The report by Ho et al. and commentary by Lewis illustrate the complexity of disentangling disease from treatment effects in evaluating the pathology of schizophrenia. This potentially confounded relationship has previously emerged in various lines of investigation such as evaluating the DA hypothesis of schizophrenia by measuring D2 receptor density in postmortem tissue.
Reductions in gray matter and increases in ventricular and subarachnoid space volume have been reported extensively in imaging and postmortem studies, including from before the introduction of antipsychotic drugs. Consequently, the neuropathology of schizophrenia, at least in part, is likely due to illness-related effects apart from pharmacologic treatment if not directly from disease pathophysiology. The study by Ho et al. suggests that the brain structural pathology is compounded, if not caused, by drug effects.
There are several limitations to the nature of the evidence that supports this conclusion. The first is the inherent, and potentially confounded, relationship between illness severity and drug treatment. The authors attempted to control for this statistically, but it is unclear how effective the illness severity measures used actually reflect the elements of the illness that might relate to its putative progressive effects on brain volume. For example, the measures used did not capture what number or duration of acute psychotic episodes were experienced by patients. In addition, treatment in the Ho study was wholly uncontrolled and thus subject to the variation of clinician practice patterns and judgment.
A second issue is that even if APDs cause reductions in brain volume, the absence of treatment in patients with schizophrenia could result in greater reductions in volume than in treated patients. In other words, two effects may be present and the absence of treatment would result in greater morphologic consequences to patients.
The same issue pertains to the preclinical evidence. The rodent and subhuman primate studies were in healthy animals, so they represent the effects of drugs on healthy brain tissue in the absence of disease pathophysiology. The suggestion by Lewis to examine APD effects in mood disorder patients is important and would be an effective way to address this question.
Primum non nocere, first do no harm, is the physician’s dictum, but the findings of this study must be interpreted (and acted upon) with caution. Clearly, judicious use of APDs in patients is absolutely warranted, but it would be a grievous error if these results lead to treatment being withheld from or unduly limited in patients.
View all comments by Jeffrey Lieberman