2011 in Review

Personalized medicine, the selection of a specific treatment for an illness based on the individual characteristics of the patient, including genotype, is often advocated as a goal for all fields of medicine. The study by Bankole Johnson and colleagues (1) is one of the first serious attempts to test a personalized genomic strategy for psychiatric illness. A number of studies have examined possible effects of genotype after a study has concluded. Johnson was the first to select patients into groups on the basis of a relevant genotype in the serotonin transporter gene for a trial of ondansetron, a medication that blocks serotonin receptors, to treat alcoholism. The selection based on genotype before the trial began prevented the confound of too many patients with one genotype in either the active drug group or the placebo group. This direct trial of gene-based treatment found that some serotonin genotypes predicted better response to ondansetron. Since there are now several alternative strategies to treat alcoholism, genotype information could be used to select patients for whom ondansetron would be the best choice. Complications inherent in the genetic approach are also revealed in the article. After the study was started, additional genetic variants within the serotonin transporter gene were discovered that also impact treatment response. The effect of these additional variants was assessed in the analysis, but the selection of patients to treatment groups had already occurred without this information. Attempts to use genetic information clinically for drug selection are not necessarily premature, but patients and doctors also need to recognize that the information is still evolving.

A major challenge facing psychiatry is the development of new treatments based on novel mechanisms of action. Currently available psychotropic medications are, in essence, pharmacological derivatives of compounds identified by astute, but serendipitous, clinical observation in the distant past. That is, the drugs we use to treat our patients today are not based on an understanding of the disease process that gives rise to the targeted signs and symptoms. In fact, most of the psychotropic medications in current clinical use are products of attempts to understand how these early drugs work and to develop new drugs with similar mechanisms of action but with greater efficacy and fewer adverse effects. As a consequence, the number of pharmacological treatments with different mechanisms of action used in clinical psychiatry today is very small. For these reasons, my pick for 2011 is the study by Raul Andero and colleagues (2), which demonstrated the beneficial effects on emotional learning of a novel small molecule (7,8-dihydroxyflavone) that crosses the blood-brain barrier. This molecule is particularly interesting because it has agonist activity at the TrkB receptor and thus mimics the effects of brain-derived neurotrophic factor (BDNF). A substantial and growing corpus of findings implicates reduced BDNF-TrkB signaling in different brain circuits in the pathophysiology of depression and anxiety, as well as in schizophrenia. Thus, although investigations with 7,8-dihydroxyflavone are still in the preclinical stages, the study by Andero and colleagues offers the much needed hope that pathophysiologically based pharmacological interventions for psychiatric illnesses, treatments with truly novel mechanisms of action, may be on the horizon.

It is unusual to select an Images in Psychiatry article as an Editor's favorite, but it offers an opportunity to discuss what we can learn from studies of the past and the value of historical research.

Paul Wender and Mallay Occhiogrosso (3) find that 300 years ago, long before Freud and even before psychiatry, Richard Steele described his poignant childhood pain at the death of his father, related it to his widowed mother's grief, and traced its derivatives in his adult character. Wender and Occhiogrosso, viewing the story through a modern psychiatric lens, add the alternative of a constitutional predisposition underlying both his childhood grief and his adult character. However, our sharpest impression comes from Steele's own conclusion: “The Mind in Infancy is, methinks, like the Body in Embrio, and receives Impressions so forcible, that they are as hard to be removed by Reason, as any Mark with which a Child is born is to be taken away by any future Application.” Since that time we may have learned more about the impact of powerful early experience on later character, and the ineffectiveness of attempts to modify the latter, but I am not certain.

The ideas are familiar to a contemporary psychiatric reader, but our understanding of the past is enriched by our current knowledge, and our conviction about that current knowledge is enhanced by this evidence that far precedes any possible contamination resulting from scientific theory influencing a compliant autobiographer.

I have always been captivated by the double meaning conveyed by the phrase “epidemiology counts.” Surely, all epidemiological studies fit the first meaning of the phrase by generating data on prevalence. However, only a unique subgroup fully meets the second meaning by changing the way in which we view illnesses. My favorite article this year, by Young Shin Kim and colleagues (4), meets both meanings as it examines autism spectrum disorders.

The study by Kim et al. really “counts” in a few respects. It focused on South Korea, a country home to no prior community-based studies of autism prevalence. Moreover, the study directly assessed a randomly ascertained sample, and such a design is rarely used to study autism prevalence. Finally, the study continued the trend of demonstrating relatively high rates of autism, much higher than the field had been led to believe before completion of a few, recent well-designed epidemiological studies. This final aspect changed the way in which we think about autism. No longer can we view the condition as a rare albeit important form of difficulty. We must confront the reality that autism affects the lives of all of us, through interactions with our peers and families, and we must therefore continue to try harder to better understand the condition.

While a review paper on experimental designs may not immediately sound like a captivating topic, the article by Wesley K. Thompson and colleagues (5) provides a compelling argument for rethinking the way we design longitudinal research.

Clinical experience tells us that the imprint of illness on the life course of an individual often spans many decades in its evolution. However, continuous observation of research subjects over 30–40 years is typically not feasible. Consequently, lifespan conclusions are often derived from cross-sectional data or data from several age groups over a shorter longitudinal period.

Thompson et al. highlight how age cohort effects may confound the interpretation of longitudinal data if not addressed, yet many analyses do not account for them. Using data from the Alzheimer's Disease Neuroimaging Initiative, the authors demonstrate how a cross-sectional analysis of age, entorhinal cortex, and APOE-ε4 fails to show an effect of APOE-ε4 that may be revealed by within-subject longitudinal data. Furthermore, when they specifically examined age cohort effects, they discovered an intriguing finding: patients with early-onset Alzheimer's disease from recent age cohorts had lower entorhinal cortical thickness relative to patients of the same age from earlier cohorts.

More systematic and refined design approaches, such as the accelerated longitudinal design, are absolutely essential to understand the course of psychiatric diseases from their prodrome to late-life expression, particularly in view of rapid changes in our environment (e.g., an epidemic of metabolic syndrome, dramatic changes in socialization through Internet media) that may exert substantial effects on course and outcome by age cohort.

Some articles report data that are so precise and revealing that they emphasize how much we have to learn before achieving real understanding. The article by Joel Swendsen and colleagues (6) was one of those. First of all, the topic is of great importance but often overlooked, namely, the comorbidity of drug abuse with schizophrenia. The question focused on the causality of drug use and symptom states in schizophrenia—with a bivalent direction tested. Do negative mood, perceived stress, and psychotic symptoms provoke drug use, or does drug use worsen these symptoms? I was not familiar with use of a “computerized ambulatory monitoring technique” nor a “personal digital assistant” (PDA) in clinical research. The article describes how these authors taught 199 people with schizophrenia to respond to random prompts six times per day for a week asking, each time, about mood, anxiety, psychosis, and drug use. This was successful, on average, 72.1% (SD=19.1%) of the time, a striking response rate. Their data show that the duration of the electronic interview fell progressively over the week, indicating a mastery of the technique by the schizophrenia volunteers. The results showed that 1) sad mood and psychosis tend to provoke drug use, 2) baseline anxiety is a risk factor for continued alcohol use, and 3) any drug use tends to result in increased anxiety and psychosis. Data like these have not been previously obtained because the course of sampling has been so slow that the cause and effect (symptoms/drug use) are effectively dissociated. The development of these clever kinds of techniques to obtain data that we have not previously seen provides hope that we will discover even more important and specific links between drug use and schizophrenia in their future use.

Editorial accepted for publication October 2011.