To the Editor: We appreciate the comments raised by Drs. Basu and Chakraborty and Dr. Somashekar et al. in reference to our article. Questions pertaining to methodological approach are always important in order to assess the strengths and limitations of all scientific efforts.
We would first like to address the comments surrounding the patient population in our study and then those referring to aspects of the study design itself. Drs. Basu and Chakraborty raise the issue of past episodes in the patient sample. These data are described in Table 1 in the article and include Young Mania Rating Scale and Montgomery-Asberg Depression Rating Scale scores for patients at entry to the pre-randomization and randomized treatment phases. From a historical perspective, in the year prior to study enrollment, patients in the intent to treat population experienced a median of three and four bipolar episodes before receiving treatment with quetiapine plus lithium/divalproex (N=310) or placebo plus lithium/divalproex (N=313), respectively.
With reference to the inclusion and exclusion criteria, Drs. Somashekar et al. raise a critical issue in trial methodology. As with all studies, inclusion and exclusion are a balance to maintain internal and external scientific validity. Our study was designed to determine among those patients responding to open-combination treatment during an episode how many would experience a new mood episode following randomization to monotherapy versus those continuing combination therapy. Effort was made to make the study as potentially generalizable as possible. The inclusion of patients with all types of index acute episodes (manic, mixed, or depressed) enhanced the generalizability of the findings to the wider population of patients with bipolar I disorder. The requirement that patients have had at least one manic, depressed, or mixed episode within the previous 2 years, in addition to the index episode related to study enrollment, was used to increase the possibility that patients were not experiencing a rare mood episode but had a history of mood episode recurrence. Thus, if a patient experienced recurrence during the blinded portion of the study, this would tell us more clearly if the patient’s stabilization while receiving the combination treatment needed to be continued or if it was enough for him or her to be treated with monotherapy after stabilization. Exclusion criteria (i.e., current anxiety disorder, intolerance to the investigational product, uncontrolled medical illness, recent history of substance dependence or abuse, and women of childbearing potential not using adequate contraception) were similar to those used in other studies of this nature and largely based on patient safety.
We would also like to confirm that the study was conducted in accordance with Consolidated Standards of Reporting Trials guidelines. Power calculations were performed in order to determine the sample size, which is reflected on page 479 of the article. In our study, quetiapine was compared with placebo as adjunct treatment to a mood stabilizer rather than directly compared with lithium or divalproex.
Drs. Basu and Chakraborty question the lack of use of a structured clinical interview such as the Structured Clinical Interview for DSM-IV. Although the nature of the index episode was not assessed using a structured interview, investigators were strongly encouraged to obtain medical documentation for prior episodes. If this was unavailable, they were asked to use their clinical judgment based on the patient’s history. This is stated on page 477 and reflects the process that we believe most clinicians use in clinical practice.
The study design allowed for the evaluation of patients who were stabilized while being treated with the combination of quetiapine and a mood stabilizer during the pre-randomization open-label treatment phase and, subsequently, the proportion of patients who became ill when quetiapine was blindly discontinued during randomized treatment. We agree with Drs. Basu and Chakraborty that this was an enriched sample of patients who stabilized and did well with this particular combination of medications. We discussed this limitation on page 483.
Drs. Somashekar et al. raise the issue of patient attrition prior to the randomization phase of the study. In fact, we expected greater attrition than that which actually occurred. The requirement that patients be stabilized for at least 12 weeks prior to randomized treatment circumvented early relapse but also increased the rate of patient discontinuation when compared with earlier maintenance trials with shorter periods of stabilization. We would disagree that this was not so much a “sample of convenience” as a sample consistent with the clinical reality in which many patients have great difficulty stabilizing while being treated with prescribed treatment, and once patients do stabilize, they often have difficulty remaining stable. This randomized withdrawal design is a useful means of evaluating whether treatments that are effective in short-term management are also applicable for long-term maintenance treatment (1).
With reference to the suggestion that the study should have included an increase in lithium/divalproex as a mood event, given that patients were already within the therapeutic range of the mood stabilizing medications and there is little evidence that increasing lithium or divalproex above the therapeutic range may be an effective strategy to treat breakthrough episodes, this was not considered appropriate. However, their comment is a good one because it nicely highlights the strengths and limitations of all controlled trials. If all things are changed, then it is not possible to assess what caused any given change.
Drs. Basu and Chakraborty suggest that median ratios would have been useful for presenting the time to recurrence of mood event. The significant difference between quetiapine and placebo in the time to event recurrence was demonstrated in Figure 3 and was also discussed. We believe that expressing the data as a median ratio would not have significantly changed or added to the results. The hazard ratios (percent of risk reduction) and proportion of patients experiencing mood events over the follow-up period were also provided on page 480.
Finally, we included as many safety data in the abstract as the word limit would permit. Weight increase and the possibility of an emergent-elevated glucose level were also included in the abstract. Further information regarding the metabolic changes observed in both treatment groups was covered in detail within the main body of the article. This article may have some of the most complete reporting in the field of bipolar disorder in which both the efficacy and safety results are thoroughly presented.
Our study examined the efficacy and safety of combination treatment with quetiapine and either lithium or divalproex in the prevention of mood event recurrence in patients stabilized while receiving the combination treatment. The results showed that quetiapine combination treatment is an effective option, for those patients who can tolerate this treatment, in the prevention of recurrent mood episodes in stabilized patients, irrespective of the nature of the index episode. The study is particularly strong in that it 1) featured a long stabilization phase prior to randomization, 2) included patients with mania, depression, and mixed index episodes, and 3) followed patients over a prolonged period of time (up to 104 weeks).
Storosum JG, van Zwieten BJ, Vermeulen HD, Wohlfarth T, van den Brink W: Relapse and recurrence prevention in major depression: a critical review of placebo-controlled efficacy studies with special emphasis on methodological issues. Eur Psychiatry 2001; 16:327–335
The authors’ disclosures accompany the original article.
Dr. Vieta is affiliated with Centro de Investigacion Biomedica En Red de Salud Mental (CIBERSAM).
This letter (doi: 10.1176/appi.ajp.2009.09050700r) was accepted for publication in July 2009.