Serum Neuron-Specific Enolase, Carnosinase, and Their Ratio in Acute Stroke
An Enzymatic Test for Predicting Outcome?
Background and Purpose Few admission variables adequately predict neuronal damage and prognosis in individual patients after stroke. Therefore, there is a need for a reliable noninvasive surrogate measure of clinical outcome.
Methods We have developed a surrogate measure of stroke outcome using the ratio of serum neuron-specific enolase (NSE) to human serum carnosinase (HSC) in 124 patients with acute ischemic or hemorrhagic stroke and 61 matched control subjects. Serum NSE is known to rise and HSC to fall after neuronal injury such as cerebral ischemia.
Results Serum NSE levels were significantly higher and HSC levels lower in the patient group. The NSE/HSC ratio was elevated in patients with stroke: median (semiquartile) hemorrhages, 0.072 (0.033); infarcts, 0.039 (0.026); and control subjects, 0.019 (0.014), P=.0001. Patients with a primary intracerebral hemorrhage had nonsignificantly higher ratios than those with an infarct (P=.082). The NSE/HSC ratio was significantly associated with 90-day outcome measured in two out of three disability and handicap scales: modified Barthel Index (rs=−.34, P=.001), modified Rankin Scale (rs=.30, P=.002), and Lindley Score (rs=.19, P=.057). Patients who died or were institutionalized had higher ratios than those who were discharged home: 0.069 (0.043) versus 0.038 (0.024), P=.011. Correlations between the NSE/HSC ratio and outcome were comparable to those between patient age or consciousness level on admission and clinical outcome.
Conclusions We believe that measurement of NSE, HSC, or their ratio may be useful in the assessment of patients with acute stroke with respect to diagnosis and prediction of clinical outcome.
Few admission variables adequately predict neuronal damage and prognosis in individual patients after stroke. Therefore, there is a need for a reliable noninvasive surrogate measure of clinical outcome. A measure in the form of serum creatine kinase already exists for myocardial infarction and is loosely related to outcome. Studies in stroke have concentrated on neuronal proteins such as NSE or S-100 within the CSF.1 2 More recently serum NSE has been evaluated after stroke in two small studies, with high concentrations correlating with an unfavorable functional outcome.3 4 Another marker, HSC activity, is reduced in stroke,5 but its relationship to outcome is unknown.
As the serum NSE level is higher, and HSC activity is lower after stroke, and since both individually exhibit considerable overlap with normal subjects, we postulated that the ratio of NSE to HSC might be useful in evaluating neuronal damage and predicting clinical outcome. This study investigates this hypothesis in a well-characterized group of patients admitted with acute ischemic or hemorrhagic stroke.
Subjects and Methods
One hundred twenty-seven consecutive stroke patients admitted to the Acute Stroke Unit at King's College Hospital6 between May 1994 and August 1995 were studied. Stroke was diagnosed on clinical grounds and the type (infarction versus hemorrhage) confirmed with the use of computed tomography (CT) neuroimaging (Siemens Somatom DRH) within 3 days of stroke onset. The study had local Research Ethics Committee approval, and both patients (or relatives if needed) and control subjects gave informed consent for venipuncture. Discharge destination or in-hospital death was noted on all patients. Eighty-one patients who were alive at 3 months were followed up in the outpatient clinic, and convalescent samples were obtained. Telephone questionnaires were performed in all the remaining patients alive at 3 months (25 patients; 1 lost to follow-up) who were unable to attend the hospital. Convalescent blood samples were obtained by their general practitioner in 3 of these patients. Functional outcome measures of disability (Barthel Index7 ) and handicap (Rankin Scale8 ), and the dependence/independence score of Lindley et al,9 were all assessed at 3 months. The Barthel and Rankin measures were modified to include all-cause mortality, death equaling −1 and 6, respectively. All assessments were performed by R.J.B. alone (who was blinded to the admission NSE and HSC values) to eliminate interobserver differences,
Sixty-one healthy age-, sex-, and race-matched control subjects were recruited from the population of a local general practice and were free from any previous history of neurological disease of any kind.
Blood samples were obtained within 48 hours of stroke onset from all 127 patients: serial samples were taken at 24, 48, 72, and 96 hours in 12 patients. Blood samples were centrifuged within 4 hours of venipuncture at 1500g for 10 minutes and serum stored at −20°C. Hemolyzed samples were not analyzed for NSE because of the presence of this enzyme within erythrocytes and platelets.10 All biochemical analyses were performed blinded to the patient's status.
NSE was measured spectrophotometrically by a two-site, solid-phase enzyme immunoassay (CIS Biointernational). The lower limit of detection of the assay is 0.5 ng/mL; the intra-assay coefficient of variation (CV) is 6% at 13.4 ng/mL and 2.8% at 24.3 ng/mL, and the interassay CV is 5.8% at 23.1 ng/mL.
Serum carnosinase activity was assayed by the method of Wassif et al.11 This is based on the fluorometric assay of histidine released by the hydrolysis of carnosine. The results are expressed as nanomoles of l-histidine formed per minute per milliliter serum (nmol/mL per minute). The intra-assay and interassay CVs are 3.1% and 4.7%, respectively, at 110 nmol/mL per minute.
Data are presented as median and SQR, since HSC, NSE, and the NSE/HSC ratio are not normally distributed (Kolmogorov-Smirnov goodness of fit test, P<.05). Statistical differences between the groups were analyzed with the use of the Mann-Whitney U test and Kruskal-Wallis one-way ANOVA by ranks for unpaired comparisons, the Wilcoxon signed rank sum test and Friedman two-way ANOVA by ranks for paired comparisons, and the χ2 test or Fisher's exact test for frequency data.12 Spearman's rank correlation coefficient (rs) was used for assessing associations. A probability value of ≤.05 was considered significant.
One hundred twenty-seven patients had blood samples taken within 48 hours of stroke onset; 103 suffered an ischemic stroke, and 21 a primary intracerebral hemorrhage. Three patients had neither a CT scan nor postmortem examination performed and have been excluded from further analyses. NSE data are available on 104 patients (the remaining samples were discarded due to hemolysis) and HSC data on 123 patients; 103 patients have data for both NSE and HSC. Sixty-one control subjects were recruited (combined NSE and HSC data are available for 57).
The patients and control subjects were matched for age (Table 1⇓). Nonsignificant trends for differences in sex and race were apparent; however, there is no evidence in the literature that serum NSE concentrations or HSC activities are affected by sex or race.
Temporal changes in serum NSE, HSC, and their ratio were evaluated in 12 patients between 24 and 96 hours after stroke. No significant changes occurred in serum NSE concentration or HSC activity over this period, although a trend was seen for the highest NSE/HSC ratio at 48 hours (Table 2⇓).
Serum NSE concentrations were higher in both the infarct and hemorrhage groups compared with the control group (Table 3⇓), with the highest levels in the hemorrhage group. HSC activity was significantly lower in both stroke groups, with the lowest activity observed in the hemorrhage group. The NSE/HSC ratio was significantly higher in both infarct and hemorrhage patients (Fig 1⇓). A nonsignificant difference existed between the infarct and hemorrhage groups for this ratio (P=.082), with the highest ratio corresponding to the hemorrhage group.
Convalescent samples were obtained at 3 months from surviving stroke patients and compared with admission samples; NSE levels were marginally lower at 3 months as compared with admission: median (SQR) (range), 7.0 (2.8) (0.5 to 21.0) ng/mL versus 7.5 (4.0) (0.5 to 46.0) ng/mL, two-tailed P=.22, n=69. HSC was slightly higher at 3 months: 146 (71) (16 to 453) nmol/mL per minute versus 135 (44) (16 to 422) nmol/mL per minute, two-tailed P=.80, n=81. The NSE/HSC ratio was also slightly lower at 3 months: 0.041 (0.023) (0.003 to 0.568) versus 0.052 (0.03) (0.001 to 0.613), two-tailed P=.21, n=66.
A surrogate measure of stroke lesion size was made through the use of the classification used in the Oxfordshire Community Stroke Project13 ; this scheme divides patients into larger (total anterior circulation infarcts, TACI) and smaller (partial anterior circulation infarcts, PACI) cortical infarcts, deep white matter small infarcts (lacunar infarcts, LACI), and posterior circulation infarcts (POCI). No significant differences in HSC activity were observed between the groups in the infarct patients. A trend between infarct size and NSE level or NSE/HSC ratio was noted with the largest infarcts (TACI) having the highest levels, and small vessel (lacunar) infarcts having lower levels (Table 4⇓).
Outcome data (disability, handicap, and death) were available for 123 patients at 3 months after stroke (1 patient lost to follow-up). Significant correlations were seen for NSE and outcome for all stroke patients (infarcts and hemorrhages) through the use of the modified Barthel Index (MBI) (rs=−.29, P=.002, n=104) and the modified Rankin Scale (MRS) (rs=.25, P=.009, n=104), although there was only a trend with the Lindley Score (LS) (rs=.16, P=.095, n=104). In each case, high NSE concentrations signaled an unfavorable outcome. Similarly, a significant correlation was observed between HSC and outcome: MBI (rs=.21, P=.02, n=122) and MRS (rs=−.20, P=.03, n=122), with low HSC activities corresponding to an unfavorable outcome. A nonsignificant trend was observed for the LS (rs=−.12, P=.18, n=122). However, the most dramatic relationship was noted between the NSE/HSC ratio and outcome: MBI (rs=−.34, P<.001, n=103), MRS (rs=.30, P=.002, n=103; Fig 2⇓) and LS (rs=.19, P=.057, n=103), with high NSE/HSC ratios predicting an unfavorable outcome. Similar correlations with outcome were assessed for two factors associated with outcome, namely age and consciousness level (Glasgow Coma scale [GCS]) on admission: Age: MBI (rs=−.33, P<.001, n=123), MRS (rs=.28, P=.002, n=123), and LS (rs=.23, P=.012, n=123); GCS: MBI (rs=.17, P=.068, n=123), MRS (rs=−.18, P=.044, n=123), and LS (rs=−.20, P=.028, n=123).
A comparison of patients who died or were discharged to an institution or supervised care (“bad outcome”) versus those who were discharged home (“good outcome”) found higher NSE levels (8.5 [3.4] versus 5.5 [2.8] ng/mL, n=29 and 75, two-tailed P=.06) and lower HSC activity (101  versus 133  nmol/mL per minute, n=35 and 88, two-tailed P=.029) in those who fared poorly. Once again, the NSE/HSC ratio provided the best discrimination, with high ratios predicting a poor outcome (0.069 [0.043] versus 0.038 [0.024], n=29 and 74, two-tailed P=.011). Similarly, comparisons were made for age and admission consciousness level. Older patients (78.3 [5.7] years versus 70.3 [7.3] years, n=35 and 89, two-tailed P=.0004) and patients with lower GCS scores (15  versus 15 , n=35 and 89, two-tailed P=.016) were more likely to suffer a bad clinical outcome. No difference in clinical outcome (good versus bad) was seen between patients in atrial fibrillation (AF) and those in sinus rhythm (Fisher's exact test, two-tailed P=.78, data for n=119). Similarly the presence of diabetes mellitus did not appear to alter clinical outcome (Fisher's exact test, two-tailed P=.77, data for n=120).
This study involving over 120 patients confirms the changes seen in blood NSE levels and HSC activity after stroke that have been reported previously in humans.3 4 5 We found that serum NSE and HSC levels were significantly different in the stroke groups compared with the control group. This was reflected even more strikingly for the NSE/HSC ratio. The highest NSE/HSC ratios were seen in patients with primary intracerebral hemorrhage. We did not see any major temporal changes in NSE or HSC levels or their ratio during the first 96 hours after stroke, and these results had not returned to normal by 3 months.
NSE is an isoenzyme of the glycolytic enzyme enolase containing two γ subunits. This form is found predominantly in the cytoplasm of neurons and neuroendocrine cells.14 Its release into the circulation after a stroke must occur due to disruption of the blood-brain barrier and also release of NSE into the CSF. It follows, therefore, that the highest levels might be expected in the largest lesions, as our results may suggest. This fits with ischemic models both in rats and humans, where CSF NSE levels and 72-hour postictus serum levels, respectively, correlate with infarct volumes.3 15
The NSE levels fell in the convalescent period but did not return to the levels seen in the control group by 3 months. This suggests that neuronal integrity is not yet complete in the absence of recurrent stroke events. Studies of stroke models in rats suggest that CSF levels normalize by 6 to 8 days after onset,15 16 but the response in humans is less clear. Cunningham and colleagues3 examined serum samples in patients up to 96 hours after stroke and did not comment on a fall by this stage. Similarly, Schaarschmidt and coworkers4 found that many patients still had serum levels greater than the control subjects at 10 days after stroke. Neither study examined patients beyond these times.
Two previous small studies (n=24 and n=43)3 4 examining clinical outcome and serum NSE after stroke have used the Glasgow Outcome Score,17 a measure of handicap similar to the Rankin Scale. In both studies, nonsignificant trends were seen, although high levels did reflect a poor outcome. However, these studies did not have sufficient power to examine the relationship between NSE level and clinical outcome.
HSC (EC 22.214.171.124) is a dipeptidase that hydrolyzes the three dipeptides carnosine (β-alanyl-l-histidine), anserine (β-alanyl-l-methylhistidine), and homocarnosine (γ−amino butyric acid [GABA]-histidine). HSC is a brain-derived enzyme synthesized within the brain and secreted through the CSF into the serum.18 The enzyme may have several functions in the central nervous system including the hydrolysis of homocarnosine producing the neurotransmitter GABA. This forms an alternative pathway to its production through the direct decarboxylation of glutamic acid. Both homocarnosine and HSC can be detected throughout the brain in close proximity by immunostaining.19 Second, carnosinase may be important in olfactory pathways as both the enzyme and the substrate carnosine are found in the olfactory areas.20
We found that HSC activities were significantly lower in stroke patients, and it is possible that death of carnosinase-producing cells reduces the secretion of this enzyme into the plasma. However, if this were the case, then the largest infarcts would be expected to have the lowest activities, and this was not seen in our study. An alternative explanation is that the regulation of HSC release is altered after neuronal insult. Within the stroke population studied, the lowest HSC activities correlated with the worst outcome. It is possible that these patients were unable to hydrolyze sufficient homocarnosine to produce enough of the inhibitory neurotransmitter GABA after a stroke. It is known that neurons produce excitatory neurotransmitters such as glutamate after an ischemic insult and that this can lead to further cell death. Thus, after stroke, limited GABA supplies secondary to low HSC activities may exaggerate glutamate-induced neuronal damage and death.
We assessed clinical outcome using the modified Barthel Index, modified Rankin Scale, Lindley Score, and discharge destination or death. While there were associations between NSE and HSC values and outcome (however measured), the strongest predictor of outcome was the NSE/HSC ratio. The observation that the NSE/HSC ratio significantly relates to clinical outcome measured by four different methods testifies to its robustness as a predictive measure of outcome. Although correlation coefficients of just over 0.3 are not initially striking, it is important to emphasize what is being correlated. We have found a significant association between a biochemical marker measured in the acute phase of stroke and functional outcome assessed 3 months later. This is the first study to report such a finding. Furthermore, the NSE/HSC ratio had comparable correlations for disability and handicap as patient age and similar or better correlations than conscious level on admission for all four outcome assessments used. Age and consciousness level are commonly considered as important predictors of clinical outcome. Similarly, patients with AF or hyperglycemia (assessed as presence of preexisting or newly diagnosed diabetes mellitus in our study) are believed to fare badly, although the presence of either in our study was not associated with a bad outcome.
In conclusion, the NSE/HSC ratio may be a useful prognostic marker of future clinical outcome when measured in patients with acute stroke. The ratio may also be of use in helping to differentiate ischemic from hemorrhagic stroke in the absence of neuroimaging information, perhaps in conjunction with clinical findings, although this requires further study. Finally, the NSE/HSC ratio could be useful in acute therapy trials, in which it will be interesting to see if potential stroke drug treatments can modulate it.
Selected Abbreviations and Acronyms
|CSF||=||cerebral spinal fluid|
|HSC||=||human serum carnosinase|
Dr Butterworth is a Stroke Association Research Fellow. Dr Bath is Wolfson Senior Lecturer in Stroke Medicine. We wish to thank Dr D. Stephens and colleagues at the Burbage Road General Practice Surgery, London, for providing the control cohort, and The Wellcome Foundation (part of Glaxo-Wellcome plc) for supporting the study. Some of the data were presented at the 4th European Stroke Conference, Bordeaux, France, June 1995.21
- Received April 22, 1996.
- Revision received July 19, 1996.
- Accepted July 26, 1996.
- Copyright © 1996 by American Heart Association
Hay E, Royds JA, Davies-Jones GAB, Lewtas NA, Timberley WR, Taylor CB. Cerebrospinal fluid enolase in stroke. J Neurol Neurosurg Psychiatry. 1984;47:724-729.
Persson L, Hårdemark H-G, Gustafsson J, Rundström G, Mendel-Hartvig I, Esscher T, Påhlman S. S-100 and neurone specific enolase in cerebrospinal fluid and serum: markers of cell damage in human central nervous system. Stroke. 1987;18:911-918.
Schaarschmidt H, Prange HW, Reiber H. Neurone specific enolase concentrations in blood as a prognostic parameter in cerebrovascular diseases. Stroke. 1994;25:558-565.
Wassif WS, Preedy VR, Summers P, Duane P, Leigh N, Peters TJ. The relationship between muscle fibre atrophy factor, plasma carnosinase activities and muscle RNA and protein composition in chronic alcoholic myopathy. Alcohol Alcohol. 1993;28:325-331.
Siegal S, Castellan NJ. Non-parametric Statistics for the Behavioural Sciences. 2nd ed. London, England: McGraw-Hill Book Co; 1988.
Butterworth RJ, Wassif WS, Gerges A, Sherwood RA, Peters TJ, Bath PMW. Serum carnosinase activity is reduced in acute stroke. Cerebrovasc Dis. 1995;5:246. Abstract.