Serum neuron-specific enolase in children's cancer.

To test its diagnostic potential and sensitivity in paediatric malignancy, serum NSE was measured at diagnosis in 191 children with solid tumours and 25 with acute leukaemia. In stages I + II, III + IV and IVs neuroblastoma median levels were 18.0, 91.0 and 24.0 ng ml-1 respectively. For Wilms' patients, median values for stages I, II, III and IV disease were 16.6, 18.0, 29.0 and 47.0 ng ml-1 respectively. High levels of NSE were also found in patients with other types of tumour. Children in clinical remission after treatment for neuroblastoma invariably had normal NSE levels (mean +/- s.d. = 9.2 +/- 3.0 ng ml-1) even though the majority had radiologically identifiable residual disease. The values rose when relapse was radiologically or clinically obvious. We conclude (a) that, though levels of greater than 100 ng ml-1 are highly suggestive of advanced neuroblastoma, caution should be exercised in using serum NSE as a diagnostic test in children with cancer and (b) that serum NSE levels are not a sensitive index of residual neuroblastoma in patients, with initially elevated levels, that are receiving treatment.

Summary To test its diagnostic potential and sensitivity in paediatric malignancy, serum NSE was measured at diagnosis in 191 children with solid tumours and 25 with acute leukaemia. In stages 1+II, III+ IV and IVs neuroblastoma median levels were 18.0, 91.0 and 24.0 ng ml -I respectively. For Wilms' patients, median values for stages I, II, III and IV disease were 16.6, 18.0, 29.0 and 47.0ngml-1 respectively. High levels of NSE were also found in patients with other types of tumour. Children in clinical remission after treatment for neuroblastoma invariably had normal NSE levels (mean+s.d.=9.2+3.0ngml-1) even though the majority had radiologically identifiable residual disease. The values rose when relapse was radiologically or clinically obvious. We conclude (a) that, though levels of > 100 ng mlare highly suggestive of advanced neuroblastoma, caution should be exercised in using serum NSE as a diagnostic test in children with cancer and (b) that serum NSE levels are not a sensitive index of residual neuroblastoma in patients, with initially elevated levels, that are receiving treatment.
Enolase, (2-phospho-D-glycerate hydrolyase or phospho- The serum samples were obtained from patients with solid pyruvate hydratase EC 4.2.1.11), a glycolytic enzyme, is tumours or acute leukaemia attending our hospitals present in the brain in three isoforms (xx, a-and 'it). The o including additional material from a bank of serum obtained form is synthesized by glial cells and most cells in the body, at diagnosis from children with a variety of solid tumours this isoform is called non-neuronal enolase (NNE) attending The Hospital for Sick Children, London. The (Marangos et al., 1980). The y form is known as neuron diagnoses of all the tumours in this series were based on specific enolase (NSE) as it is produced by neurons and histological examination. All samples were stored at -20-C.
neuroendocrine cells, and thought at one time to be specific Immunoreactive NSE is stable at least up to 4 years at this for these cell types, subsequently its demonstration in many temperature (Zeltzer et al., 1986), the distribution of NSE other cell types has indicated it lacks the specificity that the values from samples stored >2 years being similar to that of name implies (Schmechel, 1985). NSE is one of several samples analysed soon after collection. The controls were: (a) markers characteristic of cells that make up the amine Children awaiting cardiac surgery at the Hospital for Sick precursor uptake and decarboxylation (APUD) system. Children, London; (b) children with non-malignant diseases Others include the peptide hormones and L-dopa attending the Service d'Hematologie Pediatrique, Cliniques decarboxylase which are often expressed by tumours that Universitaires St. Luc, Brussels. Neuroblastomas were staged have APUD characteristics, including small cell carcinoma of according to Evans et al. (1980). Stage IV neuroblastoma is the lung, neuroblastoma, phaeochromocytoma and a variety defined as remote disease involving the skeleton, organs, or of rarer 'small round cell' tumours. Serum NSE levels can be distant lymph nodes and stage IVs defined as patients with raised in neuroblastoma and have prognostic significance localised tumour who would otherwise be stage I and II but (Ishiguro et al., 1983;Notomi et al., 1985; who have remote disease confined to one or more of the 1983; 1986). However, the experience of the Children's following sites: liver, skin or bone marrow (without radio-Cancer Study Group in the United States, indicates NSE is graphic evidence of bone metastases on complete skeletal probably not particularly useful for monitoring the treatment survey). Wilms' tumours were staged according to D'Angio of neuroblastoma as recurrence can occur without a rise in et al. (1980). the serum NSE level (Zeltzer et al., 1986).
Comparisons between the groups were analysed first by The earlier reports on serum NSE in solid tumours in Kruskal-Wallis one way analysis of variance. Differences children were limited to a few centres working with between groups were defined by Mann-Whitney test for laboratories where an assay had been produced. The skewed distributions. introduction of commercial NSE assay kits has provided a wider opportunity to evaluate the measurement of serum NSE in paediatric oncology. The NSE tests reported in this Results study use an antiserum directed against the j y sub-unit, it reacts with the o' and y7' isoforms both of which are present in the serum. (Ishiguro et al., 1983). In this paper we have  10.2 to 7,200ngmlP1, median 75ngmlF1; in two (10.5%) the level was <25ngml-P, and in 6 (31.5%) it was > 100 ngml-. including neuroblastoma, neuroganglioma, medulloblastoma, phaeochromocytoma (Triche et al., 1985;Odelstad et al., 1981) and retinoblastoma (Kivela, 1986). By contrast, focal staining for NSE has been observed in Ewing's sarcoma, rhabdomyosarcoma and lymphoma (Triche et al., 1985) and >25 ng ml' The level of > 100 ng ml1 was chosen as a in several tumours in adults including renal cell carcinoma second discriminant because it has been shown to have (Vinores et al., 1984). Biochemical analysis has shown that prognostic significance in stage IV under 1 year NSE accounts for 28% to 62.5% of the enolase activity of neuroblastoma; levels > 100 ng ml1 carry a worse prognosis neuroblastoma, but only 1% to 4.5% of that in Wilms' (Zeltzer et al., 1986).
tumours (Odelstad et al., 1982). Fractionation of the enolase Mean NSE levels in patients with neuroblastoma and indicated neuroblastomas contained the ay, and yy forms but Wilms' tumour differ significantly (P>0.0001) and both in Wilms' tumours and gliomas the aa was the dominant show a marked skew distribution, whilst those in the form with only a trace of ay and no yy (Odelstad et al., 1982; remaining groups of patients had similar NSE distributions Beemer et al., 1984;Ishiguro et al., 1983). with far less skewness. The analysis of variance showed that Our results in children with neuroblastoma and gangliothere were highly significant differences between the groups neuroblastoma mirrored those previously reported (Ishiguro (P=0.0001). When the analysis was restricted to the et al., Notomi et al., 1985;Zeltzer et al., 1986). We lymphoma, Ewing's sarcoma, soft tissue sarcoma, 'other have confirmed the direct association between stage and tumour' and control groups significant differences were still serum NSE levels (Zeltzer et al., 1986) and that patients with present (P=0.0092). The levels of NSE in Ewing's tumour stage IVs disease have relatively low NSE values. Although and 'other tumour' groups were significantly increased the majority of them have residual disease readily compared to the controls (P=0.002, and 0.009 respectively).
identifiable by radiological imaging methods, nearly all In 72 patients with neuroblastoma before treatment 57 patients in clinical remission after chemotherapy with a good (59%) of them had a serum NSE >25 ngml1 and in 31 partial or complete response according to the criteria of (44%) the value was > 100ngml 1. Levels in patients with Shafford et at. (1984) had NSE levels similar to controls, stage IVs were low compared to those with true stage IV whatever the initial stage. The sensitivity of the test is disease (Evans et at., 1971). The median levels in stages I & therefore limited.
II, III & IV, and IVs were 18.0, 91.0 and 24.0OngmlP-By contrast, the high levels of serum NSE we have respectively. Four cases of ganglioneuroma had NSE levels identified in advanced Wilms' tumours and renal carcinoma of 9.9, 14.1, 15.4 and 21.3 ngml1 respectively. Serum levels confirmed using two independent assays cannot be easily in the two other tumours of APUD cell origin in this seriesexplained and have not been reported by other investigators.
both were phaechromocytomas -were 22.4 and 16.6 ng ml 1. As previously reported histological sections did not show Nine (64%) out of 14 of patients with stages III & IV any NSE reactivity and our own studies were also negative.
Wilms' tumour had NSE levels >25 ngml-1 and raised Haimoto et at. (1986) have demonstrated that the loops of levels were also observed in two cases of renal adeno-Henle and renal collecting ducts normally contain high concentrations of NSE. The proximal tubules contain a enolase. In renal carcinoma, thought to originate from proximal tubules, there is probably an induction of the y enolase production. In Haimoto's series 20 (49%) out of patients had a raised serum NSE. Though there is no evidence for the induction of y enolase synthesis in Wilms' tumour tissue, it is possible that the renal damage caused by the tumour might lead to release enolase (NSE) into the circulation, but this does not explain why only patients with stages III & IV disease showed this phenomenon. We have observed two adults with renal carcinoma and high NSE levels 60 and 92ngmml1 (unreported data) and in one child with a benign cystic renal lesion the serum NSE was 45 ng ml -1, but renal failure is not a cause of a raised NSE (Ruibal et al., 1985). The association between renal tumours and a raised serum NSE is firm, but the mechanism could vary from one disease to another. The practical aspect of this finding is that caution must be exercised in using a serum NSE level in differential diagnosis of Wilms' tumour and neuroblastoma. It is only when the NSE is > 100 ng mlthat the test is a strong indicator of neuroblastoma, 32 (88%) out of 36 patients with an NSE level > 100 ng mlhad neuroblastomas, the others included I ALL, I renal carcinoma and 2 Wilms' tumours. Zeltzer et al. (1986) reported the NSE levels in 10 cases of acute leukaemia in children, range 12-286 ng ml-', the highest level was in a case of ALL; in 10 cases of Ewing's sarcoma the levels were 8-47 ng ml-1. A single case of hepatoblastoma had an NSE level of 176 ng ml-1. Our findings show a similar incidence of occasional raised NSE levels in most of the tumour groups not of neuroendocrine origin.
Our experience, from a large series of patients, has confirmed that serum NSE levels can be raised in a variety of childhood tumours, thus limiting the value of this test in diagnosis. Our experience of serial measurements in patients with neuroblastoma during treatment is limited but it would appear that levels can return to normal early during chemotherapy, at a time when there is still clinical and radiological evidence of residual disease. Thus, the test is not particularly sensitive and it is doubtful whether serial serum NSE measurements will be more valuable than serial catecholamine levels in monitoring of children on treatment (c.f. Dranoff & Darell, 1984).