Pharmacogenetic studies are driven by several distinct motivations: 1) determining pharmacogenetic influences on differential treatment response can help clinicians use appropriately targeted treatments for specific individuals; 2) determining predictors of adverse effects can help clinicians' decisions to avoid treatments for specific individuals; 3) deepening our understanding of how treatments work can assist with the discovery of new targets for treatment development; and 4) predicting metabolic profiles can help with medication or dosage choices.

In this issue are two similar yet very different pharmacogenetic articles. One reports evidence on treatment resistance to antipsychotics among people with schizophrenia that is potentially associated with differential splicing of a potassium channel (1). The other provides evidence that a haplotype in a set of nicotinic channel subunits is associated with differences in placebo response for people who want to quit smoking (2). Both studies use two patient populations, and both focus on candidate polymorphisms.

The findings that led to examining a potassium channel in schizophrenia nicely illustrate the progression of scientific discovery within psychiatry. In the 1960s, a set of Drosophila flies exhibited twitching, dancing legs when anesthetized with ether—the mutated gene (encoding a voltage-gated potassium channel) was ultimately named “ether a go-go” after the dance. In Drosophila, ether a go-go mutations resulted in the disruption of synaptic plasticity and memory. In the ensuing decades, two homologous human ether a go-go-related genes (hERG1 and hERG2) were eventually discovered and found to be highly expressed in the brain and heart.

In parallel, a number of antidepressants and antipsychotics were observed to lengthen cardiac QT intervals. Inhibition of hERG1 results in long QT syndrome, genetic variants of KCNH2 (the gene encoding hERG1) can cause congenital long QT syndrome, and many psychiatric compounds have the capacity to block hERG1 function. Whether genetic variants in KCNH2 may explain individual vulnerabilities for lengthened QT intervals from medications awaits delineation. Whether blockade of this potassium channel may mediate some antidepressant or antipsychotic effects is also not yet known. Certainly, genetic variability within KCNH2 may be associated with risk for schizophrenia (3).

Nonetheless, differential splicing of mRNA may result in many brain-specific isoforms. Moreover, stress and medication can influence differential splicing (4–6). The list of differentially spliced proteins now includes hERG1—for which a specific hERG1 isoform (KCNH2 3.1) is primarily expressed in the brain, lacks a domain involved in slow deactivation, has faster deactivation kinetics, and thus results in increased neuronal excitability. This 3.1 isoform is more prevalent in the brains of schizophrenia patients (7). In addition, intronic polymorphisms may influence splicing, and intronic polymorphisms in KCNH2 are associated with increased expression of the 3.1 isoform.

Would such intronic polymorphisms also be associated with differences in response to an antipsychotic? This was tested in two populations: one from an open-label study of antipsychotics (Clinical Antipsychotic Trials in Intervention Effectiveness [CATIE]), the other from a crossover placebo-controlled trial conducted in an inpatient setting. In the CATIE sample, the TT genotype (previously associated with both higher 3.1 expression and greater risk for schizophrenia) was associated with a slightly better response to treatment (primarily in positive symptoms). The effect size was small (a change of about 1 on the Positive and Negative Syndrome Scale), potentially limiting its utility for clinical guidance. However, a secondary finding was that the TT genotype (about 10% of the population, and N=10 in the examined sample of patients taking olanzapine) was associated with one-fourth to one-fifth the chance of discontinuing olanzapine.

The small placebo-controlled study was subsequently useful not just in replicating these findings but also for determining whether patients with the TT genotype improve regardless of treatment or whether there was a specific influence on medication response. In fact, there was an interaction between drug/placebo and genotype on positive symptoms, supporting the findings from CATIE. However, these results must be tempered by the fact that these patients were enrolled because of known partial resistance to previous antipsychotics, and all had limited 4-week assessment periods. Thus, the difference between placebo and antipsychotic response was fairly small. Although replicable, these pharmacogenetic findings are unlikely to immediately influence clinical care, but they deepen our understanding of how antipsychotics might work. The results raise the possibility of focusing on differential splicing or ionotropic channel kinetics to understand and treat schizophrenia (8).

For smoking cessation, several treatment modalities already exist, and some individuals manage to stop smoking on their own. The study by Chen et al. (2) focuses on a candidate region that has been the subject of a growing literature (9, 10). Three nicotinic receptor subunits are on an adjacent segment of the 15th chromosome, and there are three common haplotypes in this area that were defined using two polymorphic nucleotides (GC, GT, and AC). Genetic polymorphisms are often in linkage equilibrium with other polymorphisms (i.e., when one allele is inherited, so is the linked allele), and co-inherited blocks of a chromosome are defined as haplotypes.

The study by Chen et al. similarly investigated two populations. One was a large cohort from a community-based study (designed to examine atherosclerosis risk), and the other was from a randomized placebo-controlled treatment trial of smoking cessation. From the larger cohort study, in which participants likely employed a variety of techniques to stop smoking, the authors replicated an association between haplotype and number of cigarettes smoked per day. The answer to “How old were you when you stopped smoking?” was also associated with the haplotype (albeit with an effect size of 2 years).

Of clinical interest, though, is whether this genotype could inform the treatment recommendations for smoking cessation: no medication, bupropion, nicotine replacement, or combined medications. In the treatment trial, all three medication treatments were superior to placebo regardless of haplotype. That is, the authors did not find any evidence that this haplotype could help determine who would benefit most from which specific treatment.

However, the association of haplotype with smoking cessation was most evident in participants who received placebo. Those with the AC haplotype did poorly (only about one-quarter remained abstinent after placebo compared with about one-half on medication treatment). This effect may be large enough for a clinician to use when counseling a patient. That is, a patient may be more willing to undergo the expense and side effects of a medication if the expected benefit is anticipated to be large enough (i.e., double his or her chances of quitting). For those with the GC haplotype, the benefits of treatment were not significantly different from those of placebo (i.e., medication treatment may not be worth it).

Thus, a clinician might advise patients with the AC haplotype that their chances of quitting are doubled if they use nicotine replacement or bupropion and might advise patients with the GC haplotype that these medications may have less potential for benefit. Before this is clinically implemented, physicians will want to know the extent to which the haplotypes are associated with side effects such as nausea (10) and if the effect is long lasting. Nonetheless, these results are a clear example of how genetic results from placebo-controlled studies might directly influence psychiatric treatment decisions in the near future. The new information is not that nicotinic channels might be involved in nicotine addiction but rather that the association was observed to be treatment specific.