Antagonistic Narcissism in Women with Borderline Personality Disorder: A Case-Control Study

Abstract

Objective

Borderline personality disorder (BPD) is characterized by instability in self-image, emotions, and relationships. Features such as (auto)aggression, devaluation, and impulsivity indicate proximity to narcissistic traits, especially antagonistic aspects. While its links to grandiose and vulnerable narcissism are established, the role of antagonistic narcissism (AN) remains unclear. This study investigates AN in BPD by comparing female BPD patients with a diagnostically heterogeneous clinical control group without personality disorders and by examining its associations with symptom severity, self-harm, aggression, interpersonal problems, and empathy.

Methods

51 female BPD patients and 51 clinical control patients completed the Narcissistic Admiration and Rivalry Questionnaire and measures of borderline symptomatology, aggression, interpersonal problems, and empathy. Group comparisons were conducted using analyses of covariance, and Spearman correlations examined relationships between clinical characteristics. Interpersonal tendencies of AN were analyzed via the Structural Summary Method (SSM).

Results

BPD patients showed higher levels of AN compared to the control group (η² = 0.04), though this difference was only marginally significant (p = 0.050). AN correlated positively with overall aggression (r = 0.34, p < 0.05), particularly verbal aggression (r = 0.43, p < 0.01). SSM analysis positioned AN within the domineering-vindictive quadrant of the interpersonal circumplex, characterized by high dominance and low affiliation. No significant correlation was found with empathy.

Conclusion

Findings suggest a potential elevation of AN in females with BPD and its associations with aggression and interpersonal dysfunction, highlighting the complexity of narcissistic traits in BPD and the need for further research.

Plain Language Summary

Borderline personality disorder (BPD) often involves unstable feelings about oneself and others. It can include behaviors like anger and impulsiveness that may overlap with features of narcissism, which involves seeing oneself as more important than others and acting in antagonistic (hostile) ways. Understanding these links is important because it can help improve how professionals approach therapy for people with BPD. In our study, we looked at “antagonistic narcissism” (AN), which refers to behaving in aggressive, competitive, and unfriendly ways that can stem from feeling superior. We compared 51 female patients with BPD to 51 other female patients receiving treatment for different issues. All of them filled out questionnaires measuring BPD symptoms, narcissistic traits, aggression, problems in getting along with others, and empathy (the ability to understand and share another’s feelings). We found that women with BPD showed slightly more AN than those without BPD. Higher levels of AN were related to increased aggression and more trouble getting along with others, particularly through being more dominant and less cooperative. However, AN did not seem to be linked to levels of empathy. These results suggest that women with BPD may have more narcissistic traits linked to aggression and difficult relationships. Recognizing these connections could help guide future research and improve treatments for individuals with BPD.

Introduction

Borderline personality disorder (BPD) is characterized by a pervasive pattern of instability in self-image, emotions, and interpersonal relationships [1–3]. Clinical phenomena, such as the devaluation of others, leading to recurring interpersonal problems, as well as (auto-)aggression and impulsivity suggest a conceptual overlap with narcissistic traits, especially antagonistic aspects [1]. Despite extensive theoretical and clinical analyses of borderline and narcissistic pathologies [1], empirical research on their relationship remains scarce [4–6]. This gap may be partly attributed to the multifaceted nature of narcissism, which has led to divergent and partially inconsistent conceptualizations [7, 8]. While personality psychology focuses on “normal” narcissism as a dimensionally distributed personality trait, clinical theory, research, and practice are concerned with pathological forms such as the narcissistic personality disorder (NPD) [9–11]. Additionally, recent theories of narcissism have introduced hierarchical models, distinguishing between grandiose and vulnerable narcissism at the primary level [10]. Grandiose narcissism is marked by an inflated self-image and entitlement, whereas vulnerable narcissism is characterized by a depleted self-image and interpersonal hypersensitivity [10]. At a more detailed level, the trifurcated model distinguishes grandiose and vulnerable narcissism by their shared and distinct features: antagonism/rivalry, agentic extraversion, and narcissistic neuroticism [10, 12]. While agentic extraversion is specific to grandiose narcissism and narcissistic neuroticism to vulnerable narcissism, antagonism – also referred to as antagonistic narcissism (AN) – is present in both forms, given their shared trait of low agreeableness [12]. AN, being primarily interpersonal, is characterized by self-defensive behaviors, such as striving for supremacy (affective-motivational), devaluation of others (cognitive), and aggressiveness (behavioral) [13]. The role of AN in BPD warrants clarification, as it shares conceptual overlaps with NPD across the domains of psychopathology (i.e., HiTOP) and personality pathology (i.e., AMPD). Within the HiTOP, BPD is primarily positioned within the internalizing spectrum, specifically under the distress subfactor. However, certain features – particularly those related to interpersonal dysfunction – also overlap with the antagonistic-externalizing spectrum, where NPD is predominantly located [14, 15]. Similarly, the DSM-5 Alternative Model for Personality Disorders (AMPD) defines antagonism broadly, encompassing traits like hostility, manipulativeness, deceitfulness, grandiosity, and callousness. However, it explicitly recognizes only hostility as a core feature of BPD [2, 16, 17]. It is debatable whether the described overlaps pertain solely to antagonistic behaviors or also extend to underlying narcissism (i.e., AN) in a strict sense. Empirical research suggests distinct links between vulnerable and grandiose narcissism and BPD pathology [4–6, 18]. For instance, a study among 65 BPD patients using the Pathological Narcissism Inventory (PNI) indicated an association of grandiose narcissism with the number of diagnostic criteria for both NPD and BPD, while vulnerable narcissism was only associated with BPD [4]. In a study of 189 community adults using the Five-Factor Narcissism Inventory-Short Form to assess grandiose and vulnerable narcissism, BPD traits were significantly associated with both dimensions (r_Grandiose = 0.24; r_Vulnerable = 0.63), with a notably stronger link to vulnerable narcissism [18]. Similarly, in a sample of 1,023 students, vulnerable narcissism (measured with the PNI) was associated with non-suicidal self-injury, a core feature of BPD [19]. A comparison of NPD (N = 49) and BPD patients (N = 32) using the Personality Inventory for DSM-5 (PID-5) identified emotion dysregulation and antagonism as central features of both disorders [5].

In sum, empirical studies on the relation between BPD and narcissism have primarily examined either the grandiose vs. vulnerable dimensions of narcissism or on the comorbidity and phenomenological overlap between categorically defined BPD and NPD. As DSM-5 criteria for NPD focus primarily on grandiose traits [8], they may fail to capture possible associations between BPD and other facets of narcissism emphasized by more recent models. These theories propose antagonism as the linking element between grandiose and vulnerable narcissism [12]. A deeper understanding of the relation between AN and BPD features might be clinically helpful, as antagonistic behaviors in BPD patients present significant challenges to the therapeutic alliance [20].

Therefore, this study aimed to investigate AN in BPD by comparing patients with BPD to a clinical control group (CC) without BPD. Given the high comorbidity of BPD with other mental disorders [3], the CC consisted of patients diagnosed with disorders commonly associated with BPD, including mood, anxiety, somatoform, and eating disorders [2, 3]. To account for gender-specific differences in the associations between vulnerable and grandiose narcissism, we focused exclusively on female patients [4, 21]. Specifically, we aimed to (a) examine whether AN levels differ between BPD and CC patients while controlling for general psychopathology, as individuals with BPD commonly exhibit elevated levels of depression, anxiety, and overall distress, which could confound the observed group differences. Based on previous research, we hypothesized that BPD patients would show higher AN levels compared to the CC, reflecting the documented antagonistic traits in BPD [4, 7]. Furthermore, we aimed to (b) explore the relationship between AN and BPD pathology indicators, including borderline symptom severity and self-harm. Given the theoretical distinction between antagonistic traits and core borderline pathology, we expected only weak or inconsistent associations between AN and BPD symptom severity indicators, as antagonistic traits represent stable personality dimensions, whereas BPD pathology is characterized by affective instability and impulsivity [3]. Given our primary focus on within-group associations and the high heterogeneity of the clinical control group in both diagnoses and symptom severity, we restricted these analyses to the BPD sample to ensure more interpretable and meaningful findings. Lastly, we sought to (c) investigate associations between AN and key clinical variables, namely, aggression, interpersonal problems, and empathy. Because AN primarily reflects interpersonal antagonism [10, 13], we predicted positive correlations with aggression and interpersonal problems and a negative correlation with empathy.

Methods

Participants and Data Collection

The study sample comprised 51 female patients diagnosed with BPD and a clinical control group (CC) of 51 female patients, matched for age, and years of education. Participants were recruited from the Clinic for Psychosomatic Medicine and Psychotherapy at the University of Rostock. BPD Diagnoses were made by clinical experienced board-certified psychiatrists and confirmed using the German version of the Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD) [22], administered by the first author. All participants underwent diagnostic assessment with the German version of the Mini-DIPS [23]. Exclusion criteria included age under 18 years, body mass index below 18 kg/m², neurological diseases, psychotic symptoms, mania, drug or alcohol dependence or active substance use, NPD diagnosis, or an estimated IQ below 70, assessed by the German Multiple-Choice Vocabulary Test (MWT-B) [24]. Participants provided written informed consent and completed self-report measures. Of 119 eligible participants, 11 were excluded during diagnostic assessment for not meeting inclusion criteria (online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000545761).

Measures

Narcissistic Admiration and Rivalry Questionnaire

The Narcissistic Admiration and Rivalry Questionnaire (NARQ) is based on the eponymous theoretical model (NARC), which conceptualizes the maintenance of a grandiose self through two distinct pathways: admiration – characterized by striving for uniqueness, grandiose fantasies, and charm – and rivalry – marked by striving for supremacy, devaluation of others, and aggressiveness [13]. The NARC model aligns with the three-factor models of narcissism, linking narcissistic rivalry with antagonism and narcissistic admiration with agentic extraversion [10, 12]. In this study, when referring to AN, we specifically mean the narcissistic rivalry subscale. The NARQ consists of 18 items on a six-point Likert scale ranging from 1 (“strongly disagree”) to 6 (“strongly agree”), assessing agentic narcissism (narcissistic admiration; e.g., “I will someday be famous”) and AN (narcissistic rivalry; e.g., “I want my rivals to fail”). The questionnaire exhibits strong psychometric properties [25]. For the BPD sample, McDonald’s ω ranged from 0.90 for the admiration subscale to 0.79 for the rivalry subscale. In the CC sample, McDonald’s ω was 0.88 for admiration and 0.83 for rivalry.

Borderline Symptom List Short Version

The Borderline Symptom List short version (BSL-23) [26] is a self-report instrument consisting of 23 items that assess the severity of borderline symptoms experienced over the past week. Items are rated on a five-point Likert scale from 0 (“not at all”) to 4 (“very strong”) and are based on DSM-IV/DSM-5 diagnostic criteria. Since the short form does not include self-harm behavior, we supplemented it with the self-harm subscale from the BSL-95 version. The questionnaire demonstrated good psychometric properties, with McDonald’s ω for the total score being 0.92 in the BPD sample and 0.90 in the CC sample. For the self-harm subscale, ω values were 0.82 for the BPD group and 0.70 for the CC group.

Symptom Checklist-9 Short Version

The Symptom Checklist-9 short version (SCL-K-9) [27] is a brief self-report instrument developed to assess general psychological distress. It consists of nine items rated on a five-point Likert scale ranging from 0 (“not at all”) to 4 (“extremely”), providing a global severity index of current psychopathology. In our study, McDonald’s ω was 0.74 in the BPD sample and 0.82 in the CC group.

Aggression Questionnaire

The German version of the Aggression Questionnaire (AQ) [28] is a self-report measure that assesses dispositional tendencies toward anger and aggression with four subscales: physical aggression (e.g., “If somebody hits me, I hit back”), verbal aggression (e.g., “I often find myself disagreeing with people”), anger (e.g., “I have trouble controlling my temper”), and hostility (e.g., “I wonder why sometimes I feel so bitter about things”). The AQ comprises 29 items rated on a five-point Likert scale from 1 (“extremely uncharacteristic of me”) to 5 (“extremely characteristic of me”). McDonald’s ω for the total score in the BPD sample was 0.85, while in the CC sample it was 0.81. The subscale ω values in the BPD sample ranged from 0.69 for anger to 0.85 for physical aggression, and in the CC sample they ranged from 0.63 for verbal aggression to 0.84 for hostility.

Inventory of Interpersonal Problems-32

The Inventory of Interpersonal Problems-32 (IIP-32) consists of 32 items rated on a 5-point Likert scale from 0 (“not at all”) to 4 (“very much”) [29]. Participants were asked to indicate how much each statement described their interpersonal difficulties. Based on the interpersonal circumplex model [29], each item belongs to one of the eight octants (i.e., domineering, vindictive, cold, socially inhibited, nonassertive, exploitable, self-sacrificing, and intrusive) along the axes agency and communion. The questionnaire demonstrated satisfactory to good psychometric properties [30]. In our sample, McDonald’s ω ranged from 0.47 (vindictive) to 0.75 (domineering) for the subscales and was 0.80 for the IIP-32 total score in the BPD sample. For the CC sample, ω values ranged from 0.54 (domineering) to 0.82 (cold) for the subscales and were 0.83 for the IIP-32 total scores, indicating moderate internal consistencies in both samples.

Interpersonal Reactivity Index

Empathy was assessed by the Interpersonal Reactivity Index (IRI) [31], focusing on the subscales of empathic concern and perspective taking, which capture affective and cognitive empathy, respectively. These two domains were chosen due to their representation of central empathy dimensions particularly relevant in BPD (32–33). Participants responded to 8 items on a five-point Likert scale from 0 (“never”) to 4 (“always”). In the present study, both subscales showed acceptable internal consistencies in both samples, with McDonald’s ω in the BPD sample for empathic concern being 0.76 and 0.86 for perspective taking. In the CC sample McDonald’s ω values were 0.63 for empathic concern and 0.78 for perspective taking.

Data Analysis

All statistical analyses were conducted using R, version 4.2.3 [32], with the R code available online at https://osf.io/b3cw7/. The only study to date that investigated differences in antagonism between diagnostic groups including BPD reported a partial eta squared (η² = 0.10) for antagonism across groups [7]. Using standard conversion formulas, this corresponds to an effect size of approximately d ≈ 0.67. Additionally, the authors reported a standardized contrast (θ = 0.60, 95% CI [0.18, 1.03]) for the BPD vs. no PD comparison. Given that θ serves as a standardized effect size similar to Cohen’s d in two-group comparisons, we conservatively approximated d ≈ 0.6 as the basis for our power analysis. On this basis, we conducted a priori power analyses using GPower [33] for analyses of covariance (ANCOVA), as it was the most complex statistical test planned in our study. The results suggested that recruiting at least 46 participants per group would be sufficient to detect medium-sized effects (d ≈ 0.6) with acceptable power (1 − β ≈ 0.8). We based our sample size on these prior effect magnitude estimates to ensure adequate power for detecting similar differences, should they exist. Chi-square tests were used to compare diagnostic frequencies between subsamples. Internal consistencies of all measures were assessed using McDonald’s ω. We calculated Spearman’s correlation coefficients for all measures. To further examine the association between narcissism and interpersonal dysfunction, we employed the Structural Summary Method (SSM) [34]. This method allows to project both NARQ dimensions (i.e., rivalry, admiration) onto the circumplex structure defined by IIP-32, yielding angular displacement (θ), amplitude (D), and model fit (R²) values. Higher displacement values indicate that the variable aligns well with a specific interpersonal style, while amplitude reflects the strength of this association.

Sociodemographic variables and psychopathology (SCL-K-9) were compared using independent sample t tests. Effect sizes for these comparisons are reported as Cohen’s d (d ≈ 0.2 = small, d ≈ 0.5 = medium, d ≈ 0.8 = large). To examine group differences in all remaining primary outcomes, we conducted ANCOVA, controlling for psychopathology as measured by the SCL-K-9. For each ANCOVA, we assessed the homogeneity of regression slopes to ensure the assumptions were met. Effect sizes for ANCOVA results are reported as partial eta squared (η²), with values interpreted according to standard conventions: small (0.01), medium (0.06), and large (0.14). Statistical significance was set at p < 0.05 for all analyses.

Results

Women with BPD and CC did not differ significantly in age or years of education. As expected, women with BPD scored significantly higher on measures of borderline pathology (BSL-23 total score, self-harm, and number of BPD criteria) compared to the CC (Table 1). Additionally, BPD patients reported higher general psychopathology and had more comorbid diagnoses, as assessed by the Mini-DIPS. A detailed breakdown of comorbidities is available in the supplementary material (online suppl. Table S2).

Table 1.

Demographic and clinical characteristics of BPD (N = 51) and CC sample (N = 51)

	BPD	CC	Statistics
	M (SD)	M (SD)	t	p	d
Age	33.27 (10.22)	35.76 (12.70)	1.09	0.278	0.2
Years of education	13.86 (2.15)	14.49 (1.92)	1.56	0.123	0.3
BPD criteria (SCID-5-PD), n	6.84 (1.72)	0.35 (0.87)	−24.01	<0.001	4.8
Psychopathology (SCL-K-9)	0.39 (0.27)	0.20 (0.25)	−3.78	<0.001	0.8
Diagnoses, n	4.02 (2.01)	1.88 (1.38)	−6.27	<0.001	1.2

Controlled for psychopathology (SCL-K-9)	BPD	CC	Statistics
	M (SD)	M (SD)	F	p	η2
Borderline pathology (BSL-23)	62.69 (17.81)	43.35 (11.87)	51.62	<0.001	0.34
Self-harming behavior (BSL)	2.52 (1.07)	1.62 (0.63)	30.50	<0.001	0.24
Narcissism (NARQ)	2.00 (0.74)	1.96 (0.63)	0.09	0.772	0
NARQ – rivalry	1.95 (0.65)	1.70 (0.62)	0.05	0.050	0.04
NARQ – admiration	2.06 (0.99)	2.23 (0.83)	0.93	0.337	0.01
Aggression (AQ)	78.04 (13.57)	60.75 (10.97)	49.70	<0.001	0.33
AQ – physical aggression	17.43 (5.29)	13.96 (3.54)	15.09	<0.001	0.13
AQ – verbal aggression	12.69 (4.11)	10.49 (2.49)	10.57	0.002	0.10
AQ – anger	21.16 (4.43)	16.53 (4.04)	30.08	<0.001	0.23
AQ – hostility	26.76 (5.90)	19.76 (6.42)	33.64	<0.001	0.25
Interpersonal problems (IIP-32)	2.82 (0.43)	2.52 (0.45)	11.42	<0.001	0.10
IIP-32 – domineering (PA)	1.91 (0.86)	1.37 (0.38)	16.77	<0.001	0.14
IIP-32 – vindictive (BC)	1.80 (0.61)	1.53 (0.57)	5.37	0.023	0.05
IIP-32 – cold (DE)	2.91 (0.85)	2.09 (0.83)	23.69	<0.001	0.19
IIP-32 – socially inhibited (FG)	3.39 (0.86)	2.96 (0.96)	5.95	0.017	0.06
IIP-32 – nonassertive (HI)	3.22 (0.89)	3.32 (0.95)	0.32	0.576	0
IIP-32 – exploitable (JK)	3.35 (0.92)	3.40 (0.94)	0.07	0.796	0
IIP-32 – self-sacrificing (LM)	3.27 (0.92)	3.21 (0.86)	0.15	0.698	0
IIP-32 – intrusive (NO)	2.69 (0.99)	2.31 (0.80)	4.48	0.037	0.04
Empathy (IRI)
IRI – empathic concern	14.59 (3.37)	14.73 (2.28)	0.06	0.810	0
IRI – perspective taking	13.16 (3.47)	14.20 (2.58)	2.92	0.090	0.03

SCL-K-9, Symptom Checklist-9 short version; BPD, borderline personality disorder; CC, clinical controls; BSL-23, short version of the Borderline Symptom List; SCID-5-PD, Structured Clinical Interview for DSM-5 Personality Disorders; NARQ, Narcissistic Admiration and Rivalry Questionnaire; AQ, Aggression Questionnaire; IIP-32, Inventory of Interpersonal Problems-32; IRI, Interpersonal Reactivity Index.

Group Differences in AN

To test our first hypothesis, we examined group differences in NARQ rivalry and admiration. Results showed a marginally significant difference in NARQ rivalry, with BPD patients reporting higher levels than CC patients (p = 0.048, η² = 0.04). However, no significant difference was found for NARQ admiration.

Associations between AN and BPD Indicators

Spearman correlations between NARQ rivalry and admiration, BPD indicators, and clinical measures in the BPD sample are presented in Table 2. Consistent with our second hypothesis, NARQ rivalry showed only weak and nonsignificant associations with BPD pathology indicators, including borderline symptom severity (BSL-23), self-harming behavior (BSL-95 dimension), and number of BPD criteria (SCID-5-PD).

Table 2.

Intercorrelations between narcissism, BPD indicators, and clinical measures in the BPD sample (N = 51)

	NARQ – rivalry	NARQ – admiration
Psychopathology (SCL-K-9)	−0.12	−0.28 a
Borderline pathology (BSL-23)	−0.21	−0.34 a
Self-harming behavior (BSL)	−0.20	−0.30 a
No. of BPD criteria (SCID-5-PD)	−0.03	−0.10
Aggression (AQ)	0.34 a	0.10
AQ – physical aggression	0.22	0.03
AQ – verbal aggression	0.43 b	0.39 b
AQ – anger	0.26	0.16
AQ – hostility	0.05	−0.17
Interpersonal problems (IIP-32)	−0.01	−0.20
IRI – empathic concern	−0.23	−0.04
IRI – perspective taking	−0.10	0.12

BPD, borderline personality disorder; SCL-K-9, Symptom Checklist-9 short version; BSL-23, short version of the Borderline Symptom List; SCID-5-PD, Structured Clinical Interview for DSM-5 Personality Disorders; NARQ, Narcissistic Admiration and Rivalry Questionnaire; AQ, Aggression Questionnaire; IIP-32, Inventory of Interpersonal Problems-32; IRI, Interpersonal Reactivity Index.

^a p < 0.05.

^b p < 0.01.

Associations between AN and Clinical Variables

Consistent with our third hypothesis, NARQ rivalry was moderately correlated with aggression, particularly verbal aggression. Contrary to our expectations; however, NARQ rivalry did not show significant correlations with the IRI total score as well as empathic concern or perspective taking. The complete correlation matrix for the BPD sample is available in the supplementary materials (online suppl. Fig. S3). To further explore the interpersonal characteristics of AN in BPD, we applied the SSM. The results indicated that NARQ rivalry was primarily located within the vindictive-domineering (BC-PA) quadrant of the interpersonal circumplex (θ = 108.49°, 95% CI [80.41–131.41°], D = 0.398, R² = 0.791). In contrast, NARQ admiration aligned with the domineering-intrusive (PA-NO) quadrant (θ = 79.06°, 95% CI [45.17–98.36°], D = 0.402, R² = 0.895) (Fig. 1).

Fig. 1.

Projecting NARQ rivalry and NARQ admiration on the interpersonal circumplex model (IIP-32) in the BPD sample (N = 51). NARQ, Narcissistic Admiration and Rivalry Questionnaire; IIP-32, Inventory of Interpersonal Problems-32; PA, domineering; NO, intrusive; LM, self-sacrificing; JK, exploitable; HI, nonassertive; FG, socially inhibited; DE, cold; BC, vindictive.

Discussion

To our knowledge, this is the first study to investigate AN in BPD patients. Three key findings emerged: (a) AN was higher in females with BPD compared to a clinical control group (CC) with other mental disorders typically found in BPD [2, 3], though the observed difference was marginally significant, suggesting the need for cautious interpretation; (b) AN showed only weak and nonsignificant associations with BPD pathology indicators, including borderline symptom severity, self-harm, and the number of BPD criteria, and (c) AN was positively associated with aggression and interpersonal dysfunction (i.e., high dominance and low affiliation), but not with empathy.

Our finding of group differences with a small effect size for AN in BPD compared to CC without any personality disorder (PD) replicates and extends previous research on narcissism in BPD, suggesting that not only vulnerability but also grandiosity may be relevant [18]. However, our results indicate that antagonistic aspects, rather than agentic ones, play a role, as no significant differences were found for the Admiration scale [5]. Moreover, our findings align with the classification of BPD in both the HiTOP and AMPD, which propose only a limited overlap between BPD and antagonistic traits [15–17]. In a related vein, AN was not interrelated with any of the three BPD indicators (i.e., the BSL-23 total score, self-harm, and the number of BPD criteria). Similarly, a study of 142 patients from a psychiatric outpatient clinic specializing in PDs found low and insignificant correlations between the broader antagonism domain, its lower order traits (as defined by the AMPD), and the number of BPD criteria [35]. While some studies have reported no meaningful associations between grandiose and vulnerable narcissistic traits and self-harm [36], other findings have been contradictory [19]. This raises the question of whether individuals with BPD exhibit greater grandiose narcissism or simply higher levels of broader antagonism. While our data do not provide a definitive answer, it is important to highlight that both the NARC and trifurcated models conceptualize AN as primarily serving the maintenance of a grandiose self-concept [12, 13]. However, in the context of BPD, it is tempting to speculate that AN may function less as a means of self-enhancement and more as a response to reactive interpersonal dysfunction. Given that BPD is characterized by heightened sensitivity to rejection and intense emotional dysregulation [1–3, 18], antagonistic behaviors in BPD may be more situationally driven rather than reflecting a stable grandiose self-concept.

In line with our expectations, AN showed a significant, albeit small-to-moderate, correlation with aggression in BPD patients, particularly verbal aggression as measured by the AQ. This aligns with the conceptualization of AN as an interpersonal form of antagonism, characterized by devaluation of others and behaviors oriented toward supremacy at the cost of others [10, 13]. Consistently, verbal and physical aggression, captured by the AQ, were associated with antagonism in a sample of 122 adult outpatients with BPD [37]. However, the temporal and causal relations between narcissistic traits and aggression, especially hostility, remain complex and warrant further investigation [38]. In line with the well-known interpersonal difficulties observed in individuals with BPD [39], we found that female patients with BPD had significantly more interpersonal problems than our CC, particularly in domains involving low affiliation and high dominance. Correspondingly, the SSM results indicated that AN was primarily situated within the vindictive-domineering (BC-PA) quadrant of the interpersonal circumplex, aligning with interpersonal styles characterized by low affiliation and high dominance. This is in line with prior research indicating a moderate association between pathological narcissism and interpersonal problems psychiatric outpatients [40]. Moreover, a study among university students reported that narcissistic grandiosity was associated with a specific interpersonal profile, marked by being overly domineering in relationships, while narcissistic vulnerability was linked to a broader range of interpersonal problems [41]. Notably, these interpersonal profiles were differentially associated with the experience and expression of anger, the affective component of aggression. This suggests a complex and dynamic interplay between dimensions of narcissism, particularly antagonism, aggression, and impaired interpersonal functioning [38, 39].

While extensive evidence points to deficient cognitive and increased affective empathy in BPD individuals compared to healthy controls [42, 43], we did not find differences in empathy, as measured with the IRI, between BPD patients and those with other mental disorders. This suggests that empathy impairments can be broadly attributed to psychopathology rather than to BPD pathology. Interestingly, AN did not correlate with empathy deficits, contrary to our hypothesis based on the diagnostic criterion “lack of empathy” in NPD (DSM-5) [21]. However, recent research has provided a more nuanced understanding of the multifaceted relationship between empathy and narcissism [43, 44]. Specifically, it has been suggested that affective empathy deficits in narcissism are characterized by “affective dissonance”, i.e., contradictory affects in response to someone else’s feelings [45]. Regarding cognitive empathy in narcissism, some scholars argue it is preserved [45], whereas others suggest that it is dysfunctional rather than deficient [44, 46]. Our results enhance the understanding of AN traits in BPD, emphasizing the importance of distinguishing AN from broader narcissistic or borderline symptomatology. The distinct associations between AN, aggression, and interpersonal problems, without a connection to BPD indicators, underscore the necessity of further research.

Despite its strengths, including a priori power analyses and an age- and education-matched CC, our study has several limitations: first, we only compared AN between BPD and a CC with no PD. As we did not include a control group of patients with PDs other than BPD, it remains uncertain whether our findings are specific to BPD and NPD or may apply more broadly to personality pathology. This notion is supported by Fossati et al. [5], who found no significant differences in DSM-5 antagonism trait scores between BPD, NPD, and any other PD, suggesting that antagonism may not be unique to BPD or NPD, but rather a transdiagnostic feature of PDs. In addition, as our sample consisted solely of female patients, it is unclear whether the results can be generalized to men with BPD. Second, our sample size calculations were based on an effect size (d = 0.6) from a single study and aimed for the minimal acceptable power of 0.8. While our sample size of 51 per group meets basic statistical requirements, it may be underpowered for detecting more subtle effects. This impacts the robustness and generalizability of our findings, and future studies should consider larger sample sizes. Third, the cross-sectional design precludes causal inferences about the relationship between AN and BPD. Longitudinal studies are needed to explore their dynamic interplay, e.g., whether AN exacerbates BPD symptoms over time or arises as a consequence of BPD pathology. The low internal consistency of the IIP-32 vindictive subscale (ω = 0.47) presents a measurement limitation that may affect the stability and interpretability of related findings. Consequently, results involving this subscale should be interpreted with caution. Importantly, as the NARQ solely assesses grandiose dimensions of narcissism (rivalry and admiration), our study does not provide insights into vulnerable narcissism. Therefore, we can only conclude that the agentic (extraverted) aspect of grandiose narcissism does not appear to be characteristic of BPD, whereas its antagonistic dimension may play a more relevant role. Given that AN manifests differently in grandiose and vulnerable narcissism, our findings contribute only to the understanding of AN driven by the grandiose domain. Future research should further explore AN in BPD, particularly by distinguishing whether it arises from narcissistic mechanisms or is more closely tied to affective instability and interpersonal hypersensitivity. In the same vein, it may be fruitful to examine the associations between AN and specific BPD criteria, such as unstable relationships, self-image disturbances, and affective instability, to gain a more fine-grained understanding. Finally, regarding the current debate about appropriate concepts of pathological narcissism, our study only relied solely on the NARQ to assess AN. It is important to note that this questionnaire originates from personality psychology and has been primarily validated through studies involving healthy individuals. Integrating additional measures, such as the Pathological Narcissism Inventory with its subscales Exploitativeness and Entitlement rage [21], which have been already evaluated in clinical populations, could provide a more comprehensive understanding of antagonistic traits in BPD. Additionally, potential associations with vulnerable narcissism remain unexplored in our study. Another limitation might be the exclusive reliance on self-report measures.

Conclusion

AN was found to be marginally elevated in BPD compared to a CC group but was not significantly related to BPD indicators. Instead, AN was associated with aggression and interpersonal dysfunction, characterized by high dominance and low affiliation. Theoretically, considering AN may help clarify the heterogeneity of BPD, as evidence suggests that hostility, as an aspect of antagonism, is normal or even reduced in less severe cases but increases significantly in more severe cases [47]. Whether these differences reflect an underlying narcissistic dynamic or simply broader antagonistic trait remains an open question. While the AMPD identifies hostility as the only relevant antagonistic trait in BPD, our findings, along with others [4, 5], suggest that a wider range of antagonistic traits may be relevant. This aligns with HiTOP, which conceptualizes personality pathology dimensionally rather than through rigid diagnostic categories, emphasizing the need to further examine antagonism as a key component in specific BPD profiles.

Statement of Ethics

This research was approved by the Ethics Committee of the Medical Faculty at the University of Rostock (case No. A2021-0074) and was conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent.

Conflict of Interest Statement

The authors report that there are no competing interests to declare.

Funding Sources

P.W. received funding from the Landesgraduiertenförderung Mecklenburg-Vorpommern. The funders had no role in study design, data collection, data analysis, data interpretation, or writing of this article.

Author Contributions

P.W. conceptualized the study, performed data analyses, and drafted the first manuscript. C.S. coordinated data acquisition. P.W., N.K., C.H.L., and C.S. reviewed the manuscript critically and contributed to the manuscript.

Funding Statement

P.W. received funding from the Landesgraduiertenförderung Mecklenburg-Vorpommern. The funders had no role in study design, data collection, data analysis, data interpretation, or writing of this article.

Data Availability Statement

The R code has been made publicly available at the Open Science Framework and can be accessed at https://osf.io/b3cw7/. The raw data of this study are available from C.S. upon reasonable request.