Modeling Serum Biomarkers S100B and NSE as Predictors of Outcomes Following Out-of-Hospital Cardiac Arrest

Editor's Note: Based on Einav S, Kaufman N, Algur N, Kark JD. Modeling serum biomarkers S100 beta and neuron-specific enolase as predictors of outcome after out-of-hospital cardiac arrest: an aid to clinical decision making. J Am Coll Cardiol 2012;60:304-11.

Background

Hypoxic brain injury remains the leading cause of mortality and severe neurological impairment after cardiopulmonary resuscitation (CPR) and return of spontaneous circulation (ROSC),1,2 yet accurate prediction of these outcomes remains elusive. Serum levels of protein S100B and Neuron Specific Enolase (NSE) rise in clinical situations reflecting the three classical models of brain injury; hypoxia,3 trauma4-6 and ischemia.7-11 Commercial kits for simple measurement of blood levels of both biomarkers have become available. Nevertheless, neither biomarker is being used in clinical practice because of concerns regarding their discriminant ability.

Introduction

This study assessed the added value of biomarkers to outcome prediction through joint modeling with clinical data. The hypothesis was that clinical and laboratory characteristics, the latter determined at clinically convenient times in a non-selective population, can be used to develop models enabling clinicians to select the likelihood of misdiagnosis acceptable in their clinical settings for classifying a potential survivor as a death.

Methods

A prospective 3-year study of victims of out-of-hospital cardiac arrest (OHCA) within the Jerusalem district, where CPR is performed in accordance with American Heart Association guidelines. Following approval by the local Institutional Review Board, data were recorded on patients aged ≥18 years with ROSC at a single hospital. Blood was sampled for biomarker levels at hospital arrival and on the mornings of days 1 and 3. Treating staff and investigators were blinded to biomarker levels that were determined. Patients were followed to either hospital discharge or death, whichever occurred first. Cerebral Performance Category (CPC) was determined by death or within 24 hours before discharge. The primary outcome measure was poor versus good patient outcome at discharge (CPC 3-5 versus CPC 1-2).

Results

Of the 250 patients were screened, 55 were excluded due to their agonal state, 195 were eligible for study inclusion and 184 contributed blood samples. Participants were mostly males (66%) with an average age of 73±16 (range 19-111) years. Their presenting rhythm was most often asystole (62%), VT/VF (24%) or PEA (12%). There were 43 survivors to hospital discharge (17%).

Biomarker levels: Median S100B concentrations were consistently higher in patients with a poor outcome (7.7, 1.8 and 1.4 mcg/L at admission, day 1 and day 3, respectively) compared to patients with a good outcome (2.3, 0.3 and 0.2 mcg/L, respectively (p<0.0083 for each between group comparison, Bonferroni correction). Median NSE levels did not differ significantly between patients with poor and good outcomes at arrival, but differences emerged subsequently (35 vs. 22 mcg/L for day 1, and 61 vs. 16 mcg/L for day 3)(p<0.0083 for days 1 and 3).

Model for predicting a poor outcome upon arrival (n=158): Univariate regression modeling yielded 4 significant variables: age, presenting rhythm and the two biomarker levels. Multivariable modeling with age (youngest tertile or not) and a presenting rhythm of VT/VF (yes or no) as explanatory variables resulted in an area under the curve (AUC) of 0.880 (95% CI 0.806, 0.954). Adding the biomarker levels to the model led to retention of only three significant variables: age, VT/VF and the level of S100 at admission (AUC 0.932 (95% CI 0.887-0.976)). Comparison of the two ROC curves yielded a p-value of 0.027.

Model for predicting a poor outcome on day 3 after admission (n=74): The only variables contributing independently to the day-3 model were a presenting rhythm of VT/VF and the day 3 S100B concentration (AUC of 0.931 (95% CI 0.873-0.989)). Using the same method as above, it was determined that if the presenting rhythm was not VT/VF, S100B > 0.195 mcg/L predicts a poor outcome and if the presenting rhythm was VT/VF, serum S100B > 0.566 mcg/L predicts a poor outcome (sensitivity 86%, specificity 92%).

Conclusion

Joint modeling of clinical and biomarker data improves outcome prediction after out-of-hospital cardiac arrest. Although biomarker data independently contributed an ostensibly modest 5.2% to the arrival model AUC, they substantially reduced the probability of misclassifying a patient as non-survivor compared with using solely on clinical criteria. When a presenting rhythm of VT/VF and age were added as examples of key clinical covariates, cutoff values for S100 could be tabulated with their specificity/sensitivity characteristics and predictive values.

Commentary/Perspective

Most studies of biomarkers in post-ROSC patients seek simple cutoff points, disregarding patient demographic and clinical characteristics yet few clinicians would agree to cease ongoing resuscitation efforts based on the results of blood tests alone. Risk-stratification based upon combined clinical-laboratory data should thus be the preferred clinical approach for interpretation of biomarkers of brain damage. The model presented sought to identify those patients with a poor outcome, potentially enabling improved use of costly resources.

While researchers continue to seek the infallible brain biomarker, policy makers should determine the level of misclassification acceptable to their society within the clinical setting of CPR.

References


  1. Wijdicks EF, Hijdra A, Young GB, Bassetti CL, Wiebe S. Practice parameter: prediction of outcome in comatose survivors after cardiopulmonary resuscitation (an evidence-based review): report of the Quality Standards Subcommittee of the AA&N. Neurology 2006; 67:203-10.
  2. Geocadin RG, Buitrago MM, Torbey MT, Chandra-Strobos N, Williams MA, Kaplan PW. Neurologic prognosis and withdrawal of life support after resuscitation from cardiac arrest. Neurology 2006; 67:105-108.
  3. Nagdyman N, Komen W, Ko HK, Muller C, Obladen M. Early biochemical indicators of hypoxic-ischemic encephalopathy after birth asphyxia. Pediatr Res 2001; 49:502-6.
  4. Pelinka LE, Toegel E, Mauritz W, Redl H. Serum S100B: a marker of brain damage in traumatic brain injury with and without multiple trauma. Shock 2003; 19:195-200.
  5. Berger RP, Pierce MC, Wisniewski SR, Adelson PD, Kochanek PM. Serum S100B concentrations are increased after closed head injury in children: a preliminary study. J Neurotrauma 2002; 19:1405-9.
  6. Dimopoulou I, Korfias S, Dafni U, et al. Protein S-100b serum levels in trauma-induced brain death. Neurology 2003; 60:947-51.
  7. Martens P, Raabe A, Johnsson P. Serum S-100 and neuron-specific enolase for prediction of regaining consciousness after global cerebral ischemia. Stroke 1998; 29:2363-6.
  8. Bottiger BW, Mobes S, Glatzer R, et al. Astroglial protein S-100 is an early and sensitive marker of hypoxic brain damage and outcome after cardiac arrest in humans. Circulation 2001; 103:2694-8.
  9. Hachimi-Idrissi S, Van der Auwera M, Schiettecatte J, Ebinger G, Michotte Y, Huyghens L. S-100 protein as early predictor of regaining consciousness after out of hospital cardiac arrest. Resuscitation 2002; 53:251-7.
  10. Rosen H, Rosengren L, Herlitz J, Blomstrand C. Increased serum levels of the S-100 protein are associated with hypoxic brain damage after cardiac arrest. Stroke 1998; 29:473-7.
  11. Rosen H, Sunnerhagen KS, Herlitz J, Blomstrand C, Rosengren L. Serum levels of the brain-derived proteins S-100 and NSE predict long-term outcome after cardiac arrest. Resuscitation 2001; 49:183-91.

Keywords: Biological Markers, Diagnostic Errors


< Back to Listings