Clinical data and assessment

The questionnaire comprised questions on demographics, disease course, previous relapses (history of ON and myelitis), current therapy, spasticity, as well as pain, depression, and health-related quality of life (hr-QoL) assessment scores (PainDetect Questionnaire [PDQ], Short Form of the Brief Pain Inventory [SF-BPI], McGill Pain Questionnaire Short Form [MPQ-SF], Beck Depression Inventory II [BDI-II], and Short Form 36 Health Survey [SF-36]).

Spasticity was assessed by asking if the patient experienced attacks or short episodes (<1 min) of intensive pain, accompanied by an increased tone (increased muscular tension) and cramps in the arms/legs.

The questions of the SF-BPI consist of two categories: (i) pain severity (present, highest, least, and average pain intensity) based on a numeric rating scale from 0 (no pain) to 10 (worst pain imaginable) within the last week. The pain severity index (PSI) represents the average score of the four pain intensity scores; (ii) seven domains of pain-related interference with daily life, rated from 0 (no interference) to 10 (complete interference): general activity, mood, walking ability, working ability, relations with other people, sleep, and enjoyment of life.

The PDQ was administered to ask about pain localization and to discriminate between neuropathic (PDQ score, 19–38) and nociceptive pain (PDQ score, 0–12). PDQ score from 13 to 18 is considered to be possibly neuropathic [18]. We performed comparative analysis between patients with definitive neuropathic and nociceptive pain.

The MPQ-SF consists of 15 words describing sensory (n = 11) and affective (n = 4) components of pain. Patients rate the intensity of the respective pain quality as 0 = none, 1 = mild, 2 = moderate, or 3 = severe [19].

The BDI-II ranges from 0 (best) to 63 (worst) (<9: no depressive affect; 9–13: minimal mood disturbance; 14–20: mild depression, 21–28: moderate depression; ≥29: severe depression). Clinically relevant depression was defined by a BDI score ≥14 [20].

SF-36 measures hr-QoL. It consists of 36 items in eight subscales with components of mental and physical health. A physical component summary (PCS) and a mental component summary (MCS) were calculated using norm-based attaining values from 0 (worst) to 100 (best) [21].

Patient-reported gait function (scored from 0 [best] to 3 [worst]; 0: no restriction; 1: >500 m without rest, 2: <500 m without rest, 3: inability to walk) and visual function (scored from 0 [best] to 4 [worst]; 0: no restriction; 1: restricted when reading, writing, or working on a PC only; 2: restricted during simple activities of daily life at home; 3: very bad vision, no orientation in a foreign environment; 4: shadow vision, blindness) were assessed corresponding to the questionnaires’ relapse history (ON: visual function; myelitis: gait function).

In addition to the patient-reported questionnaire, the Expanded Disability Status Scale (EDSS) was evaluated if available at the respective study center. The EDSS was assessed within 12 months relative to the questionnaire. Therefore, the analysis was focused on self-reported gait and visual function at the time of assessment.

Testing for serum MOG-abs was performed per local protocols using established cell-based assays [2, 8].

Statistics

Statistical analysis was performed using R (version 3.4.0; The R Foundation for Statistical Computing) [22]. To investigate group differences, we used the Fisher exact test for categorical variables (attack history, BDI-II score categories, sex, group comparisons of patients with intermittent vs. permanent pain), Mann-Whitney-Wilcoxon rank sum test for not normally distributed continuous variables (age, disease duration, self-reported gait and visual function, McGill Pain Questionnaire [MPQ] scores, measures of pain intensity, number of attacks, SF-36 scores, SF-BPI scores).

To check for correlations between measures of pain intensity and depression scores and for correlations between measures of pain intensity, depression scores, self-reported physical impairment, and PCS and MCS, respectively, we performed a Spearman correlation test. We performed an additional correlation analysis of PCS and MCS with the EDSS as an objective impairment score. A robust regression model [23] was used to determine predictors of hr-QoL with SF-36 PCS and MCS score.

For all models and corresponding statistical tests, statistical significance was set at p < 0.05. This study was exploratory, without an a priori sample size calculation and adjustments for multiple testing.

RESULTS

Description of cohort

Forty-three adult patients with MOGAD were included in the analysis. Table 1 provides the main demographic and clinical characteristics of the study participants. Current immunotherapy comprised azathioprine (n = 5), glatiramer acetate (n = 1), intravenous or subcutaneous immunoglobulin (n = 3), methotrexate (n = 2), mitoxantrone (n = 1), mycophenolate mofetil (n = 1), prednisolone (n = 1), and rituximab (n = 19).

Prevalence of pain in MOGAD

Pain was a frequent symptom in our cohort. Overall, 22 of 43 (51%) patients suffered from chronic MOGAD-related pain. Eighteen patients (42%) reported painful symptoms as part of their previous attacks. Of these, nine patients had transient pain only, and nine patients had persistent pain. Attack-related pain comprised myelitis-associated pain (n = 6), ON-associated pain (n = 7), and nonspecific pain like headache (n = 4), neck pain (n = 1), and generalized body pain (n = 1).

Pain quality, intensity, and localization of relapse-independent pain

The following analysis of pain syndromes was performed in patients with relapse-independent MOGAD-related pain (n = 22). Pain was mainly described as cramping (n = 11), stabbing (n = 15), tender (n = 17), aching (n = 19) (sensory dimensions), and tiring/exhausting (n = 20) (affective dimension). Median present pain intensity was 3/10 (minimum–maximum: 0–9), median maximum and average pain intensity in the previous week were 6/10 (minimum–maximum: 1–9) and 3.5/10 (minimum–maximum: 1–9), respectively.

Of 22 patients with MOGAD-related pain, 12 suffered from permanent pain and nine from intermittent pain attacks. For one patient, no information was available. Six patients suffered from both permanent pain and pain attacks. Patients with pain attacks had the most severe pain: median PSI 4.4 (minimum–maximum: 2.5–9) in those with additional permanent pain and 4.3 (minimum–maximum: 1.5–6.5) in those without. Patients with persistent pain without attacks had the lowest median PSI: 2.1 (minimum–maximum: 1–4). Pain was localized mostly in the legs (n = 14, n = 11 bilaterally), following by back pain (n = 12), arm pain (n = 8, n = 4 bilaterally), head/neck pain (n = 6), and anterior trunk pain (n = 3).

Pain-associated factors

Demographics were similar in patients with and without pain (Table 2). However, patients with MOGAD-related pain were older at disease onset (p = 0.039) and clinically more impaired than patients without pain (self-reported gait function: p = 0.005). Moreover, clinically relevant depression was more frequent in pain sufferers (p = 0.034).

Neuropathic and nociceptive pain

Eight of 22 patients with MOGAD-related pain fulfilled the PDQ criteria for definite neuropathic pain, three patients had a possible neuropathic pain component, and 11 patients had nociceptive pain.

As expected, patients with definite neuropathic pain had higher pain intensity scores than those with nociceptive pain (Table 3). They described their pain experience more often as shooting (p < 0.001), sharp (0.014), hot–burning (p = 0.018) (sensory dimensions), and fearful (p = 0.032) (affective dimension). Sensory and affective pain scores were also higher in patients with neuropathic pain. The groups did not differ with respect to age, sex, disease duration, number of relapses, annualized relapse rate, number with ON, and number with myelitis. However, patients with definite neuropathic pain had more severe gait restriction (p = 0.032) than patients with nociceptive pain.

Spasticity-associated pain

Fifteen patients reported spasticity-associated pain (SAP). SAP was located in one (n = 6) or both legs (n = 9), or in both arms and in the back for one patient, respectively. Four patients (18.2%) presented short-lasting PTS. Patients with SAP had a trend toward higher numbers of previous relapses (p = 0.051) and surprisingly shorter disease duration (p = 0.048) than patients without SAP (Table 4). Patients with and without SAP did not differ with respect to demographics and depression prevalence.

Depression is more prevalent in pain sufferers

Eighteen patients (41.9%) had signs of clinically relevant depression that was at least moderate (BDI ≥ 21) in 10 (23.3%) cases. Clinically relevant depression was more prevalent in pain sufferers than in patients without disease-related pain (59.1% vs. 25%, p = 0.034; Table 2). Demographics, disease duration, and relapse activity did not differ between patients with and without clinically relevant depression (Table 5). The depression score did not differ between patients with nociceptive versus neuropathic pain (Table 3) or between patients with permanent and intermittent pain (odds ratio: 1.1; 95% confidence interval: 0.1–8.8; p = 1.0). However, depression severity correlated with a visual (ρ = 0.401, p = 0.009) and gait (ρ = 0.333, p = 0.041) impairment as well as pain intensity (PSI: ρ = 0.392, p = 0.010).

Pain and depression strongly affect hr-QoL and ADL in MOGAD

Both physical and mental components of the QoL were altered in patients with MOGAD, with a median MCS of 43.2 (minimum–maximum: 21.1–63.9) and a PCS of 42.4 (minimum–maximum: 9.4–65.8) compared to the general population [21]. The physical component score was lower in patients with pain than in pain-free patients (Table 2). Moreover, this score was lower in patients with neuropathic pain than in patients with nociceptive pain (Table 3). Also, patients with depression had significantly lower MCS and PCS scores than patients without depression (Table 5).

Pain strongly affected all aspects of ADL: general activity, 5.5/10 (minimum–maximum: 1.5–10); walking ability, 5.5/10 (minimum–maximum: 0–10); mood, 5/10 (minimum–maximum: 1–10); working ability, 5/10 (minimum–maximum: 1–10); sleep 4/10 (minimum–maximum: 0–9); enjoyment of life, 3.5/10 (minimum–maximum: 0–8); and relations with other people 3/10 (minimum–maximum: 0–10).

Pain severity (ρ = −0.798, p < 0.001), gait (ρ = −0.690, p < 0.001), and visual impairment (ρ = −0.370, p = 0.024), as well as depression score (ρ = −0.361, p = 0.022) correlated with PCS. In a regression model including these four factors, pain severity (B = −5.455, SE = 0.810, p < 0.001), visual function (B = −8.163, SE = 1.742, p < 0.001), and gait function (B = −5.756, SE = 1.875, p = 0.005), but not depression severity (B = 1.029, SE = 0.963, p = 0.294), were significant predictors for physical QoL. The EDSS correlated with PCS (ρ = −0.720, p < 0.001) but not with MCS (ρ = −0.223, p = 0.220).

Clinically relevant depression (B = −15.484, SE = 2.896, p < 0.001) was the only significant predictor for low mental QoL.

Symptomatic pain medication

Only 12 (54.5%) patients with MOGAD-related pain received pain medications. Treatment comprised nonsteroidal anti-inflammatory drugs and paracetamol (acetaminophen) (n = 10), anticonvulsants and/or antidepressants (n = 7), opioids (n = 2), and antispastic medication (n = 4). Five of 11 patients only retrospectively reported more than 50% reduction of pain intensity. Patients with nociceptive pain received no (n = 6), one (n = 3), or two (n = 1) different pain medications. Six of 11 patients with definite or possible neuropathic pain received three to six different types of pain medication. The PSI correlated with the number of analgesic drugs (ρ = 0.536; p = 0.012).

Four of 15 patients with SAP received an antispastic medication. All of them still reported spasticity-associated pain attacks four, 10, 12, and 60 times a month.

DISCUSSION

The present study highlights the importance of pain and depression in MOGAD. Over 40% of patients from this German multicenter cohort suffered from depression, and over 50% had nociceptive or neuropathic pain. Both comorbidities severely affected the patients’ QoL and were often untreated or resistant to treatment.

Recently, MOGAD has been acknowledged as a disease entity on its own, distinct from multiple sclerosis (MS) and NMOSD [5, 7, 8]. Underlying pathogenetic mechanisms, clinical presentation, attack severity, and remission rate are partly different in these diseases [5, 7]. Differences on pain characteristics have also been supposed; however, specific data on pain in MOGAD are scarce [4, 12, 16, 24, 25]. Previous research mentioned pain mainly as part of acute symptoms in MOGAD attacks and focused on ON-related pain and neuropathic pain during relapses [2, 4, 14, 24-27]. In our cohort, 18 patients (42%) mentioned painful symptoms as part of their previous attacks. Most importantly, even more patients (51%) suffer from chronic MOGAD-related pain. Consequently, pain prevalence in MOGAD is similar to those in MS (estimated as 50%) and lower than in NMOSD, where pain prevalence has been reported to be as high as 80% to 86% [10-12, 28]. However, estimates can differ in small studies, most likely due to heterogeneities of patient cohorts, screening instruments, and diverging pain classifications.

The prevalence of neuropathic pain in MOGAD probably exceeds that in MS [12, 29]. About 26% of our patients suffered from neuropathic pain. In patients with MS, a prevalence of 29% has been reported in a meta-analysis, but a lower prevalence of 15% was found when more specific diagnostic criteria were used [30]. A recent study using the PDQ reported a prevalence of only 5% (compared to 18% in our cohort) of definite neuropathic pain in MS after a disease duration of 4 years [30]. In our cohort, patients with neuropathic pain were particularly disabled. Although average pain intensity in the whole cohort was mild, patients with neuropathic pain suffered from more severe pains, similar to those reported in NMOSD [12, 29]. Neuropathic pain was significantly stronger and affected the patients’ general activity, walking, and working capacities as well as relations to other people, compared to nociceptive pain.

Thirty-five percent of our MOGAD cohort suffered from SAP, and PTS occurred in about 9% of the patients. These numbers are lower than in NMOSD, where PTS occurs with a prevalence of 25% to 40% [13, 31-33].

Both neuropathic pain and SAP are most likely a consequence of a spinal cord damage. Although spinal cord lesions in NMOSD are typically long and extensive [34, 35], spinal cord lesions in MOGAD can be smaller and less destructive. Still, due to the highly prevalent central lesions location, they may frequently involve relevant structures of the pro- and antinociceptive system, resulting in chronic therapy-refractory pain and/or dysesthesia [4]. As reported for patients with AQP4-IgG positive NMOSD [36], the number of previous myelitis attacks did not differ between patients with and without pain. Probably not a number of relapses, but precise axial lesion location, extension, and extent of tissue damage are decisive for chronic neuropathic and spasticity-associated pain. It has been shown that central neuropathic pain can be induced by oligodendrocyte death and axonal pathology in the spinothalamic tract [37]. Moreover, brainstem lesions can also facilitate the development of neuropathic pain by affection of ascending somatosensory fibers or descending inhibitory pathways [38, 39].

A lower overall prevalence of pain in MOGAD may be linked to a better recovery of lesions compared to NMOSD [35]. We show that patients who were older at disease onset had a higher prevalence of pain, probably due to an age-dependent decrease of neuronal plasticity.

In addition to pain, depression was a relevant factor affecting the patients’ well-being. Similar to NMOSD [12, 40], clinically relevant depression occurred in 42% of the whole cohort and was significantly more prevalent in pain sufferers (60%). Vice versa, pain was more prevalent in patients with depression. Both conditions are known to intensify each other and can worsen independently from relapses enhanced by central mechanisms [41].

Although pain severity was one of the strongest predictors of low physical QoL, clinically relevant depression predicted reduced mental QoL. Our data support the need for adequate prevention and prompt treatment of severe MOGAD relapses [5, 9], as physical impairment was directly associated with both depression and persistent—especially refractory neuropathic—pain. Of note, the relapse activity seems to also be associated with spasticity-associated pain.

We show the need of a systematic pain assessment in MOGAD, particularly as commonly used clinical assessment tools such as the EDSS do not cover pain. Moreover, our data indicate that pain is often undertreated or treated inadequately. In our cohort, patients with several pain medications still had higher pain intensity scores, highlighting the importance of a targeted specific and/or multimodal pain therapy.

Limitations

The main limitation of our study is that information was based on a self-administered questionnaire without availability of clinical interview and neurological examination. Therefore, we cannot provide a definite diagnosis of specific pain syndromes, and our results do not allow for pathophysiological conclusions. We also had no specific data on possible medication-associated pain syndromes (e.g., due to steroid-induced osteonecrosis). However, only three patients received steroids at the time of assessment, and none of them indicated having joint pain, corresponding to femoral head necrosis.

Only history of ON and myelitis were recorded, but no other possible syndromes (e.g., brainstem, cerebral). Therefore, our data on the respective type of previous attacks are incomplete. Moreover, we are aware that the questionnaire might be prone to selection bias, with patients suffering from pain having higher response rates. However, in three centers addressing 74% of the cohort, the response rate was over 75%. Last, our study remains explorative, as we could not perform an a priori sample size calculation and adjustments for multiple testing.