An exploratory study of adolescent response to fluoxetine using psychological and biological predictors

Introduction

Although serotonin-specific reuptake inhibitors (SSRIs), and in particular fluoxetine, have revolutionized the treatment of depressive and anxiety disorders, there is a sizable minority of patients who do not improve during the course of treatment. It would be of great clinical utility to be able to predict which patients might profit from a course of fluoxetine and which patients might be better treated with a different medication or some other intervention. The problem is compounded by the 4–6-week delay in clinical response in those patients who do respond. The damage to those who turn out to be non-responders includes their persisting disorders and distress, their disappointment with the failure of the treatment they were given, and the decreased conditional likelihood of the efficacy of the next medication prescribed. In children and adolescents, there is an additional consideration the risk of self-injurious behavior activation related to SSRI treatment (Elbe et al., 2014).

Although there is some argument about the relative importance of the placebo response and the actual clinical effect of fluoxetine and other SSRIs, there is general agreement that fluoxetine is the best first choice medication for depression and anxiety. The position paper of child and adolescent psychiatry for the Canadian academy (Garland et al., 2016) concludes by saying that “the risk-benefit balance for fluoxetine in child and youth depression is favorable” (p 6). Walkup (2017) comes to a similar conclusion and moreover, he estimates, based only on publicly funded research, that fluoxetine adds 25–30% to the response rate for depressed children and adolescents, beyond the 30–35% placebo response rate.

By all standards, then, fluoxetine is the first drug of choice, but by the best estimates, a sizable minority of children and adolescents, about a third, will not respond to this medication.

Type D, or “distressed” personality, was posited by Denollet (2005). It has two component traits, social inhibition (SI) and negative affectivity (NA). Type D individuals are less likely to self-disclose and to share their medical symptoms with their doctors (Schiffer et al., 2007), probably due to their discomfort with and distrust of others (SI). They are more alexythemic than non-D individuals (Zohar et al., 2011; Batselé et al., 2016). However, they are awash with negative affectivity (NA), especially moodiness and irritability. Owing to their SI, Type D individuals are less likely to seek or to profit from social support even when feeling bad. Thus the elevation of both SI and NA is the source of this enduring pattern of distress.

Being Type D is associated with increased risk for worse mental health conditions (Lee et al., 2012), including depression and anxiety. Type D personality is associated with anxiety and depression in adult cardiac patients (Kupper et al., 2013) and in non-patients in the community (Zohar et al., 2011). Type D is also associated with elevated clinical depression in adolescent out-patients (Zhang, Li & Zou, 2011) and with an elevation of symptoms of depression and anxiety in adolescents in the community (Lee et al., 2012).

At least two pathways have been suggested between personality Type D and ill-health: overactivation of the immune system and poor health behavior.

D’Acquisto (2017) suggests that negative affect and immune activation are potential responses to perceived immunological threat and/or perceived emotional stress, and that they have a synergetic effect on each other. In a large scale empirical study, Van Dooren et al. (2016) measured plasma biomarkers of inflammation and endothelial dysfunction as well as Type D and depression. Van Dooren et al. (2016) found that cytokine levels, and markers of endothelial dysfunction were related to Type D personality, and to depression, and that Type D personality had a contribution to the prediction of depression over and beyond that of the biomarkers.

Another possible connection between Type D on the one hand and depression and anxiety disorders on the other is the relative self-neglect and failure to adopt health behaviors. Zohar et al. (2011) found that Type D individuals reported less sexual activity and less subjective health; Kupper et al. (2013) found that Type D individuals were more likely to smoke and to have sedentary lifestyles; Williams, Abbott & Kerr (2016) found that Type D individuals were less likely to engage in health behaviors including eating a balanced diet, sleeping well, and getting enough physical activity.

The current study set out to identify predictors of treatment response of adolescents with affective/and or anxiety disorders to fluoxetine. The role of personality was assessed using the Type D measure because of its known association with these disorders, and because of its relationship to hyperreactivity of the immune system.

Methods

Results

Discussion

When considering the results of this study, its limitations should be kept in mind. In particular, this study includes a relatively small patient sample, and the fluoxetine trial was open label, not allowing for the assessment of placebo response. In addition, although Type D personality contributed to the prediction of non-response, its association with depressive symptoms was very significant, and the question of its temporal stability and the best time to measure personality status vs. clinical status has not been studied in a satisfactory way in the context of adolescent depression. Further studies, with a prospective design, and using large patient samples, are required to confirm our results.

The current study found a significant elevation of Type D, or distressed personality, in a clinical sample of depressed and/or anxious adolescents of psychiatric outpatients relative to community controls. This finding is consistent with previous research. In their systematic review, Mols & Denollet (2010) concluded that there was evidence for the association of Type D personality with both depression and anxiety disorders. Zhang, Li & Zou (2011) reported an elevation of Type D personality in depressed Chinese adolescents relative to healthy controls. In a prospective study, Doyle et al. (2011) found that the risk of developing depression within a year of hospitalization for acute coronary syndrome was more than three-fold for Type D personality patients. Van Dooren et al. (2016) found in an adult community sample, that Type D was associated with a 13-fold increased risk of meeting criteria for depression. However, there is a clear relationship between scoring high on both subscales of the DS14, required for classification as Type D, and depressive symptoms. This relationship raises doubt as to the stability and independence of the personality status vs. the clinical assessment. If Type D personality is associated with depression it would not be a predictor of the disorder but an indicator of its presence. Ossola et al. (2015) found, in a study of patients presenting with acute cardiac syndrome, that the NA subscale of the DS14 was so strongly associated with depressive symptoms that Type D was better conceived of as a state description than as a stable personality typology. The current study also found strong correlations between both depressive subscales used and the NA subscale of the DS14 at all three time-points.

Another related issue in using Type D status as a predictor is that of its temporal stability. The typology (i.e., classifying individuals into Type D and Non-D) is defined by scoring 10 or more on both of the DS14 subscales. Thus gaining or losing one point on either of the subscales due to an individual’s actual change or due to measurement error is enough to change the personality status. In a community sample of older adults who self-reported on the DS14 at baseline and six years later, 56.9% of individuals who were Type D at time2 had been Type D at time1 (Zohar, 2016). For children in the community (Jellesma, 2008) measured twice at a 18-month interval only 42.9% of those who were Type D at time2 had been Type D at time1. The temporal stability of caseness is considerably lower in cardiac patients. Ossola et al. (2015) found that less than a third of cardiac patients retained their Type D classification 12 months after an acute cardiac event. In using Type D as a predictor of health events, we lack a theoretical or empirical criterion for when Type D should be measured to ensure that it predates the outcome and is not just a description of the outcome. It should be noted that in the current study the differences in the DS14 scores between responder and non-responders were so large that it seems unlikely that temporal stability over the eight week trial constitutes an alternate explanation; however, to discriminate trait and state it would have been much more convincing if personality had been measured prospectively before the patients started exhibiting depression and anxiety.

Keeping these reservations in mind, the current study found that clinically depressed or anxious adolescents who have Type D personality had a nearly three-fold greater risk of non-response to the fluoxetine trial in this study. Increased risk remained for Type D personality patients to be non-responders even when controlling for initial depressive symptoms and for initial TNFα levels as shown in the binary logistic regression.

Overall, Type D patients in the current study presented more severe depression at baseline. At treatment termination, Type D adolescents were significantly worse off on all the clinical measures, anxiety scales, depression scales, and clinician rating scale. Thus there was a strong association between personality type, and fluoxetine trial outcome. This result is consistent with other medical procedures in which Type D individuals show less favorable outcomes. Zhang et al. (2016) found that Type D rectal cancer patients were less likely to respond positively to oncological treatment.

What is it about Type D personality that might be associated with less treatment response? It may be that the chronic distress of negative affectivity and social inhibition typical of Type D individuals puts them at risk of less effective immune response, and chronic inflammatory processes. This theory has extensive empirical support (Van Dooren et al., 2016; Jandackova et al., 2017). The current study measured three bio-markers of inflammatory processes, TNFα, IL-6, and IL-1b, which were measured pre- and post-treatment. Although there was a negative association between initial TNFα levels and treatment response (Amitai et al., 2016) there no association between TNFα levels and Type D personality. It does not seem that hyperactive immune function will explain the full force of association between Type D personality and non-response to fluoxetine in the current study.

An additional hypothesis proposed for the association of Type D personality with many bad health outcomes, is that Type D individuals are less likely to engage in health behavior vs. Non-D individuals. Williams et al. (2008) showed that Type D individuals were less likely to eat reasonably, to spend time outdoors every day, or to get a regular medical checkup. Williams, Abbott & Kerr (2016) showed that young adults who are Type D are less likely to engage in a healthy lifestyle and that these behaviors partially mediate the effect of Type D on self-reported (ill) health. There is also research supporting the notion that Type D individuals with cardiac conditions are less likely to complain of relevant symptoms to their physicians (Schiffer et al., 2007) and less likely to adhere to their cardiac medication (Williams et al., 2011). There is no such research on adolescent patients, and yet there is no reason to believe that adolescents with Type D personality would have as much health behavior as Non-D adolescents. It is possible that the Type D adolescents enrolled in the current study were less open with the treatment team about their relevant symptoms, or that they were less adherent to the treatment regime, or that they engaged in less health promoting behaviors that might potentially synergize the effect of the medication. More research is needed in order to understand the mechanisms at play.

Conclusions

The additional risk conferred on adolescent patients who have Type D personality suggests that taking personality assessment into consideration while treating with psychotropic medication in a medical setting might improve treatment and overall therapeutic outcome. Personality assessment in medical settings is not usually included for practical reasons. The current study used a very brief and easily scored personality assessment, thus not adding to participant burden. A bigger potential barrier for including personality assessment in medication trials has to do with disciplinary lines between medicine and psychology. The strong results associated with the personality assessment in the current study suggest that this mixed strategy might be of clinical utility and might help to expand our understanding of the processes of drug response.

Supplemental Information