Editor,—The 14-3-3 proteins are a group of highly conserved proteins involved in intracellular signalling. Detection of 14-3-3 brain protein has been described in cerebrospinal fluid (CSF) of patients with transmissible spongiform encephalopathies including both sporadic and variant Creutzfeldt–Jakob disease.^1,2 False positive results have been reported in conditions producing (sub)acute neuronal destruction, including herpes simplex encephalitis, ischaemic stroke, multi-infarct dementia, and paraneoplastic syndromes.^1–3 We postulated that 14-3-3 brain protein may be detected in CSF from patients with HIV associated dementia complex (HADC) as this condition is characterised neuropathologically by a giant cell encephalitis, leucoencephalopathy, astrogliosis and neuronal loss.

We prospectively studied 17 HIV antibody positive patients (14 men) aged 27–60 (median 37) years, with CD4 counts of 0–220 (median 20) cells ×10⁶/l, who underwent lumbar puncture for investigation of HADC (six patients), staging of lymphoma (five patients), or investigation of other conditions (six patients): epilepsy (two), cervical radiculopathy (one), chronic demyelinating polyradiculopathy (one), CMV encephalitis (one), self limiting headache (one). Of those with HADC, the severity of dementia assessed using Memorial Sloan-Kettering criteria,⁴ was mild in two and moderate in four. The degree of atrophy on cranial magnetic resonance imaging, used as a marker of neuronal loss⁵ was mild in four and moderate in two. Clinical details of those with lymphoma are given in table 1. At each lumbar puncture an aliquot of CSF (250 μl) was frozen immediately at −20°C and stored for subsequent 14-3-3 protein analysis.

CSF was routinely processed as described previously.⁶ Detection of 14-3-3 protein was done without knowledge of the patient's diagnosis, using a technique described by Hsich et al,¹ modified to use anti-14-3-3 γ polyclonal rabbit antibody.

In 14 of 17 patients CSF was negative for 14-3-3 protein. Of the three with detectable 14-3-3 protein in CSF, all had lymphoma but only one had CNS disease, the other two had only extraneural disease (table 1). These data, although from a small study population, suggest that detection of 14-3-3 protein in CSF is not useful for diagnosis of HADC. Detectable 14-3-3 protein has previously been reported in a non-HIV infected patient with CNS lymphoma,³ so this observation in our patient is not unique, although brain necrosis from coexisting cerebral toxoplasmosis provides an alternative explanation. Of the two patients with extraneural lymphoma and detectable 14-3-3 protein in CSF, one had EBV DNA in CSF and so was at high risk of developing cerebral lymphoma. This possibility could not be confirmed as necropsy was not performed. In neither of the latter two patients was there a CSF pleocytosis, so contamination by 14-3-3 protein derived from peripheral blood leucocytes is unlikely. In the final case the absence of limbic encephalitis or cerebellar degeneration³ makes it difficult to ascribe the finding to a paraneoplastic process.