGFR depends on age and sex, and the effect of weight was attenuated after calculating body surface area. Moreover, eGFR represents the kidneys’ ability to remove a certain amount of a substance per unit of time, capturing renal clearance capacity. This approach helps simplify the model and highlight the pharmacokinetic characteristics of drugs with high renal clearance rates.

This study aimed to develop a model that may help individualize drug therapy for Chinese patients with schizophrenia. A simple and accurate prediction model can support clinical decision making.

This retrospective study was naturalistic from patients hospitalized at Wuxi Mental Health Center from August 2020 to June 2021. The patients were prescribed AMI for treatment of schizophrenic psychosis (ICD 10: F20.0–20.9). Additional inclusion criteria were as follows: age range of 18–80 years; normal or abnormal liver and renal function, comedication and polypharmacy, including antipsychotic drugs and traditional Chinese medicine compound preparations. The exclusion criteria were life-threatening diseases and severe organ dysfunction.

This study did not require written informed consent from patients, but obtained verbal consent, as the data were obtained during routine hospital management and the datasets were anonymized. The study adhered to the principles of the Declaration of Helsinki, and the study design was approved by the ethics committee of Affiliated Mental Health Center of Jiangnan University (approval number: WXMHCIRB2020LLky015).

Data on demographic and clinical characteristics (age, sex, and diagnosis), comedication status, AMI treatment characteristics (daily dose, dose frequency schedule), and laboratory findings (plasma AMI concentration, serum creatinine level, and corrected QT interval) were extracted.

Among some patients with long-term hospitalizations, maintenance doses were used. Repeated measurements after steady-state plasma concentration were excluded. In cases of multiple measurements, only the latest value was included to help reduce bias introduced by repeat assessments. However, after AMI dosage adjustment, the re-steady-state trough concentration was included. A total of 263 concentration values were available, but 108 of them were excluded as duplicate entries; 2 values were excluded as incomplete information. Finally, a total of 153 concentration values were included from 128 patients. The sample size exceeds ten times the variable size.

Blood specimens and analysis method

Blood samples were prepared for drug concentration detection under steady-state conditions, i.e., maintenance dose dose had been unchanged for at least 7 days. Samples were collected at 7 a.m. before AMI administration. The upper plasma samples were aspirated from whole blood samples after centrifugation at 3000 rpm for 5 min. All plasma samples should be analyzed within 24 hours and stored at 4 ℃. In addition, 400 µL of plasma was added into 1000 µL acetonitrile and the mixture was vortexed for 1 min. After centrifuging at a high speed of 20000 rpm for 10 minutes, 1000 µL of supernatant was transferred into a 2 mL sampling bottle, 500 µL of which would be used for analysis. The 500 µL supernatant was analyzed using two-dimensional high-performance liquid chromatography with ultraviolet detection (2D-LC-UV) [13].

The sources of reagents used in 2D-LC-UV analysis include that AMI standard was supplied by the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). Acetonitrile, methanol and isopropanol were provided by Merck (Darmstadt, Germany);Phosphoric acid, perchloric acid, ammonium hydroxide, ammonium acetate, diammonium hydrogen phosphate, and ethylene glycol of analytical grade were purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China).The detected ultraviolet wavelength was at 292 nm.

The analysis method Z score was satisfactory (0.04); the method had been certified by the Laboratory of the Government Chemist (LGC) in the UK.

Statistical analysis

The statistical software was SPSS 26.0. Relationships among independent and dependent variables were examined with linear regression analysis. A VIF of < 5 indicated no multicollinearity in the mode. A Durbin-Watson value of approximately 2 indicated the absence of autocorrelation. External validation of the regression model was performed with paired sample T tests. P values of < 0.05 were indicative of statistical significance.

Data were obtained from schizophrenic patients with a mean age of 48.97 years (SD = 13.08) (Table 1). The age values were normally distributed (SK = -0.154). Dose values were normally distributed and were in the range from 100 mg to 1200 mg; the corresponding range of concentration values was wide and had a large variation. The mean dose was 493.46 mg (SK = 0.478); the most commonly prescribed daily dose was 400 mg (n = 54) (Supplemental table 1); the dose was often adjusted by 100 mg or 200 mg by clinicians because the standard dose per pill is 200 mg.

Table 1

Clinical characteristics and demographic of the enrolled concentrations
variables		enrolled concentrations (n = 153)
age	mean, std.dev, rang, year	48.97, 13.08, 21–80
gender	male(n%), female(n%)	99 (64.7), 54 (35.3)
diagnosis	schizophrenia, others	144, 9
AMI dose	mean, std.dev, rang, mg	493.46, 211.10, 100–1200
AMI concentration*	mean, std.dev, rang, ng/mL	562.94, 406.49, 5.4-2737.7
Comedications	mean, administration for each patients except for AMI	4.49, 0–12
	psychiatric / psychotic drugs	13
	antiepileptic drugs	3
	hypnotic	7
	internal medicine drugs	42
	traditional Chinese medicine	28
eGFR	mean, std.dev, rang, mL/min/1.73 m²	98.59, 16.72, 58–135, n = 153
QTc interval	mean, std.dev, rang, ms	417.87, 22.12, 350–462 n = 145
AMI = amisulpride eGFR = estimated glomerular filtration rate

The use of the National Kidney Foundation Kidney Disease Outcomes Quality Initiative and Chronic Kidney Disease Epidemiology Collaboration equation (CKD-EPI) is recommended when eGFR values are above 60 ml/min/1.73 m².

Only two patients who had AMI without any comedication. Twenty-eight patients received a single AMI administration to control psychiatric symptoms; the remaining patients received a variety of first-line atypical antipsychotic drugs. In general, patients used many kinds of concomitant drugs, including treatments for mental health symptoms, epilepsy, sleep disturbances, and other diseases. On average, each patient consumed more than four drugs.

The mean eGFR value was 98.59 mL/min/1.73 m2 (range, 58–135 mL/min/1.73 m2) (SK = -0.096), eGFR of < 60 mL/min/1.73 m2 indicated decreased renal function. Corrected QT interval values ranged from 350 to 462 ms; no pathological prolongation was found.

Table 2 showed that Model 1 was represented by the following equation: AMI concentration = -422.633 + 4.141*age (year) + 42.545*gender + 25.291*co-medication + 1.240*dose, yielding an adjusted R² of 0.372, which meant that the model accounted for 37.2% of the variation in AMI plasma levels. This model passed the F test (F = 23.480, P < 0.05), indicating that it effectively predicted AMI blood concentration. In addition, the VIF values of the model were <5, suggesting the absence of collinearity. The Durbin-Watson value was approximately 2, suggesting the absence of autocorrelation and good model validity.

Simplified model 2 was represented by the following equation: AMI concentration =838.561 - 9.169*eGFR + 1.273*dose, yielding an adjusted R² of 0.457, which means that the model accounted for 45.7% of the variation in AMI plasma levels. The predictive ability of the simplified eGFR model is still better than that of the optimized model 1.(Supplemental table 2) Figure 1 presents the histogram of simplified eGFR model standardized regression residuals, which were normally distributed. The P-P plot of model 2 is presented in Figure 2. The model passed the F test (F = 64.927, P < 0.05), suggesting that it effectively predicted AMI levels. In addition, the VIF values of the model were below 5, suggesting the absence of multicollinearity; meanwhile, the Durbin-Watson value was approximately 2, suggesting the absence of autocorrelation and good model validity. Drug dosepositively affected AMI blood concentration. In contrast, eGFR value negatively affected AMI blood concentration presented in Figure 3.

Table 2 Linear regression models and external validations

	independent	unstandardized β	P value	VIF	R²	adjusted- R²	F	D-W	compare groups	difference Mean	difference std. dev	Cohen's d
model1	constant	-422.633	0.020*	-	0.388	0.372	F=23.480, P <0.001	1.954	external concentration - model1 predicts	-90.645	302.721	0.299
	age	4.141	0.089	1.467
	gender	42.545	0.456	1.090
	co-medication	25.291	0.026*	1.345
	dose	1.240	<0.001**	1.143
model2	constant	838.561	<0.001**	-	0.464	0.457	F=64.927, P <0.001	1.917	external concentration - model2 predicts	-19.404	309.473	0.063
	eGFR	-9.169	<0.001**	1.057
	dose	1.273	<0.001**	1.057
p<0.05*P<0.01
Multiple linear regression analysis revealed dose and eGFR values explained 45.7% of the variation in blood drug concentration; this rate was higher than that achieved by the conventional four-factor model.The paired sample t-test uses Cohen's d value to indicate that critical points for distinguishing small, medium and large effects were 0.20, 0.50 and 0.80 respectively. When data difference is small at 0.2, difference between the predicted and observed values of model 2 is relatively small.

Comparisons of different administration frequencies were presented in Supplemental Table 3. The 54 data points of the 400 mg daily dose for the independent sample t test accorded with a normal distribution and square difference homogeneity. Trough concentration values were comparable between once and twice daily regimens. Partial trough concentration values were 435.18 [stddev 195.79] ng/ml and 499.17 [stddev 189.58] ng/ml, respectively for 400 mg qd and 200 mg bid.

Model 1 was valid and had certain predictive ability, but not better than simplified eGFR model 2. On one hand, among its four variables, there were two variables, age and gender, with P-values greater than 0.05. After optimization, its R² did not significantly improve. Fengli, Müller, and Bergemann reported that age affected the dose-corrected AMI plasma concentrations [6–11]. Li concluded that age was positively associated with dose-corrected rather than overall AMI plasma levels [5]. On the other hand, the constant value of model 1 was negative, and a negative predictive value would be calculated at low dose. This contradicted the actual observed values.

The adjust-R² of model 2 was greater than that of Model 1, indicating better fitting ability and prediction abilities. Moreover, external validity of model 2 was better. External validation analysis revealed that the differences between predicted and observed mean values from model 2 was smaller than that from model 1. As indicated by the Cohen D value, model 2 predicted more accurately. Model 2 was parsimonious and revealed a negative correlation between blood drug concentration and eGFR values. Specifically, when eGFR values decreased by 1 ml/min/1.73 m², AMI blood concentration increased by 9.169 ng/ml; meanwhile, when AMI dose increased by 100 mg/d, the corresponding blood concentration increased by 127.3 ng/ml. To remain below the laboratory alert level (640 ng/mL), model 2 suggests that AMI doses should remain below 300 mg/d, 500 mg/d, and 700 mg/d for patients with eGFR of 60 ml/min/1.73 m², 90 ml/min/1.73 m², and 120 ml/min/1.73 m², respectively.

Although drug doses tend to be considered continuous variables, in practice, they tend to increase by 100 mg or more than 200 mg at a time (Supplemental table 1). Consequently, the changes in the independent variable were great, while the corresponding changes in the dependent variable appeared great, too. Although previous studies yielded inconsistent findings on the association between other factors and AMI blood concentration, most studies have reported thus association.

Regimens involving doses lower than 400 mg were administered once daily. In contrast, regimens involving doses greater than 400 mg patients took administration bid followed by drug label. Only regimens involving doses of 400 mg were administered once or twice daily. In general, drugs with a short half-life and administered at the same daily dose had a steady-state trough concentration higher for twice daily than for once daily regimens. However, differences in blood drug concentration were non-significant at the AMI dose of 400 mg.

Combining different types of antipsychotics is common in schizophrenia treatment [14]. And also patients took drugs to treat physical illnesses and to improve sleep. In China, psychiatrist would prescribe traditional Chinese medicine for patients with schizophrenia.Specifically, traditional Chinese medicine preparations included many components whose pharmacokinetic properties and drug interactions remaind unclear; these unknown characteristics might been confounding factors in any analysis with the dose and concentration of AMI. These complex components had potential common pathways of metabolic enzymes and transporters in vivo, and they were likely to have pharmacokinetic interactions with each other. In model 1, each additional drug increased the blood level of AMI by 25.291 ng/ml, suggesting that concurrent medication should be considered when determining drug doses in clinical practice.

Model 2 was based on the intact nephron hypothesis and included eGFR as an independent variable. This assumes that the glomerular filtration rate (GFR) will account for all kidney drug handling. Even for drugs eliminated by tubular secretion, such as metformin, the relationships between metformin renal clearance rates and eGFR values are linear.[15] Other researchers have reported that the variability in cefepime trough concentrations are largely explained by eGFR [16, 17]. Li A et al believe that individualized use of AMI based on renal function is necessary. [18] These findings suggest that eGFR values could help assess chronic kidney disease as well as adjust drug dosing. Yet, some in vitro study shows that OCT2 (organic cation transporters of the SLC22 family ), on the kidney proximal tubules, may be important for renal elimination of AMI [19]. Well, a lot of in vivo work is required to clarify these findings for clinical implications.

This study has some limitations. First, half-life of AMI varies from 12 to 20 hours in the population, and a larger sample size may be needed to verify drug concentration with once or twice of 400mg daily dose. Second, based on challenges to the intact nephron hypothesis for renal tubular excretion drugs, metrics and biomarkers of renal tubular excretion function should be considered and applied to the concentration prediction model of AMI in the future.

In conclusion, as eGFR values capture age, sex, and renal clearance rates, model 2 had better goodness-of-fit and greater external prediction ability than model 1. AMI is used to improve both negative and positive symptoms of schizophrenia. The dose is 50-1200mg, which leads to a wide range of blood drug concentrations and often exceeds the safe concentration guidelines. The construction of eGFR model will provide a new and more accurate prediction reference model for the safe and effective use of AMI.

Consent for publication

Written informed consent was obtained from all the authors prior to the publication.

Acknowledgements

Thanks to Zhiqiang Du, Jianhong Wu and Ping Wan for their contributions to the pre-treatment of blood samples and the determination of blood concentrations.

Contributions

Conceptualization: Shushan Wang.

Data curation: Kankan Qu.

Writing – original draft: Qi Wang, Kankan Qu

Writing – review & editing: Shushan Wang.

Conflict of interest

The authors declare no conflict of interest.

Funding

N/A

Availability of Data and Materials

Data is provided within supplementary information files.

Langer SZ, 1997. 25 years since the discovery of presynaptic receptors: present knowledge and future perspectives. Trends Pharmacol Sci. Mar;18(3):95 – 9. doi: 10.1016/s0165-6147(96)01034-6. PMID: 9133779.
Linden M, Scheel T, Xaver Eich F, 2004. Dosage finding and outcome in the treatment of schizophrenic inpatients with amisulpride. Results of a drug utilization observation study. Hum Psychopharmacol Clin Exp. 19(2):111–119.
Kishimoto T, Hagi K, Nitta M, Kane JM, Correll CU, 2019. Long-term effectiveness of oral second-generation antipsychotics in patients with schizophrenia and related disorders: a systematic review and meta-analysis of direct head-to-head comparisons. World Psychiatry. Jun;18(2):208–224. doi: 10.1002/wps.20632. PMID: 31059621; PMCID: PMC6502423.
Hiemke C, Bergemann N, Clement HW, et al., 2017. Consensus guidelines for therapeutic drug monitoring in Neuropsychopharmacology: update Pharmacopsychiatry. 2018;51(1–02):9–62.
Li L, Li L, Shang DW, Wen YG, Ning YP, 2020. A systematic review and combined meta-analysis of concentration of oral amisulpride. Br J Clin Pharmacol. 86(4):668–678.
Sun F, Yu F, Gao Z, Ren Z, Jin W, 2021. Study on the relationship among dose, concentration and clinical response in Chinese schizophrenic patients treated with Amisulpride. Asian J Psychiatr. 62:102694.
Sparshatt A, Taylor D, Patel MX, Kapur S, 2009. Amisulpride - dose, plasma concentration, occupancy and response: implications for therapeutic drug monitoring. Acta Psychiatr Scand.120(6):416–428.
Bergemann N, Kopitz J, Kress KR, Frick A, 2004. Plasma amisulpride levels in schizophrenia or schizoaffective disorder. Eur Neuropsychopharmacol. 14(3):245–250.
Müller MJ, Eich FX, Regenbogen B, Sachse J, Häsignrtter S, Hiemke C, 2009. Amisulpride doses and plasma levels in different age groups of patients with schizophrenia or schizoaffective disorder. J Psychopharmacol.23:278–286.
Müller MJ, Regenbogen B, Sachse J, Eich FX, Härtter S, Hiemke C, 2006.Gender aspects in the clinical treatment of schizophrenic inpatients with amisulpride: a therapeutic drug monitoring study. Pharmacopsychiatry. 39(2):41–46.
Müller MJ, Regenbogen B, Härtter S, Eich FX, Hiemke C, 2007. Therapeutic drug monitoring for optimizing amisulpride therapy in patients with schizophrenia. J Psychiatr Res. 41(8):673–679.
Huang S, Li L, Wang Z, et al., 2021. Modeling and Simulation for Individualized Therapy of Amisulpride in Chinese Patients with Schizophrenia: Focus on Interindividual Variability, Therapeutic Reference Range and the Laboratory Alert Level. Drug Des Devel Ther.15:3903–3913.
Qu K, Wang F, Du Z, Wang S, Zhang Z, Shen Y, 2021. A novel and sensitive method for determination of amisulpride in human plasma by two-dimensional liquid chromatography. Biomed Chromatogr. 35(10):e5149.
Schmidt-Kraepelin C, Feyerabend S, Engelke C, Riesbeck M, Meisenzahl-Lechner E, Verde PE, Correll CU, Kluge M, Makiol C, Neff A, Lange C, Englisch S, Zink M, Langguth B, Poeppl TB, Reske D, Gouzoulis-Mayfrank E, Gründer G, Hasan A, Brockhaus-Dumke A, Jäger M, Baumgärtner J, Leucht S, Cordes J; COMBINE Study Group, 2022. Amisulpride and olanzapine combination treatment versus each monotherapy in acutely ill patients with schizophrenia in Germany (COMBINE): a double-blind randomised controlled trial. Lancet Psychiatry. Apr;9(4):291–306. doi: 10.1016/S2215-0366(22)00032-3. Epub 2022 Mar 8. PMID: 35276079.
Pradhan S, Duffull SB, Wilson LC, et al., 2021. Does the intact nephron hypothesis provide a reasonable model for metformin dosing in chronic kidney disease?. Br J Clin Pharmacol. 87(12):4868–4876.
Gijsen M, Bekkers B, Maertens J, et al., 2022. Prospective assessment of breakthrough infections and neurotoxicity and their association with cefepime trough concentrations in patients with febrile neutropenia. Int J Antimicrob Agents. 59(1):106472.
Falcone M, Menichetti F, Cattaneo D, et al., 2021. Pragmatic options for dose optimization of ceftazidime/avibactam with aztreonam in complex patients. J Antimicrob Chemother. 76(4):1025–1031.
Li A, Mak WY, Ruan T, et al. Population pharmacokinetics of Amisulpride in Chinese patients with schizophrenia with external validation: the impact of renal function. Front Pharmacol. 2023;14:1215065. Published 2023 Sep 4. doi:10.3389/fphar.2023.1215065
Dos Santos Pereira JN, Tadjerpisheh S, Abu Abed M, et al., 2014. The poorly membrane permeable antipsychotic drugs amisulpride and sulpiride are substrates of the organic cation transporters from the SLC22 family. AAPS J. 16(6):1247–1258.

No competing interests reported.

Download PDF

Editor invited by journal
11 Jul, 2024
Submission checks completed at journal
08 Jul, 2024
First submitted to journal
02 Jul, 2024

You are reading this latest preprint version

The eGFR parameter model helps predict blood drug concentration of amisulpride in patients with schizophrenia

Status:

Version 1

Abstract

Objective

Methods

Results

Conclusion

Figures

Introduction

Methods

Blood specimens and analysis method

Statistical analysis

Results

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1