Exploring the risk factors and prognosis of transverse myelitis in systemic lupus erythematosus

Abstract

Purpose:

We aimed to describe the clinical characteristics and outcomes of patients with transverse myelitis (TM) as a rare manifestation in systemic lupus erythematosus (SLE) and explore the risk factors and prognosis of SLE-related TM (SLE-TM).

Methods:

We conducted a retrospective case–control and cohort analysis. All patients with SLE-TM (58 patients) and 232 with SLE without TM, as a control group, were admitted to Peking Union Medical College Hospital between January 1993 and May 2021. Factors associated with the presence of SLE-TM and its prognosis were assessed using logistic regression and Cox proportional hazard models.

Results:

Multivariate analysis revealed that positive anti-Ro/Sjogren’s syndrome A (anti-Ro/ SSA) (<0.01) and increased erythrocyte sedimentation rate (ESR) (p < 0.01) were associated with SLE-TM. Regarding prognosis, methylprednisolone (MP) pulse therapy within 2 weeks of onset (adjusted hazard ratio (AHR), 2.12; 95% confidence interval (CI), 1.06–4.23; p = 0.03) was associated with short-term neurological improvement. An American Spinal Injury Association Impairment Scale (AIS) grades of A, B, or C at onset (AHR, 0.12; 95% CI 0.05–0.28; p < 0.001) and hypoglycorrhachia (AHR, 0.29; 95% CI, 0.13–0.65; p < 0.01) were associated with a short-term non-improved outcome.

Conclusions:

The positive anti-Ro/SSA antibodies and increased ESR may be associated with the presence of SLE-TM. An initial presentation with severe myelitis and hypoglycorrhachia appear to be predictors of a poor neurological outcome. Early steroid pulse therapy may improve the prognosis.

Introduction

Systemic lupus erythematosus (SLE) is a systemic autoimmune disease that can affect multiple organs and systems. Transverse myelitis (TM), a rare but serious SLE manifestation, is characterized by an acutely progressive course of paralysis, sensory deficit, and sphincter dysfunction. The estimated incidence rate of TM is 1–2% in patients with SLE,^1–3 although the reported rate varies widely between studies.

As the overall incidence rate of TM is low, the risk factors for this SLE manifestation remain unknown. The SLE-related TM (SLE-TM) prognosis also requires further elucidation since current prognostic evidence comes from case series and a few retrospective cohort studies.^4–7 Understanding the risk and prognostic factors of TM may increase recognition of this low-incidence disease and facilitate early diagnosis, allowing administration of timely aggressive treatment, which may reduce disability in patients with SLE-TM.^1,6

Therefore, this study aimed to investigate the risk factors and prognosis for SLE-TM in a large clinical population to inform risk factor modification, prognostication, and treatment strategies.

Methods

Research design

To study the risk factors of SLE-TM, we used a case–control analysis. In addition, we chose a retrospective cohort study design to explore the prognostic factors and relapse rates of SLE-TM. The study followed guidance from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ⁸ The Research Ethics Committee of the Peking Union Medical College Hospital approved this study on September 1, 2021 (approval no. S-F1745).

Study participants

Patients with SLE-TM were identified from the Peking Union Medical College Hospital between January 1993 and May 2021. All patients with SLE-TM met the Transverse Myelitis Consortium Working Group ⁹ definition of myelitis. Those with TM related to central nervous system infections, multiple sclerosis, and structural lesions, including tumor metastasis, herniated disk, or vertebral fracture, were excluded. Furthermore, all patients fulfilled the 1997 American College of Rheumatology (ACR) revised criteria ¹⁰ or the ACR/Systemic International Collaborating Clinics (SLICC) 2012 criteria. ¹¹ The patients in the control group were extracted by random sampling technique from the SLE patients without TM, which was confirmed by a follow-up period starting at the first admission in our hospital and ending at the date of the latest medical record. Sample size was calculated considering that the sample size should be >10–20 times the number of independent variables to be added into the model ¹² in the SLE-TM and control groups, as well as considering α = 0.5 and power = 80%. Finally, data from 58 patients with SLE-TM and 232 controls (1:4) were analyzed. Our study de-identified all patients’ details, and we obtained patients informed written consent in the choice of treatments, such as glucocorticoids pulse therapy or immunosuppressants.

Definitions

At the time of admission and follow-up visits, neurological deficits were assessed by a neurologist and a rheumatologist using the American Spinal Injury Association Impairment Scale (AIS), which is stratified from A (no sensory or motor function below the level of spinal injury) to E (normal sensory and motor function). ¹³ The Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2K) was adopted to evaluate SLE disease activity at the time of TM diagnosis. ¹⁴ The SLICC/ACR Damage Index (SDI), which includes 12 organ systems, was used to evaluate organ damage of SLE patients. ¹⁵ Longitudinal extensive TM (LETM) was defined as a spinal cord lesion extending over ⩾3 vertebral segments. ¹⁶ Patients were considered to have neuromyelitis optica spectrum disorder (NMOSD) simultaneously with SLE-TM when the 2015 international consensus diagnostic criteria for NMOSD were fulfilled. ¹⁷ Definition of high-risk anti-phospholipid (aPL) antibodies’ profiles was that the presence (in two or more occasions at least 12 weeks apart) of lupus anticoagulant (LA), or of double (any combination of LA, anticardiolipin (aCL) antibodies or anti-β2-glycoprotein I (anti-β2GPI) antibodies) or of triple (all three subtypes) aPL positivity, or the presence of persistently high aPL titers. Low-risk aPL profile included isolated aCL or anti-β2GPI antibodies at low–medium titers, particularly if transiently positive. ¹⁸ Hyperproteinorachia was defined as a protein level >0.45 g/L in the cerebrospinal fluid (CSF). Hypoglycorrhachia was defined as a glucose level <2.8 mmol/L or <50% of the blood glucose level in the CSF. AIS classifications of A, B, or C at initial presentation were classified as severe myelitis. ⁶ The occurrence of relapse of myelitis during the follow-up period, defined as TM-compatible neurological symptoms confirmed by magnetic resonance imaging (MRI) after a period of neurological improvement, was also evaluated. The follow-up period was defined as the time from first onset of myelitis to the last follow-up or relapse. All patients were closely followed up during the first 3 months. The main endpoint was neurological impairment (AIS grade) during the first 3 months. An unfavorable neurological outcome was defined as an AIS grades A, B, or C at follow-up. In addition, patients with at least one-grade improvement in AIS after treatment were classified as improved outcomes, while the non-improved outcome included patients with permanent neurological impairment.

Data for analysis

Data were extracted from the medical records for analysis. Clinical data for SLE-TM patients and non-SLE-TM patients were retrospectively collected at the time of admission, including demographic data (sex, age at enrollment, SLE duration, and TM duration), systemic manifestations, physical signs (including neurological signs), and laboratory examinations that included erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), complement, and self-antibodies status. CSF was collected, and biochemical analyses and MRI were performed. Information concerning treatment and prognosis was also compiled. Anti-nuclear antibody (ANA) and anti-double-stranded DNA (anti-dsDNA) were tested by indirect immunofluorescence and enzyme-linked immunosorbent assay (ELISA). Anti-Ro/Sjogren’s syndrome A (Anti-R0/SSA), anti-La/Sjogren’s syndrome B (anti-La/SSB), anti-Smith (anti-Sm), anti-ribosomal P protein (anti-rRNP), and anti-U1 ribonucleoprotein (anti-RNP) antibodies were tested by dot blotting. Serum titers of aPL, including aCL and anti-β2GPI antibodies were measured using standardized commercial ELISA kits. LA activity was tested by the integrated activated partial thromboplastin time test.

Statistical analysis

The characteristics of patients with SLE-TM and those with SLE without TM (controls) are summarized as mean (SD), median (first and third quartiles), or number (percentage). Given the retrospective collection of data and the observational nature of the study, some data were incomplete. Missing data were not imputed. The control population was selected randomly, reducing the risk of selection bias. We used nonparametric analysis because the continuous variables were not normally distributed. Associations between categorical or continuous variables and TM were examined using the chi-square and Mann–Whitney U tests. We used multivariable logistic regression models to calculate the odds ratios (OR) and 95% confidence intervals (CIs) and assess the association between specific factors and SLE-TM. A logistic regression model was used with all the potential factors to ascertain the factors for the risk of SLE-TM. We explored factors associated with short-term (3-month) prognosis of SLE-TM using a logistic regression model and Cox proportional hazard regression to calculate the OR and hazard ratio (HR). Regarding short-term neurological improvement, we used Cox proportional hazard regression to estimate HR adjusted for the underlying confounding variables. The Kaplan–Meier method was used to draw the remission curve and calculate the cumulative remission rate. A p value < 0.05 was set for statistical significance.

All statistical analyses were performed using the 19th version of SPSS software (IBM Inc.) and STATA 15 (Stata Corporation, College Station, TX, USA).

Results

Demographic and clinical characteristics

Fifty-eight patients with SLE-TM were included in the study group. The median age at TM onset was 34.5 years (interquartile range (IQR), 25.75–45.25 years). TM presented as one of the initial manifestations of SLE in 8/58 (13.8%) cases. The most common neurological presentation of TM was symmetrical flaccid paraparesis (39.7%), with anesthesia or hypoesthesia (31.0%), and sphincter dysfunction (70.7%). SLE-TM patients with an initial TM presentation were classified according to their neurological deficits, by AIS grade: grade A, 10/58 (17.2%) patients; grade B, 8/58 (13.8%); grade C, 5/58 (8.6%); and grade D, 35/58 (60.3%). Concomitant NMOSD was present in 25/49 (51.0%) cases. The clinical characteristics, treatment regimens, and outcomes of these patients are described in Tables 1 and 2.

Table 1.

Demographic and clinical features of SLE-TM.

	N = 58 (%)
Features of SLE
Female	56 (96.6)
Age at myelitis onset (years)	34.50 (25.75–45.25)
SLE duration (years)	2.00 (0.00–7.25)
TM after SLE diagnosis	50 (86.2)
Cutaneous involvement	8/55 (14.5)
Musculoskeletal involvement	10/55 (18.2)
Pleuritis	9/55 (16.4)
Pericarditis	8/55 (14.5)
Renal involvement	24 (41.4)
Hematological involvement	25 (43.1)
SLEDAI-2K	9.50 (4.00–16.00)
SDI	1.00 (1.00–1.00)
Features of TM
TM as the first symptom	8 (13.8)
Paraparesis	23 (39.7)
Sensory deficit	18 (31.0)
Sphincter dysfunction	41 (70.7)
Severe myelitis at onset	23 (39.7)
Concomitant NMOSD	25/49 (51.0)

SLE, systemic lupus erythematosus; TM, transverse myelitis; NMOSD, neuromyelitis optica spectrum disorder; SDI, Systemic International Collaborating Clinics/American College of Rheumatology Damage Index; SLEDAI-2K, the Systemic Lupus Erythematosus Disease Activity Index 2000.

Table 2.

Laboratory, imaging, and treatment features of SLE-TM.

	N = 58(%)
Laboratory results
Hyperproteinorachia	40 (69.0)
Hypoglycorrhachia	16 (27.6)
Imaging results
LETM	40/56 (71.4)
Affected segments
Thoracic	18/56 (32.1)
Cervical	15/56 (26.8)
Treatment
MP pulse	53 (91.4)
MP pulse within 2 weeks	36 (62.1)
MP pulse within 4 weeks	43 (74.1)
CTX	42 (72.4)
RTX	12/52 (23.1)
Follow-up time (years)	2.00 (0.25–5.00)

CTX, cyclophosphamide; LETM, longitudinal extensive transverse myelitis; MP, methylprednisolone; RTX, rituximab.

Hyperproteinorachia refers to a protein level < 0.15 g/L in the cerebrospinal fluid. Hypoglycorrhachia was defined as a glucose level <2.8 mmol/L or <50% of the blood glucose level in the cerebrospinal fluid.

Laboratory and imaging results

Laboratory test results, including CSF results, and imaging findings are presented in Tables 2 and 3. A positive aCL screen occurred in 11/56 (19.6%) patients, anti-β2GPI in 9/51 (17.6%), and LA antibodies were present in 15/54 (27.8%). The most frequent positive antibody at the time of TM diagnosis was anti-Ro/SSA, identified in 37/58 (63.8%) patients, followed by anti-dsDNA in 36/58 (62.1%) patients. Spinal cord MRI was performed in 56/58 patients at baseline. LETM was observed in 40/56 (71.4%) of the patients, with the thoracic spine being the most common site of involvement (18/56), followed by the cervical spine (15/56).

Table 3.

Characteristics between the SLE cases with transverse myelitis and controls.

	TM patients (N = 58) (%)	Controls (N = 232) (%)	p value
Age at SLE onset (years)	29.00 (21.50–41.50)	27.12 (21.51–35.88)	0.41
Female	56 (96.6)	213 (91.8)	0.34
Elevated ESR	49 (84.5)	136 (58.6)	<0.001
Elevated CRP	19 (32.8)	59/175 (33.7)	0.89
Hypocomplementemia	34 (58.6)	138 (59.5)	0.91
Autoantibody profiles
Anti-dsDNA	36 (62.1)	118 (50.9)	0.13
Anti-Sm	8 (13.8)	59 (25.4)	0.06
Anti-Ro/SSA	37 (63.8)	78 (33.6)	<0.001
Anti-La/SSB	15 (25.9)	23 (9.9)	0.001
Anti-RNP	20 (34.5)	45 (19.4)	0.014
Anti-rRNP	11 (19.0)	48 (20.7)	0.77
aPL and different aPL profiles
aCL	11/56 (19.6)	35 (15.1)	0.40
Anti-β2GP1	9/51 (17.6)	50 (21.6)	0.53
LA	15/54 (27.8)	35 (19.3)	0.18
SLEDAI-2K	9.50 (4.00–16.00)	8.00 (3.00–13.00)	0.14
SDI	1.00 (1.00–1.00)	0.00 (0.00–0.00)	<0.001

SLE, systemic lupus erythematosus; TM, transverse myelitis; aPL, anti-phospholipid antibodies; aCL, anti-cardiolipin antibodies; anti-β2GP1, anti-beta 2 glycoprotein 1 antibodies; LA, lupus anticoagulation; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; SDI, Systemic International Collaborating Clinics/American College of Rheumatology Damage Index; SLEDAI-2K, the Systemic Lupus Erythematosus Disease Activity Index 2000.

Treatments and outcomes

Intravenous methylprednisolone (MP) pulse therapy was administered to 53/58 (91.4%) patients, 500–1,000 mg for 3–5 days. This treatment was provided within 2 weeks of TM onset in 36/58 (62.1%) patients, within 4 weeks of TM onset in 43/58 (74.1%), and after >4 weeks of TM onset in 15/58 (25.9%). Cyclophosphamide (CTX) was administered to 42/58 (72.4%) patients. Other treatments provided included immunoglobulin therapy in 27/58 (46.6%) patients and rituximab in 12/52 (23.1%). All 58 patients had a recorded AIS grade at the 3-month follow-up, with neuropsychiatric functional improvement attained in 45/58 (77.6%) patients and non-improvement in 13/58 (22.4%). In addition, an unfavorable outcome (AIS grades A, B, or C) was observed in 18/58 (31.0%) patients and a favorable outcome in 40/58 (69.0%).

Risk factors associated with SLE-TM

We constructed an analysis of 58 SLE-TM cases and 232 controls. Compared with the controls, SLE-TM patients had a higher prevalence of increased ESR, SDI and positive anti-Ro/SSA, anti-La/SSB, and anti-RNP antibodies (p < 0.05) (Table 3). The multivariate analysis (Table 4) revealed that anti-Ro/SSA positive (OR, 2.68; 95% CI, 1.35–5.31; p < 0.01) and increased ESR (OR, 3.73; 95% CI, 1.71–8.17; p < 0.01) were considered as risk factors for the presence of SLE-TM.

Table 4.

Factors associated with the risk of SLE-TM.

	Univariate logistic regression		Multivariate logistic regression
	OR (95% CI)	p value	OR (95% CI)	p value
Characteristics
Age at SLE onset (years)	1.01 (0.99–1.04)	0.39
Elevated ESR	3.84 (1.80–8.20)	<0.001	3.73 (1.71–8.17)	<0.01
Elevated CRP	0.96 (0.51–1.80)	0.89
Hypocomplementemia	0.97 (0.54–1.73)	0.91
Autoantibody profiles
Anti-dsDNA	1.58 (0.88–2.85)	0.13
Anti-Sm	0.47 (0.21–1.05)	0.07
Anti-Ro/SSA	3.48 (1.91–6.34)	<0.001	2.68 (1.35–5.31)	<0.01
Anti-La/SSB	3.17 (1.53–6.57)	<0.01	1.99 (0.85–4.62)	0.11
Anti-RNP	2.19 (1.16–4.11)	0.02	1.98 (0.99–3.84)	0.06
Anti-rRNP	0.90 (0.43–1.86)	0.77
aPL and different aPL profiles
aCL	1.38 (0.65–2.92)	0.41
Anti-β2GP1	0.78 (0.36–1.71)	0.54
LA	1.60 (0.80–3.23)	0.19
SLEDAI-2K	1.02 (0.98–1.06)	0.27

SLE, systemic lupus erythematosus; TM, transverse myelitis; aPL, anti-phospholipid antibodies; aCL, anti-cardiolipin antibodies; anti-β2GP1, anti-beta 2 glycoprotein 1 antibodies; LA, lupus anticoagulation; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; SDI, Systemic International Collaborating Clinics/American College of Rheumatology Damage Index; SLEDAI-2K, the Systemic Lupus Erythematosus Disease Activity Index 2000.

Prognostic factors for SLE-TM patients

The analysis of prognostic factors included all 58 patients with SLE-TM whose neurological outcome (AIS grade) had been assessed at 3 months (Table 5). An AIS grades of A, B, or C at onset (p < 0.001) and hypoglycorrhachia (p < 0.001) were associated with an unfavorable prognosis. MP pulse therapy within 2 weeks of onset (p < 0.01) was considered a protective factor for neurological outcomes. In the multivariate logistic regression analysis, no obvious factors were associated with an unfavorable neurological outcome of SLE-TM.

Table 5.

Prognostic factors for unfavorable neurological outcome ^a of transverse myelitis.

	Univariate logistic regression
	OR (95% CI)	p value
Characteristics
Age at myelitis onset	0.95 (0.90–1.00)	0.05
SLEDAI-2K	1.02 (0/94–1.1)	0.67
SDI	0.28 (0.03–2.44)	0.25
Neurological impairment
AIS A/B/C at onset	152.00 (19.67–1175.01)	<0.001
Laboratory findings
Hyperproteinorachia	3.00 (0.74–12.11)	0.12
Hypoglycorrhachia	18.00 (4.33–74.76)	<0.001
Hypocomplementemia	1.64 (0.51–5.23)	0.83
Increased CRP	2.11 (0.66–6.73)	0.21
Increased ESR	4.25 (0.49–36.87)	0.19
Autoantibody profiles
Anti-dsDNA positive	0.67 (0.22–2.10)	0.49
Anti-Sm	0,71 (0.13–3.91)	0.69
Anti-Ro/SSA	0.60 (0.19–1.88)	0.38
Anti-La/SSB	0.75 (0.20–2.80)	0.67
Anti-RNP	1.32 (0.42–4.20)	0.64
Anti-rRNP	1.35 (0.34–5.35)	0.67
aPL and different aPL profiles
aCL	0.44 (0.09–2.32)	0.34
Anti-β2GP1	0.25 (0.03–2.20)	0.21
LA	0.93 (0.24–3.54)	0.91
Low-risk aPL profile	2.38 (0.31–18.36)	0.41
High-risk aPL profile	0.53 (0.15–1.92)	0.34
Spinal cord MRI
LETM	2.60 (0.64–10.64)	0.18
Affected segments
Thoracic	0.80 (0.28–2.26)	0.67
Treatment
MP pulse within 2 weeks	0.17 (0.05–0.56)	<0.01
CTX	1.50 (0.41–5.51)	0.54

AIS, American Spinal Injury Association Scale; CRP, C-reactive protein; CTX, cyclophosphamide; ESR, erythrocyte sedimentation rate; LETM, longitudinal extensive transverse myelitis; MP, methylprednisolone; SDI, Systemic International Collaborating Clinics/ American College of Rheumatology Damage Index; SLEDAI-2K, the Systemic Lupus Erythematosus Disease Activity Index 2000; TM, transverse myelitis.

^aAn unfavorable neurological outcome was defined as an AIS grade A, B, or C at follow-up.

When considering neurological improvement at 3 months (Table 6), age at TM diagnosis (p = 0.02), an initial presentation with severe myelitis (p < 0.001) (Figure 1), hypoglycorrhachia (p = 0.001), and MP pulse within 2 weeks (p < 0.01) (Figure 2) were significantly associated with neurological improvement. Finally, the analysis where we adjusted for age at myelitis showed that MP pulse therapy within 2 weeks of onset (adjusted hazard ratio (AHR), 2.12; 95% CI, 1.06–4.23; p = 0.03) was associated with short-term neurological improvement. An American Spinal Injury AIS grades of A, B, or C at onset (AHR, 0.12; 95% CI 0.05–0.28; p < 0.001) and hypoglycorrhachia (AHR, 0.29; 95% CI, 0.13–0.65; p < 0.01) were associated with a short-term non-improved outcome.

Table 6.

Prognostic factors for improvement outcome ^a of myelitis within 3 months of follow-up.

	HR (95% CI)		Adjusted b	p value
	Unadjusted	p value
Characteristics
Age at myelitis onset	1.03 (1.00–1.06)	0.02	1 (reference)	1 (reference)
SLEDAI-2K	0.99 (0.95–1.03)	0.57
SDI	1.41 (0.65–3.04)	0.38	1.07 (0.48–2.40)	0.86
Neurological impairment
AIS A/B/C at onset	0.11 (0.05–0.26)	<0.001	0.12 (0.05–0.28)	<0.001
Laboratory findings
Hyperproteinorachia	0.60 (0.32–1.12)	0.11
Hypoglycorrhachia	0.26 (0.12–0.59)	0.001	0.29 (0.13–0.65)	<0.01
Hypocomplementemia	0.733 (0.40–1.33)	0.31
Increased CRP	0.84 (0.43–1.60)	0.59
Increased ESR	0.56 (0.26–1.22)	0.14
Anti-dsDNA positive	0.87 (0.48–1.57)	0.64
Anti-Sm positive	0.81 (0.36–1.81)	0.60
Anti-Ro/SSA positive	1.22 (0.65–2.30)	0.54
Anti-La/SSB positive	1.10 (0.57–2.13)	0.78
Anti-RNP positive	0.61 (0.33–1.14)	0.12
Anti-rRNP positive	0.88 (0.41–1.89)	0.74
aPL and different aPL profiles
aCL positive	1.02 (0.49–2.13)	0.96
Anti-β2GP1 positive	1.30 (0.60–2.85)	0.50
LA positive	0.98 (0.50–1.91)	0.96
Low-risk aPL profile	0.42 (0.10–1.75)	0.23
High-risk aPL profile	1.12 (0.60–2.09)	0.73
Spinal cord MRI
LETM	0.68 (0.36–1.30)	0.25
Affected segments
Thoracic	1.68 (0.90–3.13)	0.11
Treatment
MP pulse within 2 weeks	2.44 (1.25–4.76)	<0.01	2.12 (1.06–4.23)	0.03
CTX	0.80 (0.42–1.52)	0.49

AIS, American Spinal Injury Association Scale; CRP, C-reactive protein; CTX, cyclophosphamide; ESR, erythrocyte sedimentation rate; LETM, longitudinal extensive transverse myelitis; MP, methylprednisolone; SDI, Systemic International Collaborating Clinics/ American College of Rheumatology Damage Index; SLEDAI-2K, the Systemic Lupus Erythematosus Disease Activity Index 2000; TM, transverse myelitis.

^aAn improved neurological outcome was defined as at least one-grade improvement in AIS after treatment.

^bAdjusted HR indicates hazard ratio adjusted for age at myelitis.

Figure 1.

Kaplan–Meier survival curves of systemic lupus erythematosus patients with transverse myelitis for severe myelitis-cause cumulative improvement rate.

Figure 2.

Kaplan–Meier survival curves of systemic lupus erythematosus patients with transverse myelitis for methylprednisolone pulse within 2-week-cause cumulative improvement rate.

Relapse rates of SLE-TM patients

The median follow-up time of all the patients is 2.00 years (IQR, 0.25–5.00 years). The 1-, 3- and 5-year relapse rates were 18.42% (7/38) (95% CI, 0.08–0.34), 37.04% (10/27) (95% CI, 0.19–0.58) and 56.25% (9/16) (95% CI, 0.30–0.80), respectively.

Discussion

As SLE-TM is rare, our current understanding of the underlying prognostic factors of TM is based on small sample-sized studies, and to the best of our knowledge, we are the first to explore the factors for SLE-TM risk. This study is the largest to date to have evaluated underlying factors for risk and prognosis of SLE-TM. Anti-Ro/SSA positivity and increased ESR may be associated with the presence of SLE-TM. Moreover, an initial presentation with severe myelitis (AIS grades A, B, or C at onset), hypoglycorrhachia, and delayed steroid pulse treatment appear to predict a worse neurological prognosis.

Previous studies revealed that most of the patients developed TM after the diagnosis of SLE,^1,4,19–21 which is easy to identify the cause. Nevertheless, in our studies, TM was the initial complaint in 8/58 (13.8%) patients. In these cases, it would be essential to diagnose the underlying SLE for prompt and adequate treatment. Hence, immunological screening for SLE, including ANA, anti-dsDNA, anti-Sm, and aPL antibodies screening, should be considered for patients presenting with TM.

The current understanding of the pathogenic mechanism of SLE-TM is limited to injuries resulting from vascular pathologies, including inflammatory or embolic/thrombotic/ischemic conditions. Some researchers believe that an autoimmunologic process is dominant, while others have postulated on the presence of thrombosis, fibrinoid arteries, perivasculitis, spinal cord softening, and peripheral white-matter degeneration at multiple spinal cord levels as causes of SLE-TM.^22,23 The presence of aPL antibodies always were considered as a possible etiology of SLE-TM through their thromboembolic effects on the microcirculation of the spine. ²⁴ However, the clinical value of aPL in SLE-TM remains controversial.^5,7,25 In our review of the literature, although the presence or absence of aPL was examined in the development of SLE-TM, the individual profiles of aPL (LA, aCL, and anti-β2GPI) have not been considered. Nevertheless, we found that aPL and different aPL profiles results were not predictive of the presence of the TM and short-term TM prognosis. Therefore, the role of aPL and different aPL profiles as risk and prognostic factors for SLE-TM is not yet clear and needs to be further investigated.

Anti-Ro/SSA positivity may be potential risk factors for the presence of SLE-TM in our study. Precious study showed that the target protein of anti-Ro/SSA antibodies also played an important role in the regulation of inflammation in lupus. ²⁶ Similarly, the presence of anti-Ro/SSA antibodies in SLE-TM has been shown to predict a relapse in the disease course. ²³ Some studies have associated the presence of anti-Ro/SSA antibodies with the development of neuropsychiatric lupus erythematosus (NPSLE). ²⁷ The value of anti-Ro/SSA positivity in the pathogenesis of SLE-TM remains unclear and anti-Ro/SSA antibodies requires further study.

Hypoglycorrhachia appears to be associated with an unfavorable neurological prognosis in this study. Undoubtedly, glucose metabolism is of great importance in central nervous function. Studies have found focal infiltration of the spinal cord by monocytes and CD4+ and CD8 + T lymphocytes, accompanied by activation of astrocytes and microglia in TM patients. ²⁸ CD4+ T-cells and proinflammatory microglia exert biological function by glycolysis, which relies on high levels of glucose uptake.^29–32 Thus, hypoglycorrhachia is considered to be a hallmark of inflammatory activity. It may give an indication of the underlying mechanisms of TM-related nerve injury in SLE. High-quality evidence is limited, and the exact pathogenesis awaits further exploration in future studies.

The initial severity of myelitis may be a strong predictive factor of SLE-TM prognosis. Most recovery occurs within the first 3 months from symptom onset, but improvement could continue for ⩾1 year. Our study could therefore provide risk stratification for TM management for those patients with severe myelitis and/or in the presence of hypoglycorrhachia who are at high risk of poor prognosis^4,5 may need to be treated more aggressively in the early stage. Notably, we did not identify a strong correlation between anti-dsDNA, hypocomplementemia, or SLEDAI scores and TM prognosis, which were reported in previous studies.^4,6,25 Differences in the sample populations across studies may have contributed to the noted contradiction.

MRI of the spine is very helpful as a diagnostic method to confirm SLE-TM.^33,34 In our study, the most commonly affected region was the thoracic spine.^19,35 We noticed that abnormality on the MRI was not always consistent with clinical manifestations. In our analysis, LETM was not found to be associated with a worse prognosis. However, the exact value of MRI findings, currently, in patients with SLE-TM is inconsistent, with heterogeneous findings having been reported, including ‘normal’ MRIs in patients with severe SLE-TM.^21,36,37

Glucocorticoids combined with immunosuppressants provide a basis to treat the inflammatory manifestations of SLE. However, there is no strong evidence for the optimal treatment because most are derived from case studies or extrapolation from trials involving TM patients with other etiologies. Nevertheless, we still propose that early aggressive treatment may be crucial for a favorable response.^38,39 The possible window period for the likelihood of a better outcome of SLE-TM was within 2 weeks of the onset of TM symptoms, using glucocorticoids pulse therapy. Early induction therapy, using high doses of glucocorticoids combined with CTX, has been used as the standard first-line treatment for SLE-TM. ⁴⁰ Saison et al. ⁶ reported that non-use of CTX was associated with unfavorable neurological outcomes in SLE-TM. However, we did not identify a specific effect of CTX on the prognosis of neurological outcomes in our study. Apart from the differences in sample populations, the time window from the onset of TM to the initiation of treatment, as well as appropriate administration of CTX at an adequate dose at TM onset may be an explanatory factor. Concerning anticoagulation therapy, a previous study did not show a positive effect on neurological outcomes of SLE-TM, even among aPL-positive patients. Furthermore, since the characteristics of TM (ischemic or vasculitic) are not easy to define, the effects of anti-aggregation or anticoagulation treatment would be a worthwhile subject for further study more carefully before drawing any solid conclusion.

Approximately, half of the SLE-TM patients have a combination of NMOSD in our cohort. However, the exact relationship between SLE-TM and NMOSD remains controversial. An international expert panel concluded that SLE in NMOSD is a coexistence rather than a complication. ¹⁷ This conclusion was mainly based on the fact that in patients with NMOSD and SLE, the central nervous system lesions are secondary to an astrocytopathy from AQP4-IgG, rather than the vasculitis found in SLE. Nevertheless, this hypothesis is inconsistent because NMOSD could occur in the context of established SLE, ⁴¹ which is also demonstrated in our study. Therefore, the relationship between NMOSD and SLE warrants further study before definite conclusions are reached.

TM is typically monophasic but relapsing TM within several months of the first event in 21–55% of patients. ⁴² Our studies have revealed that the 1-, 3- and 5-year relapse rates were 18.42% (95% CI, 0.08–0.34), 37.04% (95% CI, 0.19–0.58), and 56.25% (95% CI, 0.30–0.80), respectively. It can be seen clearly that the recurrence rate is relatively high in lupus myelitis, worth prompt treatment and monitoring the conditions closely.

Strengths

Our study has several strengths. First, it is the largest SLE-TM cohort to have been evaluated to date. Second, this is the first study to explore the risk factors for lupus-related myelitis. Finally, our findings may yield hypotheses on the risk and prognosis of SLE-TM, contributing to evidence on the prediction of risk and prognosis of SLE-TM to inform future research and clinical practice.

Limitations

First, this was an observational retrospective study in nature. Second, this was a relatively small sample size. Hence, the statistical power could have been weakened, especially in prognostic analysis. Third, both the case and the control groups included inpatients. Therefore, the potential selection bias was inevitable. In addition, the lack of widespread use of scales, such as AIS, especially by non-neurologists, would lead to the absence of long-term neurological prognosis assessment.

Conclusion

The positive anti-Ro/SSA antibodies and increased ESR may be risk factors for the presence of SLE-TM. An initial presentation with severe myelitis and hypoglycorrhachia might be predictors of a poor neurological outcome. Early aggressive treatment with MP pulse may improve the prognosis.