Recurrent herpes simplex virus encephalitis secondary to carbamazepine induced hypogammaglobulinaemia

Herpes simplex encephalitis (HSE) is frequently complicated by seizures. Immunosuppression is a recognised side effect of anticonvulsant therapy. We present a case of recurrent HSE where host immunocompetence was impaired because of hypogammaglobulinaemia, most likely induced by carbamazepine therapy.

A 40‐year‐old man presented to his local hospital with a 5 day history of rigors, nausea, vomiting and photophobia. He reported formed visual hallucinations and his wife complained that over the preceding week he had been uncharacteristically short tempered. A tonic–clonic seizure was witnessed in the accident and emergency department. A cranial CT scan was normal. CSF analysis showed 120 white cells/mm³ (lymphocytes 100%), 3 red cells/mm³ and no organisms on gram stain. The CSF opening pressure was 36 cm water and protein concentration 770 mg/l (150–450). He was treated with intravenous cefotaxime, aciclovir (10 mg/kg) and phenytoin, with a presumptive diagnosis of encephalitis. Increased signal in the right medial temporal lobe was noted on an MRI scan. An EEG revealed abnormal excess theta/delta waves over both hemispheres, consistent with diffuse cerebral dysfunction. Dexamethasone (4 mg three times daily) was started on day 2 of admission. Subsequently, herpes simplex virus (HSV) DNA type 1 was detected by PCR analysis of CSF. Serum immunoglobulin levels were normal: IgA 2.47 g/l (1.0–5.0), IgG 12.5 g/l (6.8–15.6) and IgM 0.74 g/l (0.5–2.8), as was serum electrophoresis. He completed 3 weeks of intravenous aciclovir therapy and was discharged on oral phenytoin.

One month after discharge, he re‐presented with a maculopapular rash and mouth ulcers. A skin biopsy confirmed erythema multiforme and carbamazepine was substituted for phenytoin.

Seven months after his initial presentation, he was admitted to our hospital following a tonic–clonic seizure. His wife reported that he had been increasingly agitated over the preceding 10 days. He was pyrexial, confused and non‐compliant with anticonvulsant drug therapy. A cranial CT scan demonstrated atrophic changes in the right temporal lobe. The CSF opening pressure was 38 cm water, white cell count 20 cells/mm³ (82% lymphocytes, 18% polymorphs), protein concentration 600 mg/l and glucose 3.9 mmol/l (blood glucose 7.9 mmol/l). He was restarted on intravenous aciclovir (10 mg/kg) and cefotaxime. An MRI scan confirmed an area of scarring in the right medial temporal lobe. EEG findings were consistent with viral encephalitis. Immunoglobulin levels revealed profound hypogammaglobulinaemia; IgA 0.85 g/l (0.8–2.8), IgG 1.1 g/l (6–16) and IgM 0.16 g/l (0.5–1.9). (The immunoglobulin reference ranges vary between the two hospitals.) A full blood count demonstrated lymphopenia; total lymphocyte count 0.9×10⁹/l. Lymphocyte subsets were normal on immunophenotyping. CSF PCR detected the presence of HSV DNA type 1.

Hypogammaglobulinaemia was attributed to carbamazepine therapy; this was cautiously withdrawn and substituted with sodium valproate. Antibiotic prophylaxis and twice weekly intravenous pooled human immunoglobulin replacement therapy (Flebogamma 0.4 g/kg) (IVIg) were commenced. He completed 3 weeks of intravenous aciclovir therapy (10 mg/kg). Regular immunoglobulin was continued for 6 months. The IgM level was monitored during this time as a marker of immune recovery (IVIg contains negligible levels of IgA and IgM). Immunoglobulin levels were normal 6 weeks after starting IVIg therapy (IgA 1.29 g/l, IgG 11.0 g/l and IgM 0.39 g/l). However, after 8 months (when IVIg replacement had been discontinued), immunoglobulin levels fell again (IgG 3.6 g/l, IgM 0.32 g/l with normal IgA) and he received two further infusions of IVIg (0.4 g/kg) over a 6 month period. It is now 9 months since he was last treated with IVIg and his immunoglobulin levels remain normal off treatment (IgG 7.05 g/l). No alternative cause for hypogammaglobulinaemia has since emerged. He has been seizure free for 2 years. Psychometric testing highlights difficulty with tasks requiring complex attention skills and significant anterograde memory impairment. He has returned to part‐time employment.

The incidence of recurrent HSE following aciclovir therapy has been reported as up to 5%.¹ Low levels of antibody generally predispose to extracellular bacterial infection but susceptibility to viral infection,² including enteroviral meningo‐encephalitis, is recognised.

Anticonvulsants have been reported to cause immunosuppression, typically IgA deficiency, although carbamazepine has also been associated with low levels of IgG.³,⁴,⁵ There are insufficient data to be sure of the mechanism underlying hypogammaglobulinaemia induced by carbamazepine, but the majority of the few reported cases cite low levels of circulating B cells with normal cell mediated immunity.³,⁵,⁶ Fortunately, immunosuppression attributed to anticonvulsants is generally reversible following drug withdrawal.

The true prevalence of immunosuppression secondary to anticonvulsant therapy is difficult to estimate. One study reported that 60% of general hospital patients on phenytoin and 47% of those on carbamazepine fail to mount delayed hypersensitivity reactions to antigens or make antibodies to Salmonella typhi and tetanus toxoid.⁴ Both visceral leishmaniasis⁶ and reactivation of human herpes virus 6⁷ have been reported in patients with hypogammaglobulinaemia secondary to carbamazepine therapy but, to our knowledge, recurrent HSE has not been previously attributed to anticonvulsants.

Erythema multiforme is a recognised side effect of phenytoin,⁸ thought to be the causative agent in this case. However, erythema multiforme is associated with HSV⁹ and recurrent erythema multiforme with hypogammaglobulinaemia has also been reported.¹⁰ Other clinical signs of HSE and immunoglobulin levels were not recorded when this patient presented with erythema multiforme.

Given that seizure activity is a frequent complication of encephalitis and that many of these patients require long term anticonvulsant treatment, our finding of recurrent HSE associated with hypogammaglobulinaemia due to carbamazepine has treatment and monitoring implications. It may be prudent to consider monitoring for hypogammaglobulinaemia in these patients for whom viral reactivation can have such potentially disastrous consequences. In the absence of routine monitoring, recurrent viral or bacterial infection certainly necessitates measurement of immunoglobulin levels.

Acknowledgements

CMR is funded by a Patrick Berthoud Clinical Research Fellowship.