Attribution of psychiatric manifestations to systemic lupus erythematosus in Chinese patients: A retrospective study

Abstract

BACKGROUND

Not all neuropsychiatric (NP) manifestations in patients with systemic lupus erythematosus (SLE) are secondary to lupus. The clarification of the cause of NP symptoms influences therapeutic strategies for SLE.

AIM

To understand the attribution of psychiatric manifestations in a cohort of Chinese patients with SLE.

METHODS

This retrospective single-center study analyzed 160 inpatient medical records. Clinical diagnosis, which is considered the gold standard, was used to divide the subjects into a psychiatric SLE (PSLE) group (G1) and a secondary psychiatric symptoms group (G2). Clinical features were compared between these two groups. The sensitivity and specificity of the Italian attribution model were explored.

RESULTS

A total of 171 psychiatric syndromes were recorded in 138 patients, including 87 cases of acute confusional state, 40 cases of cognitive dysfunction, 18 cases of psychosis, and 13 cases each of depressive disorder and mania or hypomania. A total of 141 (82.5%) syndromes were attributed to SLE. In contrast to G2 patients, G1 patients had higher SLE Disease Activity Index-2000 scores (21 vs 12, P = 0.001), a lower prevalence of anti-beta-2-glycoprotein 1 antibodies (8.6% vs 25.9%, P = 0.036), and a higher prevalence of anti-ribosomal ribonucleoprotein particle (rRNP) antibodies (39.0% vs 22.2%, P = 0.045). The Italian attribution model exhibited a sensitivity of 95.0% and a specificity of 70.0% when the threshold value was set at 7.

CONCLUSION

Patients with PSLE exhibited increased disease activity. There is a correlation between PSLE and anti-rRNP antibodies. The Italian model effectively assesses multiple psychiatric manifestations in Chinese SLE patients who present with NP symptoms.

Key Words: Systemic lupus erythematosus; Neuropsychiatric systemic lupus erythematosus; Attribution; Italian attribution model; Referral consultation

Core Tip: This retrospective study of 160 Chinese patients with systemic lupus erythematosus (SLE) focused on the attribution of psychiatric manifestations. SLE-related psychiatric syndromes tend to occur in patients with more severe disease activity and were associated with the presence of anti-ribosomal ribonucleoprotein particle antibodies. The Italian attribution model, which was initially developed and tested in European cohorts, also effectively assessed multiple psychiatric manifestations in Chinese SLE patients.

INTRODUCTION

Systemic lupus erythematosus (SLE), when it affects the nervous system, can induce neuropsychiatric (NP) manifestations termed NPSLE, which frequently signifies considerable disease severity[1-3]. The etiopathogenesis of NPSLE remains incompletely understood. Potential pathological mechanisms involve ischemic and neuroinflammatory pathways[4]. Not all NP manifestations in individuals with SLE are secondary to lupus, and other pathological mechanisms or psychosocial factors must be considered[5]. The clarification of the cause of NP symptoms influences therapeutic strategies for SLE; hence, this clarification is a major focus of clinical research[1]. In clinical settings, the attribution of NP symptoms is intricate and typically requires multidisciplinary collaboration. This complexity is also reflected in the study outcomes, which drastically differ among study cohorts, methodologies, and diagnostic modalities, ranging from 14% to 95%[5]. Some of these studies reflect the incidence of NP manifestations rather than NPSLE.

Extensive research has indicated that not all pathologic alterations in NPSLE can be discerned through laboratory examination[5-7]. Thus, clinical diagnosis continues to be regarded as the gold standard for NPSLE. Over the years, specialists have formulated instruments for clarifying NP symptoms in SLE patients. On the basis of the American College of Rheumatology (ACR) Nomenclature of 19 NPSLE syndromes in 1999[8], Ainiala et al[9] differentiated between mild and severe manifestations in 2001, where mild manifestations encompass anxiety, headache, mild depression, mild cognitive impairment, and polyneuropathy unconfirmed by electromyography. The Systemic Lupus Collaborating Clinics (SLICC) attribution model proposed by Hanly et al[10] incorporates the relative chronology of NP symptoms and SLE, the types of NP manifestations, and exclusion/correlation factors. The Italian Study Group model presented by Bortoluzzi et al[11] focused on four dimensions: (1) The chronological association of NP symptoms with the diagnosis of SLE; (2) The incidence of minor or major NP symptoms according to Ainiala et al[9]; (3) The recognition of confounding factors; and (4) The inclusion of favoring factors. This model transitioned NP symptom attribution from qualitative analysis to quantitative examination. The Italian model can be employed to attribute both minor and major NP manifestations to the underlying SLE. Prior investigations using this attribution model indicated that the sensitivity and specificity were optimal when the threshold value was established at 7[11,12].

To our knowledge, there has not been any research exploring NP symptom attribution in Chinese SLE patients. This study aimed to retrospectively analyze the attribution of psychiatric manifestations in a cohort of Chinese patients with SLE, followed by a comparison of differences between psychiatric SLE (PSLE) and secondary psychiatric symptoms.

MATERIALS AND METHODS

The present single-center retrospective study was conducted at the Peking Union Medical College Hospital (PUMCH), a tertiary general hospital situated in Beijing, China, which functions as a referral center for critically ill patients within proximate areas. Using psychiatric consultation records for inpatients at PUMCH from April 2013 to July 2020, a total of 160 cases were identified, all of which exhibited psychiatric manifestations potentially indicative of PSLE prior to consultation.

The diagnosis of SLE was established by treating rheumatologists on the basis of the 2012 SLICC criteria[13]. Psychiatric syndromes were diagnosed by experienced consulting psychiatrists integrating the International Statistical Classification of Diseases and Related Health Problems 10^th Revision[14] and the 1999 ACR NPSLE nomenclature[8] to ascertain whether they met the syndrome diagnosis of acute confusional state (ACS), psychosis, mood disorders (inclusive of depression and mania or hypomania), and anxiety disorders. Cognitive dysfunction (CD) is clinically diagnosed if the patient exhibits impairment in at least two of the five cognitive domains during mental status examination: Attention; memory; executive function and motor performance; verbal function and language skills; and concept formation and reasoning. A comprehensive consideration of the patient's clinical state, laboratory analysis, and psychiatric consultation resulted in the final diagnosis of PSLE by the attending rheumatologist, in accordance with the 2010 European League Against Rheumatism recommendations for the management of NPSLE[15].

We analyzed the medical records of the included patients and acquired demographic and clinical details including age, sex, age at illness onset, course of illness, investigations upon admission (including comprehensive blood tests, autoantibodies, head imaging results, and electroencephalogram findings), SLE Disease Activity Index-2000 (SLEDAI-2K) scores upon admission[16], and medications prescribed during hospitalization.

The subjects were categorized into the PSLE group and the secondary psychiatric syndrome group on the basis of diagnosis, and the disparities in the clinical parameters between the two groups were evaluated. The Italian model constructed by Bortoluzzi et al[11] was examined as a methodology for attributing psychiatric symptoms. The attribution outcomes were contrasted with the gold standard of clinical diagnosis, yielding the specificity and sensitivity of the model in this demographic. Psychiatric symptoms were deemed attributed to SLE if the subject attained a score ≥ 7 points. We also scrutinized the model using a threshold value of 6.

The study was conducted in accordance with the Helsinki Declaration of the World Medical Association. The design of the study was reviewed and approved by the Ethics Committee of PUMCH, and informed consent was waived because of the observational/non-interventional design.

All the statistical analyses were performed via IBM SPSS Statistics 21.0.0.0. (IBM Corp., Armonk, NY, United States). Categorical variables are described as frequencies (percentages). Quantitative variables are described as medians [interquartile ranges (IQRs)]. Student’s t test, the Mann-Whitney U test, and Fisher’s 1-way ANOVA with the Bonferroni correction were used for between-group comparisons of continuous normally and nonnormally distributed variables. The χ² test and Fisher's exact test were used for between-group comparisons of categorical variables. A two-tailed P value < 0.05 was considered statistically significant.

RESULTS

Among this cohort of 160 SLE patients exhibiting psychiatric manifestations, 145 were female (90.6%). The median age was 28 years (IQR 23-39) with a median duration of disease of 1.5 years (IQR 0.3-6). A total of 121 (75.6%) patients demonstrated multiorgan or multisystem involvement. The most affected systems included the nervous (98.8%), hematological (78.8%), and renal systems (62.5%). The median SLEDAI-2K score was 20 (IQR 12-24). Cerebrospinal fluid tests were administered to 146 patients, all of whom tested negative for central nervous system (CNS) infection. In total, 132 patients underwent head magnetic resonance imaging (MRI), and abnormal results were noted in 85 patients (64.4%) (Supplementary Table 1). Electroencephalography was performed on 21 patients, with two-thirds demonstrating epileptiform waves. Throughout the course of hospitalization, all patients underwent glucocorticoid (GC) and immunosuppressive (IS) therapy. The most frequently used ISs were cyclophosphamide (66.3%) and hydroxychloroquine (61.3%). Antipsychotics and sedatives were prescribed to 103 (64.3%) and 52 (32.5%) individuals, respectively. The median duration of hospitalization was 30 days (IQR 22-43). Ten (6.2%) patients died during hospitalization.

A total of 171 psychiatric syndromes were diagnosed in 138 patients. There were 87 patients with ACS, 18 patients with psychosis, 40 patients with CD, and 26 patients with mood disorders (13 with depression and 13 with mania/hypomania). According to the clinical diagnosis, 141 (82.5%) syndromes were identified as PSLE manifestations in 110 patients, whereas the remaining syndromes were deemed psychiatric syndromes secondary to alternative factors. Forty-seven patients (42.7%) with PSLE exhibited psychiatric manifestations with the onset of SLE, whereas 18 patients (64.3%) presented with secondary psychiatric syndromes. The psychiatric manifestations of the remaining 22 patients did not meet the diagnostic criteria for any specific mental syndrome.

Table 1 delineates disparities between PSLE patients (group 1) and those with secondary psychiatric manifestations (group 2). Compared with group 2, group 1 had higher SLEDAI-2K scores (21 vs 12, P = 0.001). Group 2 had a greater prevalence of anti-beta-2-glycoprotein 1 (β2GP1) antibodies (8.6% vs 25.9%, P = 0.036) and a lower prevalence of anti-ribosomal ribonucleoprotein particle (rRNP) antibodies (39.0% vs 22.2%, P = 0.045). The mortality rates were comparable between the two groups. The sex ratio, age, age at illness onset, course of illness, organ involvement, other autoantibodies, MRI findings, and electroencephalogram results exhibited no significant disparities. Compared with group 2, group 1 demonstrated significantly greater utilization rates of cyclophosphamide (78.7% vs 53.3%, P = 0.005), antipsychotics (80.9% vs 43.3%, P < 0.001), and sedatives (43.3% vs 10.0%, P < 0.001). The maximum dosage of antipsychotics was also elevated in group 1 (equivalent to olanzapine 8.7 mg vs 3.9 mg, P = 0.004).

Table 1 Comparisons between psychiatric systemic lupus erythematosus and secondary psychiatric syndromes, n (%).

Parameters n (percentage) or median (IQR)	PSLE (141)	Secondary psychiatric syndromes (30)	P value
Female sex	129 (91.5)	27 (90.0)	0.555
Age	32 (30-34)	31 (24-38)	0.865
Age of disease onset	28 (26-30)	27 (21-32)	0.530
Disease duration (year)	3.5 (2.8-4.2)	4.7 (2.6-6.8)	0.186
Multiorgan/system involvement	111 (78.7)	20 (66.7)	0.135
SLEDAI-2K	21 (20-22)	12 (5-17)	0.001
Autoantibodies
ANA	137 (97.2)	29 (96.7)	0.824
Anti-dsDNA	80 (56.7)	13 (43.3)	0.128
Anti-Sm	42 (29.8)	5 (16.7)	0.105
Anti-RNP	59 (41.8)	10/29 (34.5)	0.301
Anti-SSA	94 (66.7)	18/29 (62.1)	0.670
Anti-SSB	22 (15.6)	5/29 (17.2)	0.785
Anti-Scl70	4 (2.8)	0 (0)	0.470
Anti-Jo1	1 (0.7)	0 (0)	0.829
Anti-rRNP	55 (39.0)	6/29 (22.2)	0.045
Anti-PCNA	3 (2.1)	0 (0)	0.568
Anti-AHA	30 (21.3)	6/29 (20.7)	0.583
Anti-Ro52	58 (41.1)	15/29 (51.7)	0.199
Anti-PM-Scl	2 (1.4)	1/29 (3.4)	0.432
Anti-ANuA	48 (34.0)	8/29 (27.6)	0.329
Anti-CENPB	0 (0)	1/29 (3.4)	0.171
Anti-AMA2	17 (12.1)	2/29 (6.9)	0.333
Anticardiolipin	17/129 (13.2)	6/27 (22.2)	0.443
Anti-β2GP1	11/128 (8.6)	7/27 (25.9)	0.036
Abnormal MR results	75/122 (61.5)	17/25 (68.0)	0.354
Abnormal EEG results	11/18 (61.1)	7/9 (77.8)	0.339
Use of IS
Cyclophosphamide	111 (78.7)	16 (53.3)	0.005
Hydroxychloroquine	98 (69.5)	20 (66.7)	0.458
Use of antipsychotics	114 (80.9)	13 (43.3)	< 0.001
Maximum dosage of antipsychotics (as in olanzapine)	8.7 (7.6-9.7)	3.9 (2.1-5.8)	0.004
Use of sedatives	61 (43.3)	3 (10.0)	< 0.001
Death	8 (5.7)	2 (6.7)	0.551

PSLE: Psychiatric systemic lupus erythematosus; SLEDAI-2K: SLE disease activity index-2000; ANA: Antinuclear antibody; RNP: Ribonucleoprotein; SSA: Sjogren syndrome A; SSB: Sjogren syndrome B; rRNP: Ribosomal ribonucleoprotein particle; PCNA: Proliferating cell nuclear antigen; AHA: Antihistone antibody; ANuA: Anti-nucleosome antibody; CENPB: Centromere protein B; AMA2: Antimitochondrial antibody; β2GP1: Beta-2-glycoprotein 1; MR: Magnetic resonance; EEG: Electroencephalogram.

In contrast to the gold standard, the Italian model exhibited a sensitivity of 95.0% and a specificity of 70.0% when the threshold was set at ≥ 7 (area under the curve score 0.81, P < 0.001). When the threshold was adjusted to ≥ 6, the sensitivity increased to 96.5%, although the specificity decreased to 50.0%. Table 2 presents the comprehensive attribution results of various psychiatric syndromes. Employing the Italian model utilizing a threshold value of ≥ 7, the model demonstrated excellent diagnostic consistency for ACS, psychosis, mood disorders, and CD. A total of 144 (83.7%) cases of psychiatric syndromes were attributed to SLE. The majority of cases of ACS (93.1%), CD (87.5%), psychosis (77.8%), and mania/hypomania (76.9%) were attributed to SLE, as illustrated in Figure 1.

Figure 1 Attribution of psychiatric syndromes to systemic lupus erythematosus. ACS: Acute confusional state; CD: Cognitive dysfunction; PSLE: Psychiatric systemic lupus erythematosus.

Table 2 Attribution results of psychiatric syndromes using the Italian model compared with those of clinical diagnoses.

Psychiatric syndromes	Clinical diagnoses		Italian model (threshold value 7)		Italian model (threshold value 6)
	PSLE	Secondary	PSLE	Secondary	PSLE	Secondary
ACS	81	6	84	3	86	1
Psychosis	14	4	15	3	17	1
Mania/hypomania	10	3	11	2	11	2
Depression	1	12	1	12	4	9
CD	35	5	32	8	33	7

ACS: Acute confusional state; CD: Cognitive dysfunction; PSLE: Psychiatric systemic lupus erythematosus.

DISCUSSION

In this study, we retrospectively analyzed single-center data to evaluate symptom attribution among Chinese SLE patients presenting with psychiatric symptoms and to contrast disparities in clinical characteristics between patients presenting with PSLE and those manifesting secondary psychiatric symptoms. We also utilized the Italian Study Group's attribution model and evaluated its diagnostic efficacy within our study population.

The elucidation of the mechanisms associated with secondary mental disorders may improve the understanding of the etiological mechanisms underlying primary mental disorders. Subsequent to decades of meticulous research, it is hypothesized that the clinical manifestations of NPSLE correlate with multiple mechanisms. A distinctive characteristic of SLE is the generation of autoimmune antibodies, with numerous antibodies that have been documented to be related to NPSLE manifestations. To date, over 100 antibodies have been documented in SLE, yet the mechanism underlying the distinct NP manifestations of SLE instigated by a single autoantibody remains unresolved[6,17]. Additionally, inflammatory and immune-mediated mechanisms have been implicated. There is an imbalance in the production and function of cytokines in NPSLE. Augmented levels of pro-inflammatory cytokines can induce inflammation in the brain, thereby compromising the blood-brain barrier, enabling immune cells and antibodies to access the CNS more readily and intensifying the inflammatory response within the brain, culminating in neuronal dysfunction and NP symptomatology. Furthermore, activated immune cells, which include T lymphocytes and macrophages, can discharge cytotoxic substances and cytokines, thereby directly harming neurons and glial cells and contributing to the progression of NP disorders[3,18]. Numerous studies have detected vascular dysfunction. SLE-induced inflammation can induce dysfunction in endothelial cells lining cerebral blood vessels, resulting in aberrant vasoconstriction or vasodilation, modified blood flow patterns, and increased susceptibility to thrombosis, all of which can impair the standard functioning of the brain and culminate in NP morbidities[3,5].

When the characteristics of patients with PSLE and secondary psychiatric manifestations were compared in this study, patients with PSLE exhibited more severe disease conditions. PSLE is indeed considered a severe manifestation of SLE and has been associated with mortality[4,5]. Nonetheless, in our research, patients who experienced secondary psychiatric symptoms also exhibited multisystemic involvement, and a median disease activity score of 12 further indicated moderate to severe lupus activity[16]. Several prior investigations have indicated that antiphospholipid antibody syndrome (APS), macrophage activation syndrome, and other concomitant disorders associated with SLE may result in multiple NP manifestations[19,20], suggesting that secondary NP symptoms, similar to NPSLE, also necessitate careful clinical observation. β2GP1 is a plasma protein engaged in phospholipid binding. In a clinical context, the anti-β2GP1 antibody serves as a vital serological marker for APS and is correlated with an increased risk of both arterial and venous thrombosis[21]. In NPSLE, antiphospholipid antibody-induced thrombosis is considered a possible cause of cerebrovascular disease and a potential mechanism underlying CD and mood disorders[22-24]. Our results, in which patients who developed secondary psychiatric syndromes exhibited a higher level of anti-β2GP1 antibodies than did patients with PSLE, suggest the potential of anti-β2GP1 antibodies as a differential marker for psychiatric manifestations. Previously, the anti-rRNP antibody has was associated with NPSLE[6,25], which mirrors our current study's observations. In patients with SLE, cell damage and other conditions may lead to exposure to rRNPs. Once B cells recognize the exposed rRNP self-antigen, its activation results in proliferation and differentiation and ultimately the production of many antibodies against rRNP[26]. In a large-scale cohort study conducted in China, the presence of anti-rRNP antibodies at baseline predicted accumulated NP damage (adjusted HR = 3.8, 95%CI: 2.7-5.7) during the follow-up period of 11.5 years[27]. Another cross-sectional study revealed that anti-rRNP serum levels were independently associated with depressive symptoms[28]. However, the exact mechanism by which anti-rRNP antibodies contribute to these clinical features remains under investigation. It is hypothesized that these proteins may interact with ribosomal proteins in cells, leading to disruptions in protein synthesis and subsequent cellular dysfunction[27,28].

The psychiatric management of NPSLE is symptomatic and adheres to the same principles as the management of secondary psychiatric manifestations[29]. Our findings suggest that patients with PSLE are more inclined to receive prescriptions for antipsychotics and sedatives, and that antipsychotics are prescribed at higher dosages. Owing to the absence of retrospective psychiatric evaluation data, our proposed explanation for this phenomenon is that patients with PSLE potentially experience more profound psychiatric symptoms than do those with secondary symptoms and thus necessitate more potent antipsychotic therapy. We suggest that future studies should evaluate the magnitude of psychiatric symptoms to establish a foundation for refining symptom-oriented treatment of PSLE.

The sensitivity and specificity of the Italian attribution model incorporating a threshold value of 7 within this Chinese SLE cohort were 95.0% and 70.0%, respectively. In a multicenter validation study of 243 patients utilizing the model with the same cutoff value, the sensitivity and specificity for first-episode NPSLE were 87.9% and 82.6%, respectively[12]. Discrepancies in diagnostic accuracy could be related to variances in study contexts, as our research exclusively concentrated on psychiatric manifestations. Study population variances are also crucial influencing elements. Prior investigations have hypothesized that ethnicity may be one of the influencing factors for NPSLE[30,31], and the majority of NP symptom attribution models, including the Italian model, have been developed and validated mainly in Caucasian patients. Whether ethnicity adaptation is warranted in clinical implementations requires broader multinational explorations.

A significant benefit of the Italian model compared with preceding attribution models is its capacity not to disregard mild NP presentations[11], thereby decreasing the underdiagnosis of NPSLE. Previous studies have indicated that this model has inferior performance with mild NP function[12]. The probable cause is correlated with the scoring methodology of the Italian model, as mild cognitive impairment is a minor NP symptom; this item is not scored, and it is less frequently attributed to SLE in the scoring paradigm than moderate or severe cognitive impairment. However, in our findings, the Italian model for CD did not exhibit a significantly higher misdiagnosis rate than other mental disorders did. This is likely because in our research design, clinically diagnosed CD excluded mild cognitive impairment, thereby minimizing the misdiagnosis rate.

According to the Italian model, a substantial proportion (83.6%) of psychiatric symptoms in this study were attributed to SLE. In the attribution investigation by Pawlak-Buś et al[32] involving 164 Polish SLE patients, which utilized the same model, only 19% of syndromes were attributed to SLE among approximately 80 psychiatric/neuropsychological manifestations. The significant disparity could be attributed to variations in the demographic features of the study population, with SLE patients in our investigation having a younger mean age (28 years) and high disease severity (mean SLEDAI score of 20), with more than three-quarters of patients presenting with multisystemic involvement and more than three-fifths necessitating multiple IS therapies alongside GCs. Our findings also demonstrated discrepancies in attribution with different psychiatric syndromes. ACS, psychosis, mania, and cognitive disorders are predominantly attributed to SLE, whereas depression is primarily attributed to non-SLE elements. Notably, the psychiatric literature generally indicates that psychosocial and environmental factors contribute to a greater proportion of the pathogenesis of depression than bipolar disorder does[33,34], suggesting that bipolar disorder may be perceived as a more "organic" mood disorder. In subsequent studies, whether similar trends exist in different settings should be further explored.

Our research has several limitations. First, owing to its retrospective nature, some clinical data were absent, potentially causing some attribution biases. Moreover, despite focusing on the psychiatric treatment of patients exhibiting diverse characteristics, due to the absence of comprehensive psychiatric evaluations, we could only make preliminary inferences regarding variations in treatment conditions. Second, given its single-center design, the study cohort is a unique group of critically ill patients requiring psychiatric consultation, potentially limiting the broader generalizability of the findings. However, this is also one of the strengths of this study, as it can offer a more comprehensive depiction of psychiatric disorders.

CONCLUSION

In conclusion, in this study, we replicated previous findings indicating that patients with PSLE exhibit severe disease conditions, and that PSLE is associated with anti-rRNP antibodies. Concurrently, we discovered that ACS, psychosis, mania, and CD are more likely to be attributable to SLE. We also validated the relevance of the Italian attribution model within Chinese SLE patients and found that it exhibited favorable sensitivity to multiple psychiatric symptoms, which may serve as a crucial supplement to the clinical diagnosis of PSLE in the future. Larger, multicenter studies and studies focusing on single NP syndromes are needed, to further investigate the attribution of NPSLE.

ACKNOWLEDGEMENTS

The authors would like to thank colleagues in the Department of Rheumatology and Clinical Immunology for their diligent work and participation in this study.