A synopsis of current treatment options shows only moderate effect sizes. They could be improved considerably if individual predictors were available.

Previous attempts to identify predictors for the treatment response in alcoholism have mainly concentrated on social and personality variables. Project MATCH was such an attempt which finally failed. The same holds true for similar attempts in pharmacotherapy. Therefore, we set out for a large oligocentre trial “Project PREDICT”. 432 patients are randomly assigned to treatment with acamprosate, naltrexone or placebo. At baseline patients are assessed with a battery of interviews, questionnaires and biological examinations (e.g. genetics). Specific emphasis is put on patients’ individual pathways into relapse. It is determined whether relapse drinking respresented a positive reinforcer (“reward craving”) or a negative reinforcer (“relief craving”). This is assessed with questionnaires, the startle reflex and fMRI. We hypothesize, that patients who are a priori identified as “positive reinforcers” better respond to naltrexone. Negative reinforcers should benefit most from acamprosate.

All patients have been included by now. Preliminary analyses suggest that it is possible to distinguish between the two craving types. The equivalent of positive reinforcement in the startle reflex correlates with fMRI responses to cues with a positive valence of about 0.7. These methods might offer a platform for a targeted pharmacotherapy in alcoholism.