Association of Personality Profiles with Depressive, Anxiety, and Cancer-related Symptoms in Patients Undergoing Chemotherapy

INTRODUCTION

The Five Factor model of personality structure is strongly established.¹ The Big Five personality dimensions or traits (i.e., neuroticism, extraversion, openness to experience, agreeableness, conscientiousness)² describe enduring, cross-culturally validated, individual traits that are known to influence numerous important health outcomes, including disease burden,³ self-rated health,^{4, 5} impact of illness events,⁶ and mortality.^7–10 In addition, personality traits, particularly high neuroticism and low conscientiousness, are associated with greater risk of depression and anxiety symptoms and syndromes.^11–16 Personality is associated with physical symptoms in both clinical and non-clinical populations.¹⁷

An extensive literature has documented the risk for depressive and anxiety symptoms in oncology patients undergoing treatment,^18–20 as well as in cancer survivors.^20–23 These symptoms exert deleterious effects not only on quality of life, but also on many important outcomes, including adherence, physical symptoms, functioning, and possibly mortality.^{20, 24, 25} Substantial efforts have been made to understand risk factors for depression and anxiety in patients with cancer, in order to identify higher risk patients, provide treatment for those with elevated symptom levels, and elucidate mechanisms of action for interventions.

Despite the well-established relationship between personality traits and depression and anxiety, relatively few studies were identified that examined this relationship in oncology patients.^26–32 These studies suggest that higher levels of neuroticism increase the risk for depression and anxiety in patients with various cancer types. For example, among women with breast cancer who underwent surgical treatment (n=210), neuroticism increased the risk for depression.²⁸ Similarly, among patients with lung cancer (n=1334), higher neuroticism, coping characterized by helplessness/hopelessness, and female gender were associated with higher levels of anxiety.²⁷ A limited number of studies, with inconsistent findings, have examined the relationship between personality and physical symptoms (e.g., pain, fatigue) in cancer patients.^33–36

No studies were found that utilized combinations of traits, rather than single personality traits (e.g., neuroticism), to identify profiles of personality dimensions that may increase risk for depressive, anxiety, and physical symptoms among patients undergoing cancer treatment. Person-centered approaches to examining data³⁷ enable the identification of latent classes (subgroups) of individuals with distinct profiles of personality dimensions. Such approaches, which include cluster analysis, latent profile analysis (LPA), and latent class analysis, complement variable-centered approaches by conceptualizing personality as “an interrelated system of several traits.” Subgroups or latent classes can then be evaluated for differences on a wide range of characteristics.

Such person-centered methods can be viewed as complementary approaches to trait-based, variable-centered approaches to examining personality.^37–40 Examination of multivariate distributions of patterns in personality traits can help identify previously unobserved patterns of personality and compare these patterns or types across samples and studies.³⁸ As noted by Specht, “The aim of the typological approach is to identify a preferably parsimonious number of personality types that allow for broad categorizations of individuals.”³⁹

Numerous typological studies of personality have been conducted in non-medically ill populations (reviewed in Specht³⁹). Caspi posited, based on a number of studies, that there are three major personality types (labeled generally as “resilients,” “overcontrollers,” and “undercontrollers”),⁴¹ and Asendorpf and colleagues confirmed this hypothesis in four studies of children and (primarily young) adults.³⁸ Their broadly categorized prototypes differed somewhat depending on the specific sample and method of personality assessment, but the overall generalizability of the three major types was confirmed. Thus, to date, a substantial body of literature exists, primarily in non-medically ill populations (i.e., general population samples, college students, adolescents), that has identified three broad classes of personality based on distributions of patterns of dimensional traits.³⁹

In addition, several studies have utilized latent class methods to examine associations between membership in personality profile classes and psychological measures and outcomes. For example, Merz and Roesch utilized LPA to examine personality profiles in a sample of university students (n=371), using the International Personality Item Pool,⁴² a measure based on the Five Factor Model.⁴³ A three-class solution provided the best fit to the data. Based on the mean levels of each of the five personality dimensions, the classes were characterized as: “well-adjusted,” “reserved,” and “excitable.” Relationships among the classes and measures of affect, self-esteem, depression, anxiety, and coping efficacy were examined. Compared to both the reserved and excitable classes, the well-adjusted class (i.e., low on neuroticism, high on extraversion, agreeableness, and openness) reported better psychological functioning in terms of positive affect, negative affect, depression, anxiety, self-esteem, and coping. The reserved and excitable groups differed on anxiety, with the excitable group (i.e., high neuroticism, high extraversion) reporting generally higher anxiety than the reserved group (i.e., moderate neuroticism, low extraversion, agreeableness and openness).

Hori and colleagues utilized LPA among outpatients with major depression to identify personality profiles using a different personality measure (the Temperament and Character Inventory⁴⁴), with the goal of better characterizing the heterogeneity of symptoms in major depressive disorder.⁴⁵ They identified three latent profiles that they termed “neurotic,” “adaptive,” and “socially detached.” The three profiles differed on a number of characteristics salient to diagnosis, treatment, and outcomes of major depression (e.g., depressive symptomatology, anxiety symptom, psychotropic medication use, and social functioning),⁴⁵ suggesting the utility of latent class methods for uncovering important contributors to heterogeneity among clinically-characterized populations.

To our knowledge, no prior studies have utilized latent class methods to examine the relationship between personality profiles and psychological or cancer-related symptoms in patients with cancer. Therefore, the purposes of this study were to: 1) identify, using LPA, latent classes of cancer patients with distinct personality profiles based on the Five Factor Model,⁴⁶ and evaluate for differences among the latent classes in demographic and clinical characteristics; and 2) examine differences among the latent classes in trait and state anxiety, depressive symptoms, and cancer-related symptoms. Based on prior work demonstrating associations between personality and depression, anxiety, and cancer-related symptoms,^{11–16, 27, 28, 33, 34} we hypothesized that classes with personality profiles higher on neuroticism and lower on conscientiousness would exhibit higher levels of anxiety, depression, and cancer-related symptoms and symptom-related distress.

METHODS

RESULTS

DISCUSSION

This study is the first to use LPA to identify latent classes of cancer patients based on the Big Five personality dimensions, complementing prior studies that used trait-based approaches to examine personality in relation to cancer patients’ symptoms.^26–32 As hypothesized, we found that personality profiles characterized by higher levels of neuroticism and lower levels of conscientiousness endorsed higher levels of anxiety, depression, cancer-related symptoms and symptom-related distress. Although we did not explicitly predict a dose-dependent pattern in these associations, we found that the Distressed class had higher symptom levels than the Normative class, who in turn had had higher symptom levels than the Resilient class. The consistency of this pattern suggests that these findings are robust.

Our findings are partially consistent with an LPA analysis conducted by Specht et al., who identified three latent classes of personality profiles in a German sample (n=14,718), and four latent classes in their Australian sample (n=8,315).³⁹ The authors emphasized the similarity of the three types identified in the German sample to a substantial prior literature—i.e., the “resilient” class (56% of the sample) had low neuroticism and high extraversion, openness, agreeableness and conscientiousness; “undercontrollers” (22%) had relatively lower levels of agreeableness and conscientiousness; and “overcontrollers” (23%) had low levels of extraversion and openness and slightly higher neuroticism. The Australian sample resulted in a four-class solution—i.e., “average” (41% of the sample;); “resilients” (36%; low neuroticism; high extraversion, agreeableness, and conscientiousness; average openness); “undercontrollers” (13%; high neuroticism, lower agreeableness and conscientiousness); and “overcontrollers” (high neuroticism, high openness). Our Resilient class resembles the “resilients” in both samples and our Distressed class resembles the “undercontrollers” in both samples.

Our findings validate and extend the work of Merz and Roesch,⁴³ who identified three latent classes based on Big Five personality profiles in 371 university students. They characterized these three classes as “well-adjusted” (49.3% of the sample; relatively low neuroticism, and relatively high on the other four facgtors), “reserved” (moderate neuroticism and conscientiousness, lowest extraversion, and relatively low agreeableness and openness) (28.3%) and “excitable” (22.4%; highest neuroticism and extraversion, and relatively high openness). Their latent classes shared some similarities with our findings in terms of the combinations of traits. For example, our “Resilient” class resembles their “well-adjusted” class (i.e., these classes demonstrated low levels of neuroticism and relatively high levels of other personality dimensions). In addition, the classes that scored highest on neuroticism were the smallest classes (i.e., their “Excitable” group, 22.4%; our “Distressed” group, 14.3%). Their “Excitable” group was characterized by higher levels of anxiety, higher emotional instability, higher negative affect, and lower coping self-efficacy compared to their “Well-adjusted” group. Similarly, our “Distressed” group had the highest levels of state and trait anxiety as well as the highest levels of each domain of depressive symptoms. The possible mediating role of coping self-efficacy in the relationship between personality profiles and anxiety, as well as other symptoms commonly reported by cancer patients, should be investigated.

However, the personality profiles identified in each study do not completely map onto one another. For example, whereas our “Distressed” class scored highest on neuroticism and lowest on extraversion, agreeableness, and conscientiousness, their “Excitable” class scored highest on neuroticism but also highest on extraversion.⁴³ These differences may be explained by differences in the sample size and composition, and the use of different measures to assess personality dimensions. Moreover, whereas we found statistically significant differences among the three classes on all five dimensions (essentially a validity check for separation of the identified classes), Merz and Roesch described the classes in terms of relative levels of each personality dimension, but did not report statistical significance for differences in personality dimensions.⁴³

In terms of demographic characteristics, patients who were younger, had lower education, were not married or partnered, provided care for another adult, did not exercise regularly, or smoked were more likely to be in the Distressed class. In terms of age, our findings, while cross-sectional, are consistent with longitudinal studies that used variable-centered or person-centered approaches. For instance, hierarchical linear modeling analyses of age trends in personality among participants in the Baltimore Longitudinal Study of Aging found that neuroticism, extraversion, and openness tended to decline over time, while agreeableness and conscientiousness tended to increase over time.⁸³ The longitudinal study by Specht et al. (described above) that used LPA to examine personality profiles found that older adults were more likely to have a “resilient” personality profile.³⁹ The psycho-oncology literature has consistently shown that younger patients are more likely to be distressed (higher levels of depression and anxiety). Could this reflect, in part, underlying age differences in personality profiles? Other authors have argued that age-related differences in distress may be explained by social, occupational, and relational pressures and expectations.⁸⁴

Additionally, compared to White patients, Non-white patients were more likely to be classified in the Distressed versus the Resilient class. In addition, compared to Non-Hispanic patients, Hispanic patients were more likely to be classified in the Distressed versus the Resilient or Normative classes. Prior literature utilizing LPA to identify latent personality classes did not describe findings related to ethnicity, so comparisons cannot be made.^{39, 43} However, other studies by our group found higher levels of distress (i.e., depressive and anxiety symptoms) in Non-white oncology patients. However, these studies did not specifically examine personality.^{85, 86}

While it is plausible that some differences among the personality profile classes may reflect differences in disease or treatment-related characteristics, we did not find significant differences in terms of cancer diagnoses, number of metastatic sites, number of prior treatments, types of prior treatments, time since diagnosis, or types of metastatic sites, arguing against the idea that disease or treatment-related characteristics were responsible for differences in personality profiles. However, the number and impact of comorbidities did differ across the groups. A number of plausible explanations exist for these differences. For example, patients with higher levels of neuroticism and lower levels of conscientiousness were more likely to have more comorbidities, as well as more likely to have functional impairments. This hypothesis is supported by our finding that patients in the Distressed class exercised less and were more likely to smoke. The findings related to differences in smoking and exercise, as well as functional status, are consistent with literature on personality and health behaviors and outcomes.^87–89 However, caution is warranted when inferring causality from this cross-sectional analysis, as it is plausible that patients with lower functional status or more comorbidities were more likely to endorse specific patterns on the personality assessment that would place them in the more “Distressed” class. Thus, further work is needed to examine the possible influence of personality profiles, as well as of individual traits and facets, on functioning and comorbidity in cancer patients, and to identify possible mechanisms that explain the observed associations.

As predicted, patients in the class that scored highest on neuroticism reported the highest levels of anxiety and depressive symptoms. These findings are consistent with prior studies in both cancer and non-cancer populations that demonstrated positive relationships between trait neuroticism and these symptoms.^{11, 12, 27, 30, 90–92} Notably, the Distressed class had a mean CES-D score (24.1) that was above the accepted cutoff for clinically meaningful depression,⁹³ while the Normative class had a mean level of depressive symptoms that might be considered subsyndromal.⁹⁴ Our results for STAI-T and STAI-S are similar.⁸⁶

Additionally, as hypothesized, the classes with lower levels of conscientiousness (i.e., the Normative and Distressed classes) reported higher levels of depressive and anxiety symptoms, in a dose-dependent manner. These findings are consistent with a substantial literature documenting the association betweem lower levels of conscientiousness and worse mental health outcomes and quality of life.^{87, 95–97} Evidence suggests that conscientiousness exerts its effects on health outcomes (including physical health outcomes) through its effects on health behaviors.⁸⁹

Personality profiles were also associated with different levels of physical symptoms and symptom-related distress, as measured by the MSAS. Although no prior studies were found that examined MSAS scores in relation to Big Five personality profiles, these findings are consistent with several studies that found associations between specific traits and symptoms (e.g., higher neuroticism and fatigue^{33, 34}). However, findings regarding the relationship between other traits (extraversion, openness, conscientiousness) and self-reported pain and fatigue are inconsistent.^{35, 36} Such inconsistent findings may be explained in part by different sample characteristics and measures. However, it is possible that the integrated LPA approach used in the present study helps clarify the trait-based approaches used in prior studies.

Several study limitations should be noted. First, the sample may not accurately represent the distribution of personality traits among patients undergoing CTX, as the most common reason for declining participation was feeling overwhelmed. Also, because personality was assessed during (but not before) CTX, it is possible that trait levels may have been affected by this experience. Literature on personality provides evidence that personality is stable over time⁹⁸ and is a predictor rather than an outcome of how patients cope with stressful life events.¹⁷ Nevertheless, future research using prospective designs is needed to delineate further the relationship between personality traits and profiles and psychological and physical symptoms in patients with cancer and other illnesses.

Despite these limitations, the present findings suggest that personality traits are important to consider when evaluating the symptom experience of cancer patients. The personality profile in the Distressed class may represent a combination of traits that, in this particular context (undergoing CTX), predisposes to higher levels of depressive and anxiety symptoms. In contrast, the Resilient class showed a predisposition to lower psychological symptom levels, and the Normative class was intermediate in terms of psychological symptoms. These findings should not be construed to suggest that some traits or combinations of traits are inherently better, more advantageous, or more desirable. Rather, the extensive literature on personality traits and psychological functioning suggests that certain traits, combinations of traits, and facets (sub-domains of traits), in different contexts, situations, and life circumstances, may predispose to different thoughts, behaviors, and actions, as well as different psychological experiences and outcomes.¹⁷

Future research should examine the influence of personality profiles on other symptoms commonly experienced by cancer patients (e.g., sleep disturbance, fatigue, pain). Approaches such as LPA may be useful to researchers examining the relationship of personality to symptoms, illness responses, coping, and quality of life in patients with medical illnesses. Moreover, investigation of the mechanisms by which personality profiles may influence psychological and physical symptoms should be investigated, as a deeper understanding of these mechanisms would help investigators and clinicians develop and test screening and intervention strategies that target specific strengths and vulnerabilities that may influence important health-related outcomes.