Article Text

Download PDFPDF

Research paper
CLIPPERS and its mimics: evaluation of new criteria for the diagnosis of CLIPPERS
  1. Guillaume Taieb1,
  2. Patricia Mulero2,
  3. Dimitri Psimaras3,
  4. Bob W van Oosten4,
  5. Jörg D. Seebach5,
  6. Romain Marignier6,
  7. Fernando Pico7,8,
  8. Valérie Rigau9,
  9. Yuji Ueno10,
  10. Claire Duflos11,
  11. Vera Fominykh12,
  12. Vincent Guiraud13,
  13. Christine Lebrun-Frénay14,
  14. Jean-Philippe Camdessanché6,15,
  15. Philippe Kerschen16,
  16. Guido Ahle17,
  17. Nieves Téllez18,
  18. Alex Rovira2,
  19. Khe Hoang-Xuan3,
  20. Jean Pelletier19,
  21. Pierre Labauge1
  22. in cooperation with the French CLIPPERS group
    1. 1 Department of Neurology, University Hospital of Montpellier, Montpellier, France
    2. 2 Centre d'Esclerosi Mútiple de Catalunya, Hospital Universitari Vall d'Hebron, Barcelona, Spain
    3. 3 Department of Neurology, AP-HP Pitié-Salpêtrière, Service de Neurologie Mazarin, Paris, France
    4. 4 Department of Neurology, VU University Medical Center, Amsterdam, Netherlands
    5. 5 Department of Immunology and Allergology, University Hospital of Genève, Geneva, Switzerland
    6. 6 Centre de Recherche en Neurosciences de Lyon, INSERM U1028 and CNRS UMR5292, Equipe FLUID, Lyon, France
    7. 7 Department of Neurology, Centre Hospitalier de Versailles, Versailles, France
    8. 8 University of Versailles Saint Quentin en Yvelines et Paris Saclay, Versailles, France
    9. 9 Department of Neuropathology, University Hospital of Montpellier, Montpellier, France
    10. 10 Department of Neurology, Juntendo University School of Medicine, Tokyo, Japan
    11. 11 Clinical Research and Epidemiology Unit, DIM, University Hospital of Montpellier, Montpellier, France
    12. 12 Department of Neurology, University Hospital of Moscow, Moscow, Russia
    13. 13 Department of Neurology, University Hospital of Saint-Etienne, Saint-Etienne, France
    14. 14 Department of Neurology, Hospital of Nice, Nice, France
    15. 15 Department of Neurology, University Hospital of Sainte Anne, Paris, France
    16. 16 Department of Neurology, University Hospital of Luxembourg, Luxembourg, Luxembourg
    17. 17 Department of Neurology, Hospital of Colmar, Colmar, France
    18. 18 Department of Neurology, Hospital Clinico de Valladolid, Valladolid, Spain
    19. 19 APHM, Hôpital de la Timone, Pôle de Neurosciences Cliniques, Service de Neurologie, Marseille, France
    1. Correspondence to Dr Guillaume Taieb, Department of Neurology, CHU Montpellier, Hôpital Guy de Chauliac, Montpellier 34295, France; taiebguillaume{at}gmail.com

    Abstract

    Objective To evaluate the accuracy of the recently proposed diagnostic criteria for chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids (CLIPPERS).

    Methods We enrolled 42 patients with hindbrain punctate and/or linear enhancements (<3 mm in diameter) and tested the CLIPPERS criteria.

    Results After a median follow-up of 50 months (IQR 25–82), 13 out of 42 patients were CLIPPERS-mimics: systemic and central nervous system lymphomas (n=7), primary central nervous system angiitis (n=4) and autoimmune gliopathies (n=2). The sensitivity and specificity of the CLIPPERS criteria were 93% and 69%, respectively. Nodular enhancement ( ≥ 3 mm in diameter), considered as a red flag in CLIPPERS criteria, was present in 4 out of 13 CLIPPERS-mimics but also in 2 out of 29 patients with CLIPPERS, explaining the lack of sensitivity. Four out of 13 CLIPPERS-mimics who initially met the CLIPPERS criteria displayed red flags at the second attack with a median time of 5.5 months (min 3, max 18), explaining the lack of specificity. One of these four patients had antimyelin oligodendrocyte glycoprotein antibodies, and the three remaining patients relapsed despite a daily dose of prednisone/prednisolone ≥ 30 mg and a biopsy targeting atypical enhancing lesions revealed a lymphoma.

    Conclusions Our study highlights that (1) nodular enhancement should be considered more as an unusual finding than a red flag excluding the diagnosis of CLIPPERS; (2) red flags may occur up to 18 months after disease onset; (3) as opposed to CLIPPERS-mimics, no relapse occurs when the daily dose of prednisone/prednisolone is ≥ 30 mg; and (4) brain biopsy should target an atypical enhancing lesion when non-invasive investigations remain inconclusive.

    • CLIPPERS
    • primary central nervous system vasculitis
    • primary central nervous system lymphoma
    • lymphomatoid granulomatosis
    • anti-myelin oligodendrocyte glycoprotein related diseases
    • autoimmune glial fibrillary acidic protein astrocytopathy
    • diagnostic criteria

    Statistics from Altmetric.com

    Introduction

    CLIPPERS (the acronym for chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids) is a chronic central nervous system (CNS) inflammatory disorder of unknown origin.1 Since its first description in 2010 by Pittock and colleagues,1 about 60 cases have been reported throughout the world.2–4 The mean age at onset is 50 years (range 13–86 years), with male predominance (male to female ratio 3:1). As mentioned in its acronym, CLIPPERS is defined by steroid-responsive symptomatic hindbrain punctate and curvilinear enhancing lesions. These pontine and cerebellar enhancing lesions could spread rostrally up to the juxtacortical areas, and caudally up to the conus. White and deep grey matter are both involved, while the cortex is spared. Relapses usually occur in the absence of long-term steroid therapy, which explains its chronic course. Enhancing lesions seen on MRI are related to perivascular lymphohistiocytic infiltrates with no evidence of necrotising vasculitis, granuloma, remote demyelinating lesion or lymphoma. Pittock and colleagues1 suggested that brain biopsy is not mandatory in the presence of other typical CLIPPERS features and in the absence of differential diagnosis on non-invasive investigations. Since its first description, well-characterised diseases such as multiple sclerosis and myelin oligodendrocyte glycoprotein (MOG) antibody-associated diseases may mimic clinical and radiological CLIPPERS features.5 6 Furthermore, cases of primary angiitis of the CNS (PACNS), primary CNS lymphoma (PCNSL) and CNS lymphomatoid granulomatosis (CNS-LYG) initially presenting with all CLIPPERS features, including histological characteristics, have been reported.7–9

    In 2017, Tobin and colleagues10 proposed a new set of criteria for the diagnosis of CLIPPERS. However, these new criteria have not yet been evaluated, especially over a long period of follow-up.

    Methods

    Patients and brain imaging

    We conducted an international retrospective study to evaluate the 2017 CLIPPERS criteria. Since 2011, University Hospitals of Montpellier gathered data of patients presenting with hindbrain punctate and/or curvilinear gadolinium enhancements and collected clinical, radiological and histological data. Forty-two out of 62 patients presenting with hindbrain perivascular gadolinium enhancements suggestive of CLIPPERS were enrolled (France (n=30), the Netherlands (n=5), Spain (n=4), Switzerland (n=1), Luxembourg (n=1) and Russia (n=1)). The 20 remaining patients were excluded due to lack of data. Among the 42 enrolled patients, including 33 male and 9 female, all had brainstem signs and/or symptoms. Seventeen out of 42 patients have already been reported elsewhere, but none of them have been used to elaborate the 2017 CLIPPERS criteria.4 11–18 The median age of onset was 48.5 (IQR 41.25–56.75). All the patients of our study were assessed clinically and examined with a 1.5-Tesla or a 3-Tesla MRI. Brain MRIs were obtained using the following sequences: precontrast and postcontrast T1-weighted images (WI), T2WI, fluid-attenuated inversion recovery, T2*-weighted gradient-echo images (T2*WI), diffusion-weighted imaging, apparent coefficient diffusion map (ADC), and magnetic resonance angiography (MRA). Spinal MRIs were obtained using the following sequences: precontrast and postcontrast T1WI and T2WI. According to the radiological definitions, punctate gadolinium enhancement (Gd+) is a dot-like enhancement with a diameter less than 3 mm, nodular Gd+ is a round homogeneous enhancement with a diameter ranging from 3 mm to 9 mm, and large Gd+ has a diameter greater than 9 mm.11 19 Patchy (ie, cloudy) Gd+ is an ill-delimited enhancement. Brain, hindbrain and/or spinal cord ‘on fire’ is defined by linear Gd+ along the deep medullary veins, spreading from the ependymal surface to the subpial areas without interruption.20 A brain CT scan was systematically performed in the presence of T2* hypointensities on brain MRI, and digital subtraction angiography (DSA) was made when vascular involvement was present on MRI (ie, brain infarction, haematoma and arterial narrowing on MRA).

    Assessment of alternative diagnoses

    Investigations performed to exclude alternative diagnoses are specified in box 1. Detection of anti-MOG antibodies in the serum was performed according to the method described by Cobo-Calvo and colleagues,21 and detection of antiglial fibrillar acidic protein (GFAP) alpha antibodies in the serum and cerebrospinal fluid (CSF) was performed using cell-based assays according to the method previously described by Fang and colleagues.22 When stereotactic brain biopsy was performed, histological sections were stained with H&E, Bielschowsky stain, Perls’ iron stain, and the following antigens: CD20, CD3, CD4, CD8, granzyme B, CD68, CD 38, CD 138, CD1a, beta-amyloid, neurofilament and glial fibrillar acid protein. A myelin stain (Klüver-Barrera) and a proliferation marker (Ki-67) were also used. In situ hybridisation studies were performed for Epstein-Barr virus early RNA-1.

    Box 1

    Investigations in patients referred for a suspicion of CLIPPERS

    • Routine blood tests (including C reactive protein level, serum protein electrophoresis), serum lactate deshydrogenase level and IgE level.

    • Serological tests for infections (HIV, hepatitis B virus, hepatitis C virus, Epstein-Barr virus, Lyme and syphilis).

    • Serological tests for connective tissue diseases (antinuclear antibodies, (anti-double stranded DNA), extractable nuclear antigens, rheumatoid factor, complement C3 and C4, lupus anticoagulant, anticardiolipin and antiphospholipid antibodies), systemic vasculitis (antineutrophil cytoplasm antibodies and cryoglobulinaemia), sarcoidosis (ACE), demyelinating disorders (aquaporin-4 antibody and MOG-IgG), antineuronal and antiganglioside antibodies (in 42/42 patients).

    • CSF analysis (in 42/42 patients) including flow cytometry and B cell and T cell receptor rearrangements in case of lymphocytic pleocytosis (in 11/42 patients).

    • Bone marrow biopsy and aspiration (in 11/42 patients).

    • Whole body CT or PET scan (in 42/42 patients).

    • GFAP-alpha antibody in serum and CSF (2/42 patients, cases 1 and 7).

    • Electromyogram and nerve conduction studies (1/42 patients, case 1).

    • CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids; CSF, cerebrospinal fluid; GFAP, glial fibrillar acidic protein; MOG, myelin oligodendrocyte glycoprotein; PET, positron emission tomography.

    The diagnosis of lymphoma was made according to the WHO classification, and the diagnosis of PACNS was made according to the Calabrese and Mallek criteria (ie, presence of an acquired and otherwise unexplained neurological deficit, with the presence of either classic angiographic or histopathological features of angiitis within the CNS and no evidence of systemic vasculitis or any condition that could elicit the angiographic or pathological features).23 24

    Initial and final diagnoses

    At each attack during follow-up, patients were assigned in three categories according to the 2017 CLIPPERS criteria specified in box 2: non-CLIPPERS (ie, patients did not meet the clinical, radiological and/or histological criteria), probable CLIPPERS (ie, patients met the clinical and radiological criteria) and definite CLIPPERS (ie, patients met the clinical, radiological and histological criteria). The initial diagnosis was the diagnosis at first attack, and the final diagnosis was the diagnosis at the end of follow-up.

    Box 2

    Diagnostic criteria for CLIPPERS10

    • Clinical.

      • Subacute pontocerebellar dysfunction, with or without other CNS symptoms such as cognitive dysfunction and myelopathy.

      • CNS symptoms responsive to corticosteroid therapy.

      • Absence of peripheral nervous system disease.

      • Lack of alternative better explanation for clinical presentation.

    • MRI.

      • Homogeneous, gadolinium enhancing nodules without ring enhancement or mass effect predominating in the pons and cerebellum, measuring <3 mm in diameter.

      • Marked improvement in abnormal gadolinium enhancement with corticosteroid treatment.

      • Homogeneous T2 signal abnormality where the degree of T2 does not significantly exceed the size of the area of postgadolinium enhancement.

      • Spinal cord lesions with similar T2 and gadolinium enhancing lesions as above.

    • Neuropathology.

      • Dense lymphocytic inflammation with perivascular predominance and parenchymal diffuse infiltration; both white matter and grey matter could be involved.

      • T cells predominating infiltration (CD4 > CD8) with variable macrophage components.

      • Absence of myelin loss or focal secondary myelin loss.

      • Lack of alternative better explanation for pathological presentation.

    • Definite CLIPPERS: patients fulfilling all clinical, radiological and neuropathological criteria.

    • Probable CLIPPERS: patients fulfilling all clinical and radiological criteria without available neuropathology.

    • CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids; CNS, central nervous system.

    Data analysis

    Sensitivity and specificity of the initial diagnosis (test procedure) were calculated, relatively to the final diagnosis (reference procedure). For this calculation, we pooled probable and definite CLIPPERS in the CLIPPERS category, because they are eligible to the same follow-up with no evidence of an alternative diagnosis on our ancillary investigations. The lack of histological confirmation in probable CLIPPERS is due to the risk of the cerebral biopsy and no other consideration. Thus, we had two groups of patients at each assessment: CLIPPERS and non-CLIPPERS.

    We also identified red flags, by gathering all features of patients, at all attacks. Each clinical, radiological and histological feature that did not match the CLIPPERS criteria was defined as atypical. Atypical features only present in patients who were non-CLIPPERS at the final diagnosis were referred as red flag. For this analysis, we named the non-CLIPPERS at the final diagnosis ‘CLIPPERS-mimics’.

    The qualitative data are presented as number and the quantitative data are presented as median (IQR).

    Results

    Diagnostic values of the initial diagnosis

    Among the 42 patients, 11, 20 and 11 were, respectively, classified as non-CLIPPERS, probable CLIPPERS and definite CLIPPERS at first attack (test procedure; figure 1, tables 1 and 2). Patients (n=42) were followed during a median of 50 months (IQR 25–82). The median number of attacks per patient was 3 (IQR 2–4). At the second attack, 2 out of 20 patients with probable CLIPPERS and 2 out of 11 patients with definite CLIPPERS had atypical clinical and/or radiological features and were reassigned to the non-CLIPPERS category. Conversely, atypical radiological findings disappeared in 2 out of 11 non-CLIPPERS patients (n=13), leading to reclassification of each of them to probable (n=19) and definite (n=10) CLIPPERS, respectively. No other category changes were observed after the second attack (reference procedure; figure 1). The median time to reach the definitive category was 11 months (min 3, max 18). Considering that 6 out of 42 patients were misclassified at their first attack, the sensitivity and specificity of the proposed 2017 criteria were 93% and 69%, respectively. Figure 2 provides the MRI section of the hindbrain at the first evaluation in patients with atypical MRI features during their disease course (ie, 13 CLIPPERS-mimics and 2 patients without alternative diagnosis to CLIPPERS).

    Figure 1

    Classification of patients according to the CLIPPERS criteria during the follow-up period. CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids.

    Figure 2

    MRI section of the hindbrain at the first evaluation in patients presenting with atypical MRI features during their disease course, including 13 patients with CLIPPERS-mimics (cases 1–13) and 2 patients with CLIPPERS (cases 20 and 33). The final diagnosis for each patient is indicated between brackets. CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids; CNS, central nervous system; CNS-LYG, CNS lymphomatoid granulomatosis; GFAP, glial fibrillar acidic protein; HL, Hodgkin’s lymphoma; MOG, myelin oligodendrocyte glycoprotein-related disease; PACNS, primary angiitis of the CNS; PCNSL, primary CNS lymphoma.

    Table 1

    Clinical characteristics, radiological findings, treatment and outcome concerning patients with CLIPPERS-mimic

    Table 2

    Clinical characteristics, radiological findings, treatment and outcome concerning patients with probable and definite CLIPPERS

    Description of alternative diagnoses at the final diagnosis

    The 13 non-CLIPPERS patients, coined CLIPPERS-mimics, all had an alternative diagnosis: CNS lymphomas (n=6; 4 CNS-LYG, 2 PCNSL), systemic Hodgkin’s lymphoma (n=1), PACNS (n=4), and glial-autoimmune diseases with seropositivity for MOG (n=1) and GFAP-alpha (n=1) autoantibodies (table 1). Conversely, none of the patients with probable and definite CLIPPERS (n=29) had an alternative diagnosis (table 2).

    Patients moving from the probable or definite CLIPPERS category at the initial diagnosis to the non-CLIPPERS category at the final diagnosis

    Concerning the two patients who fulfilled the 2017 criteria for probable CLIPPERS at the initial diagnosis (patients 10 and 11), the final diagnoses were grade III CNS-LYG and MOG antibody-associated disease. Atypical clinical and MRI features occurred at the second attack 4 and 7 months after disease onset, respectively (table 1, figure 3). In the latter patient, anti-MOG antibodies were only performed at the second attack, in the presence of atypical findings.

    Figure 3

    Atypical radiological findings in CLIPPERS-mimics axial contrast T1WI showing large pontine round (A) or trident-shaped (B) Gd+, Gd+ lining the pial surface of the pons with trigeminal nerve involvement (C), ring-like Gd+ with necrotic centre (D), cerebellar patchy Gd+ (E), and perivenular spaces Gd+ with ‘hindbrain on fire’ (F). Contrast T1WI showed ‘brain on fire’ with perivenular spaces Gd+ on axial (G) and sagittal (H) sections. Axial contrast T1WI showed cortical nodular (3–9 mm) Gd+ (I) matching with hypointense lesion on T2*WI (J), and sulcal Gd+ (arrowhead, K) matching with hypointense lesion on T2*WI (arrowhead, L) related to microaneurysm on DSA (arrowhead, M). In two other patients, pontine hypointensities on T2*WI (N) matched with microcalcifications on cranial CT scan (O), and DSA showed stenosis involving the first segment of the left anterior cerebral artery supplying both the anterior cerebral artery territories (arrowhead, P), and bilateral frontal infarctions were seen on DWI (Q). The final diagnosis for each patient is indicated between brackets. CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids; CNS, central nervous system; CNS-LYG, CNS lymphomatoid granulomatosis; DSA, digital subtraction angiography; DWI, diffusion-weighted imaging; Gd+, gadolinium enhancement; GFAP, glial fibrillar acidic protein; PACNS, primary angiitis of the CNS; T1WI, T1-weighted image; T2*WI, T2*-weighted gradient-echo image.

    About the two patients who fulfilled the 2017 criteria for definite CLIPPERS at the initial diagnosis (patients 12 and 13), the final diagnoses were PCNSL. Atypical clinical and MRI features occurred at the second attack 3 and 18 months after disease onset, respectively (table 1, figure 3).

    Patients moving from the non-CLIPPERS category at the initial diagnosis to the probable or definite CLIPPERS category

    Two patients had spinal cord enhancing lesion extending from the C1 to C2 level at their first attack (patients 20 and 33; table 2, figure 4). They were assigned to the non-CLIPPERS category because the size of the spinal cord enhancing lesion exceeded 3 mm (Box 2). A nodular spinal cord enhancing lesion in both patients disappeared under steroid therapy and did not relapse during the follow-up (ie, 71 months with 2 further attacks in patient 20 and 229 months with 3 further attacks in patient 33). At their second attack, patients 20 and 33 met the 2017 criteria for probable and definite CLIPPERS, respectively. At the end of follow-up, clinical course, radiological features and investigations, and brain biopsy in patient 33 remained compatible with the diagnosis of CLIPPERS in both patients.

    Figure 4

    Spinal cord MRI in CLIPPERS and its mimics. In two patients with CLIPPERS, sagittal postcontrast T1WI showed confluent enhancing lesions in the cervical spinal cord that extended from C1 to C2 level (A, C), matching with T2WI hyperintensities with bright spotty lesions (B, D). In another patient with CLIPPERS, punctate enhancing lesions were seen in the medulla oblongata and in the cervical spinal cord on sagittal postcontrast T1WI (E), together with punctate T2WI hyperintensities in the corresponding areas (F). Sagittal postcontrast T1WI of the spinal cord revealed Gd+ along the ependymal canal (ie, ‘spinal cord on fire’, G) matching with cervical and thoracic longitudinal extensive transverse myelitis on sagittal T2WI (H) in a patient with autoimmune GFAP astrocytopathy. The final diagnosis for each patient is indicated between brackets. CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids; Gd+, gadolinium enhancement; GFAP, glial fibrillar acidic protein; T1WI, T1-weighted image; T2WI, T2-weighted image.

    Identification of red flags

    Clinical criteria in patients with CLIPPERS and CLIPPERS-mimic

    All patients with CLIPPERS (n=29) met the clinical criteria defined by steroid-responsive brainstem signs and symptoms, without peripheral neuropathy, and no evidence of differential diagnosis. Conversely, 9 out of 13 patients with CLIPPERS-mimic did not meet the clinical criteria, at their first (n=5) or at their second (n=4) attack. Aphasia was present in one patient with PACNS, clinical unresponsiveness to a high dose of steroids was seen in two patients with CNS-LYG and one patient with PCNSL, and axonal polyneuropathy was diagnosed in the patient presenting with an autoimmune GFAP astrocytopathy. As defined above, all clinical atypical features could be considered as clinical red flags. Investigations providing alternative diagnosis to CLIPPERS were anti-MOG and anti-GFAP antibodies (n=2), DSA showing arterial stenosis (n=2) or multiple sulcal microaneurysms (n=1), and positron emission tomography (PET) scan-guided lymph node biopsy revealing Hodgkin’s lymphoma (n=1) (tables 1 and 3).

    Table 3

    Discriminative findings between CLIPPERS and its mimics during attacks

    MRI criteria in patients with CLIPPERS and CLIPPERS-mimic

    Typical punctate enhancing lesion (as defined in the CLIPPERS criteria) is characterised by (1) its size (ie, <3 mm), (2) its shape (ie, homogeneous and not ring-shaped), (3) its location (ie, pontocerebellar and spinal cord), (4) its lack of mass effect, (5) its response to steroids and (6) the presence of small foci of T2 hyperintensity in the corresponding area (Box 2). All patients with CLIPPERS except two (n=27) met the MRI criteria. These two patients had cervical spinal cord nodular Gd+ (ie, ranging from 3 mm to 9 mm), but not during their relapses. Conversely, all patients with CLIPPERS-mimic except one (n=12) did not meet the MRI criteria, at their first (n=8) or at their second (n=4) attack. Furthermore, once present, atypical MRI features relapsed in the case of new attacks.

    Atypical MRI features in CLIPPERS-mimics included (1) unusual size with large Gd+ and (2) non-homogeneous Gd+ with ring-shaped Gd+ and patchy Gd+. There were also (3) unusual locations as subpial and cranial nerve Gd+, cortical Gd+, subarachnoid round Gd+ related to microaneurysm, and diffuse linear Gd+ along the perivenular spaces, involving either the whole CNS (ie, ‘CNS on fire’), or limited to the brain and/or hindbrain (‘brain on fire’ and ‘hindbrain on fire’); (4) mass effect; (5) lack of steroid responsiveness; and (6) large T2 hyperintensities. Thus, all atypical MRI features except nodular Gd+ could be considered as MRI red flags (tables 1 and 3, figure 3).

    Neuropathological criteria in patients with CLIPPERS and CLIPPERS-mimic

    All patients with CLIPPERS who had brain biopsy (n=11) met the neuropathological criteria defined by perivascular and parenchymal lymphohistiocytic infiltrates, with predominance of CD4 cells, in the absence of remote demyelination and no evidence of alternative diagnosis. Conversely, all brain biopsies of patients with CLIPPERS-mimic which concerned MRI red flags did not meet the neuropathological criteria and revealed an alternative diagnosis (n=7). In three of them, the first brain biopsy of typical punctate Gd+ showed characteristic histological findings, while the second brain biopsy of MRI red flags revealed a lymphoma. Anti-CD20 staining associated with hybridisation for Epstein-Barr virus and Perls’ iron stain were useful to reveal CNS lymphomas (PCNSL (n=2), CNS-LYG (n=4)), and CNS small vessel angiitis (PACNS with normal DSA (n=1)), respectively (table 3, online supplementary file).

    Supplemental material

    The following are additional discriminating features between CLIPPERS and its mimics (table 3): CLIPPERS had a chronic course with relapses in the absence of long-term steroid therapy or immunosuppressive drugs. When glucocorticoid therapy (ie, prednisone/prednisolone) was ≥ 30 mg/day, while relapses occurred in 5 out of 13 patients with CLIPPERS-mimic, none of our patients with CLIPPERS relapsed. These five patients had PACNS (n=1) and CNS lymphomas (CNS-LYG (n=2) and PCNSL (n=2)).

    Besides postcontrast T1WI and T2WI abnormalities, diffusion restriction on ADC map, arterial stenosis on MRA and T2*WI hypointensities were only present in patients with CLIPPERS-mimic. Arterial stenosis (confirmed on DSA) and diffusion restriction were seen in PACNS (n=2), and T2*WI hypointensities were present in PACNS (n=3) but also in the patient with anti-MOG-related disease (n=1) and in the patient presenting with Hodgkin’s lymphoma relapse (n=1). T2*WI hypointensities were related to gadolinium enhancing lesions in anti-MOG-related disease, Hodgkin’s lymphoma, and one out of three PACNS. In the other two patients with PACNS, T2*WI hypointensities matched with microcalcifications on CT scan and sulcal microaneurysms on DSA, respectively (figure 3).

    CLIPPERS criteria discriminated CLIPPERS from its mimics only at the second attack in 4 out of 13 patients with CLIPPERS-mimic. In these patients, red flags occurred at the second attack with a median time of 5.5 months (min 3, max 18). Additional discriminating features described above were also only present during the second attack: relapse occurred despite glucocorticoid therapy ≥ 30 mg/day in three of them, and the presence of T2*WI hypointensities in the patient with Hodgkin’s lymphoma. The number of red flags in patients with CLIPPERS-mimic increased with the time of follow-up. At the end of follow-up, all patients with CLIPPERS-mimic but one (n=12) had two or more red flags for CLIPPERS. Demographic data (ie, gender and age of onset) and CSF findings (ie, blood cell counts, protein levels and presence of oligoclonal bands) did not distinguish CLIPPERS from its mimics.

    Discussion

    We describe the most important cohort of patients presenting with symptomatic hindbrain perivascular enhancements. At the end of follow-up, about one-third of patients had an alternative diagnosis to CLIPPERS, including lymphomas, either systemic or restricted to the CNS, autoimmune gliopathies, and PACNS (ie, CLIPPERS-mimics). Furthermore, the present study has evaluated the recently published CLIPPERS criteria at different times and provides some clues to improve their accuracy.10 When assessed at the first attack, the sensitivity and specificity of this new set of criteria were 93% and 69%, respectively. All atypical clinical, MRI and neuropathological features were exclusively associated with CLIPPERS-mimics and could be considered as red flags, except enhancing lesion ≥ 3 mm in diameter. In our study, Gd+ ≥ 3 mm is present in CLIPPERS-mimics (n=4) but also in CLIPPERS (n=2), explaining why the sensitivity does not reach 100%. Tobin and colleagues10 described these MRI findings in 7 out of 12 non-CLIPPERS and in 3 out of 23 CLIPPERS. Interestingly, by contrast to our patients with CLIPPERS, Gd+ ≥ 3 mm were larger (ie, >9 mm) and systematically associated with one or more additional red flags in our patients with CLIPPERS-mimic. Thus, to increase the sensitivity, nodular Gd+ (ie, 3–9 mm) should be considered as an atypical MRI feature more than an MRI red flag. In addition, while pons, cerebellum and spinal cord are described as usual areas of punctate Gd+ in CLIPPERS, white matter and deep grey matter in the supratentorial compartment are omitted and should be added into the MRI criteria. Indeed, supratentorial punctate Gd+ is frequently reported in CLIPPERS and was observed in more than half of our patients with CLIPPERS (table 2).

    Regarding the specificity, our results highlight that some MRI red flags and additional discriminating radiological features may suggest a specific diagnosis. Ring-shaped Gd+ with mass effect, pial and cranial nerve Gd+, and steroid-resistant Gd+ imply lymphoma (PCNSL or CNS-LYG). Sulcal and cortical Gd+ are observed in PACNS, and diffuse linear Gd+ (‘CNS on fire’) suggests autoimmune GFAP astrocytopathy, as described in the literature.22 25 Before the knowledge of anti-GFAP antibodies, some patients presenting with these similar characteristic diffuse perivenular enhancements were considered as CLIPPERS. Although speculative, these patients might actually have an autoimmune GFAP astrocytopathy.26 27 Concerning additional discriminating radiological features, diffusion restriction on ADC map, arterial stenosis on MRA (confirmed by DSA) and T2*WI hypointensities were only seen in CLIPPERS-mimics and suggest PACNS. To our knowledge, only one reported case with clinical and radiological features of CLIPPERS (probable CLIPPERS according to the CLIPPERS criteria) had foci of T2*WI hypointensities matching with previous punctate enhancements.28 Thus, ADC map and MRA may be useful to detect CLIPPERS-mimics; however, it is probably too early to consider T2*WI hypointensity as an MRI red flag.

    Considering alternative diagnoses of CLIPPERS, in the presence of red flags, non-invasive investigations should include ADC map, T2*WI, MRA, DSA in case of arterial stenosis on MRA, CSF analysis, whole body CT or PET scan, and anti-MOG and anti-GFAP antibodies. When non-invasive investigations are inconclusive, neuropathological examination should be performed and target MRI red flags. Anti-CD20 staining associated with hybridisation for Epstein-Barr virus and Perls’ iron stain should be used to search for CNS lymphoma (Epstein-Barr virus-mediated or not) and PACNS, respectively.

    CLIPPERS has a relapsing remitting course in the absence of long-term steroid therapy or immunosuppressive drugs.1–4 10 Our results show that a daily dose of 30 mg of glucocorticoids or above prevents further relapses in patients with CLIPPERS but not in patients with CLIPPERS-mimic. Thus, the presence of relapse despite a glucocorticoid therapy ≥ 30 mg/day might be considered as a clinical red flag for the diagnosis of CLIPPERS.

    Finally, our study demonstrates that red flags could only occur at the second attack in CLIPPERS-mimics, with a median time of 5.5 months (min 3, max 18). The delay of red flag emergence explains the lack of specificity of the CLIPPERS criteria. Our results underscore the necessity for re-evaluating the CLIPPERS criteria at each new attack, especially in the first 18 months. In the absence of a specific biomarker for CLIPPERS, a time criterion challenging the diagnosis of CLIPPERS at each new attack may help to increase the specificity of the CLIPPERS criteria.

    In keeping with our results, we propose a practical approach in the presence of symptomatic hindbrain punctate Gd+ (figure 5).

    Figure 5

    Practical approach of symptomatic hindbrain punctate gadolinium enhancements. CLIPPERS, chronic lymphocytic inflammation with pontine perivascular enhancement responsive to steroids; CNS, central nervous system; CNS-LYG, CNS lymphomatoid granulomatosis; DSA, digital subtraction angiography; Gd+, gadolinium enhancement; MRA, magnetic resonance angiography; PCNSL, primary CNS lymphoma; T2WI, T2-weighted image.

    The question of whether CLIPPERS is a disease or a syndrome remains unsolved. We believe that three hypotheses are conceivable. First, CLIPPERS is a syndrome including prestage or unusual presentation of well-determined diseases such as autoimmune gliopathies, small vessels PACNS, systemic and restricted CNS lymphomas, and probably other diseases. Second, CLIPPERS is a prelymphoma state, which may transform into a high-grade lymphoma with a similar or close pathophysiological process described in lymphomatoid granulomatosis.29 Third, CLIPPERS could correspond to an unknown inflammatory disease. In this issue, we cannot exclude that a subset of CLIPPERS could be triggered by lymphomas. Interestingly, Tobin and coworkers10 did not exclude the diagnosis of CLIPPERS in a patient developing a PCNSL and in patients with ongoing systemic lymphomas.10 However, considering that red flags arise relatively early in the course of the disease (at the first or the second attack), but also that treatment guidelines exist for lymphomas in contrast to CLIPPERS, we have chosen to keep lymphomas in the CLIPPERS-mimic category.30–32 We assume that our choice may partly explain the decrease in the specificity of the 2017 CLIPPERS criteria. Recently, Blaabjerg and colleagues33 showed activation of complement in the pathophysiological process of CLIPPERS. In addition, high levels of vascular cell adhesion protein-1 in CSF were present in their five patients with CLIPPERS in contrast to patients with multiple sclerosis. These results should be confirmed and tested in other CLIPPERS-mimics. Indeed, our results may be biased by several mechanisms. First, as our study is retrospective, all cases of brainstem punctate Gd+ may not have been referred to us. Nevertheless, we think that the missing cases were patients with a more obvious alternative diagnosis, and that our study population comprises probably the most difficult cases. Thus, the sensitivity and specificity in our study would be reduced by this selection bias. Second, as differential diagnoses must be excluded before making a diagnosis of CLIPPERS, the diagnostic values of the proposed criteria will depend on the accuracy of the tests used to assess alternative diagnoses. However, in our study population, final alternative diagnoses were confirmed by many ways, which reduces the probability of this particular bias. Third, although we did not observe the apparition of differential diagnoses after 18 months, we cannot exclude that some alternative diagnoses would appear later.

    Pending specific biomarkers for CLIPPERS, the current criteria are necessary. However, as nodular enhancements could be seen in CLIPPERS and its mimics, this MRI finding should be considered more as an unusual radiological feature than a radiological red flag. Finally, as clinical, MRI and neuropathological red flags could be initially absent in CLIPPERS-mimics, a time criterion challenging the diagnosis of CLIPPERS along the disease course should be added. These considerations may help to increase both sensitivity and specificity of the current CLIPPERS criteria.

    Acknowledgments

    Benjamin Taieb helped design the figures, and Michael Drake helped edit the manuscript.

    References

    Footnotes

    • Collaborators Elsa Kaphan, Bertrand Audoin, MD (Department of Neurology, APHM, Hôpital de la Timone, Pôle de Neurosciences Cliniques, Service de Neurologie, Marseille, France, Site investigator); David Brassat, MD, PhD (Department of Neurology, University Hospital of Toulouse, France, Site investigator); Emmanuelle Uro-Coste, MD, PhD (Department of Neuropathology, University Hospital of Toulouse, Site investigator); Thibaut Allou, MD (Department of Neurology, Hospital of Perpignan, France, Site investigator); Jean-Christophe Ouallet, Patrice Desbordes, MD (Department of Neurology, University Hospital of Bordeaux, France, Site investigator); Giovanni Castelnovo, Eric Thouvenot, MD (Department of Neurology, University Hospital of Nîmes, France, Site investigator); Caroline Arquizan, Xavier Ayrignac, Nicolas Gaillard, MD (Department of Neurology, University Hospital of Montpellier, France, Site investigator); Nicolas Menjot de Chamfleur, MD (Department of Neuroradiology, University Hospital of Montpellier, France, Site investigator); Thierry Vincent, MD, PhD (Department of Immunology, University Hospital of Montpellier, France, Site investigator); Nadège Limousin, Cecilia Rousselot, MD (Department of Neurology, University Hospital of Tours, France, Site investigator); Delphine Loussouarn, MD (Department of Neurology, University Hospital of Nantes, France, Site investigator); Laurent Magy, MD, PhD (Department of Neurology, University Hospital of Limoges, France, Site investigator); Romain Lefaucheur, MD (Department of Neurology, University Hospital of Rouen, France, Site investigator); Jerome De Sèze, Christine Tranchant, MD, PhD (Department of Neurology, University Hospital of Strasbourg, France, Site investigator); Yannick Béjot, MD (Department of Neurology, University Hospital of Dijon, France, Site investigator).

    • Contributors GT and PL had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: GT and PL. Acquisition, analysis or interpretation of data: all authors. Drafting the manuscript: GT, BWvO, FP, VG, CL, J-PC, GA, JP and PL. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: CD.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Patient consent for publication Parental/guardian consent obtained.

    • Ethics approval The Ethics Committee of Montpellier University Hospital Research approved the study, and individual consent was obtained from all patients.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Linked Articles