Paraneoplastic neurological syndromes are remote complications of systemic malignancy, characterised clinically by subacute progression to profound neurological disability and pathologically by destruction of neurones, with or without an accompanying inflammatory response.¹ Onset of symptoms may occur more than 2 years prior to diagnosis of the accompanying neoplasm. An important finding over recent years has been that many affected patients have serum and cerebrospinal fluid (CSF) antibodies that are reactive with internal neuronal antigens.^1,2 In some cases, there is strong correlation between patterns of antibody reactivity and specific tumours, and tumours from some affected patients have been found to express proteins that immunologically cross-react with neuronal proteins.^3,4 Thus, detection of antineuronal autoantibodies in a patient with a progressive neurological disorder may allow the diagnosis of a paraneoplastic syndrome and also provide information about the nature of the underlying neoplasm.^1,2,4

The major paraneoplastic neurological syndromes had been largely categorised by 1965, and this same categorisation is still used for diagnosis today, supported by detection of antineuronal autoantibodies. In some instances, however, determination as to whether a given patient has a paraneoplastic syndrome may prove difficult. Firstly, not all patients with paraneoplastic neurological syndromes have antineuronal antibodies, nor do all patients with antineuronal antibodies have paraneoplastic neurological syndromes. Secondly, there is a subset of patients in whom no tumour can be detected despite a classical clinical presentation and high titres of antineuronal antibodies. Finally, uniform diagnostic criteria for paraneoplastic neurological syndromes, incorporating both clinical and serological data, have not been established, nor is there an established framework for categorisation of newly discovered syndromes. This same lack of standardised criteria for diagnosis of these disorders also hinders development of collaborative therapeutic trials.

Graus et al, in their paper on page 1135,⁵ present criteria by which patients with suspected paraneoplastic syndromes can be divided into “definite” and “probable” categories, based upon immunological and clinical data. “Definite” syndromes include (a) class-ical presentations (such as cerebellar degeneration or encephalomyeloneuritis) accompanied by cancer developing within 5 years of diagnosis of the neurological disorder; (b) a non-classical presentation that remits following cancer treatment without concomitant immunotherapy; (c) non-classical presentations with onconeural antibodies and cancer developing within 5 years of neurological diagnosis; and (d) a neurological disorder, classical or otherwise, in which well characterised onconeural antibodies are present but no cancer can be found. “Possible” syndromes include (a) classical presentations without onconeural antibodies or identified cancer but with high risk of an underlying neoplasia; (b) a neurological syndrome (classical or otherwise) with partially characterised onconeural antibodies and no cancer; and (c) a non-classical neurological condition without onconeural antibodies, but with cancer identified within 2 years of neurological presentation.

Graus and his colleagues have performed a significant service for neurologists by providing a system for assessing whether or not a given patient has a paraneoplastic neurological disorder. The criteria that they have developed should allow categorisation of newly identified paraneoplastic syndromes and, of even greater importance, should be of real benefit in standardising both research protocols and multi-institutional therapeutic trials.