# Exploratory analysis of non-linear clozapine dose-concentration relationship in real-life hospital database

## TL;DR

This study explores how clozapine dose and blood concentration are not always linear, using real-life data from hospital patients.

## Contribution

The study identifies a non-linear dose-concentration relationship in clozapine, with a potential hinge point at 400 mg/day.

## Key findings

- Specialists identified cases of unexpected clozapine concentrations during dose adjustments.
- A MARS model found a non-linear relationship with a hinge point around 400 mg/day.
- Clozapine dose and smoking were key predictors, but only explained 25% of concentration variability.

## Abstract

Clozapine (CLO) remains the gold standard for the treatment of drug-resistant schizophrenia. It is commonly accepted that there is a linear relationship between CLO dose and blood concentration, although deviations from this pattern are frequently observed in clinical practice. The aim of the present naturalistic study was to further investigate this relationship using a real-world database of CLO therapeutic drug monitoring (TDM) samples, with a particular focus on: i) identifying cases of “unexpected” CLO levels during repeated within-subject blood sampling in the process of CLO dose adjustment and ii) assessing linearity of the cross-sectional, between-subject CLO dose-concentration relationship and identifying potential breakpoints.

The study was based on a single-center TDM database derived from routine monitoring of CLO concentration in psychiatric inpatients, supplemented with data from medical records. The database was reviewed independently by a laboratory medicine specialist and psychiatrist to identify individual cases of “unanticipated” dose-concentration relationships. The study also employed the Multivariate Adaptive Regression Splines (MARS) to detect non-linear relationships in the prediction of CLO levels, as determined by high-performance liquid chromatography. Analyses incorporated variables such as daily CLO dose, smoking status, age, sex, and co-medications.

Individual cases of “unanticipated” CLO concentrations supporting the partially non-linear within-subject dose-concentration relationship were unambiguously identified by both specialists. The MARS model revealed a breakdown in the between-subject CLO dose-concentration linear relationship identifying a hinge point around 400 mg/day, below which CLO concentrations were less dose-dependent. CLO dose and smoking were the most important predictive factors, but the model explained only about 25% of CLO concentration variability.

Our data suggest that the non-linear relationship between CLO concentration and its daily dose can be a real-life clinical problem with CLO doses of ∼400 mg/day as the provisional hinge point. Although preliminary, the present results warrant further investigations on non-linear aspects of CLO pharmacology, including “unexpected” CLO concentrations, toxicity, and lack of therapeutic activity.