Exploration of the Value of Optical Coherence Tomography (OCT) and OCT Angiography in Differentiating Between Major Depressive Disorder and Borderline Personality Disorder

ABSTRACT

Background

Major depressive disorder (MDD) and borderline personality disorder (BPD) are psychiatric conditions with overlapping clinical features but distinct pathologies. This study aims to explore the microvascular changes in superficial retinal capillary plexus (SRCP) in patients between MDD and BPD using Optical Coherence Tomography (OCT) and Optical coherence tomography angiography (OCTA), as well as determine their clinical diagnostic value.

Methods

The study involved 35 MDD patients, 36 BPD patients, and 36 healthy controls (HCs). OCTA and OCT were used to assess macular vessel density (MVD) and peripapillary vessel density (PVD), retinal nerve fiber layer (RNFL), and central retinal thickness (CRT). The study also evaluated the correlation of these ocular parameters with clinical symptoms.

Results

Both MDD and BPD patients showed significant reductions in MVD, RNFL and CRT compared to HCs (all p < 0.05). BPD patients exhibited more pronounced decreases in the foveal area, RNFL and CRT than MDD patients (all p < 0.05). These ocular changes were associated with BPD symptom severity (all p < 0.05). The RNFL and CRT demonstrated potential as diagnostic markers for BPD (AUROCs of 0.625 and 0.695, p < 0.05).

Conclusions

OCT and OCTA might differentiate between MDD and BPD by identifying specific ocular changes, suggesting distinct neurobiological mechanisms for each disorder. These technologies hold promise for improving diagnostic accuracy.

1. Introduction

Major Depressive Disorder (MDD) and Borderline Personality Disorder (BPD) are two common mental disorders. MDD is characterized by persistent sad mood, anhedonia, and cognitive impairment (Uher et al. 2014), while BPD is characterized by emotional instability, impulsivity, Non‐suicidal self‐injury behavior (NSSI) and interpersonal sensitivity (Bohus et al. 2021). However, it is worth noting that these two mental disorders have similar clinical manifestations, for example, people with MDD also have descriptions of emotional instability (Thompson et al. 2012), and people with BPD may also have chronic emotional dysregulation (Chapman et al. 2022). This can lead to very easy confusion and misdiagnosis of the two. MDD and BPD can both be considered as dysfunction of emotion processing (Rao and Broadbear 2019). Neuroimaging studies have found that these two diseases may share some common neurobiological mechanisms, such as prefrontal cortex dysfunction and hyperactivity in the limbic system (De la Pena‐Arteaga et al. 2021).

Early and accurate differential diagnosis of MDD and BPD is crucial because the treatment methods and prognosis for these two diseases are different. For example, MDD is typically treated with antidepressants and cognitive behavior therapy while BPD requires more specific psychotherapeutic approaches such as Dialectical Behavior Therapy (DBT) and Good Psychiatric Management (GPM) (Gunderson et al. 2018; Rao and Broadbear 2019). Misdiagnosis of BPD as MDD and receiving incorrect treatment may lead to treatment inefficacy, worsening of the condition, or potential side effects (Paris 2018). Therefore, early and accurate differential diagnosis of MDD and BPD can ensure that patients receive appropriate treatment and management, thereby improving their quality of life and prognosis. Currently, the differentiation between the two mainly relies on subjective psychiatric interviews, which can be prone to misdiagnosis, especially in the early stages of the disease (Rao and Broadbear 2019).

Comparing the neuroimaging data of MDD and BPD directly can help to outline their specificities and common features. Magnetic resonance imaging (MRI) studies have found some structural and activation difference in the limbic system and frontal regions in BPD compared to MDD (Cao et al. 2022; De la Pena‐Arteaga et al. 2021; Depping et al. 2018), but the number of neuroimaging studies directly comparing the two is limited. There is an urgent need to find a new objective biomarker to discriminate between MDD and BPD, in order to better understand the shared and different neural mechanisms of these two diseases, improve diagnostic accuracy, and develop more effective targeted therapies.

Some studies have found that both MDD and BPD patients may have chronic inflammatory responses and neuronal damage, which may be related to abnormalities in the hypothalamic‐pituitary‐adrenal (HPA) axis (Drews et al. 2019; Mikulska et al. 2021). These chronic inflammations have been shown to lead to changes in retinal thickness and cerebral microvascular changes (Sekeryapan et al. 2022; Tan et al. 2020). As an extension of the brain neurons, the retina provides an easily observable window. The physiological and anatomical characteristics of the retina are similar to those of the brain, and changes in the structure and microvascular parameters of the retina may reflect similar changes in the brain (Haddow et al. 2018; Hart et al. 2016; Muhammed 2018). Exploring changes in the structure and microvasculature of the retina provides a new perspective for us to identify the common and different biological mechanisms of MDD and BPD, and to find an objective biological marker.

Optical Coherence Tomography (OCT) and Optical coherence tomography angiography (OCTA) utilizes optical coherence tomography technology to generate images through interferometric reflection of light, measuring tissue density and structure at each level of the retina, and displaying various sizes of vascular networks including capillaries (Spaide et al. 2018). Compared to traditional retinal imaging techniques such as fluorescein angiography, OCT and OCTA do not require contrast agents to be injected and does not cause any adverse reactions in patients (Lains et al. 2021). As a non‐invasive imaging technology, OCT and OCTA has the advantage of evaluating vascular abnormalities in a variety of diseases, and therefore has broad applications in the medical field (Kashani et al. 2017). In recent years, OCT and OCTA has been explored as a biomarker for psychiatric disorders (bipolar disorder, MDD and schizophrenia) (Gediz et al. 2021; Koman‐Wierdak et al. 2021; Liu et al. 2022; Xiao et al. 2023). The studies of MDD found that compared to the healthy control group, MDD patients had lower retinal vascular density (including radial peripapillary capillary density, superficial and deep capillary plexus density), thinner choroidal thickness, and worse visual acuity (Liu et al. 2022; Xiao et al. 2023). Currently, there is a lack of OCT and OCTA studies on BPD. Furthermore, the differences in retinal vascular and structural changes between MDD and BPD are unknown.

This study is the first to use OCT and OCTA to explore the differences between MDD and BPD, recruiting 35 MDD patients, 36 BPD patients and 36 healthy control subjects (HCs), and to explore the commonalities and differences of retinal indicators between MDD and BPD. In order to exclude the influence of the comorbidities on the interpretation of the results, we deliberately excluded the condition of MDD and BPD comorbidities when we included the subjects, that is, if the subjects met the diagnostic criteria of both MDD and BPD, we excluded them. We predict that some differences between MDD and BPD patients can be found in the specific retinal indicators.

2. Methods

2.1. Participants

Our study was approved by the institutional review board (IRB: 202301005) of our hospital, and all participants provided written informed consent. A total of 112 participants were enrolled in this research, comprising 37 patients with MDD, 38 patients with BPD and 37 HCs. After excluding 2 MDD patients, 2 BPD patients and 1 HCs due to poor image quality caused by eye movement, the analysis included 35 MDD patients, 36 BPD patients and 36 HCs. Participants with MDD and BPD were consecutively recruited from the Psychiatric Clinics at the Mental Health Center of Xiangya Hospital between December 2021 and May 2023. HCs were recruited through local advertisements during the same period. Refer to Figure S1 for enrollment process.

For the MDD group, participants had to be between 18 and 45 years old and meet the diagnostic criteria for MDD in the DSM‐IV (Lobbestael et al. 2011). To be eligible for the BPD group, participants had to be between 18 and 45 years old, meet the diagnostic criteria for BPD outlined in the DSM‐IV (Lobbestael et al. 2011), and have stable symptoms for more than two years with no better explanation by DSM‐IV Axis I psychiatric or neurodevelopmental disorders. HCs had to be between 18 and 45 years old with no history of any psychiatric disorders or psychotropic medication.

Exclusion criteria for all participants were a history of schizophrenia spectrum disorder, bipolar spectrum disorder, post‐traumatic stress disorder, neurodevelopmental disorders, alcohol or drug dependence, neurological disorders, or an intelligence quotient (IQ) of ≤ 80. The MDD group also had to exclude any comorbidities of personality disorders of Class A and Class B, while the BPD group specially had to rule out comorbid MDD. All participants had not taken any psychotropic, vascular, or ophthalmic medications for at least 4 months. Additionally, all participants were instructed to avoid alcohol or psychotropic substances for 24 h prior to the OCTA. Participants excluded from the study exhibited diverse ophthalmic conditions, such as epiretinal membrane, age‐related macular degeneration, glaucoma, macular hole, refractive error beyond ± 3 diopters (D) sphere and ± 2D cylinder, macular drusen, and pathological myopia. Additionally, exclusion criteria encompassed individuals with diabetes and concurrent neurological disorders, uncontrolled hypertension, pulmonary, severe cardiac, renal diseases, hepatic conditions, any types of tumors, or insufficient clinical data.

2.2. Structured Interview

To assess for Axis I mental disorders including MDD, we used the Structured Clinical Interview for DSM‐IV Axis I Disorders (SCID‐I) (Lobbestael et al. 2011). In order to confirm a diagnosis of BPD, we conducted a structured interview using the BPD section of the Structured Clinical Interview for DSM‐IV Personality Disorders (SCID‐II) (Lobbestael et al. 2011). Confirmation of the BPD diagnosis required the presence of five or more of the nine diagnostic items. We administered the same structured interviews to the HCs to rule out any personality disorders and Axis I mental disorders. The final diagnosis was made by a clinical team consisting of two senior psychiatrists (Q.X and F.J) with 10 and 8 years of subspecialty experience in psychiatry, respectively. A comprehensive clinical evaluation was performed, and family history was utilized to collect information about psychiatric disorders in first‐degree relatives of the participants.

2.3. Symptom Assessment

In this study, we utilized three scales: the Self‐Rating Depression Scale (SDS), the Borderline Symptom List‐23 (BSL‐23) and the Difficulties in Emotion Regulation Scale (DERS). The SDS is used to assess depressive mood (Jokelainen et al. 2019). It is a self‐report scale that evaluates the severity of depression in patients. The total score of the SDS ranges from 20 to 80, with higher scores indicating more severe symptoms of depression. The SDS cut‐off is 41. The BSL‐23 is used to measure the severity of BPD symptoms (Bohus et al. 2009). Each item is scored on a scale from 0 (none) to 4 (very strong), with a total score range of 0–92. Higher scores indicate more severe borderline symptoms. According to Kleindienst et al. (2020), the BSL‐23 not only has comparable psychometric properties but also saves time and significantly reduces the burden on participants (Kleindienst et al. 2020). The DERS is used to evaluate emotion regulation abilities (J. Li et al. 2018). It includes multiple subscales covering difficulties such as lack of emotional awareness, lack of emotional clarity, and non‐acceptance of emotional responses. The total score of the DERS ranges from 36 to 180, with higher scores indicating greater difficulties in emotion regulation.

By using the SDS, BSL‐23 and DERS scales, we can comprehensively assess depressive mood and emotion regulation abilities in patients, providing valuable data for exploring the associations between ocular indicators and these psychological characteristics. We also evaluated the participants' Intelligence Quotient (IQ) using the Abbreviated Wechsler Intelligence Scale (Wechsler 1999). A skilled psychiatrist (FJ) with 8 years of experience in psychiatric evaluation conducted the clinical assessment. All interviews and scale evaluations were conducted on the same day as the OCTA examination. Demographic information and clinical symptoms are presented in Table 1.

TABLE 1.

Clinical characteristics of all included eyes.

	MDD (n = 35)	BPD (n = 36)	HC (n = 36)	p‐value
Sex				0.941
Male	20	21	22
Female	15	15	14
Age [y]	26.6 $\pm$ 7.5	25.1 $\pm$ 3.4	25.8 $\pm$ 3.9	0.520
Education	14.7 $\pm$ 2.2	14.5 $\pm$ 2.6	13.6 $\pm$ 0.4	0.075
BCVA [logMAR]	−0.088 $\pm$ 0.07	−0.086 $\pm$ 0.07	−0.100 $\pm$ 0.06	0.545
IOP [mmHg]	15.8 $\pm$ 2.1	15.3 $\pm$ 2.0	15.4 $\pm$ 2.1	0.479
Illness duration [y]	4.0 $\pm$ 2.2	3.5 $\pm$ 2.3	—	0.348
Family history	5	11		0.101
SDS	67.7 $\pm$ 8.9	73.0 $\pm$ 8.6	—	0.009
BSL‐23	43.3 $\pm$ 14.2	68.7 $\pm$ $14.3$	—	0.000
DERS	113.1 $\pm$ 15.5	126.4 $\pm$ 22.4	—	0.005

Abbreviations: BCVA, best corrected visual acuity; BPD, borderline personality disorder; BSL‐23, Borderline Symptom List‐23; DERS, Emotion Regulation Scale; HC, healthy controls; IOP, intraocular pressure; MDD, major depressive disorder; SDS, The Self‐Rating Depression Scale.

2.4. Optical Coherence Tomography Angiography Imaging

To image the microvasculature of the retina in all participants, a technique called optical coherence tomography angiography (OCTA) was utilized. All participants underwent retinal microvascular imaging using the Cirrus 5000 HD‐OCT system (Carl Zeiss Meditec Inc. Dublin, CA, United States). The analysis focused on the macular vessel density (MVD) and peripapillary vessel density (PVD) of the superficial retinal capillary plexus (SRCP) in a 6 mm × 6 mm area centered on the macula and peripapillary region. The vessels included in the analysis ranged from the inner limiting membrane (ILM) to the inner plexiform layer (IPL). In addition, vessel density was measured in nine subfields of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid. The ETDRS grid consisted of a central 1‐mm circle, as well as inner and outer rings with diameters of 3 and 6 mm, respectively, that were further divided into four quadrants (superior, nasal, inferior, and temporal). The software used for OCTA automatically segmented and quantified the foveal avascular zone (FAZ) area. The process of determining central retinal thickness (CRT) included the computation of the mean thickness within the 1 mm ring positioned at the retinal center (Xu et al. 2022). Measurements of the peripapillary retinal nerve fiber layer (RNFL) were acquired through the utilization of an automated algorithm for OCT analysis. In our study, to ensure the accuracy and objectivity of manual image checks, we adopted a standardized double‐rater blind evaluation protocol. Specifically, two independent researchers (with 5 and 7 years of experience in ophthalmic image analysis, respectively) who were unaware of the patients' clinical information and grouping results conducted manual review of all included images. The review criteria mainly covered image clarity (whether the key structures such as the retinal nerve fiber layer and macular fovea could be clearly identified), the presence of obvious artifacts (excluding motion artifacts temporarily), and the integrity of the imaging area. Motion artifacts were quantified via the built‐in motion‐tracking algorithm of the OCT/OCTA device, images were excluded if they had a signal strength index < 7 or displayed visible motion artifacts.

2.5. Statistical Analysis

Statistical analyses were conducted using SPSS version 25.0 (SPSS Inc., Chicago, IL, USA). Gender differences between groups were compared using the Pearson Chi‐square test. For continuous variables with normal distributions, means ± standard deviations and were analyzed using the Student's t‐test, the differences among three groups were assessed using one‐way ANOVA. To compare differences between the subgroups, the results of this study were analyzed with one‐way ANOVA based on LSD Post‐Hoc test. The Benjamini–Hochberg method was applied to calculate adjusted p‐values, aiming to control the false discovery rate (FDR) which was specified at 0.05. Continuous variables that did not follow a normal distribution were expressed as median (25% quartile, 75% quartile) and were analyzed using the Mann‐Whitney U test. Correlations between data were tested using Pearson's correlation. Receiver operating characteristics curves were plotted using Medcalc software, and area under the curve was calculated. Statistical significance was defined as p < 0.05.

3. Results

3.1. Participant Recruitment and Clinical Characteristics

This study included 70 MDD eyes (35 MDD patients), 72 BPD eyes (36 BPD patients)and 72 HCs eyes (36 HCs). There were no significant differences in gender (p = 0.941), age (p = 0.520), education (p = 0.075), illness duration (p = 0.348), intraocular pressure (IOP) (p = 0.479), best‐corrected visual acuity (BCVA) (p = 0.545) and visual acuity (p = 0.759) (Table 1). The mean age of individuals with MDD is 26.6 ± 7.5 years, with the proportion of females is 42.9%. The mean age of individuals with BPD is 25.1 ± 3.4 years, with the proportion of females is 41.7%. In the healthy control group, the mean age is 25.8 ± 3.9 years, with the proportion of females is 38.9% (Table 1).

In a comparative study, significant differences were observed between individuals with MDD and those with BPD. The Self‐Rating Depression Scale (SDS) scores were 67.7 ± 8.9 for MDD and 73.0 ± 8.6 for BPD, indicating more severe symptoms in BPD (p = 0.009). The Borderline Symptom List‐23 (BSL‐23) showed scores of 43.3 ± 14.2 for MDD and markedly higher at 68.7 ± 14.3 for BPD, suggesting more pronounced borderline symptoms (p = 0.000). For the Difficulties in Emotion Regulation Scale (DERS) scores were 113.1 ± 15.5 for MDD and 126.4 ± 22.4 for BPD, again higher in BPD, denoting greater emotional regulation difficulties (p = 0.005). Table 1 provides an overview of the demographic characteristics, ophthalmic features, and clinical parameters of all participants.

3.2. Macular Vessel Parameters

Table 2 presents compared with HC groups, the macular vessel density (MVD) parameters in MDD and BPD patients declined noticeably in all subfields in superficial retinal capillary plexus (SRCP) of all participants, including superior inner (SI), inferior inner (II), nasal inner (NI), temproal inner (TI), superior outer (SO), inferior outer (IO), nasal outer (NO) and temporal outer (TO) area (all p < 0.05). Moreover, quadrantal analyses indicated the fovea area decreasing significantly in the SRCP of the BPD patients in comparison with the MDD group [8.5 $\pm$ $1.5$ mm² in MDD eyes versus $7.8$ $\pm$ 2.8 mm² in BPD patients, p < 0.005]. In addition, results suggested that MDD and BPD patients showed fovea area noticeably declining earlier than other values. As revealed from the results of OCT parameters analysis, the RNFL, CRT showed significantly decreased in BPD patients when compared with MDD patients [106.1 $\pm$ 9.7 mm² of RNFL in MDD eyes versus 103.2 $\pm$ 7.4 mm² of RNFL in BPD patients, p < 0.005] [285.9 $\pm$ 10.8 mm² of CRT in MDD eyes versus 281.6 $\pm$ 12.9 mm² of CRT in BPD patients, p < 0.005]. To sum up, when comparing BPD patients directly with MDD patients, BPD individuals showed more pronounced decreases in the MVD of the fovea, RNFL, and CRT. However, the MVD of the fovea, RNFL, and CRT decreased significant in BPD patients compared to healthy controls (p < 0.005), while MDD patients only differed in fovea measures compared to healthy controls (p > 0.005).

TABLE 2.

The OCTA and OCT parameters in the MDD patients, BPD patients and healthy controls (HC) eyes.

	MDD	BPD	HC	p‐value	MDD versus HC	BPD versus HC	MDD versus BPD
Fovea	8.5 $\pm$ $1.5$	$7.8$ $\pm$ 2.8	9.3 $\pm$ 1.6	0.000	0.020	0.000	0.038
SI	17.7 $\pm$ 1.3	17.6 $\pm$ 2.1	18.3 $\pm$ 1.1	0.036	0.036	0.019	0.820
II	17.3 $\pm$ 1.8	17.4 $\pm$ 2.3	18.4 $\pm$ 1.1	0.000	0.000	0.002	0.672
NI	17.9 $\pm$ 1.7	17.8 $\pm$ 2.0	18.5 $\pm$ 1.1	0.021	0.038	0.008	0.584
TI	17.4 $\pm$ 1.8	17.3 $\pm$ 2.5	18.4 $\pm$ 1.0	0.001	0.004	0.001	0.616
SO	18.9 $\pm$ 0.9	18.6 $\pm$ 1.1	18.9 $\pm$ 0.7	0.046	0.043	0.025	0.834
IO	17.9 $\pm$ 1.4	18.0 $\pm$ 1.5	18.8 $\pm$ 0.7	0.000	0.000	0.000	0.618
NO	19.4 $\pm$ 2.3	19.3 $\pm$ 1.0	19.9 $\pm$ 0.7	0.018	0.029	0.008	0.644
TO	16.9 $\pm$ 1.8	16.8 $\pm$ 2.1	17.5 $\pm$ 1.1	0.024	0.043	0.009	0.574
FAZ	0.27 $\pm$ 0.1	0.27 $\pm$ 0.1	0.29 $\pm$ 0.1	0.289	0.198	0.155	0.901
Inner ring	17.6 $\pm$ 1.5	17.5 $\pm$ 2.1	18.4 $\pm$ 0.9	0.003	0.007	0.002	0.663
Outer ring	18.2 $\pm$ 1.1	18.2 $\pm$ 1.1	18.8 $\pm$ 0.6	0.000	0.001	0.000	0.932
RNFL [μm]	106.1 $\pm$ 9.7	103.2 $\pm$ 7.4	106.5 $\pm$ 7.2	0.028	0.765	0.029	0.014
CRT [μm]	285.9 $\pm$ 10.8	281.6 $\pm$ 12.9	288.7 $\pm$ 10.7	0.001	0.155	0.000	0.029

Note: The Benjamini‐Hochberg Adjusted p‐value was calculated to defend the false discovery rate (FDR), and the FDR was set at 0.05.

Abbreviations: BPD, borderline personality disorder; CRT, Central retinal thickness; FAZ, Foveal avascular zone; HC, healthy controls; II, Inferior inner; IO, Inferior outer; MDD, major depressive disorder; NI, Nasal inner; NO, Nasal outer; OCT, Optical Coherence Tomography; OCTA, Optical Coherence Tomography Angiography; RNFL, retinal nerve fiber layer; SI, Superior inner; SO, Superior outer; TI, Temporal inner; TO, Temporal outer.

However, statistically significant difference was not identified in FAZ for comparation of MDD or BPD subgroups with HC (Table 2), moreover, the MVD in all grids in MDD and BPD patients were significantly lower than those in HC. As revealed from the results, statistically significant difference was identified in all grids for comparisons between the MDD group and the HC group, as well as between the BPD group and the HC group (p < 0.05) (Table 2).

3.3. Peripapillary Vessel Density Parameters

The peripapillary vessel density parameters (PVD) parameters of all participants are presented in Table 3. As indicated from the result, no significant difference of PVD was found in all the subfields of the peripapillary scan among three groups (p > 0.05) (Table 3). However, the MVD in all grids in patients with BPD and MDD were lower than those in HC.

TABLE 3.

Peripapillary capillary layer (RPCP) densities in the MDD patients, BPD patients and HC eyes.

	MDD	BPD	HC	p‐value	MDD versus HC	BPD versus HC	MDD versus BPD
Center	$6.2\pm$ 3.4	6.3 $\pm$ 3.6	6.8 $\pm$ 2.8	0.561	0.325	0.392	0.893
SI	18.7 $\pm$ 1.3	19.0 $\pm$ 1.0	19.0 $\pm$ 0.8	0.233	0.149	0.929	0.129
II	18.4 $\pm$ 2.0	18.6 $\pm$ 1.4	18.7 $\pm$ 1.2	0.470	0.223	0.636	0.456
NI	17.9 $\pm$ 2.7	18.5 $\pm$ 1.5	18.6 $\pm$ 1.3	0.082	0.046	0.916	0.060
TI	17.3 $\pm$ 3.1	17.8 $\pm$ 3.4	17.9 $\pm$ 3.0	0.497	0.251	0.751	0.408
SO	18.9 $\pm$ 2.0	19.1 $\pm$ 1.4	19.3 $\pm$ 0.7	0.228	0.086	0.390	0.388
IO	19.2 $\pm$ 1.6	19.3 $\pm$ 1.5	19.3 $\pm$ 0.9	0.849	0.568	0.441	0.794
NO	17.4 $\pm$ 3.4	17.7 $\pm$ 2.1	18.0 $\pm$ 1.5	0.331	0.137	0.473	0.441
TO	18.6 $\pm$ 3.1	18.8 $\pm$ 1.9	19.5 $\pm$ 1.0	0.185	0.163	0.085	0.752
Inner ring	18.2 $\pm$ 1.4	18.5 $\pm$ 1.3	18.5 $\pm$ 1.1	0.273	0.123	0.717	0.238
Outer ring	18.5 $\pm$ 2.3	18.7 $\pm$ 1.5	19.0 $\pm$ 0.7	0.157	0.058	0.234	0.472

Note: The Benjamini‐Hochberg Adjusted p‐value was calculated to defend the false discovery rate (FDR), and the FDR was set at 0.05. one‐way ANOVA based on LSD Post‐Hoc test.

Abbreviations: BPD, borderline personality disorder; HC, healthy controls; II, Inferior inner; IO, Inferior outer; MDD, major depressive disorder; NI, Nasal inner; NO, Nasal outer; SI, Superior inner; SO, Superior outer; TI, Temporal inner; TO, Temporal outer.

3.4. Correlation Between OCT and OCTA Parameters and Severity of BPD Symptoms

The correlation of MVD of patients with BPD with clinical parameters was analyzed, Spearman's correlation coefficients and corresponding p‐value are presented in Figure 1. As indicated from the results, CRT, RNFL, TI grid, SO grid, Outer ring and Inner ring indicated a significant negative correlation with BLS‐23 (p < 0.05). The TI and SO grids were the subfilds of Outer ring and Inner ring.

FIGURE 1.

Correlation analysis between OCT and OCTA parameters and the scores for BLS‐23 in the patients with borderline personality disorder (BPD). (A) Negative correlation between the CRT and BLS‐23. (B) Negative correlation between inner ring and BLS‐23. (C) Negative correlation between outer ring and Negative. (D) Negative correlation between RNFL and BLS‐23. BSL‐23, Borderline Symptom List‐23; CRT, Central retinal thickness; OCT, Optical Coherence Tomography; OCTA, Optical Coherence Tomography Angiography; RNFL, retinal nerve fiber layer.

3.5. Diagnostic Abilities of Vessel Density for MDD and BPD

The areas under the receiver operating characteristic curves (AUROCs) for distinguishing BPD and HC eyes in OCTA were observed for inner ring MVD (0.501), followed by outer ring MVD (0.527), however, the difference were not significant (p > 0.05). OCT were the highest for CRT (0.695), followed by RNFL (0.625) (p < 0.05) (Figure 2).

FIGURE 2.

The areas under the receiver operating characteristic curves (AUROCs) for distinguishing BPD and HC eyes. (A) AUROCs of RNFL. (B) AUROCs of CRT. CRT, Central retinal thickness; RNFL, retinal nerve fiber layer.

4. Discussion

In this study, using Optical Coherence Tomography (OCT) and Optical coherence tomography angiography (OCTA), we compared two psychiatric disorders with similar clinical presentations—Major Depressive Disorder (MDD) and Borderline Personality Disorder (BPD)—and found similarities and differences in ocular indicators between these conditions. Compared with structural neuroimaging techniques such as MRI, OCT/OCTA offers several practical and complementary advantages for clinical and research applications. OCT/OCTA is significantly less costly and time‐consuming than MRI (Windsor et al. 2018), and is suitable for outpatient and community settings, facilitating large‐scale and repeated measurements (Green et al. 2022). The spatial resolution of OCT/OCTA for retinal layers and microvasculature is much higher than conventional neuroimaging, making it particularly sensitive to subtle microstructural and microvascular changes (Green et al. 2022). Furthermore, OCT/OCTA is entirely non‐invasive, requiring no ionizing radiation or intravenous contrast, which makes it well‐suited for longitudinal follow‐up (Yang et al. 2025). Recent developments of portable, low‐cost OCT devices and remote/self‐administered imaging solutions further enhance the accessibility and cost‐effectiveness of retinal monitoring (Song et al. 2021). Nevertheless, retinal imaging should be regarded as a complementary peripheral window into CNS pathology, ideally interpreted alongside MRI, clinical assessments, and other biomarkers (Almonte et al. 2020; Manogaran et al. 2016).

In our comparison of MDD and BPD patients with healthy controls (HC), we observed common decreases in certain ocular indicators, the macular vessel density (MVD) parameters in MDD and BPD patients declined noticeably in all subfields in superficial retinal capillary plexus (SRCP) of all participants, including SI, II,NI,TI,SO,IO,NO and TO area. Moreover, quadrantal analyses indicated the fovea area decreasing significantly in the SRCP of the BPD patients in comparison with the MDD group. In addition, results suggested that MDD and BPD patients showed fovea area noticeably declining earlier than other values. As revealed from the results of OCT parameters analysis, the RNFL, CRT showed significantly decreased in BPD patients when compared with MDD patients. However, statistically significant difference was not identified in FAZ for comparation of MDD or BPD subgroups with HC, moreover, the MVD in all grids in MDD and BPD patients were significantly lower than those in HC. As revealed from the results, statistically significant difference was identified in all grids for comparisons between the MDD group and the HC group, as well as between the BPD group and the HC group.

These ocular changes may reflect an underlying neurobiological mechanism shared between the eye and brain tissues, as both originate from the neuroectoderm, suggesting neurodegenerative changes in both psychiatric conditions, indicating similarities between these diseases. As the retina serves as a “gateway” to the brain, there exists a parallel pathophysiological mechanism between them. OCTA's ability to measure changes in retinal microvasculature may indicate reduced retinal blood flow or microvascular abnormalities, assumed mechanisms in various psychiatric and neurodegenerative diseases (Shin et al. 2020).

Building upon these findings, we also examined the peripapillary vessel density (PVD), which did not show significant differences between patients and controls. This finding suggests that microvascular alterations associated with BPD and MDD may first emerge in the macular region, whereas PVD changes may not represent an early feature. One possible explanation is the anatomical difference: the macula is mainly supplied by the central retinal artery without collateral circulation (Baldoncini et al. 2019), whereas the peripapillary radial peripapillary capillary (RPC) plexus receives dual perfusion from both the central retinal artery and short posterior ciliary arteries (Campbell et al. 2017; Taylor et al. 2024), providing greater hemodynamic compensation. Moreover, the overlapping multilayer structure of the RPC may limit OCTA sensitivity in detecting subtle capillary loss (Jia et al. 2017; Xu et al. 2022). A similar temporal pattern of “macular‐first involvement with delayed peripapillary changes” has been reported in Alzheimer's disease, Parkinson's disease, and schizophrenia (Budakoglu et al. 2021; Jin et al. 2021; Koman‐Wierdak et al. 2021; Xu et al. 2022; Zhang et al. 2019). Thus, we speculate that PVD may be less sensitive to early microvascular damage in BPD and MDD, with reductions becoming more apparent at later disease stages or under conditions of impaired systemic vascular reserve. Alternatively, the non‐significant PVD finding may also reflect insufficient statistical power due to the limited sample size. Future longitudinal studies, incorporating choroidal blood flow and stress‐related biomarkers, are warranted to further clarify these mechanisms.

It is noteworthy that we also found significant differences between the two psychiatric disorders. When comparing BPD patients directly with MDD patients, BPD individuals showed more pronounced decreases in the MVD of the fovea, Retinal Nerve Fiber Layer (RNFL), and Central Retinal Thickness (CRT), suggesting more severe abnormalities in BPD. For all three ophthalmic measures, the MVD of the fovea, RNFL, and CRT decreased in BPD patients compared to healthy controls, while MDD patients only differed in fovea measures compared to healthy controls.

The fovea, a small depression in the macular area of the retina, is responsible for high acuity and central vision. Changes or damage to the fovea can lead to decreased vision and visual distortion (Provis et al. 2013). The foveal avascular zone (FAZ), positioned within the fovea and encircled by the uninterrupted capillary network of the retina, exhibited an absence of capillary structure. These specific areas housed dopaminergic (DA) neurons in the human retina. The reduction in the FAZ area is proposed to result from the impairment of foveal DA neurons, thereby fostering vasculogenesis from the surrounding capillary network (Miri et al. 2015; Xu et al. 2022). A diminished size of the FAZ, coupled with a decrease in foveal MVD, suggests impairment and an incomplete parafoveal capillary arcade in the macula. The key pathological regions in MDD involve abnormal dopaminergic pathways within the midbrain cortical and limbic system (Y. Li et al. 2022). Friedel RO reported that an imbalance in the homeostasis of DA receptors and transporters may contribute to the pathophysiology of BPD disorder (Friedel 2004). These research findings indicate a certain correlation between MDD and BPD and a reduction in intracerebral DA levels, consistent with our study results that the diminished size of the FAZ, coupled with a decrease in foveal vessel density, suggests impairment and an incomplete parafoveal capillary arcade in the macula.

The RNFL contains the neural pathways from the retina to the brain and is crucial for transmitting visual information. Damage or thinning of the RNFL is often associated with vision loss, particularly in neurodegenerative eye diseases like glaucoma (Costello et al. 2008). CRT refers to the thickness of the central area of the retina, encompassing the macula, key to vision function. A decrease in CRT may indicate damage or atrophy of retinal cells, affecting vision quality and sensitivity (Puell et al. 2021). The reduction in these three ocular indicators in BPD patients might reflect damage or functional decline in retinal cells, especially ganglion cells and photoreceptors, part of the central nervous system and playing a crucial role in information processing (Sanes and Masland 2015).

This is the first study to explore OCT and OCTA ocular markers of BPD compared with MDD. Until now, there has been a lack of OCTA studies on BPD. Recent studies suggest abnormal OCT findings in bipolar disorder (BD) patients (Ayik et al. 2022). Although that research primarily focuses on BD, it reveals potential pathological changes in the RNFL of the retina, which might be related to the more pronounced retinal changes observed in BPD, given the clinical similarities such as emotional instability and impulsivity shared between BPD and BD. The reduction in RNFL thickness could indicate the loss of axons in the retina, potentially related to pathological changes in the central nervous system. In fact, thinning of the RNFL has been observed in various neurodegenerative diseases, considered potential biomarkers for conditions like Alzheimer's (Zhang et al. 2019) and Parkinson's diseases (Polo et al. 2016). For the three eye indicators of fovea, RNFL and CRT, BPD not only decreased compared with the control group, but also showed a significant decrease compared with MDD. These findings demonstrate the neurobiological uniqueness of BPD relative to MDD. The specific neurobiological underpinnings of BPD are less understood, but these ocular changes might also reflect systemic neurobiological alterations.

The OCT and OCTA abnormalities are more pronounced in BPD than in MDD, possibly because BPD involves greater mood fluctuations and more pronounced anxiety and agitation (Dixon‐Gordon et al. 2017), which may lead to more severe microvascular damage and more pronounced immune abnormalities. One study highlighted that the characteristics of BPD are associated with elevated blood pressure in response to social stressors, a known cardiovascular risk factor (Grove et al. 2017). Long‐term emotional stress and anxiety may cause the body to be in a chronic stress state and affect the vascular regulatory function (Malan et al. 2023; Ruan et al. 2020). In addition, psychological stress and mood fluctuations have been shown to be associated with increased inflammatory responses, possibly leading to endothelial dysfunction (Toda and Nakanishi‐Toda 2011) and immune abnormalities (O'Leary 1990). The pathophysiology of BPD may involve increased levels of inflammatory cytokines and oxidative stress substances that exacerbate neuronal damage (Forte et al. 2023). Complex interactions between the immune system and the nervous system may play a role in the pathophysiology of BPD, possibly reflected by abnormalities in OCTA indicators.

Our study further found that in BPD patients, the RNFL, CRT, inner ring, outer ring, TI and SO negatively correlate with the Borderline Symptom List‐23 (BSL‐23) scores, a valuable finding. BSL‐23, a standardized tool to assess the severity of symptoms in BPD patients, covers issues related to emotional instability, self‐image, relationships, and self‐harm/suicidal behaviors. It is noteworthy that, in addition to the previously discussed RNFL and CRT, a decline in the inner ring, outer ring, temporal‐inferior (TI), and superior‐outer (SO) may also indicate retinal problems or macular dysfunction (Araki et al. 2017). The negative correlation with RNFL, CRT, inner ring, outer ring, TI and SO indicators might mean that the structural changes in the retina are related to the core symptoms of BPD. For the two ocular indicators RNFL and CRT, the decline is more significant in BPD compared to MDD, and these indicators are closely associated with the severity of symptoms in BPD. This reflects the potential importance of RNFL and CRT in differentiating between MDD and BPD. For MDD, our previous studies have found a negative correlation between the fovea indicator and depressive mood in MDD (Xiao et al. 2023), suggesting that the findings of our study might indicate unique neurobiological alterations distinguishing BPD from MDD.

In our study, we found that RNFL and CRT demonstrate moderate diagnostic performance (AUROC values 0.6–0.7) in the identification of BPD. This result provides only an initial perspective on the potential role of ophthalmic imaging in BPD assessment, as current diagnosis still primarily relies on clinical evaluations and self‐reported symptoms. Although RNFL and CRT may reflect structural and functional changes in neural circuits related to emotional regulation, stress response, and cognitive functions, this interpretation remains speculative because we did not directly measure neurobiological markers such as dopaminergic activity or inflammation. Our findings should be regarded as exploratory. At this stage, RNFL and CRT cannot serve as clinically applicable biomarkers, but they may offer a foundation for future longitudinal and multimodal studies to further evaluate their potential as auxiliary diagnostic indicators for BPD.

This investigation has multiple limitations. It's credibility is diminished by a limited participant number (only 35–36 participants per group), challenges in deciphering data, and the absence of uniform guidelines for chosen metrics, potentially skewing results toward incorrect negatives or positives. The narrow participant range—predominantly young (mean age ∼ 25–26 years), female (∼40%), and recruited from a single Chinese center—fails to sufficiently represent populations (e.g., older adults, males, other ethnicities) or illness durations, limiting generalizability. In addition, participants with common comorbidities (e.g., co‐occurring MDD and BPD) were excluded during recruitment; given the high real‐world comorbidity between MDD and BPD (e.g., MDD with BPD prevalence estimates and high rates of MDD among individuals with BPD (Więdłocha et al. 2024)), this exclusion may reduce the ecological validity of our findings. Although strict exclusion criteria were applied to minimize potential confounding—such as no psychiatric, vascular, or ophthalmic medication use within the past 4 months, and exclusion of major eye diseases, uncontrolled hypertension, diabetes with neurological complications, severe systemic diseases, and recent substance abuse—residual confounding cannot be entirely ruled out. First, prior medication use before the 4‐month window may have long‐term effects on retinal structure. Second, lifestyle factors such as smoking history, sleep quality, and caffeine intake could subtly influence retinal microvasculature. Third, subclinical ophthalmic parameters (e.g., minor refractive errors within inclusion thresholds, differences in axial length) and disease‐related variables (e.g., symptom severity, number of episodes, psychosocial stress exposure) may also affect retinal metrics.

Variations in individual vascular structures and blood circulation could also influence discrepancies in participant responses (Grewal et al. 2019). Our findings didn't capture variations linked to the duration of the condition. Contrasting reports exist regarding the inverse relationship between RNFL thickness and the length of psychiatric disorders, with some studies confirming it (Chu et al. 2012; Mehraban et al. 2016) and others finding no such link (Ascaso et al. 2015; Silverstein et al. 2021). Because of the cross‐sectional design, our study cannot establish causality; thus, the retinal alterations observed should not be interpreted as either a cause or a consequence of psychiatric disorders, and concluding that RNFL or CRT has diagnostic specificity is premature. Longitudinal evidence in depression suggests potential directionality: for instance, thinner baseline RNFL predicted a higher incidence and worsening of clinically significant depressive symptoms over time, independent of demographic and cardiometabolic factors (van der Heide et al. 2021). Similarly, prospective retinal vascular studies have linked baseline microvascular characteristics with subsequent depressive symptom trajectories, although findings regarding arterial and venular calibers remain inconsistent (van Gennip et al. 2024, 2022). These findings raise the possibility that neurodegenerative and microvascular processes may precede or contribute to depressive symptoms in some individuals, but whether such mechanisms extend to our relatively young clinical sample, particularly in BPD, remains uncertain. Relevant literature in BPD is limited, and to our knowledge, no longitudinal OCT/OCTA studies have been published. To clarify the temporal sequence and potential state–trait effects, we plan to conduct larger, multicenter longitudinal studies using repeated OCT/OCTA assessments across different disease stages and treatment exposures. Concluding that RNFL and CRT have diagnostic specificity for BPD is premature at this stage. Future research is needed to validate the diagnostic effectiveness of RNFL and CRT across different populations, as well as their roles in various subtypes of BPD and the disease progression. Combining ocular indicators with traditional psychiatric assessment methods may provide a more comprehensive diagnostic picture for BPD.

5. Conclusion

This study found that compared to healthy controls, patients with MDD and BPD exhibit similar ocular indicator changes in OCT and OCTA studies. However, the damage in BPD is more severe in the fovea, RNFL and CRT compared with MDD, and is associated with core features of BPD. The OCT and OCTA technologies can detect not only changes in the thickness of the RNFL but also abnormalities in retinal microvasculature. Further research is needed to explore the causal relationship between retinal changes and difficulties in emotional regulation in patients with MDD and BPD, and whether these changes could serve as biomarkers for monitoring disease progression or treatment response. Future studies should also investigate whether decreases in retinal indicators like fovea, RNFL and CRT are specific to BPD or if they also occur in other psychiatric disorders, which might help us better understand and differentiate these conditions.

Author Contributions

Qian Xiao: conceptualization, funding acquisition, methodology, validation, writing – original draft preparation, writing – review and editing. Liying Shen: data curation, formal analysis, investigation, validation, writing – original draft preparation, writing – review and editing. Fangling Li: investigation, writing – original draft preparation. Furong Jiang: investigation, writing – review and editing. Bei Xu: conceptualization, resources, supervision, writing – review and editing.

Funding

This study received financial support from the National Natural Science Foundation of China, General Program (Grant 32571264), the Natural Science Foundation of Hunan Province (Grant 2025JJ50591), and the National Natural Science Foundation of China, Youth Fund (Grant 82201702).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Illustrates the enrollment process for participants with major depressive disorder (MDD), borderline personality disorder (BPD) and healthy control (HC) participants.

Xiao, Qian , Shen Liying, Li Fangling, Jiang Furong, and Xu Bei. 2025. “Exploration of the Value of Optical Coherence Tomography (OCT) and OCT Angiography in Differentiating Between Major Depressive Disorder and Borderline Personality Disorder,” International Journal of Methods in Psychiatric Research: e70043. 10.1002/mpr.70043.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.