Introduction

Neuromyelitis optica spectrum disorder (NMOSD) is a type of autoimmune inflammatory central nervous system (CNS) disease characterized by optic neuritis and/or acute myelitis as well as a group of brain syndromes.¹ NMOSD, different from multiple sclerosis (MS), is primarily associated with aquaporin-4 immunoglobulin G (AQP4-IgG).² In MS, the presence of new asymptomatic lesions on MRI aids in diagnosing the disease in patients with clinically isolated syndromes and can be used for therapeutic monitoring and prognostic assessment.³ In contrast, disability accumulation in NMOSD seems closely linked to attacks, with asymptomatic lesions indicating ongoing disease activity found in only a small fraction of NMOSD patients.^4-10 These asymptomatic lesions can emerge during attacks or remission, suggesting MRI's heightened sensitivity compared to clinical symptoms. Performing a one-off MRI scan following an acute episode seems reasonable, allowing a reestablishment of the imaging baseline and enabling comparisons with future relapses.^{11, 12} The Latin American NMOSD Consensus recently recommended annual brain MRIs for NMOSD patients after starting immunosuppressive therapy to complement clinical follow-ups.¹³

Acute myelitis, a primary core symptom of NMOSD, frequently presents as longitudinally extensive transverse myelitis (LETM) in about 70% of cases and as short transverse myelitis (STM) in roughly 20% of cases.^{1, 13, 14} Studies on asymptomatic optic nerve and brain lesions are predominant in NMOSD, whereas fewer studies have been conducted especially on asymptomatic spinal lesions (ASLs) concerning the incidence, implications for treatment failure, relationship with subsequent relapses, and management strategy. Currently, there is no global consensus regarding the use of spinal MRI in NMOSD follow-ups (unless a clinical relapse occurs). Furthermore, a previous study showed that in AQP4-IgG-positive NMOSD (AQP4-NMOSD), reductions in the mean upper cervical cord area could be detected even without a prior history of clinical relapses or spinal cord lesions on MRI, suggesting that there may be a subtle or subclinical inflammatory pathological process within the spinal cord.¹⁵ Given accumulating evidence of asymptomatic tissue damage in NMOSD, the role of MRI in routine patient monitoring warrants reconsideration. Therefore, this study aims to investigate the frequency and influencing factors of ASLs and their impact on subsequent relapses by retrospectively collecting clinical and spinal MRI data from Chinese AQP4-NMOSD patients in a real-world setting.

Methods

Results

Baseline clinical features

Over the span of 9 years, a total of 282 NMOSD patients were identified. Among them, 129 AQP4-NMOSD patients with available spinal MRI imaging were ultimately enrolled (Fig. 1), including 19 males and 110 females, with a median onset age of 42 years (range: 8–78) and a median enrollment age of 44 years (range: 19–78). Out of these, 22 patients presented concurrent autoimmune diseases, including 6 with systemic lupus erythematosus (1 with Sjogren's syndrome, 2 with Hashimoto's thyroiditis, and 1 with anticardiolipin antibody syndrome), 12 with Sjogren's syndrome (1 with autoimmune hepatitis), 2 with myasthenia gravis, 1 with Grave's disease, and 1 with autoimmune hepatitis and primary biliary cholangitis.

Acute myelitis (72 patients, 55.8%) and optic neuritis (38 patients, 29.5%) constituted the most prevalent initial core clinical symptoms, followed by area postrema syndrome (APS) (9 patients, 7%) and a combination of optic neuritis and acute myelitis (4 patients, 3.1%). Furthermore, a minor percentage of patients exhibited diverse core clinical symptoms such as APS combined with acute myelitis, cerebral syndrome, acute brainstem syndrome, acute mesencephalic syndrome, and acute myelitis combined with brainstem and cerebral syndrome (Table 1). The median number of attacks before enrollment was 1 (range: 0–11). Up until 30 June 2023, 3 patients were lost to follow-up, and 1 patient had passed away. The median follow-up period for the 129 AQP4-NMOSD patients was 42.8 (38.7) months.

Table 1. Cohort characteristics of 129 AQP4-NMOSD patients.

Age, median (range), years
At onset	42 (8–78)
At enrollment	44 (19–78)
Gender, n (%)
Female	110 (85.3)
Male	19 (14.7)
Concomitant autoimmune diseases, n (%)	22 (17.1)
Number of attacks before enrollment, median (range)	1 (0–11)
Follow-up duration, median (IQR), months	42.8 (38.7)
Initial core clinical symptoms, n (%)
Optic neuritis	38 (29.5)
Acute myelitis	72 (55.8)
APS	9 (7)
Cerebral syndrome	1 (0.8)
Acute mesencephalic syndrome	1 (0.8)
Acute brainstem syndrome	1 (0.8)
Diverse core clinical symptoms	7 (5.4)
Period of onset at enrollment, n (%)
Acute phase	60 (46.5)
Relapse phase	61 (47.3)
Remission phase	8 (6.2)
Number of attack spinal MRIs during follow-up	173 (121)
Number of remission spinal MRIs during follow-up	89 (68)
Therapy at enrollment, n (%)
None	91 (70.5)
Steroids (including prednisone and methylprednisolone)	27 (20.9)
AZA ± steroids	6 (4.7)
MMF ± steroids	3 (2.3)
RTX ± steroids	2 (1.6)

• APS, area postrema syndrome; AZA, azathioprine; IQR, interquartile range; MMF, mycophenolate mofetil; RTX, rituximab.

Cox proportional hazard analysis of determinants of subsequent relapses in AQP4-NMOSD patients

After a median follow-up of 13.3 (20.6) months among 68 AQP4-NMOSD patients, 14 experienced relapses. In patients with ASLs, the median time from remission scan to subsequent relapse was 15.9 months (range: 3.5–31.4 months). Kaplan–Meier curve analysis indicated a notable distinction in relapse risk between those with ASLs and those without (p = 0.003, log-rank test) (Fig. 2A). In Cox proportional hazard analysis, patients with ASLs displayed a significantly higher relapse risk compared to those without (HR = 4.658, 95% CI: 1.519–14.285, p = 0.007). Similarly, differing times to subsequent relapse were evident across various immunosuppressive agents (p < 0.001, log-rank test) (Fig. 2B). In contrast to individuals not using immunosuppressive agents, those using AZA exhibited a reduced but statistically nonsignificant relapse risk (HR = 0.320, 95% CI: 0.087–1.182, p = 0.087), whereas MMF (HR = 0.027, 95% CI: 0.003–0.260, p = 0.002) and RTX users (HR = 0.035, 95% CI: 0.006–0.203, p < 0.001) experienced a significantly lower relapse risk. Following adjustment for factors like the number of attacks before enrollment and immunosuppressive agents, the presence of ASLs exhibited no statistically significant impact on subsequent relapses (HR = 2.297, 95% CI: 0.652–8.085, p = 0.195). However, the number of attacks before enrollment exhibited a significant capacity to increase the risk of subsequent relapses (HR = 1.299, 95% CI: 1.019–1.657, p = 0.035) (Table 3).

Table 3. Cox proportional-hazard analysis of determinants of subsequent relapses.

Variables	Unadjusted HR (95% CI)	p value	Adjusted HR (95% CI)	p value
Age at onset	1.016 (0.981–1.052)	0.366
Female, n (%)	0.784 (0.175–3.520)	0.751
Concomitant autoimmune diseases	0.208 (0.027–1.599)	0.131
Initial core clinical symptoms, n (%)
Optic neuritis	Ref
Acute myelitis	5.666 (0.741–43.355)	0.095
APS	–	–
Acute mesencephalic syndrome	–	–
Acute brainstem syndrome	–	–
Number of attacks before enrollment	1.158 (0.959–1.397)	0.127	1.299 (1.019–1.657)	0.035
Period of onset at enrollment, n (%)
Initial attack phase	Ref
Relapse phase	0.941 (0.329–2.689)	0.909
Remission phase	–	–
Spinal cord segment involved
None	–	–
STM	0.885 (0.197–3.983)	0.873
LETM	Ref
Remission ASLs	4.658 (1.519–14.285)	0.007	2.297 (0.652–8.085)	0.195
Therapy at remission MRI scan
None	Ref		Ref
AZA ± steroids	0.320 (0.087–1.182)	0.087	0.335 (0.082–1.366)	0.127
MMF ± steroids	0.027 (0.003–0.260)	0.002	0.027 (0.003–0.282)	0.003
RTX ± steroids	0.035 (0.006–0.203)	<0.001	0.037 (0.005–0.275)	0.001

• HR, hazard ratio; 95% CI, 95% confidence interval; APS, area postrema syndrome; ASLs, asymptomatic spinal lesions; AZA, azathioprine; LETM, longitudinally extensive transverse myelitis; MMF, mycophenolate mofetil; RTX, rituximab; STM, short transverse myelitis.

Management strategies and follow-up after ASL detection

Among 8 patients with ASLs, Case 1 presented an extension of a preexisting C2 lesion to C3–4 without enhancement, continued on prednisone, and relapsed after 3.5 months. In Cases 2 and 3, an enlarged preexisting lesion and a new thoracic segment lesion were observed, respectively, without enhanced scans. Acute phase strategies were not applied, but they persisted on AZA, later altering prednisone dosages for maintenance. Unfortunately, both experienced relapses over a year later. Enhanced ASLs were identified in Cases 4–8. Case 4 received only 40 mg of intravenous methylprednisolone (IVMP) and later switched to MMF, relapsing after 5.8 months. Cases 5 and 7 remained relapse-free with sequential MMF following methylprednisolone pulse therapy (MPPT). In Case 6, significant atrophy and enhancement of the initial spinal lesion transpired; she did not relapse following MMF up-dosing with prednisone. Case 8 was treated with CTX at the time of ASL detection. He continued to use CTX without receiving IVMP and solely used prednisone following the CTX course, relapsing 14.8 months after ASL detection (Fig. 3 and Table 4).

Table 4. Management strategies and outcomes in 8 patients with ASLs.

Case No.	Therapy at remission MRI scan	Time since reference MRI, months	ASL location and features	Therapy after ASL detection	Outcomes
1	Prednisone 10 mg once daily	8.8	Extension of a preexisting C2 lesion to C3–4 (without enhancement)	Prednisone 10 mg once daily	T-segment myelitis relapse 3.5 months later
2	AZA 50 mg twice daily	7.6	Enlargement of previous C2–3 lesion (without enhanced MRI scan)	AZA 50 mg twice daily and prednisone 10 mg once daily	C-segment myelitis relapse 18.4 months later
3	AZA 50 mg twice daily and prednisone 10 mg once daily (irregularly)	22.2	New T5–6 lesion (without enhanced MRI scan)	AZA 50 mg twice daily and prednisone 30 mg once daily and tapered	C- and T-segment myelitis relapse 17 months later
4	AZA 50 mg twice daily and prednisone 5 mg once daily	11	C- and T-segmental spinal lesions with speckled enhancement	IVMP 40 mg; MMF 0.5 g twice daily and prednisone 5 mg once daily	Optic neuritis relapse 5.8 months
5	AZA 25 mg twice daily and prednisone 5 mg once daily	19.2	T1–6 lesion with a slight enhancement	IVMP 1000 mg; MMF 0.5 g twice daily and prednisone 60 mg once daily and tapered	No relapse
6	MMF 0.5 g twice daily and pred 5–10 mg once daily	10.8	T9–T12 spinal cord atrophy with lesion enhancement	MMF 0.75 g twice daily and prednisone 10 mg once daily	No relapse
7	MMF 0.75 g twice daily and prednisone 5 mg once daily	53.3	C- and T-segmental spinal lesions with C3–6 enhancement	IVMP 500 mg; MMF 0.75 g twice daily and prednisone 10 mg once daily	No relapse
8	CTX 0.6 g every month and prednisone 5 mg once daily	6.1	Medulla oblongata and upper C-segmental lesions with a slight enhancement	Prednisone 10 mg once daily following the CTX course	C-segment myelitis relapse 14.8 months later

• ASL, asymptomatic spinal lesion; AZA, azathioprine; CTX, cyclophosphamide; IVMP, intravenous methylprednisolone; MMF, mycophenolate mofetil.

Discussion

In this retrospective cohort study conducted in a real-world setting, we found that ASLs may be observed among patients with AQP4-NMOSD, occurring in approximately 5.3% (14 out of 262) of cases. The presence of ASLs was found to be linked to the use of immunosuppressive agents, primarily observed in patients on conventional ones, whereas no instances of ASLs emerged among patients using RTX. Kaplan–Meier curve analysis indicated a significantly elevated risk of relapse in patients with ASLs compared to those without. However, subsequent analysis after adjusting for variables like the number of attacks before enrollment and immunosuppressive agents demonstrated that the presence of ASLs did not exhibit a statistically significant impact on subsequent relapses. This suggests that the primary determinant of ASLs and subsequent relapse is the choice of immunosuppressive agents.

ASLs refer to newly detected silent lesions or enlargement or sustained enhancement of pre-existing lesions on spinal MRI, even in the absence of new neurological deficits. In this context, we encompass enhanced lesions on remission spinal MRI 6 months or longer from the last episode as ASLs, as enhanced lesions frequently denote potential disease activity, even if clinical symptoms are lacking. The actual occurrence rate of ASLs in AQP4-NMOSD patients remains uncertain. Paolilo et al.⁴ monitored MRI findings of 16 pediatric AQP4-NMOSD patients, identifying ASLs in one case (6.3%), indicating a potentially higher incidence of ASLs. However, in a cohort of 220 AQP4-NMOSD cases, attack-associated silent lesions were mainly localized in the brain, with only three attack ASLs discovered among 331 spinal MRIs (0.9%); new ASLs during remission were identified in 1 of 116 patients (0.9%) and 2 of 200 spinal cord remission MRIs (1%).⁷ In the present study, we observed six ASLs in 6 patients (6 out of 173, 3.5%) out of 121 patients with 173 spinal MRIs during acute attacks, as well as eight ASLs in 8 patients (8 out of 89, 9%) among 68 patients with 89 spinal MRIs during remission. Nonetheless, the infrequency of regular MRI scans might lead to the oversight of certain ASLs. A preliminary analysis from the N-MOmentum trial (inebilizumab study) indicated that nearly 18% of NMOSD patients displayed ASLs on gadolinium contrast-enhanced MRI during inter-attack surveillance MRI.¹⁷ These findings, combined with prior research, suggest that the actual incidence of ASLs might be higher than initially perceived, albeit with reduced concern.

The occurrence of ASLs in NMOSD patients can be influenced by various factors, including the frequency of remission MRI scans, short-segmental spinal cord lesions, the type of immunosuppressants used, and non-standardized treatment approaches. As previously noted, systematic and rigorous surveillance tends to reveal asymptomatic lesions more frequently.¹⁷ Among asymptomatic lesions, ASL ranks as the second most prevalent type in NMOSD, trailing behind asymptomatic optic nerve lesions occurring during relapse-free periods and attacks, particularly in cases with short-segmental spinal cord lesions.¹⁷ Flanagan et al. documented three NMOSD patients displaying ASLs, all of whom exhibited enhanced cervical or thoracic cord lesions of a short segment.⁵ In our study, new or enlarged lesions were mostly short-segmented (Cases 1, 2, and 3), accounting for 37.5% of all ASLs, while the persistent enhancement of preexisting spinal lesions was not necessarily short-segmented, being mainly related to the length of the previous lesions. A study involving 165 individuals reported 27 MRIs (8%) from 24 patients (15%) displaying acute asymptomatic NMOSD-typical brain lesions; most asymptomatic brain lesions occurred before immunosuppressive therapy, with a few arising post-immunotherapies.⁹ The study also revealed the detection of five asymptomatic brain lesions from 5 patients during interferon treatment (for MS).⁹ These findings imply that appropriate and standardized immunosuppressive therapy may significantly diminish the likelihood of asymptomatic lesions. Our study identified a higher proportion of ASLs among AQP4-NMOSD patients who were either not employing immunosuppressants, using conventional agents like AZA or CTX, or irregularly taking such medications. Conversely, no ASLs were detected in patients using RTX, indicating that the utilization of more potent monoclonal antibody analogs notably mitigates the risk of ASLs. Additionally, irregular immunosuppressant usage emerged as a significant factor affecting ASLs. For instance, in Case 3, irregular AZA intake after 1 year of clinical stabilization led to the discovery of a thoracic lesion on repeat spinal MRI, despite the absence of new clinical symptoms. Similarly, Case 5, who only took AZA 25 mg twice daily after approximately 2.5 years of clinical stabilization, exhibited T1–6 spinal enhanced lesions during routine MRI reviews. This suggests that subtherapeutic dosages could contribute to ASL occurrence. Presently, no laboratory or radiological biomarkers exist to predict an inadequate response to immunosuppressive treatment.¹⁸ The potential of using ASLs as an imaging marker for evaluating subclinical disease activity and monitoring immunosuppressive treatment response merits further investigation.

Prior research has highlighted that AQP4-NMOSD patients may not remain “silent” in the presence of asymptomatic lesions during remission, which may have implications for future relapses.⁷ Recent findings from an NMOSD cohort study indicated that asymptomatic lesions during remission often anticipated imminent relapses, with the location of these lesions providing insight into subsequent relapse symptoms. Notably, patients with asymptomatic lesions experienced relapses in a median time of 5 months after the remission scans (with relapses occurring at 1, 2, 3, 5, 6, 6, and 7 months). In contrast, patients without asymptomatic lesions had a substantially longer median time of 85 months from remission scans to subsequent relapses. This suggests that the presence of asymptomatic lesions significantly elevates the risk of future relapses (HR = 21.23, 95% CI: 8.05–53.65).⁷ Consistently, our current study also demonstrated a markedly increased risk of relapse in patients with ASLs compared to those without (HR = 4.658, 95% CI: 1.519–14.285). This underscores the possibility that ASLs may signify an underlying inflammatory activity that could potentially evolve into a clinical relapse if not identified and addressed on time. However, the presence of ASLs did not exhibit a statistically significant impact on subsequent relapses in multivariate analyses. Another study also showed that the presence of new asymptomatic MRI lesions during the relapse-free period was not associated with a shorter time to develop subsequent relapses.¹⁰ We speculated that this might be related to treatments and immunosuppressant switches after ASL detection. Furthermore, previous investigations have indicated that a higher frequency of prior relapses is linked to an elevated risk of subsequent relapses.^19-21 In alignment with this, our study also revealed that the number of attacks experienced before enrollment was significantly associated with a higher risk of subsequent relapses. In this context, considering ASLs as manifestations of underlying subclinical acute attacks could elucidate the notable increase in the risk of subsequent relapses.

Undoubtedly, the management of ASLs during acute attacks aligns with acute phase treatment strategies following primary symptomatic attacks. However, limited research exists on how to effectively manage ASLs during remission. In our investigation, we delved into the treatment of NMOSD patients after ASL discovery and its influence on subsequent relapses. Intriguingly, we observed that patients who received high-dose IVMP following ASL detection did not experience relapses. Conversely, those who did not undergo this therapy faced a heightened risk of subsequent relapses (as illustrated in Table 4). Additionally, the risk of relapse was notably diminished in patients on MMF in comparison to those on AZA. This risk reduction was further evident in patients who switched to MMF, a more potent immunosuppressant, as exemplified by Cases 5 and 7. Following ASL identification, Case 6 remained relapse-free after augmenting MMF to the high therapeutic dose. In contrast, Case 4, characterized by recurrent relapses, experienced a relapse shortly after switching to MMF at a lower dosage. In light of these findings, it appears reasonable to tailor ASL management in NMOSD patients based on acute phase therapy principles if enhanced MRI confirms that the ASL represents an active lesion (reflecting the blood–brain barrier destruction). Conversely, if an enhanced MRI indicates an inactive lesion (e.g., absence of enhancement), the ASL may not necessitate management according to acute phase therapy guidelines. Nonetheless, both scenarios might signal the inadequacy of the previous immunotherapy regimen in controlling the condition, warranting dose adjustments or medication switches.²² Flanagan et al.⁵ also documented three cases with ASLs. In the first two cases, short-segmental cervical enhanced lesions were observed during acute episodes of optic neuritis and APS, respectively. These lesions disappeared following MPPT treatment, as verified by subsequent MRIs conducted 2 months and 10 days later. In the third case, the patient was not treated with MPPT but started on AZA after the detection of short-segment thoracic enhanced lesions; unfortunately, the patient developed a symptomatic long-segmental cervical lesion after 2 months. These observations suggest that the post-ASL relapse occurrence might hinge on the subsequent management strategy. However, further longitudinal investigations involving larger sample sizes are imperative to validate these findings.

This study represents the first comprehensive analysis of ASL incidence, associated factors, and their influence on subsequent relapses, meanwhile offering preliminary insights into ASL management. Nevertheless, several limitations warrant consideration. First, being a retrospective cohort study, its findings should be approached with caution due to inherent biases and confounders. Second, the study predominantly examined spinal MRI data evolution across varying scanners and medical centers, each with distinct magnetic field strengths, parameters, and intervals, potentially introducing variability. Furthermore, due to irregular scans and inaccessible MRIs from external institutions, some ASL instances might have been overlooked. Third, our focus was solely on ASLs, without considering asymptomatic optic nerve or brain lesions, which could also impact subsequent relapses, an aspect we could not exclude. Last, the lack of a standardized ASL management strategy poses an additional consideration in subsequent relapses; however, due to the limited ASL cases, further subgroup analyses on its impact were unfeasible, allowing only preliminary observations from real-world clinical practice.

Conclusions

In summary, our findings underscore the relatively frequent occurrence of remission ASLs, primarily prevalent in AQP4-NMOSD patients using conventional immunosuppressive agents, thereby heightening the risk of subsequent relapses. Given the pronounced activity of NMOSD, assessing immunotherapy efficacy solely based on clinical data might inadequately capture disease activity and progression. The presence of ASLs may signify an underlying inflammatory activity due to insufficient immunotherapy. Hence, extensive multicenter longitudinal studies, incorporating regular follow-up spinal MRIs, are imperative to elucidate ASL incidence during follow-up, their implications for subsequent relapses and disability progression, and the formulation of ASL management strategies, including detecting potential subclinical disease activity and optimizing treatment shifts.

Author Contributions

CS and ZY: study design, patient selection, analysis and acquisition of data, and writing the manuscript; WX: study design, patient selection, and acquisition and interpretation of data; DJ: patient selection, analysis of imaging data, and revising the manuscript; XS: acquisition of data; CP: analysis of imaging data; LQ and XM: interpretation of data and revising the manuscript; TY and XQ: study design, interpretation of data, and revising the manuscript.

Funding Information

This study was supported by grants from the Natural Science Foundation of Jiangsu Province (BK20211075) and the Key Research and Development Plan of Jiangsu Province (BE2019666).

Conflict of Interest

The authors have no conflicts of interest to declare.