Pharmacologic Treatment of SchizophreniaJohn M. KaneThe pharmacologic treatment of schizophrenia remains acritical component in the short- and long-term manage-ment of this disease. Considerable progress has beenmade in delineating different domains of this illness,ranging from positive and negative symptoms to cognitivedysfunction and psychosocial vulnerabilities. Increas-ingly, treatments are being studied in relation to a varietyof different outcome measures with functional ability andquality of life achieving appropriate emphasis.The introduction of a new generation of antipsychoticdrugs has helped to raise optimism and expectations.Overall, second-generation drugs do provide clear advan-tages in terms of reducing adverse effects (particularlydrug-induced Parkinsonism, akathesia, and, hopefully,tardive dyskinesia). Advantages in alleviating refractorysymptoms, negative symptoms, depression, and suicidalbehavior are found in some reports; however, muchremains to be done methodologically in establishing therelative merits of specific drugs in the multiple domains ofinterest. Biol Psychiatry 1999;46:1396–1408 © 1999Society of Biological PsychiatryKey Words: Schizophrenia, treatment, psychopharmacol-ogy antipsychotic medicationIntroductionThe modern-age of pharmacologic treatment of schizo-phrenia began in the mid-1950s with the developmentof chlorpromazine. Subsequent to that development, anumber of different chemical classes of antipsychoticdrugs were introduced into clinical practice. Over the nextthree decades, considerable progress was made in severalareas which influenced subsequent development and uti-lization of antipsychotic drugs as well the over all care ofpatients suffering from schizophrenia.The observations that antipsychotic drugs were associ-ated with a variety of neurologic side effects, mostcommonly drug-induced Parkinsonism, but also tardivedyskinesia and tardive dystonia, helped contribute supportfor a dopamine hypothesis of antipsychotic drug activityand, in turn, a dopamine hypothesis of schizophrenia.At the same time, further developments in establishingthe validity and reliability of psychiatric diagnosis resultedin a narrowing of the diagnosis of schizophrenia and tosome extent a reevaluation of drug efficacy in moreselectively defined patient populations.It took many years after the introduction of antipsychot-ics to begin to establish the role of these medications inlong-term treatment, with the intention initially of reduc-ing rates of relapse and rehospitalization. Even afterrelevant studies began to be conducted, very few lastedmore than 1 year, and they tended to involve multi-episodeor chronically ill patients. At the same time that more andmore data emerged supporting the value of maintenancetreatment (Davis 1975), increasing knowledge and con-cern was developing regarding the long-term risks associ-ated with antipsychotic drug treatment (Kane and Smith1982).Not surprisingly, outcome measures in clinical trials ofantipsychotic drugs initially tended to focus more onpositive symptoms than negative symptoms. Positivesymptoms were more likely to be associated with disrup-tive or troublesome behavior, more likely to result inhospitalization (the focus for most initial clinical trials ofdrug efficacy), and to some extent were felt to be easier tomeasure.With this background, for the first two decades ofwidespread clinical use of antipsychotic medication, it wasgenerally accepted that among available agents there wereno significant differences in clinical efficacy. In 100comparisons of different conventional antipsychotic drugs,only one study reported a significant difference (Klein andDavis 1969; Janicak et al 1993). This result did notnecessarily mean that an individual patient was equallylikely to respond to any drug, but rather in group compar-isons similar proportions of patients responded regardlessof which drug was involved. The notion that if a particularpatient did not respond to one drug, then he or she mightrespond to another was widely assumed, however, as wewill discuss below, rarely tested.The discovery that enormous variations in bioavailabil-ity and metabolism occurred with these drugs led torenewed hope that outcome could be improved substan-tially by measuring blood levels following the develop-From the Department of Psychiatry, Hillside Hospital, Division of Long IslandJewish Medical Center, Glen Oaks, New York and Department of Psychiatryand Neuroscience, Albert Einstein College of Medicine, Bronx, New York.Address reprint requests to John M. Kane, MD, Hillside Hospital, 75-59 263rdStreet, Glen Oaks, NY 11004.Received October 12, 1998; revised February 17, 1999; accepted March 5, 1999.© 1999 Society of Biological Psychiatry 0006-3223/99/$20.00PII S0006-3223(99)00059-1ment of steady state and adjusting them up or downaccordingly (Kane et al 1976). Despite a fairly extensiveliterature suggesting relationships between blood levelsand clinical response, very few studies confirmed thevalue of these measures by subsequently manipulatingblood levels into a putative therapeutic range, undercontrolled conditions and demonstrating a clinically andstatistically significant effect (Volavka et al 1992). Thereality is that blood levels have not come to be usedroutinely as a reliable and valued guide to clinical man-agement. (Whether this is the consequence of lack ofknowledge transfer, impediments related to cost and fea-sibility, or an insufficiently compelling data base could bedebated.)Another important development in the pharmacologictreatment of schizophrenia was the introduction of long-acting injectable (depot) medications. Compliance in oralmedication-taking over long periods of time is an enor-mous problem (Kane 1985). Initially, the mirror image andcontrolled trials with these medications led to considerableenthusiasm regarding their potential to improve long-termoutcomes, by reducing rates of relapse and rehospitaliza-tion. Interestingly, a number of trials comparing oral anddepot medication were not successful in demonstrating assignificant a reduction in rates of relapse as expected(Glazer and Kane 1992). We have argued elsewhere (Kaneand Borenstein 1985) that, to some extent, this is likely aresult of the methodology employed in these trials. Spe-cifically, relatively compliant patients were selected giventhe nature of the study design (double-blind oral and depotdrug, frequent assessments, need for consent, etc.) Perhapsmost