Clinical features, treatment and outcome in neurosarcoidosis: systematic review and meta-analysis

Background Neurosarcoidosis is a rare variant of sarcoidosis and is only described in small cohort studies. We define clinical features, treatment and outcome of patients with neurosarcoidosis over the last 35 years. Methods We performed a systematic review and meta-analysis of studies on neurosarcoidosis published between 1980 and 2016. Studies were included if they reported at least 5 cases. Studies describing one specific neurological presentation were excluded. Results We identified 29 articles describing 1088 patients diagnosed between 1965 and 2015. Neurosarcoidosis occurred in 5% of patients with systemic sarcoidosis. Mean age at presentation was 43 years and neurological symptoms were the first clinical manifestation of sarcoidosis in 52%. The most commonly reported feature of neurosarcoidosis was cranial neuropathy in 55%, with the facial and optic nerve most commonly affected, followed by headache in 32%. Pleiocytosis and elevated CSF protein were found in 58 and 63%. MRI of the brain showed abnormalities in 70%. Chest X-ray, chest CT, or gallium-67-scintigraphy showed findings consistent with sarcoidosis in 60%, 70% and 69%, respectively. First line therapy with corticosteroids was initiated in 434 of 539 patients (81%). Second and third line therapy was started in 27 and 9%. Outcome consisted of complete remission in 27%, incomplete remission in 32%, stable disease in 24%, deterioration in 6% and death in 5%. Conclusion Neurosarcoidosis has a heterogeneous clinical presentation and the diagnosis can be difficult because of low sensitivity of ancillary investigations. New treatments have emerged, but nevertheless one third of patients do not respond to treatment. Prospective cohort studies and RCTs on treatment are urgently needed. Electronic supplementary material The online version of this article (doi:10.1186/s12883-016-0741-x) contains supplementary material, which is available to authorized users.


Background
Sarcoidosis is a multisystem granulomatous inflammatory disease of unknown aetiology, that typically affects young adults [1]. The incidence varies throughout the world, but is estimated to be between 10 and 20 per 100.000 population [2]. It mainly affects lung, skin and eyes, and has been reported to involve the nervous system in 5-20% [3,4]. Neurologic manifestations described are cranial nerve palsy, aseptic meningitis, peripheral neuropathy and myopathy [5]. Data on neurosarcoidosis are mostly derived from single-centre retrospective studies and vary considerably between studies [6]. Treatment is based on expert opinion, and no randomized controlled trials have been done comparing treatment in patients with neurosarcoidosis [7].
Diagnostic criteria for neurosarcoidosis have been proposed based on a clinical presentation suggestive of neurosarcoidosis, results of ancillary investigation and exclusion of other diagnoses [5,[8][9][10]. A definite diagnosis of neurosarcoidosis is met in only a minority of patients because this needs histologic confirmation of non-caseating granulomas of affected nervous system tissue. A probable diagnosis is defined as evidence of nervous system inflammation on magnetic resonance imaging (MRI) or cerebrospinal fluid (CSF; elevated protein, cells, immunoglobin G indices, or presence of oligoclonal bands) in combination with evidence of systemic sarcoidosis with histological confirmation and/or at least two of the indirect indicators consisting of fluorodeoxyglucose positron emission tomography (FDG-PET), gallium scan, chest imaging, and serum angiotensin-converting enzyme. Possible neurosarcoidosis is defined a clinical suspicion and exclusion of other diagnoses, but above mentioned criteria are not met.
We performed a systematic review and meta-analysis to determine clinical features, treatment, and outcome of patients with neurosarcoidosis over the last 35 years.

Methods
PubMed and Embase were searched using the search terms "neurosarcoidosis", "sarcoidosis" and "nervous system". Studies written in Dutch, English, French, German, or Spanish published between 1980 and March 2016 were considered for inclusion. Studies were included in the meta-analysis if they reported at least 5 cases of neurosarcoidosis, involving at least intracranial manifestations other than isolated hypothalamo-pituitary neurosarcoidosis. Studies were excluded if they: 1) reported only on a specific subset of neurosarcoidosis manifestations (e.g., neuro-ophthalmic or spinal cord neurosarcoidosis); or 2) were duplicate publications. Clinical features described in at least five case studies are reported. Data on study characteristics, demographic features, clinical manifestations, ancillary investigations, treatment, and outcome were systematically scored by DF and MB. Therapy was classified as first line, second line, or third line therapy. First line therapy consists of corticosteroid treatment, second line treatment consists of immunosuppressive therapy with methotrexate, azathioprine, mycophenolate mofetil, cyclosporine A, or (hydroxyl) chloroquine, and third line therapy either consists of cyclophosphamide or immunomodulatory medication (tumor necrosis factor-alpha inhibitors (TNF-alpha) or B-cell targeted therapy) [3]. Remission was defined as complete improvement, without residual symptoms. Favourable outcome was defined as remission, either complete or incomplete, and no need for alternative immunosuppressants.
We performed a pooled reanalysis of all published data in the included studies. Because of heterogeneity between studies, all data is presented as a number for which a certain characteristic is present out of the total number of patients for which it was described (n/N [%]) and a 95% confidence interval (95% CI) was used. For continuous data the standard deviation (SD) was used. Heterogeneity between studies was calculated for all reported variables using the Cochrane Q and I 2 statistical tests.

Outcome
Treatment response is reported in

Discussion
Our data show that neurosarcoidosis is a diverse illness, with heterogeneous clinical presentation, varying results of ancillary investigation, and considerable inter-individual differences in treatment response. Four clinical presentations  (15) a n/N: number for which a certain characteristic is present out of the total number of patients for which it was described b As reported in articles. The normal values of serum calcium were not reported in studies. The normal values of serum ACE varied between > 52 IU/L and >100 IU/L c As reported in the articles. The normal values of CSF glucose, CSF IgG index and CSF ACE varied between studies (CSF glucose between <35 mg/dL and <0.50 mg/dL, IgG index between >0.6 and >0.7 and CSF ACE between > 1U/L and 3 U/L) d Total histopathological confirmation, either in the CNS or systemic disease could be distilled from our meta-analysis: those presenting with cranial neuropathy, spinal cord inflammation, peripheral neuropathy or myopathy, or (chronic) meningitis. Neurological symptoms were the first clinical manifestation of sarcoidosis in about half of patients, and of those known with sarcoidosis about 5% eventually develop neurosarcoidosis. Although new treatments, such as TNF-alpha antagonists, are increasingly used, the mortality rate among patients with neurosarcoidosis remains 5% and about one third does not have a substantial clinical improvement on treatment.
Our systematic review and meta-analysis has limitations. First, all but one study were retrospective, introducing selection bias. This limits the external validity of our results and may overestimate effect of treatment in this disease. Second, most studies were performed in tertiary referral centres in European countries or the USA, which is also reflected in the ethnicity of included patients, again introducing selection bias. Tertiary care patients might be more severely affected and perhaps less responsive to treatment. Third, studies used heterogeneous inclusion criteria and not all items were reported for all patients. Fourth, patients were included over a period of 30 years. Within this time period, alternative diagnoses mimicking neurosarcoidosis have been described which previously would be classified as probable or even definite neurosarcoidosis. Examples of such mimics are IgG4-related disease or the POEMS syndrome [37][38][39]. These rare diseases may present with similar clinical features and histopathology. It may well  (39) Between third line 7/23 (30) a n/N: number for which a certain characteristic is present out of the total number of patients for which it was described Fig. 2 Proportion of treatment used according to timeframe be that different aetiologies may exist for a disease that we now call neurosarcoidosis. Finally, when performing the meta-analyses we found there frequently was significant heterogeneity between studies on the reported data. This confirms the influence of the abovementioned biases.
Making a diagnosis of neurosarcoidosis can be difficult. A definite diagnosis was made only in one out of four patients. The far majority of patients were classified as probable neurosarcoidosis. Diagnostic classification for neurosarcoidosis includes MRI, CSF and serum inflammatory markers, FDG-PET, gallium scan, and chest imaging [10]. Our study shows that none of these diagnostic markers seems to have a good accuracy for the diagnosis. Multicentre prospective diagnostic studies in patients with suspected neurosarcoidosis are needed to test diagnostic accuracy of these tests in the diagnostic process. These studies should determine the value of markers noted in the diagnostic criteria as well as other promising new markers, such as soluble interleukin-2receptor [35,40].
Clear guidance for the treatment of patients with neurosarcoidosis is lacking. Most treatment strategies are extrapolated from few studies on pulmonary sarcoidosis. In our meta-analysis 83% of patients received corticosteroids, which can be considered the mainstay of treatment in neurosarcoidosis. However, 24% of the patients initially treated with first line therapy were switched to second or third line therapy. In addition, corticosteroidassociated side effects occur frequently and can be severe [8,20]. Over time, second and third line therapy emerged into the treatment of neurosarcoidosis, most recently TNF-alpha antagonists. Value of TNF-alpha antagonists has been suggested only in a phase II randomised controlled trial (RCT) in pulmonary sarcoidosis [41]. This trial including 138 patients showed that infliximab therapy resulted in a statistically significant improvement in percentage of predicted forced vital capacity at week 24. The value of TNF-alpha antagonists is based on sporadic case series, and it is unclear if patients benefit from a combined therapy of TNF-alpha antagonists and first and second line therapy. It is also unclear for how long TNF-alpha antagonists should be continued and whether this therapy is associated with side effects. RCTs on treatments or treatment strategies in this disease are urgently needed.

Conclusions
Neurosarcoidosis develops in 5% of the sarcoidosis patients, and in more than half of these patients neurological symptoms are the primary presentation of sarcoidosis. The diagnosis is difficult due to the heterogeneous clinical presentation and low sensitivity of ancillary investigations. Despite the increasing use of second and third line medication, still one-third of patients do not improve or deteriorate. These data stress the need for prospective cohort studies and treatment trials.

Additional file
Additional file 1: Table S1. Study descriptives. This includes a summary of all studies included concerning the first author, date of publication, country of study, study design, whether it is a single or multi center study, number of included patients and inclusion period.