Anaplastic astrocytoma mimicking progressive multifocal leucoencephalopathy: a case report and review of the overlapping syndromes

Multiple sclerosis (MS) is a disabling inflammatory demyelinating disease of the central nervous system (CNS). Early initiation of immunotherapy and its adjustment in view of ongoing inflammatory disease activity is desirable [1]. The main treatment goals aim at terminating inflammation and at reducing axonal damage [1]. Establishing MS diagnosis and decisions about the initial and ongoing treatments should not be made until other disorders that could better explain neurological symptoms and signs are excluded [1]. Although glial brain tumors (GT) have occasionally been reported in patients with MS, with only about 80 cases described in medical literature so far [2‐5], this co-occurrence is uncommon since MS is caused by putative CNS autoimmune mechanisms whereas brain neoplasms may depended on a subclinical immunosuppressive state [6]. However, the last 15 years have seen increased use of immunomodulatory therapies (IMT) for relapsing MS, and considerable progress in the development of new, much stronger IMT for MS [7]. That rises questions about long-term safety of IMT as well as their risks and benefits [7]. Currently we know that several IMT make patients more susceptible to developing dangerous brain infection caused by John Cunningham virus (JCV) called progressive multifocal leukoencephalopathy (PML) [8‐12]. However, MRI findings of tumefactive demyelinating lesions (TDL), PML and GT can overlap [12‐19]. The first appearance of atypical brain lesion in an MS patient should lead to more extensive investigation in order to exclude another disease [15‐19]. However, in some cases only biopsy is capable to reveal the cause of atypical MRI lesion [20, 21].

Here we describe the case of a malignant GT in a patient with early onset of MS. To the best of our knowledge, anaplastic astrocytoma misdiagnosed as PML has not yet been described. We discuss diagnostic tools that can help in differential diagnosis.

Here we report on a 27-year-old woman with the first neurological symptoms suggesting MS in 1999 at the age of 12. Her medical and psychosocial history was negative, but her family history was positive. Patient’s father is also treated for MS. The patient’s clinical timeline follows (Figs. 1, 2, 3, 4, 5, 6 and 7):

Our case shows that MS can have variable presentation therefore concurrence of MS with brain tumours may remain undetected for some time. One possible explanation is that an association of MS with overall risk of cancer has not been proved [22, 23]. One meta-analysis even identified a small but significant reductions of total cancer risk in patients with MS (odds ratio [OR] 0.92; 95% CI 0.87–0.97, p = 0.004) [24]. Another meta-analysis suggested lower overall co-occurrence of cancer in patients with MS [24]. On the other hand, a recent large systematic study evaluating risk and survival of brain tumours among patients with autoimmune disorders found higher standardised incidence of brain tumours in MS (OR 2.14; 95% CI 1.65–2,73) than in other autoimmunity disorders of the CNS [25]. The most frequent brain tumours were gliomas with OR 1.49; 95% CI 0.91–2.31. In this study, the risk of glioma-associated deaths in MS patients was relatively high (OR 2.3; 95% CI 1.47–3.61). In addition, the data on survival showed the same decrease for both benign and malignant types of tumours co-occurring with MS [25]. Curiously, MS was the only autoimmune disease for which the overall brain cancer and especially glioma-specific survival appeared to decrease in recent years [25]. One explanation could be that comorbidity weakens patient’s physical condition. Moreover in MS, glial tumours interfere with mortality burden due to reduced treatment options and lower capacity of the affected brain to resist tumour growth. Another explanation could be that brain tumours remain hidden in demyelinated brain tissue and therefore they are diagnosed later, in advanced stages, which was also our case.

At the beginning of our patient’s disease the pathological lesion was connected to the frontal pole of lateral ventricle stretching to the periphery dispersing into the white matter. It was T2- and FLAIR- hyperintense and its localisation near the ventricle favoured TDL over GT (Fig. 1b,c). MRI of the atypical lesion fulfilled the reported characteristics of TDL [15‐18]. However, MRI signs of TDL and GT can overlap easily [15, 17, 18]. Brain CT examination may be helpful in differentiation between the TDL and malignant GT [26] but we did not perform it in the early stage. Later, growth activity of the atypical lesion in the frontal lobe, good response to steroids, and absence of clinical signs suggesting tumour [15‐18] led to treatment escalation. The effect of natalizumab and fingolimod to the lesion growth was evident and its MRI characteristics evoked suspicion of PML [11‐13]. Although CSF examination of DNA JCV is a reliable test of PML, there were also reported cases of PML in JCV-PCR CSF negative patients in early PML as well as during Immune Reconstitution Inflammatory Syndrome (IRIS) [27].

In our patient, minimal mass effect and absence of post-contrast enhancement supported PML diagnosis [11‐13]. Double Inversion Recovery (DIR) (Fig. 4c) confirmed cortical involvement but did not add new information. Cortical lesions are typical for advanced forms of MS but they can be found in PML as well as GT [11‐13, 17‐20]. Diffusion Weighted Imaging (DWI) hyperintensities (Fig. 3d) favoured PML [11‐13]. ¹H–MRS revealed decreased creatine to choline ratio in several areas of the lesion. We did not find changes of glutamate and glutamine peaks, as was reported by Yamashita [14]. ¹H–MRS is able to identify brain tumours or demyelination [11, 28‐31] although several authors recommend caution when interpreting ¹H–MRS measured concentrations of metabolites in isolation [11, 31]. As our lesion was not well defined its precise place was estimated. Its localisation and size was finally revealed by ¹¹C MET PET. Increased uptake of ¹¹C–MET, i.e. high proliferation index of ¹¹C–MET indicating malignancy [32], was found in the right cortical frontal area. Slightly increased uptake of ¹¹C–MET was also found in left frontal and occipital cortex.

Cortical clinical symptoms, including disorientation, confusion, epileptic seizures, behavioural changes and headaches appeared in our patient in terminal stages of GT progression. They can also be found in patients with PML [8‐11]. Moreover, symptomatic seizures correlate with cortical demyelination in advanced forms of MS [33], and could be associated with TDL [15, 16].

Basic CSF examination in our patient showed normal proteinorhachia and cell count, but increased number of oligoclonal bands type 2 (14 bands) and IgG index 1.44. These findings suggest active demyelinating processes associated with MS [15, 34] or PML-IRIS [10, 11]. This reaction is unusual in GT, where increased lactate in CSF and hyperproteinorrhachia would be expected [6]. However, we did not prove it.

Retrospective analysis of the blood cellular immunity over the years showed chronic increase of CD4+ (50–66%) and mild deficit of CD8 (11–12% with higher CD4/8 index (4.5–5.0) at the beginning of her disease (2007–2009). During interferon beta I treatment, CD4+ (55%) and CD8 (10%) remained unchanged, while CD16 + CD56 Natural Killers (NK) (4%, 66 abs) fell, and CD19 rose (31%, 513 abs). Natalizumab chronically decreased CD8 subpopulations of T lymphocytes. Fingolimod reduced CD4+ but significantly increased % of NK (49%). Humoral immunity remained normal and unchanged during all that time. Deficit of NK cells in our patient could decreased resistance against brain tumour growth, as NK cells can play an important role in anti-tumour immunity [35]. Moreover, it is known that in MS, autoimmune conditions are mediated mostly by CD4+ T-cells with a proinflammatory Th1 and Th17 phenotype, causing inflammation and demyelinating lesions in the CNS [36]. The dominancy of glial tumours in MS leads to a hypothesis that chronic hyperactivation of glial cells via Th1/Th17 pathways could cause their neoplastic transformation in demyelinated lesions [34]. On the other hand, if MS-associated proinflammatory Type17 T-cells mediate potent antitumour immunity [34], suppression and sequestration of those cells could potentially cause its breakdown. We did not test our patient’s Th17 lymphocytes, the mechanism could only be hypothesised. During natalizumab and fingolimod treatment we found lymphocytes decreased in periphery but we are not able to describe changes in the brain of our patient. This selective deficiency could have been involved in inefficient antitumour immune surveillance and tumour progression [37, 38].

Treatment by IMT could potentially trigger a variety of immunologic abnormalities that are typical for patients with malignant brain tumours [38, 39]. IMT may targeted many factors such as impaired responsiveness of peripheral blood lymphocytes to mitogens [39], failure of the T cells mediating adaptive immune responses within the local tumour micro-environment [40], and induction of regulatory T cells [41]. Development of GT can potentially be influenced by immunosuppressive cytokines (such as IL-10, TGF-β, and prostaglandin E2), and by down modulation of co-stimulatory molecules by antigen presenting cells (APCs) resulting in loss of T cell effector function [41]. Long-term monitoring of these markers in patients treated by highly effective IMT could be beneficial.

In tumour vessels, aggressive endothelial proliferation [42, 43], increases CD34+, a marker of endothelial progenitor cells [42]. Although natalizumab treatment results in an increase in CD34+ progenitor cells in both the bone marrow and the blood [44], it is not clear whether it can enhance tumour angiogenesis in a brain tumour. Moreover, increased angiogenesis was also described in PML lesions [11].

Indeed, the CNS biopsy remains the most useful method for defining the histological type of an atypical brain lesion susceptive of tumour. In our patient, Vimentin’s over-expression in cancer correlates well with increased tumour growth, invasion and poor prognosis [45]. OLIG2 is highly expressed in all diffuse gliomas [46] and was found expressed in our patient’s anaplastic astrocytoma samples. Tumourous cells were Nogo positive, Nogo-A exerts a growth inhibitory function leading to restricted axonal regeneration [47]. In our case Ki-67 was ≤5% but >2%. Worsened survival has been associated with Ki-67 > 2% [48]. Mutation of IFH1 and positive p53 in our patient suggests secondary nature of the astrocytoma [49].

In our patient we did not prove the histopathological triad of PML (multifocal demyelination, hyperchromatic, enlarged oligodendroglial nuclei, and enlarged bizarre astrocytes with lobulated hyperchromatic nuclei) [11, 27], which would have been a rather convincing evidence of the disorder. In Table 1, we summarise the differential diagnoses of tumefactive demyelinated lesions, malignant glial tumours, PML and PML associated with IRIS. We hypothesise that IMT may have transformed the glial cells into malignant GT. However, in a descriptive study among 22,563 French MS patients, including patients receiving IMT, brain tumours were not detected in total of 253 patients (1.2%) with a history of cancer [45]. In the SENTINEL trial of natalizumab in combination with interferon beta-1a, the incidence rate of cancer in the combination group (n = 589) was 1% compared to 2% in the interferon beta-1a alone group (n = 582) [50]. In interim analysis of TOP study, including 4821 patients from 16 countries, the incidence of malignancies was 0.5%. There were 24 patients with 12 types of malignancies. Glioblastoma was detected in 1 patient, while breast cancer was the most common, affecting seven patients (all female) [51].

Our patient was the carrier of human leukocyte antigens (HLA)-DRB 1*15 and DRB 1*11 alleles, and heterozygote of single nucleotide polymorphism rs3135388, which is typical for multiple sclerosis [52]. Although expression of HLA antigens is important for the immune response against infectious agents and malignant cells, there is an information gap about the link between HLA antigens and brain glial tumours. However, one prospective study, focused on HLA typing of German Caucasian patients, found that HLA-DRB1*15 in combination with HLA-DRB1*11 was associated with higher (a 13.4-fold increased) risk of glioma than was found for other HLA alone or in other combination [53]. It might explain the occurrence of astrocytoma Grade III in our patient.

We suppose that the anaplastic astrocytoma in our patient, developed after the diagnosis of MS, could have arisen in demyelinative plaques with reactive gliosis and could have evolved from a low grade glioma to anaplastic astrocytoma over time. The role of the immunomodulatory treatment as well as other immunological factors in malignant transformation of the tumour can only be hypothesised. The association between gliomas and MS is uncommon but it must be kept in mind when an atypical tumefactive lesion develops in a patient with MS. In our case, it is the first time when malignant glioma was misdiagnosed as PML.