In 2018, the Neurosarcoidosis Consortium Consensus Group proposed a table of possible differential diagnoses and their investigations, depending on the clinical context6. The diagnostic workup for the patients included in this study has largely followed this approach since 2007.
The most common symptom in our cohort was cranial neuropathy, mainly presenting as optic neuropathy, hydrocephalus, and motor signs. These findings are consistent with previously reported clinical manifestations of neurosarcoidosis17,18. Strikingly, 17% of the patients presented with isolated pituitary/hypothalamic dysfunction. A recent study reported an involvement of the hypothalamus/pituitary axis in 9% of patients with suspected neurosarcoidosis19.
Hypothalamus, pituitary gland, and optic chiasma are among the predilection sites for neurosarcoidosis. Patients belonging to this category usually present with features of hypopituitarism such as central diabetes insipidus, hyperprolactinemia, and other endocrinopathies arising from inflammation of the pituitary gland or a sellar mass lesion20,21. Gadolinium enhanced MRI of the pituitary gland shows either asymmetrical pituitary enlargement with stalk deviation, as in the case of adenoma, or symmetrical enlargement of the pituitary gland and pituitary stalk, as in the case of lymphocytic hypophysitis. Another scenario is a normal MRI of the pituitary gland which makes the diagnosis of neurosarcoidosis even more challenging. As was the case for our patients, the histopathology of hypophysitis was due to sarcoidosis (n = 2), IgG4-related disease (n = 5), xanthogranulomatous (n = 1), or lymphocytic hypophysitis (n = 2) affecting different parts of or the entire pituitary gland22.
All our patients with large- and small-fiber neuropathy (7%) had systemic disease, raising the suspicion of peripheral nervous system involvement. This is in line with the decision of the 2018 Neurosarcoidosis Consortium Consensus Group, which ultimately included the peripheral nervous system as a feature of neurosarcoidosis6.
Given the rarity of isolated neurosarcoidosis, an open-minded discussion of the differential diagnosis of each MRI pattern is crucial in every single case, as is correlating the findings to patient history and other laboratory findings. The lack of reliable biomarkers compelled us to rely heavily on repeated MRI studies (normally twice per year) in combination with clinical findings, both at the onset and later on, for monitoring disease activity. The MRI studies in this cohort revealed a wide range of findings, including leptomeningeal, dural, and intraparenchymal lesions with CE, as well as involvement of the pituitary gland, hypothalamus, cranial nerves, and hydrocephalus. All these findings were in agreement with previous reports19,23. Spinal involvement was observed in 23% of our sample, mostly as leptomeningeal enhancement24,25. Notably, the deep medullary vein sign on MRI was absent in all our patients with histological features of neurosarcoidosis (n = 14), despite a reported sensitivity of 71.4% and a specificity of 92.3% for the diagnosis of neurosarcoidosis26,27.
A careful assessment of the systemic manifestations is an important part of the diagnostic work-up. This justifies the use of whole body 18F-FDG PET/CT for visualization of extra- neural sites of inflammatory active sarcoidosis suitable for diagnostic biopsy, for detection of clinically silent lesions such as ocular or cardiac sarcoidosis, and for evaluation of the treatment response28,29. In our study, FDG PET/CT was part of the investigation in 33% of the cases. The results were used to identify and obtain biopsies from sites with hypermetabolism.
The most striking laboratory findings in our cohort was elevated levels of WCC, protein, T lymphocyte CD4+/CD8+ ratios, ACE, and OCBs in the CSF, suggesting a neuroinflammatory disease. The frequency of OCBs in our cohort was consistent with previous reports (27–37%)5. Recent data showed that a combined CD4+/CD8+ ratio ≥ 5 and elevated WCC had a negative predictive value of 88%, with a specificity of 95% for neurosarcoidosis30,31. By contrast, serum ACE measurements lack both sensitivity and specificity, as ACE levels can also be elevated in other diseases, such as diabetes mellitus, hypothyroidism, and lymphoma32. Overall, serum biomarkers have only limited value in the diagnostic work-up of neurosarcoidosis. The ACE levels in the CSF, although more specific (94–95%) than in the serum, are rather insensitive (24–55%). The level increases in proportion to that of CSF protein and is reported to increase in CNS infections and malignant tumors33.
One study showed elevated levels of lysozyme and B2-microglobulin in the CSF in a small sample of patients with neurosarcoidosis, but those findings were not confirmed later, and the diagnostic value of both serum and CSF lysozyme levels remains doubtful34. Given that 23% of our patients had normal CSF and that no specific pattern is diagnostic, our CSF analyses were conducted mainly to confirm neuroinflammation.
Because the other tests were never conclusive enough to make a definitive diagnosis of neurosarcoidosis, performing a greater number of CNS biopsies will increase the chance of achieving a definite diagnosis. However, CNS biopsies were not the first option unless patients were first given a standardized noninvasive investigation (e.g., blood, CSF, MRI, PET imaging) and/or if a very aggressive or atypical course was present. Examples included hydrocephalus, rapid deterioration of neurological functions, mass lesions, multiple cranial nerve palsies, and pituitary enlargement, which signaled a clear indication for CNS biopsy. Initiating treatment with high-dose steroids for a subset of patients with severe neurological symptoms before all investigations had been completed led to a diagnosis of non-specific chronic inflammation on biopsy, thereby complicating the histopathological confirmation of neurosarcoidosis.
The diagnosis of “probable neurosarcoidosis” is the most common scenario in clinical practice, as soon as the investigation reveals signs of systemic sarcoidosis and excludes other possible diagnoses. Only four cases belonging to this category were biopsied, compared to the “possible neurosarcoidosis” group, most likely because the diagnostic certainty was considered higher and the CNS manifestations were attributed to systemic sarcoidosis. A high suspicion of neoplasm in solitary intracranial or intraspinal mass lesions, a threatening hydrocephalus in need of a ventriculoperitoneal shunt, or the presence of cortical tumor-like lesions, especially in the temporal lobe, causing medically refractory epilepsy, were among the clinically precarious situations that justified the need for CNS biopsy despite histologically confirmed active systemic sarcoidosis.
In reality, the most difficult scenario was investigating isolated neurosarcoidosis. A comprehensive diagnostic workup had two purposes: first, to reduce the rate of incorrectly diagnosing patients with “possible neurosarcoidosis”, and second, to identify other CNS disease mimics that needed urgent treatment.
We recognize a number of limitations of this study. One is its retrospective nature; another is the small number of patients, which is explained by the low overall prevalence of neurosarcoidosis. A risk of selection bias also probably existed because the more severe cases were referred to our center. Furthermore, this study spanned the years between 1990 and 2021, and diagnostic tools have vastly improved over that period.