Schizophrenia
What is Schizophrenia?
Disease definition & pathophysiology
Schizophrenia is a chronic neuropsychiatric disorder characterized by altered perception of reality, cognitive dysfunction, and impaired emotional regulation. Pathophysiology involves dysregulated dopamine neurotransmission, particularly hyperactivity in mesolimbic pathways (causing symptoms such as hallucinations and delusions (positive symptoms)) and hypoactivity in mesocortical pathways (causing loss of motivation and emotional expression (negative symptoms) and cognitive symptoms). Glutamate and serotonin dysregulation are also involved.
Commonly used Pharmacodynamic (PD) Models
- Dopamine receptor occupancy models predicting clinical response.
- Disease progression models capturing symptom severity over time (PANSS scale).
- Indirect-response models describing symptom modulation by antipsychotics (delayed treatment response).
Patient characteristics
Typical patient population
Schizophrenia commonly emerges during late adolescence to early adulthood (16–30 years). Incidence is slightly higher in males, with earlier onset compared to females. Global prevalence is about 0.5–1%.
Risk factors & disease progression indicators
Risk factors include genetics (family history), prenatal complications, neurodevelopmental disturbances, substance use (especially cannabis), and psychosocial stressors. Indicators of disease progression include severity of positive and negative symptoms, cognitive deficits, level of functioning, and frequency of psychotic relapses.
Diagnosis & biomarkers
Key Clinical Biomarkers
Diagnosis is primarily clinical, based on symptom presentation and patient history (Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria). Biomarkers are limited, but neuroimaging (MRI, PET) may show structural brain changes or altered dopamine receptor activity. No specific laboratory tests currently exist for schizophrenia diagnosis.
Disease Severity Classification
Severity often assessed by clinical scales such as the Positive and Negative Syndrome Scale (PANSS):
- Mild: PANSS total score ~58–74
- Moderate: PANSS total score ~75–94
- Severe: PANSS total score ≥95
How can Schizophrenia be treated?
Treatment aim (PD-targets)
Main therapeutic aim is symptom reduction through modulation of neurotransmitter imbalance, primarily targeting dopamine (D2 receptors) and serotonin (5-HT2A receptors).
Common Drug Classes & Regimens
Antipsychotics sit in ATC N05 Psycholeptics, where D2 receptor occupancy as a biomarker of efficacy is described.
First-generation (typical) antipsychotics (historically first-line, now less preferred)
Examples include Haloperidol (N05AD01), Chlorpromazine (N05AA01).
MoA: Strong dopamine D2 receptor antagonism reduces positive symptoms but carries high risk of movement-disorder (extrapyramidal) side effects.
Second-generation (atypical) antipsychotics (current first-line treatment)
Examples include Risperidone (N05AX08), Olanzapine (N05AH03), Quetiapine (N05AH04), Aripiprazole (N05AX12), Clozapine (N05AH02, reserved for treatment-resistant cases).
MoA: Mixed serotonin (5-HT2A) and dopamine D2 receptor antagonism; improves negative and cognitive symptoms, with lower risk of extrapyramidal side effects.
Dose Adjustments
Dose adjustments typically required for renal or hepatic impairment. Dose initiation and titration vary according to tolerability. Lower starting doses recommended in elderly patients or patients sensitive to side effects. Pediatric use is limited and requires cautious dose adjustment.
Commonly used PK-models
- Population PK models describing variability in exposure due to genetic factors (CYP2D6, CYP3A4 polymorphisms), smoking status, and patient demographics.
- One- or two-compartment models for antipsychotics.
- PK-PD models linking antipsychotic exposure (plasma levels, receptor occupancy) to changes in PANSS scores or side-effect incidence.
Clinical studies
Under the European Medicines Agency (EMA) guideline, short-term efficacy trials are double-blind, randomized and parallel-group, preferably 6 weeks long, and a placebo arm is recommended, since without one a trial’s ability to detect a difference (assay sensitivity) cannot be guaranteed. At least 20% of the patients should have had the disease for less than 5 years. The primary endpoint is the change from baseline to the end of the trial in the total score of a composite symptom scale, such as the PANSS or the Brief Psychiatric Rating Scale (BPRS). Responder analyses should also be presented, and a reduction of at least 30% in the PANSS total is generally accepted as a response in acutely ill patients. A continued need for treatment is best shown against placebo in a randomized withdrawal trial, in which patients stable after at least 12 weeks of open-label treatment are randomized to the drug or placebo, and the primary endpoint is the proportion who relapse. Withdrawals, and the missing data they leave, are a particular problem in both short- and long-term trials [1].
Disease severity in schizophrenia trials is scored on rating scales such as the PANSS, which has 30 items, each rated by a psychiatrist from 1 (absent) to 7 (extreme), in positive, negative and general subscales. An analysis of three 6-week trials of paliperidone (N05AX13) used item response theory (IRT) instead, because analyzing the total score as continuous data discards information and violates the numerical nature of the scale. IRT models each item as ordered categorical data, driven by an unobserved disease severity. The model described the placebo and drug effects at the item, subscale and total score levels, and 41.5% of the patients had dropped out by the end of the trials [2].
In a pooled analysis of placebo arms from 16 schizophrenia trials with 1436 patients, the time course of the PANSS score was best described by a Weibull model or an indirect response model. The probability of dropping out was associated with a high observed PANSS score. Dropout was modeled as a hazard in a time-to-event model, fitted jointly with the PANSS model, and exponential, Weibull and Gompertz hazards described it equally well [3].
In a simulation study of informative dropout, where dropout depended on the efficacy variable being measured, ignoring it biased parameter estimates by up to 21% with the FOCE-I estimation method, against less than 5% with a dropout model in the base scenario. Even without a dropout model, the bias translated into only small effects on predictions of the efficacy variable, but the dropout model was crucial for simulating realistic trial outcomes [4].
Clinical pharmacology studies
For long-acting injections (depot preparations), an appendix to the EMA guideline sets the aims of the development: to establish the full pharmacokinetics of the new formulation, including its release properties, to compare its bioavailability with that of the oral form, and to assess how long it keeps an acceptable level of the drug. The release rate over time, the drug remaining at the injection site and accumulation may be estimated by pharmacokinetic modeling of the oral data together with single-dose depot data. Clinical studies comparing efficacy with the oral form and justifying the dosing interval are needed in principle, unless a clear pharmacokinetic/pharmacodynamic relationship has been shown for the oral formulation. Because patients are first stabilized on the oral drug, the switch to the depot (the matching depot dose, and whether the oral dose can be stopped at once or should be phased out) should be justified by pharmacokinetic and clinical data [1].