Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Feb 1.
Published in final edited form as: J Consult Clin Psychol. 2011 Nov 21;80(1):78–86. doi: 10.1037/a0026433

Interpersonal Pathoplasticity in the Course of Major Depression

Nicole M Cain 1, Emily B Ansell 2, Aidan GC Wright 3, Christopher J Hopwood 4, Katherine M Thomas 4, Anthony Pinto 5, John C Markowitz 5, Charles A Sanislow 6, Mary C Zanarini 7, M Tracie Shea 8, Leslie C Morey 9, Thomas H McGlashan 2, Andrew E Skodol 10, Carlos M Grilo 2
PMCID: PMC3265649  NIHMSID: NIHMS339544  PMID: 22103955

Abstract

Objective

The identification of reliable predictors of course in major depressive disorder (MDD) has been difficult. Evidence suggests that the co-occurrence of personality pathology is associated with longer time to MDD remission. Interpersonal pathoplasticity, the mutually influencing non-etiological relationship between psychopathology and interpersonal traits, offers an avenue for examining specific personality vulnerabilities that may be associated with depressive course.

Method

This study examined 312 participants with and without a co-occurring personality disorder diagnosis who met criteria for a current MDD episode at baseline and were followed for 10 years in the Collaborative Longitudinal Personality Disorders Study (CLPS).

Results

Latent profile analysis (LPA) identified six interpersonal groups (extraverted, dominant, arrogant, cold, submissive, and unassuming) and circular statistical profile analysis confirmed group interpersonal distinctiveness. No significant differences between groups were found in comorbid Axis I disorders or baseline MDD severity. Chronicity and functioning analyses found significantly greater chronicity and poorer functioning in individuals with a submissive interpersonal style over 10 years.

Conclusions

These findings support the relevance of interpersonal pathoplasticity in depressive course and that this heterogeneity has clinical significance. This study is the first to use LPA and circular profiles to examine interpersonal heterogeneity within a diagnostic group. The implications of these findings for therapeutic intervention, interpersonal functioning, and psychopathological course are discussed.

Keywords: major depressive disorder, interpersonal circumplex, pathoplasticity

The National Comorbidity Survey – Replication (NCS-R; Kessler, Berglund, et al., 2005) identified major depressive disorder (MDD) as the most common mental disorder in the United States with a lifetime prevalence rate of 16.6%. MDD often follows a chronic course, substantially impairing psychosocial functioning. One issue in the assessment and treatment of MDD is the need to identify reliable predictors of MDD course and outcome. Previous research has identified several potential predictors of longer time to remission and lower rates of remission such as female gender (Kornstein et al., 2000), presence of dysthymia (Keller, Shapiro, Lavori, & Wolfe, 1982), and Axis I comorbidity (Keller et al., 1992). Grilo et al. (2005, 2010) investigated the influence of comorbid personality disorder (PD) diagnosis on the prospective course and outcome of MDD and found that patients with MDD with existing PD pathology had a significantly longer time to remission from MDD than patients with MDD without any PD, even when controlling for prognostic predictors. The presence of a PD at baseline or recurrent MDD but not gender or dysthymic disorder significantly predicted time to relapse. These studies demonstrate the importance of examining predictors over long-term follow-up and underscore the need to delineate specific baseline characteristics that influence course and outcome of MDD.

Interpersonal functioning may prove clinically useful in predicting the course of MDD. Pincus and Wright (2010) argued that evaluating interpersonal functioning is an essential part of the diagnostic process, beyond symptom assessment. Empirical and theoretical formulations have emphasized interpersonal functioning in the etiology and maintenance of major depression (see Joiner & Timmons, 2009 for a review). In particular, research has associated interpersonal dependency with increased vulnerability for major depression (see Zuroff, Mongrain, & Santor, 2004 for a review) and increased depressive symptoms over time (Mongrain, Lubbers, & Struthers, 2004). Although research suggests that dependent interpersonal styles may represent a risk factor for poorer course and functioning in depression, the current study aimed to assess whether a specific interpersonal style represents a negative prognostic risk factor over 10 years.

One method for examining interpersonal functioning as a predictor of MDD course and outcome is the interpersonal circumplex (IPC; Leary, 1957). The IPC is rooted in interpersonal theory, which posits one’s interpersonal style can be described using two orthogonal dimensions: warmth and dominance. This model permits the description of individual or group data by locating them in the two-dimensional space created by the orthogonal dimensions (see Figure 3 for an example of the interpersonal circumplex and its eight octants). The IPC model of interpersonal style is applicable to understanding pathoplasticity in MDD course. Pathoplasticity is characterized by a mutually influencing non-etiological relationship between psychopathology and personality (Widiger & Smith, 2008). In this way, psychopathology and personality influence the expression of each other, but neither exclusively causes the other, as might occur in an etiological or spectrum relationship. Pathoplasticity recognizes that psychopathology occurs in the larger context of an individual’s personality, making it unreasonable to assume that the expression of pathology would not be influenced by one’s characteristic manner of perceiving, thinking, feeling, and relating to the environment (Widiger & Smith, 2008). Interpersonal pathoplasticity describes the observed heterogeneity in the phenotypic expression of psychopathology within the IPC model (e.g. Cain, Pincus, Grosse & Holtforth, 2010) and can predict variability in response to psychotherapy within a disorder (e.g. Borkovec et al., 2002).

Figure 3.

Figure 3

Open in a new tab

Circumplex locations of the predominant interpersonal trait reported by the interpersonal subtypes in depression

Note: An example of the 8 octants found in the Interpersonal Circumplex (IPC) adapted from Leary (1957). Octants are labeled with the alphabetical notation originally provided by Leary (1957) (e.g., PA, BC, DE, etc.). Circumplex locations for the whole sample of depressed patients (n = 312) located at 191.62°, Group 1: Extraverted (n = 69) located at 35.86°, Group 2: Dominant (n = 10) located at 91.53°, Group 3: Arrogant (n = 60) located at 127.55, Group 4: Cold (n = 29) located at 194.68°, Group 5: Submissive (n = 54) located at 256.33°, and Group 6: Unassuming (n = 90) located at 307.76°, All circumplex locations are approximate.

To demonstrate interpersonal pathoplasticity in MDD, three criteria should be met: (a) the identification of distinct and homogeneous interpersonal groups of depressed individuals, (b) interpersonal group classification is not accounted for by features of psychopathology (e.g., symptom severity or Axis I comorbidity), and (c) evidence of differential expression of the disorder (e.g., chronicity) across subgroups. Accordingly, we first aimed to identify distinct, prototypical interpersonal subgroups in subjects with MDD by applying latent profile analysis (LPA) to IPC octant scores. Second, we tested for differences on baseline depression symptom severity and Axis I comorbidity. Third, we examined interpersonal subgroup differences in MDD chronicity and functioning over a 10-year period while controlling for the presence of PD diagnosis. This study serves as an important first step in determining whether a specific interpersonal style is a reliable predictor of MDD course and outcome, over and above PD diagnosis, over a 10-year period. Identifying baseline characteristics predicting differential MDD outcomes are an important area of much needed research. If interpersonal style can account for some of the observed heterogeneity in MDD course, then further research on baseline assessments may prove useful in personalizing treatment interventions.

Method

Participants and Procedures

Study participants were drawn from the Collaborative Longitudinal Personality Disorders Study (CLPS), a multi-site, prospective, naturalistic study designed to assess the course and outcome of 668 patients aged 18–45 years old diagnosed with one of four PDs (schizotypal PD, borderline PD, avoidant PD, and obsessive compulsive PD) and a comparison group of patients diagnosed with current MDD but no PD.1 Details of CLPS methods and participants have been previously reported (e.g., Gunderson et al., 2000).2

The current study includes 312 participants assessed over a 10-year period who met criteria for current MDD at baseline and completed the self-report measures (see Table 1 for demographic and clinical characteristics). Participants provided written informed consent. Interviewers were experienced clinicians with masters or doctoral degrees in mental health disciplines who underwent training across sites to achieve and maintain reliability in diagnostic measures. Participants were interviewed at baseline, 6 and 12 months, and yearly thereafter.

Table 1.

Demographic and Clinical Characteristics

Sex N %
  Female 199 63.8
  Male 113 36.2
Mean SD
Age 33.27 6.92
Ethnicity N %
  Caucasian 230 73.7
  African American 47 15.1
  Hispanic 28 9.0
  Asian 4 1.3
  “Other” 3 0.9
Axis I Comorbidity N %
  Substance Abuse/Dependence 137 43.9
  Alcohol Abuse/Dependence 122 39.1
  Post Traumatic Stress Disorder 106 34.0
  Panic Disorder 99 31.7
  Social Phobia 76 24.4
  Generalized Anxiety Disorder 74 23.7
  Dysthymic Disorder 55 17.6
  Obsessive Compulsive Disorder 50 16.0
MDD Specifier N %
  Melancholic Type 105 33.7
  Atypical Type 65 20.8
CLPS Group Assignment N %
  No PD/positive for MDD 77 24.7
  Borderline PD 75 24.0
  Avoidant PD 62 19.9
  Obsessive Compulsive PD 56 17.9
  Schizotypal PD 42 13.5
Open in a new tab

Measures

At baseline, interviewers administered the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV; American Psychiatric Association, 1994) Axis I Disorders –Patient Version (SCID-I/P; First et al., 1996) to assess current and lifetime Axis I psychiatric disorders. The kappa coefficient for MDD inter-rater reliability was 0.80 and the test-retest kappa was .61 (Zanarini et al., 2000). Course of MDD was assessed using the Longitudinal Interval Follow-up Evaluation (LIFE; Keller et al., 1987) across 12 assessments. The LIFE is a semi-structured interview rating system for assessing the longitudinal course of mental disorders.3 In this study, the LIFE measured the presence and severity of MDD over the 10-year follow-up period using weekly ‘psychiatric status ratings’ (PSRs) ranging from PSR=1, no symptoms to PSR=6, severe symptoms and dysfunction.4 The LIFE also assesses monthly Global Assessment of Functioning (GAF) scale ranging from 0–100.

The Diagnostic Interview for DSM-IV Personality Disorders (DIPD-IV; Zanarini et al., 1996) assessed PD diagnosis at baseline. Participants also completed the self-report version of the NEO Personality Inventory – Revised Form R (NEO-PI-R; Costa & McCrae, 1992) at baseline. The current study used a scoring procedure for the NEO-PI-R that derives IPC octant scores (Figure 1) (see Traupman et al., 2009).5 In the current sample, Cronbach alphas for the NEO-PI-R IPC octants were .58 (JK), .63 (HI), .67 (PA), .69 (FG), .70 (LM), .73 (BC), .74 (DE), and .78 (NO), consistent with prior research.

Figure 1.

Figure 1

Open in a new tab

An example of a circumplex structural summary

Note. X axis = circumplex angle in degrees; Y axis = standard (z) score on NEO-PI-R IPC octant; Angular displacement = the person’s interpersonal “central tendency,” signifying the individual’s “typology” (Leary, 1957). Amplitude = measure of profile differentiation. It is viewed as a measure of the profile’s “structured patterning,” or degree of differentiation, indicating the extent then to which the predominant trend “stands out.” An amplitude value of 0 indicates a flat (i.e., undifferentiated) profile; high amplitude indicates a profile with a clear interpersonal peak (and trough). Elevation = an index of stylistic response to IPC measures which do not contain a general substantive factor (Ansell & Pincus, 2004).

Data Analysis

Latent profile analysis (LPA) was conducted using Mplus 6 (Muthén & Muthén, 2010) to classify depressed individuals into latent groupings based on their interpersonal profile. LPA is a person-specific technique that establishes latent groups of individuals who share a similar profile on a given set of observed variables. The IPC octant scales were used as the observed variables for the LPA models. Model fit was compared using the Akaike (AIC) and Bayes information criteria (BIC), smaller values of which indicate better fit to the data.

The structural summary approach for circumplex data models an interpersonal profile of octant scores with a cosine-curve function. As Figure 1 shows, the parameters of this curve are its (a) angular displacement; (b) amplitude; and (c) elevation. The goodness-of-fit of the modeled curve to the actual scores can be evaluated by calculating an R2 value, which quantifies the degree to which the profile conforms to prototypical circumplex expectations. Detailed descriptions of the structural summary, procedures for solving for the various parameters, and interpretive guidelines that relate each of these summary features to clinical hypotheses have been reported (Ansell & Pincus, 2004; Wright, Pincus, Conroy, & Hilsenroth, 2009).

External validation of interpersonal pathoplasticity was conducted and analysis of covariance (ANCOVA) with Bonferroni post hoc analyses was employed to examine depression chronicity and functioning over the 10-year period while controlling for PD diagnosis.

Results

Using the structural summary method, an interpersonal profile was calculated for the sample of 312 depressed participants at baseline. The interpersonal profile of this sample was, on average, located in the DE octant (191.62°), reflecting a cold interpersonal style. However, the structural summary parameters of amplitude (0.22) and R2 (.32) indicate that the overall group exhibits low interpersonal differentiation and prototypicality. Therefore, to test the nature of the interpersonal heterogeneity, participants’ IPC octant scores were subjected to LPA. Models were estimated ranging from one to eight profiles. Table 2 summarizes model fit and entropy statistics. Six- and seven-profile solutions were given close consideration and we retained a six-profile solution based on fit and subsequent circumplex analyses.6 The entropy for the six-profile solution (0.81) suggested well-differentiated profiles for subsequent classification. The resulting subgroups of the six-profile solution were subjected to circumplex group comparison techniques (Wright et al, 2009) (see Table 3). All groups exhibited highly prototypical circumplex profiles (all R2 values > .90 and all amplitude values > .60). Figure 2 provides a visual representation of the interpersonal profiles for each group. Figure 3 depicts the circumplex locations of the predominant interpersonal traits reported by the whole sample, as well as each interpersonal LPA group. For between-group statistical comparisons of interpersonal groups, circular means, circular variances, and 95% circular confidence intervals (CI) were calculated for each group.7 Table 3 presents the circular means, variances, and 95% CIs for the six interpersonal MDD groups. The angular CIs of the interpersonal groups do not overlap, demonstrating that individuals within each of these groups reported distinct interpersonal styles. For further evidence of interpersonal pathoplasticity, we compared the six interpersonal groups on Axis I comorbidity and baseline depression severity (Table 4). There were no significant differences among the six groups on Axis I comorbidity or baseline depression severity such as number of previous major depressive episodes, age of onset of MDD, and current MDD severity, providing support for interpersonal pathoplasticity in depression.

Table 2.

Latent Profile Analysis Model Fit Indices and Entropy Statistic

AIC BIC Entropy
One-profile solution 7055.82 7115.71 --
Two-profile solution 6687.12 6780.70 .80
Three-profile solution 6542.97 6670.23 .79
Four-profile solution 6456.37 6617.32 .82
Five-profile solution 6390.84 6585.48 .83
Six-profile solution 6351.22 6579.54 .81
Seven-profile solution 6319.74 6581.75 .84
Eight-profile solution 6295.47 6591.17 .83
Open in a new tab

Note. AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; Entropy is a measure of classification certainty with values > .80 reflecting acceptable certainty. Dashes indicate that no entropy is calculated for a one-profile solution (i.e., classification certainty is perfect by definition). Bold type indicates preferred model. Nine-profile models failed to converge.

Table 3.

Probability of Most Likely Latent Class Membership and Interpersonal Characteristics of the Groups

Structural summary parameters Circular Statistics
Group N Probability Angle Elevation Amplitude R2 Circular
Mean		 Circular
Variance		 95% Circular CI
Whole Sample 312 -- 191.62° −0.09 0.22 0.32 -- -- --
Group 1 (Extraverted) 69 0.87 35.86° −0.08 0.99 0.98 36.56° 22.11° 31.34° – 41.77°
Group 2 (Dominant) 10 0.93 91.53° 0.31 1.91 0.97 90.76° 14.89° 81.50° – 99.97°
Group 3 (Arrogant) 60 0.87 127.55° 0.03 0.63 0.95 127.47° 34.65° 118.70° – 136.24°
Group 4 (Cold) 29 0.90 194.68° −0.13 1.4 0.99 194.89° 14.72° 189.53° – 200.25°
Group 5 (Submissive) 54 0.84 256.33° 0.04 1.16 0.98 253.94° 21.04° 248.33° – 259.56°
Group 6 (Unassuming) 90 0.87 307.76° −0.03 0.62 0.98 303.56° 33.89° 296.58° – 310.58°
Open in a new tab

Note. Probability in the table refers to the average latent class probability for most likely latent class membership. Angle = circumplex location of the predominant interpersonal trait in degrees; Elevation = average octant endorsement; Amplitude = a measure of profile differentiation; R2 = interpersonal prototypicality. Circular Mean = the average of the angular displacements for each individual within the group; Circular Variance = the dispersion of the angular displacements of individuals within a group around the circular mean; 95% Circular CI’s = 95% circular confidence intervals which identify reliable differences in circular means.

Figure 2.

Figure 2

Open in a new tab

Circumplex structural summary profiles of the interpersonal subtypes in depression

Note. X axis = IPC octant; Y axis = standard (z) score on NEO-PI-R IPC octant; PA (ambitious-dominant), BC (arrogant-calculating), DE (cold-quarrelsome), FG (aloof-introverted), HI (lazy-submissive), JK (unassuming-ingenuous), LM (warm-agreeable), NO (gregarious-extraverted)

Table 4.

Comparisons of the Interpersonal Subgroups in Depression on Circular Statistics, Diagnostic Comorbidity and Baseline MDD Severity

Group 1:
Extraverted
(n = 69)		 Group 2:
Dominant
(n = 10)		 Group 3:
Arrogant
(n = 60)		 Group 4:
Cold
(n = 29)		 Group 5:
Submissive
(n = 54)		 Group 6:
Unassuming
(n = 90)
Comorbidity N (%) N (%) N (%) N (%) N (%) N (%) χ2 (df) V
Mood D/O’s 3 (4.3) 2 (20.0) 15 (25.0) 6 (20.7) 18 (33.3) 11 (12.2) 1.58 (10) .16
Anxiety D/O’s 6 (8.7) 1 (10.0) 12 (20.0) 6 (20.7) 9 (16.7) 26 (28.9) 2.37 (20) .14
Alcohol/Substance Use D/O’s 11 (15.9) 1 (10.0) 8 (13.3) 4 (13.8) 8 (14.8) 21 (23.3) 1.63 (10) .16
MDD Melancholic Type 23 (33.3) 2 (20.0) 21 (35.0) 9 (31.0) 22 (40.7) 28 (31.1) 2.79 (5) .10
MDD Atypical Type 11 (15.9) 2 (20.0) 11 (18.3) 9 (31.0) 7 (13.0) 25 (27.8) 7.39 (5) .16
MDD Severity Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) F (5, 306) η2
# of previous MDE 40.96 (45.27) 25.70 (39.36) 46.93 (46.74) 46.97 (47.85) 46.83 (46.15) 46.14 (46.69) 0.54 .01
age of onset 18.63 (7.81) 17.30 (9.56) 17.87 (8.93) 18.21 (7.30) 19.80 (10.08) 19.39 (10.00) 0.41 .01
current severity 2.42 (1.41) 2.20 (1.55) 2.62 (1.40) 2.55 (1.30) 2.70 (1.18) 2.64 (1.15) 0.54 .01
Open in a new tab

Note. Comorbidity = Axis I comorbidity diagnosed using SCID-I/P; Melancholic Type and Atypical Type = specifiers of MDD assessed in SCID-I/P; MDD Severity = baseline depression severity; MDE = major depressive episode; V = Cramer’s V, measure of effect size in Chi-Square analysis; η2 = measure of effect size in ANOVA

Finally, to demonstrate the incremental clinical value of assessing interpersonal traits along with DSM–IV diagnosis, we compared the six interpersonal groups on depression symptom chronicity as measured by number of weeks spent at a 5 or 6 on the LIFE PSR scale and on functioning as measured by number of months with a LIFE GAF score below 70 over the 10-year period. We performed between-subjects ANCOVAs controlling for the presence of PD diagnosis using a single dummy coded variable (0 = no PD, 1 = PD) with Bonferroni post hoc tests. There were significant group differences for chronicity after controlling for the effect of PD diagnosis [F (5, 305) = 5.21; p < .001; η2 = 0.08] (Table 5). The submissive group spent significantly more weeks at a 5 or 6 PSR (μ = 177.09) compared to all groups except the cold group (μ = 166.69). Similarly, we found significant group differences on functioning after controlling for the effect of presence of PD diagnosis [F (5, 305) = 4.81; p < .001; η2 = 0.07] (Table 5) with the submissive group spending significantly more months with a GAF score below 70 (μ = 52.13) than the extraverted and unassuming groups.

Table 5.

Mean Comparisons of the Interpersonal Subtypes in Depression on Depression Symptom Chronicity and Functioning Over a 10-Year Period While Controlling for the Presence of a PD Diagnosis

Group1:
Extraverted
(n = 69)		 Group 2:
Dominant
(n = 10)		 Group 3:
Arrogant
(n = 60)		 Group 4:
Cold
(n = 29)		 Group 5:
Submissive
(n = 54)		 Group 6:
Unassuming
(n = 90)
Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) F (5, 306) η2
Chronicity 62.99c (13.80) 49.00b,c (36.24) 97.57b,c(14.79) 166.69a,b (21.28) 177.09a (15.60) 114.84b,c(12.08) 5.21*** .08
Functioning 30.45a (2.52) 32.70a,b,c,d (6.62) 43.03b,c,d (2.70) 50.17b,c,d (3.88) 52.13b,c (2.85) 40.67d (2.21) 4.81*** .07
Open in a new tab

Note. Chronicity: measured by number of weeks spent at a 5 or 6 on the Psychiatric Status Rating (PSR) scale; Functioning: measured by number of months with a Global Assessment of Functioning (GAF) score below 70; η2 = measure of effect size in ANCOVA;

***

p<.001; Post hoc significant differences noted by different superscripts.

Discussion

The current study represents an important step in identifying reliable predictors of MDD course and is the first to use IPC octants and LPA to examine interpersonal pathoplasticity in MDD course. The identification of six distinct, homogeneous interpersonal groups of depressed individuals that do not differ on baseline symptom severity provides evidence supporting the importance and applicability of interpersonal pathoplasticity in this clinical sample. Additionally, we found significant differences in depression symptom chronicity and on a measure of functioning, with individuals endorsing a submissive interpersonal style reporting a more chronic depressive course and poorer functioning over the 10-year follow-up period, even after controlling for the effects of PD diagnosis. This finding extends prior research on dependency and depression by clarifying the specific interpersonal style associated with the worst outcomes in prospective course and functioning within depression (e.g. Mongrain et al., 2004; Pincus & Wilson, 2001). A submissive interpersonal style may represent a fundamental risk factor for poorer outcomes in major depression, thus providing a first step toward identifying a reliable predictor for MDD course. While Grilo et al. (2005, 2010) found that MDD combined with baseline PD pathology results in significantly longer time to remission from MDD, the current study delineates the specific interpersonal style, submissiveness, most associated with poorer MDD outcome, and establishes that interpersonal style increments PD pathology in predicting course of MDD symptoms and functioning.

Our results suggest that using the IPC model to assess interpersonal functioning may increment diagnostic categories in explaining heterogeneity in course of psychiatric disorder. The interpersonal styles associated with MDD are notably diverse (i.e., a single interpersonal style does not adequately describe the potential diversity in MDD presentation), and the average interpersonal style, cold/unaffiliative, is not the interpersonal style associated with the poorest outcome. These findings highlight the importance of examining the pathoplastic expression of interpersonal style within depression and the incremental information garnered by examining outcomes at the interpersonal level, rather than by diagnosis alone. Further research is needed to determine whether interpersonal heterogeneity is associated with pathoplastic effects on functioning across diagnoses.

The current study and its conclusions have several limitations. First, the majority of participants had a co-occurring PD diagnosis, limiting generalizability to MDD samples without co-occurring PD pathology. However, we believe this heterogeneity may accurately reflect the variety of interpersonal functioning found in treatment-seeking patients with MDD. Second, two of the subgroups identified by LPA had a small number of participants (e.g., dominant subgroup = 10; cold subgroup = 29). Despite limited statistical power, we nonetheless detected large effects across our chronicity and functioning analyses; however, future studies should include a larger sample size to ensure that each interpersonal group has an adequate number of subjects. Third, this study used a naturalistic design of treatment-seeking individuals, yielding greater variability among participants than what a controlled treatment protocol might offer. However, our findings may not generalize to depressed individuals who are not treatment-seeking or who refuse to participate in research. The ethnic composition of our sample was mostly Caucasian (73.7%) and female (63.8%); however, we highlight that the ethnic distribution of our participants reflects the geographic areas sampled, and our gender distribution reflects treatment-seeking populations. Finally, although well validated and reliable, the LIFE methodology may be vulnerable to bias or inaccurate recall of symptoms.

In conclusion, our results identified six distinct, interpersonally prototypical groups in MDD. The six distinct groups reported differential depression symptom chronicity and differential functioning over a 10-year follow-up period not attributable to differences in symptom severity or diagnostic comorbidity. These findings indicate that interpersonal style may account for some of the heterogeneity observed in the course of MDD. Future studies of MDD should integrate analyses testing interpersonal styles as predictors and moderators of outcomes and should investigate psychotherapy interventions that specifically target at-risk interpersonal styles in depressed patients. Baseline assessments of interpersonal style may inform personalized interventions targeting specific interpersonal mechanisms thereby reducing the chronicity of depressive symptoms and improving functioning over time. For example, depressed patients with a submissive interpersonal style may require a therapeutic focus on assertiveness skills and encouragement for self-direction before undertaking behavioral activation to reduce depressive symptom chronicity and improve functioning. Accounting for distinct interpersonal styles underlying a common Axis I diagnosis may improve treatment and outcome across diagnoses.

Acknowledgments

Supported by NIH grants MH 50837, 50838, 50839, 50840, 50850; DA029641; MH016545; MH073708; MH080221; MH087053.

Footnotes

Publisher's Disclaimer: The following manuscript is the final accepted manuscript. It has not been subjected to the final copyediting, fact-checking, and proofreading required for formal publication. It is not the definitive, publisher-authenticated version. The American Psychological Association and its Council of Editors disclaim any responsibility or liabilities for errors or omissions of this manuscript version, any version derived from this manuscript by NIH, or other third parties. The published version is available at www.apa.org/pubs/journals/ccp

This publication has been reviewed and approved by the Publications Committee of the Collaborative Longitudinal Personality Disorders Study.

1

Participants were assigned to a primary PD diagnostic group on the basis of a semi-structured diagnostic interview (DIPD-IV; Zanarini et al, 1996) with support from at least one of two other PD instruments. The MDD comparison group consisted of participants who met criteria for current MDD according to the SCID-I/P, had no more than 2 criteria of any PD diagnosis, and less than 15 PD criteria in total.

2

The CLPS co-occurrence patterns (e.g. Grilo et al., 2005) are similar to those reported for other clinical samples (Becker et al., 2000) thus increasing confidence in the generalizability of this sample to other clinical samples.

3

The LIFE has served as the primary measure for major longitudinal studies of psychopathology (e.g. Bruce et al., 2005) and has demonstrated good-to-excellent reliability (Warshaw et al., 2001). The LIFE developers and official training staff at the Brown site trained and certified interviewers across sites and provided ongoing training and consultation for the interview and ratings. These methods maintain long-term reliability and prevent drift (Warshaw et al., 2001).

4

The complete 6 point rating scale for MDD: PSR=1, no symptoms; PSR=2, residual symptoms, but less than full diagnostic criteria; PSR=3, mild symptoms, partial remission; PSR=4, marked symptoms but not full diagnostic criteria; PSR=5, symptoms meeting full diagnostic criteria; PSR=6, severe symptoms and dysfunction.

5

Previous research by Traupman et al. (2009) identified a subset of NEO-PI-R items to identify IPC octants, and established the internal consistency of the NEO-PI-R IPC octant scales and their goodness of fit to circumplex structure. Previous research using the NEO-PI-R IPC octant scales reported Cronbach alphas ranging from .58 (JK) to .77 (BC).

6

As is commonly the case in LPA modeling, the AIC and BIC differ in the model they suggest is best. Moving vertically down the columns in Table 2 the AIC and BIC decrease as profiles are added (suggesting improved fit), with the BIC increasing again after 6. However, the increase to seven profiles is very small. Here the BIC supports a model with fewer profiles as compared to the AIC. This is due to the fact that the BIC provides a more conservative estimate of fit than the AIC because of the more stringent penalty imposed for more estimated parameters. The eight-profile solution resulted in some very small classes (i.e., less than 2% of the sample), and therefore was not given serious consideration. We closely examined the classes that emerged from the six- and seven-profile solutions using the circumplex modeling techniques. In the seven profile solution, a smaller (n = 8) Cold (DE Octant) class emerged that was not overlapping but highly redundant with a larger class. Therefore, we chose to retain the six-profile solution based on both fit and class structure.

7

It is important to note that the angular locations of each group as defined by a circular mean will differ slightly from the angular displacement given by the structural summary method. The reason is that circular means are calculated using only angular locations and not the vector length from the origin of the circle. By not taking vector length into account, all angles are accorded equal weight in the equation. The structural summary method accounts for data that not only differ in angular location but also vector length, thus according differing weights to each subject’s angle when calculating the overall displacement for the group. In defining groups based on circular statistics some of the information given by the structural summary method is lost, but what is gained is the ability to statistically compare separate groups (Wright et al., 2009).

References

  1. American Psychiatric Association. Diagnostic and statistical manual of mental disorders (4th ed.) Washington DC: Author; 1994. [Google Scholar]
  2. Ansell EB, Pincus AL. Interpersonal perceptions of the Five-Factor Model of personality: An examination using the structural summary method for circumplex data. Multivariate Behavioral Research. 2004;39:167–201. doi: 10.1207/s15327906mbr3902_3. [DOI] [PubMed] [Google Scholar]
  3. Becker DF, Grilo CM, Edell WS, McGlashan TH. Comorbidity of borderline personality disorder with other personality disorders in hospitalized adolescents and adults. American Journal of Psychiatry. 2000;157:2011–2016. doi: 10.1176/appi.ajp.157.12.2011. [DOI] [PubMed] [Google Scholar]
  4. Borkovec TD, Newman MG, Pincus AL, Lytle R. A component analysis of cognitive-behavioral therapy for generalized anxiety disorder and the role of interpersonal problems. Journal of Consulting and Clinical Psychology. 2002;70(2):288–298. [PubMed] [Google Scholar]
  5. Bruce SE, Yonkers KA, Otto MW, Eisen JL, Weisberg RB, Keller MB. Influence of psychiatric comorbidity on recovery and recurrence in generalized anxiety disorder, social phobia, and panic disorder: A 12-year prospective study. American Journal of Psychiatry. 2005;162:1179–1187. doi: 10.1176/appi.ajp.162.6.1179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cain NM, Pincus AL, Grosse Holtforth M. Interpersonal subtypes in social phobia: Diagnostic and treatment implications. Journal of Personality Assessment. 2010;92:514–528. doi: 10.1080/00223891.2010.513704. [DOI] [PubMed] [Google Scholar]
  7. Costa PT, McCrae RR. Revised NEO personality inventory (NEO-PI-R) and NEO five-factor inventory (NEO-FFI): Professional manual. Odessa, FL: PAR; 1992. [Google Scholar]
  8. First MB, Spitzer RI, Gibbon M, Williams JBW. Structured Clinical Interview for DSM-IV Axis I Disorders – Patient Version (SCID I/P) New York: New York State Psychiatric Institute; 1996. [Google Scholar]
  9. Grilo CM, Sanislow CA, Shea MT, Skodol AE, Stout RL, McGlashan TH. Two-year prospective naturalistic study of remission from major depressive disorder as a function of personality disorder comorbidity. Journal of Consulting and Clinical Psychology. 2005;73:78–85. doi: 10.1037/0022-006X.73.1.78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Grilo CM, Stout RL, Markowitz JC, Sanislow CA, Ansell EB, McGlashan TH. Personality disorders predict relapse after remission from an episode of Major Depressive Disorder: A 6-year prospective study. Journal of Clinical Psychiatry. doi: 10.4088/JCP.08m04200gre. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gunderson JG, Shea MT, Skodol AE, McGlashan TH, Morey LC, Keller MB. The Collaborative Longitudinal Personality Disorders Study: development, aims, design, and sample characteristics. Journal of Personality Disorders. 2000;14:300–315. doi: 10.1521/pedi.2000.14.4.300. [DOI] [PubMed] [Google Scholar]
  12. Joiner TE, Timmons KA. Gotlib IH, Hammen CL. Handbook of depression 2nd ed. NY: Guilford: 2009. Depression and its interpersonal context; pp. 322–339. [Google Scholar]
  13. Keller MB, Lavori PW, Friedman B, Nielsen E, Andreasen NC. The Longitudinal Interval Follow-up Evaluation: A comprehensive method for assessing outcome in prospective longitudinal studies. Archives of General Psychiatry. 1987;44:540–548. doi: 10.1001/archpsyc.1987.01800180050009. [DOI] [PubMed] [Google Scholar]
  14. Keller MB, Lavori PW, Mueller TI, Endicott J, Coryell W, Shea MT. Time to recovery, chronicity, and levels of psychopathology in major depression: A 5-year prospective follow-up of 431 subjects. Archives of General Psychiatry. 1992;49:809–816. doi: 10.1001/archpsyc.1992.01820100053010. [DOI] [PubMed] [Google Scholar]
  15. Keller MB, Shapiro RW, Lavori PW, Wolfe N. Relapse in major depressive disorder : analysis with the life table. Archives of General Psychiatry. 1982;39:911–915. doi: 10.1001/archpsyc.1982.04290080031005. [DOI] [PubMed] [Google Scholar]
  16. Kessler RC, Berglund P, et al. Lifetime prevalence and ago of onset distribution of DSM-IV disorders in the National Comorbidity Survey Replication (NCS-R) Journal of the American Medical Association. 2005;289:3095–3105. [Google Scholar]
  17. Kornstein SG, Schatzberg AF, Thase ME, Yonkers KA, McCullough JP, Keitner GI, et al. Gender differences in chronic major and double depression. Journal of Affective Disorders. 2000;60:1–11. doi: 10.1016/s0165-0327(99)00158-5. [DOI] [PubMed] [Google Scholar]
  18. Leary T. Interpersonal diagnosis of personality. New York: Ronald Press; 1957. [Google Scholar]
  19. Mongrain M, Lubbers R, Struthers W. The power of love: Mediation of rejection in roommates of dependents and self-critics. Personality and Social Psychology Bulletin. 2004;30:94–105. doi: 10.1177/0146167203258861. [DOI] [PubMed] [Google Scholar]
  20. Muthén LK, Muthén BO. Mplus User’s Guide. Sixth Edition. Los Angeles, CA: Muthén&Muthén; 1998–2010. [Google Scholar]
  21. Pincus AL, Wilson KR. Interpersonal variability in dependent personality. Journal of Personality. 2001;69:223–251. doi: 10.1111/1467-6494.00143. [DOI] [PubMed] [Google Scholar]
  22. Pincus AL, Wright AGC. Interpersonal diagnosis of psychopathology. In: Horowitz LM, Strack S, editors. Handbook of Interpersonal Psychology: Theory, Research, Assessment, and Therapeutic Interventions. New York: Wiley; 2010. pp. 359–381. [Google Scholar]
  23. Traupman EK, Smith TW, Uchino BN, Berg CA, Trobst KK, Costa PT. Interpersonal circumplex octant, control, and affiliation scales for the NEO-PI-R. Personality and Individual Differences. 2009;47:457–463. doi: 10.1016/j.paid.2009.04.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Warshaw MG, Dyck I, Allsworth J, Stout RL, Keller MB. Maintaining reliability in a long-term psychiatric study:An ongoing inter-rater reliability-monitoring program using the longitudinal interval follow-up evaluation. Journal of Psychiatric Research. 2001;35:297–305. doi: 10.1016/s0022-3956(01)00030-9. [DOI] [PubMed] [Google Scholar]
  25. Widiger TA, Smith GT. Pervin L, John OP, RW . Handbook of personality: Theory and research. 3rd Ed. New York: Guilford: 2008. Personality and psychopathology; pp. 743–769. [Google Scholar]
  26. Wright AGC, Pincus AL, Conroy DE, Hilsenroth MJ. Integrating methods to optimize circumplex description and comparison of groups. Journal of Personality Assessment. 2009;91:311–322. doi: 10.1080/00223890902935696. [DOI] [PubMed] [Google Scholar]
  27. Zanarini MC, Frankenburg FR, Sickel AE, Yong L. Diagnostic Interview for DSM-IV Personality Disorders (DIPD-IV) Belmont, MA: McLean Hospital; 1996. [Google Scholar]
  28. Zanarini MC, Skodol AE, Bender D, Dolan R, Sanislow C, Gunderson JG. The Collaborative Longitudinal Personality Disorders Study: reliability of axis I and II diagnoses. Journal of Personality Disorders. 2000;14:291–299. doi: 10.1521/pedi.2000.14.4.291. [DOI] [PubMed] [Google Scholar]
  29. Zuroff DC, Mongrain M, Santor DA. Conceptualizing and measuring personality vulnerability to depression: Comment on Coyne and Whiffen (1995) Psychological Bulletin. 2004;130:489–511. doi: 10.1037/0033-2909.130.3.489. [DOI] [PubMed] [Google Scholar]

RESOURCES