This work may not be copied, distributed, displayed, published, reproduced, transmitted, modified, posted, sold, licensed, or used for commercial purposes. By downloading this file, you are agreeing to the publisher’s Terms & Conditions.


Randomized Controlled Trials of Olanzapine Treatment of Borderline Personality Disorder: Two Similar Studies With Different Results

H. George Nurnberg, MD

Published: April 5, 2011

See related article by Zanarini, et al.

Randomized Controlled Trials of Olanzapine Treatment of Borderline Personality Disorder: Two Similar Studies With Different Results

Borderline personality disorder, with origins in the psychodynamic literature, was a controversial addition to the official psychiatric nomenclature of DSM-III and became the most frequently diagnosed and extensively studied Axis II category. As defined in DSM nosology, borderline personality disorder is a prototypal construct with the following characteristics: (1) different polythetic criteria combinations constitute category membership heterogeneity (5 of 9 criteria lead to 225 potential combinations), (2) some individual criteria or combinations have a greater weight for making a diagnosis, (3) category members may simultaneously meet defining criteria for other Axis I and Axis II disorders, (4) homogeneity for diagnostic covariates is not expected, and (5) boundaries are imprecise. Borderline personality disorder is one of the more common and difficult-to-treat disorders in psychiatric practice and adversely affects treatment outcomes of other Axis I disorders when concurrent with them. Although there are no Food and Drug Administration (FDA)-approved pharmacologic treatments for DSM-IV borderline personality disorder, approximately 80% of patients in outpatient and inpatient treatment take medication (eg, antipsychotics, antidepressants, mood stabilizers, and sedatives) for the different facets of the pathological spectrum, which includes affective instability, impulsivity, dissociative states, cognitive difficulties, or disturbed interpersonal relations.1 Treatment recommendation guidelines on pharmacologic treatment of borderline patients have predominantly relied on "expert" opinion (which diverges from more recently reported meta-analyses of small studies) and revision.1,2

The randomized placebo-controlled trial3 of fixed-dose (2.5 mg/d and 5-10 mg/d) olanzapine treatment of borderline personality disorder reported in this issue of the Journal is the second report on 2 large, concurrently conducted industry-sponsored pharmacotherapy trials using essentially the same protocols and outcome measures and differing primarily on fixed-dose versus flexible-dose medication administration for treatment.

The results of the current fixed-dose trial3 with 451 subjects ( Identifier NCT000880364) need to be considered together with the prior 2.5- to 20-mg/d flexible-dose report5 published in 2008 involving 314 patients ( Identifier NCT000916506). The olanzapine 5- to 10-mg/d fixed-dose group, in comparison to the 2.5-mg/d fixed-dose and placebo groups, showed a statistically significant difference in the baseline-to-endpoint outcome measure (Zanarini Rating Scale for Borderline Personality Disorder [ZAN-BPD]7 score) for improvement of overall borderline psychopathology (−8.5 vs −8.0 vs −6.8, respectively; P = .01), with a small clinical effect size (0.29; 95% CI, 0.06-0.52).3 The magnitude of change (mean improvement in baseline-to-endpoint ZAN-BPD total scores) versus placebo was less in the prior 2.5- to 20-mg/d flexible-dose trial (0.31) than in this 5- to 10-mg/d fixed-dose trial (1.71), without a significant difference between treatment groups (−6.56 for olanzapine vs -6.25 for placebo; P = .66).5 In the current fixed-dose trial, olanzapine 2.5 mg/d did not produce significant results (change in ZAN-BPD score) relative to placebo (−8.0 vs −6.8; P = .06) but had a 0.19 (95% CI, -0.04 to 0.42)3 effect size that exceeded the effect size of 0.03 (95% CI, −0.20 to 0.25)5 in the 2.5- to 20-mg/d flexible-dose trial. The differences in treatment outcomes between the 2 reports are interesting and are not explained simply by the differences in medication dose; 5 mg/d was the most common daily dose of olanzapine during both double-blind trial periods,3,5 with 7.09 mg/d being the mean modal dose in the 2.5- to 20-mg/d flexible-dose group,5 compared to a 6.7-mg/d mean modal dose in the 5- to 10-mg/d fixed-dose group (distinct from the 2.5-mg/d fixed-dose group).3

On secondary last-observation-carried-forward outcome measures, the rate of response (response defined as a ≥ 50% decrease from baseline ZAN-BPD total score) for the olanzapine 5- to 10-mg/d fixed-dose treatment relative to the placebo treatment was 73.6% vs 57.8%, which was significant, with a number needed to treat (NNT) of 6 (95% CI, 4-21; P = .03),3 and greater than the response rate for the 2.5- to 20-mg/d flexible-dose olanzapine treatment relative to placebo treatment (64.7% vs 53.5%; P = not significant), with an NNT of 9 (95% CI, 7-24).5 For other secondary outcome measures, such as the Modified Overt Aggression Scale irritability, Sheehan Disability Scale family life, and Symptom Checklist-90-Revised hostility scores, both studies were generally consistent for significant mean baseline-to-endpoint improvements in the olanzapine treatment groups; however, there were no significant depression score changes between groups according to the Montgomery-Asberg Depression Rating Scale.

Measures of individual items for the 9 DSM-IV criteria for borderline personality disorder showed significant last-observation-carried-forward mean scale reductions compared to placebo for the inappropriate anger item, with effect sizes of 0.26 (95% CI, 0.08-0.44; P = .002) and 0.13 (95% CI, 0.05-0.41; P = .08) for the olanzapine 5- to 10-mg/d and 2.5-mg/d fixed-dose groups, respectively, and 0.23 (95% CI, 0.05-0.41; P = .01) for the 2.5- to 20-mg/d flexible-dose group.5 The 5- to 10-mg/d fixed-dose design showed additional, significantly reduced mean change for affective instability and paranoid dissociation items, with smaller effect sizes (d = 0.18 and d = 0.28, respectively), but similar findings did not occur in the prior flexible-dose trial. Additionally, in the flexible-dose trial, the placebo group showed a greater reduction (improvement) (not statistically significant) for suicidality/self-harm in comparison to the olanzapine 2.5- to 20-mg/d group (−0.6 vs −0.3, respectively). However, in the fixed-dose trial, the mean change of −0.3 for the 2.5-mg/d and 5- to 10-mg/d groups was greater than the mean change of −0.2 for the placebo group on suicidality/self-harm.

Assessment for effectiveness must include evaluation of adverse events and premature treatment discontinuation. Premature discontinuation due to adverse effects favors placebo relative to active drug treatment outcomes and becomes an additional consideration vis-à -vis benefit to understand differences between treatment outcomes and whether it is beneficial to treat with medication or not. In 12 weeks of double-blind treatment, a 35% overall premature discontinuation rate occurred among the 3 fixed-dose groups3 compared to a 48% rate in the flexible-dose groups, with a number needed to harm (NNH) of 8 (95% CI, 6-21; P = .02).3,5 Incidence of ≥ 7% weight gain, a concern with use of second-generation antipsychotic agents, was significant for fixed-dose olanzapine 5-10 mg/d (30.6%) or 2.5 mg/d (20.3%) over placebo (4.8%; P = .002, P = .045, respectively), with a relative risk (RR) of 4.26 (95% CI, 1.93-9.38) and an NNH of 6.45 (95% CI, 4.38-12.23) for olanzapine 2.5 mg/d and an RR of 6.42 (95% CI, 2.99-13.77) and an NNH of 3.88 (95% CI, 2.94-5.71) for olanzapine 5-10 mg/d.3 Therefore, the cost-benefit ratio is derived from likelihood of harm (NNH = 4) divided by likelihood of benefit (NNT = 6.3), yielding a cost-benefit ratio of 0.63 (95% CI, 0.46-0.67), which indicates a 3:2 greater likelihood of ≥ 7% weight-gain harm over the modest treatment response (effect size, 0.29; 95% CI, 0.06-0.52) with fixed-dose olanzapine 5-10 mg/d. Treatment-emergent weight gain ≥ 7% over baseline incidence was significant and greater for individuals treated with flexible-dose olanzapine 2.5-20 mg/d relative to placebo (34.2% vs 2.6%, respectively; P = .001), with an NNH of 2.6 (95% CI, 2-3), resulting in a less effective cost-benefit consideration of 0.33 compared to fixed-dose olanzapine 5-10 mg/d.5

It is understandable that the efficacy data and adverse effects data together would not support application for an FDA-approved indication to treat overall global borderline personality disorder psychopathology with olanzapine. However, it must be underscored that the failure to prove efficacy should not be taken as proof of failure of olanzapine to treat not all but some borderline patients effectively. These 2 studies provide a rich and valuable new data source to inform clinicians in practice on more specific applications available for treatment of borderline patients. However, one size does not fit all, and by further examination and understanding of the data, patient selection can be better matched to selected potential treatment benefits.

For example, absent a gold standard for identifying borderline personality disorder, patient selection for inclusion in the studies required borderline personality disorder diagnosis according to the Diagnostic Interview for DSM-IV Personality Disorders (DIPD-IV), together with a minimum total severity score of 9 on the ZAN-BPD,7 a semistructured interview derived from the DIPD-IV with anchored ratings from 0 (no symptoms) to 4 (severe symptoms) on each of 9 items corresponding to the 9 DSM-IV criteria for borderline personality disorder. Given inherent construct heterogeneity with differences of up to 50%8 in diagnostic agreement between different scales and instruments for diagnosis of borderline personality disorder (eg, the Structured Clinical Interview for DSM-IV Axis II Personality Disorders, the DIPD-IV, the Personality Disorder Examination, and others), patient selection can be expected to result in highly variable borderline personality disorder samples with different severity ratings that will influence outcomes across studies. For example, heterogeneity of illness severity in this trial,3 with a treatment effect size of 0.29 (95% CI, 0.06-0.52), indicates that half of the patients were lower-half treatment-effect responders, at an effect size of 0.06-0.29, while the other half had a more robust treatment effect of 0.30-0.52.

More stringent entry criteria (eg, higher severity cutoff) or selection of subjects with more specific target criteria (eg, high anger ratings and high paranoid ratings) can result in higher differential treatment effectiveness than would occur by selecting a borderline group with other symptoms (eg, fear of abandonment, impulsivity, and feelings of emptiness) or with lower severity, for which olanzapine efficacy effectiveness (increased NNT) would be lower than the adverse-event impact (lower NNH)—and comparatively less effective overall. It is to be expected in borderline personality disorder, in which 5 of 9 criteria lead to 225 potential combinations and heterogeneity, that large randomized controlled trials will show wide variability in demonstrating global efficacy. Broad heterogeneity poses a challenge to demonstrating global treatment efficacy by a drug with specific effects.

In conclusion, for clinicians in practice who are treating patients with borderline personality disorder, it would have been more parsimonious to have results of these 2 important trials presented together in the same journal instead of 2+ years apart in different journals. While the evidence from the 2 randomized controlled trials does not support an indication for overall efficacy of olanzapine for borderline personality disorder treatment, the evidence is consistent with suggestions that pharmacotherapy should be targeted at specific symptoms and that second-generation antipsychotics can be effective for treating a number of core symptoms and their associated psychopathology. The updated clinical guidelines by Oldham9 reflect a suggestion for more specific treatments of borderline personality disorder by focusing on matching different drug classes to more specific symptom clusters rather than focusing on one drug for all. An evidence-based psychopharmacology is emerging, enabling clinicians to be able to better match patients by subsets of specific criteria to different classes of agents (mood stabilizers, selective serotonin reuptake inhibitors, atypical antipsychotics) rather than relying on opinion-based guidelines that need revision in concordance to new data on pharmacologic treatment of borderline personality disorder. This industry-sponsored study provides an important contribution supporting clinicians’ use of atypical treatment for borderline personality disorder in the absence of an FDA indication. The planned revision for DSM-5 suggests an additional caveat: the expected revisions in diagnostic criteria can be expected to influence whether the generalizability of prior drug study results will be concordant and apply to patients identified in the new classification system.

Drug names: olanzapine (Zyprexa).

Author affiliation: Department of Psychiatry, University of New Mexico School of Medicine, Albuquerque.

Potential conflicts of interest: None reported.

Funding/support: None reported.


1. American Psychiatric Association. Practice Guideline for the Treatment of Patients with Borderline Personality Disorder. Am J Psychiatry. 2001;158(10 suppl):1-52. PubMed

2. Binks CA, Fenton M, McCarthy L, et al. Pharmacological interventions for people with borderline personality disorder. Cochrane Database Syst Rev. 2006;(1):CD005653. PubMed

3. Zanarini MC, Schulz SC, Detke HC, et al. A dose comparison of olanzapine for the treatment of borderline personality disorder: a 12-week randomized, double-blind, placebo-controlled study. J Clin Psychiatry. Epub ahead of print April 5, 2011.

4. Eli Lilly and Company. Clinical Study Summary: Study F1D-MC-HGKK. Efficacy and safety of olanzapine in patients with borderline personality disorder: a randomized double-blind comparison with placebo. Verified February 18, 2011.

5. Schulz SC, Zanarini MC, Bateman A, et al. Olanzapine for the treatment of borderline personality disorder: variable dose 12-week randomised double-blind placebo-controlled study. Br J Psychiatry. 2008;193(6):485-492. PubMed doi:10.1192/bjp.bp.107.037903

6. Eli Lilly and Company. Clinical Study Summary: Study F1D-MC-HGKL. Efficacy and safety of olanzapine in patients with borderline personality disorder: a randomized, flexible-dose, double-blind comparison with placebo. Verified February 21, 2011.

7. Zanarini MC, Vujanovic AA, Parachini EA, et al. Zanarini Rating Scale for Borderline Personality Disorder (ZAN-BPD): a continuous measure of DSM-IV borderline psychopathology. J Pers Disord. 2003;17(3):233-242. PubMed doi:10.1521/pedi.

8. Kavoussi RJ, Coccaro EF, Klar HM, et al. Structured interviews for borderline personality disorder. Am J Psychiatry. 1990;147(11):1522-1525. PubMed

9. Oldham JM. Guideline watch: practice guideline for the treatment of patients with borderline personality disorder. Arlington, VA: American Psychiatric Association; 2005. Verified February 21, 2011.

Submitted: January 6, 2011; accepted January 13, 2011.

Online ahead of print: April 5, 2011 (doi:10.4088/JCP.11com06844).

Corresponding author: H. George Nurnberg, MD, Department of Psychiatry, University of New Mexico School of Medicine, 2400 Tucker NE, Albuquerque, NM 87131 (

J Clin Psychiatry 2011;72(10):1363-1365 (doi:10.4088/JCP.11com06844).

© Copyright 2011 Physicians Postgraduate Press, Inc.

Related Articles

Volume: 72

Quick Links: