Progressive multifocal leukoencephalopathy associated with carbamazepine-induced immune dysfunction and recovery after carbamazepine cessation

Background

Progressive multifocal leukoencephalopathy (PML) is a usually fatal demyelinating disease of the brain. It is caused by reactivation of latent John Cunningham virus (JCV), usually in the context of severe immunosuppression as a consequence of AIDS, haematological malignancies, organ transplant recipients or treatment with immunosuppressant drugs. Rarely, PML may occur in individuals with minimal or occult immunosuppression.1 We report a case of PML in a patient treated on long-term carbamazepine, highlighting the extent to which antiepileptic drugs can disrupt normal immune function.

Case presentation

A woman in her mid-60s with epilepsy presented with 1 month’s insidious-onset gait impairment and left-hand incoordination. Shortly before onset, they had a brief influenza-like illness with sore throat, diarrhoea and lethargy. The patient had well-controlled epilepsy, secondary to grey matter heterotopia, on carbamazepine 400 mg two times per day and levetiracetam 250 mg two times per day. The medical history also included migraine and periodontitis. She had no relevant family history, had never smoked or travelled abroad and consumed alcohol infrequently. Figure 1 summarises the illness course.

Examination demonstrated normal cranial nerves, tone, power and reflexes. She had an intention tremor, past-pointing and dysdiadochokinesia, left more than the right. Sensation was normal. Gait was slow and broad-based.

Investigations

Investigations revealed lymphopenia (0.8×10⁹/L (normal range 1.5–4.0); T-lymphocytes 0.63 (0.78–2.24), CD4+ count 0.50 (0.49–1.64), CD8+ count 0.17 (0.17–1.20); B-lymphocytes 0.25 (0.08–0.49)) and hypogammaglobulinaemia (IgG (3.4 g/L (normal range 6.0–16.0)), IgM (0.29 g/L (normal range 0.5–3.0) (see figure 1), normal IgA). The lymphopenia was longstanding (also present 1 and 4 years previously). Immunoglobulins had not previously been tested. Cerebrospinal fluid (CSF) testing showed normal CSF protein (319 mg/L), normal CSF glucose (2.7 mmol/L vs 4.5 in plasma), 0 white blood cells per µL, no growth on culture and absence of oligoclonal bands. MRI head revealed peritrigonal grey matter heterotopia, asymmetrical cerebellar T2 and fluid attenuated inversion recovery hyperintensity (left more than right), with gadolinium ring-enhancement in the left middle cerebellar peduncle (figure 2).

Differential diagnosis

On suspicion of postinfectious cerebellitis, 1 month after presentation, we prescribed prednisolone 60 mg daily for 2 weeks by a rapid taper. Carbamazepine was gradually down-titrated and stopped, despite normal carbamazepine levels, in case it was contributing to their ataxia; levetiracetam was increased. Amoxicillin was trialled to cover central nervous system listeriosis.

Unfortunately, the patient deteriorated over 2 months, developing worsened asymmetrical dysmetria, hypermetric saccades, dysarthria, with incomprehensible speech and cognitive impairment. She required assistance of two to mobilise and needed help for all activities of daily living (ADLs). Montreal cognitive assessment was 17/30. MRI repeated after 1 month showed worsening cerebellar changes (figure 2). CT chest–abdomen–pelvis, fluorodeoxyglucose (FDG)-positron emission tomography (PET) and two further CSF examinations were unremarkable. Investigations for several infective, inflammatory and metabolic causes were normal or negative. These included negative HIV test, negative ANA, non-specific antineutrophil cytoplasmic antibodies staining with negative MPO and PR3, normal serum angiotensin-converting enzyme, negative serology for syphilis, borreliosis, brucellosis and toxoplasmosis, and normal imaging as detailed above.

Outcome and follow-up

The following month, JCV DNA was detected in blood (3903 copies/mL) and in CSF (1636 copies/mL) from samples taken 2 months prior, confirming a diagnosis of PML. The diagnosis was clinically definite according to the 2013 criteria of Berger et al (a combination of compatible clinical and imaging findings together with positive JCV PCR in CSF).2 The delay in diagnosis was due to PML not been suspected initially.

We did not test antimyelin oligodendrocyte glycoprotein (MOG) antibodies, although anti-MOG antibody associated disease (MOGAD) can cause similar imaging findings.3 We do not consider MOGAD is plausible in this case given (1) the gradual onset over 3 months; (2) the lack of typical clinical features or other radiological features of MOGAD such as optic neuritis or transverse myelitis; (3) the older age than is typical for MOGAD and (4) the positive CSF PCR for JCV.

By the time of confirmation of diagnosis, the patient had begun to improve: cognition was improving, speech was clearer and she was able to mobilise with a stick. The clinical improvement subsequently plateaued and mild dysarthria, left upper limb dysmetria and gait ataxia persisted at further follow-up visits and continued to be stable at the time of our last contact four and half years after presentation. She regained ability to perform almost all ADLs independently. Lymphocytes, IgG and IgM subsequently normalised (figure 1). Serial imaging showed gradual improvement (figure 2). CSF JCV was undetectable 6 months after presentation.