Pro-gastrin-releasing peptide, neuron specific enolase and chromogranin A as serum markers of small cell lung cancer

Abstract

Background: Small cell lung cancer (SCLC) yields neuroendocrine properties. Pro-gastrin-releasing peptide (ProGRP), neuron specific enolase (NSE) and chromogranin A (CGA) are therefore putative serum markers in this disease. Aim: To assess any difference in the sensitivity–specificity relationship between these neuroendocrine markers regarding various control populations and disease extent. Method: A total of 146 patients were prospectively assessed clinically and biologically. Serum marker titrations were performed using commercial immunoradiometric assays (NSE, CGA) or ELISA (ProGRP). Areas under receiver operating characteristic curves (AUC-ROC) were calculated in order to assess the sensitivity–specificity relationship of each marker and to compare marker accuracy. Maximum Youden indices were used to determine marker thresholds able to produce the best overall diagnostic information. Results: Assessing the sensitivity in the SCLC population and the specificity in benign lung disease, ProGRP demonstrated the best sensitivity relationship in as much as its AUC-ROC was significantly greater than the ones calculated using NSE and CGA (respective values, 0.95, 0.89, 0.70; two-tailed Z-test <0.05). The ProGRP threshold value, which offered the best sensitivity–specificity relationship was 53 pg/ml corresponding to a 0.80 sensitivity and a 0.96 specificity. In addition, when specificity was assessed in NSCLC and again the sensitivity in the whole SCLC population, ProGRP continued to demonstrate a greater AUC-ROC in comparison with other markers. Using the 53 pg/ml threshold the specificity of this marker was excellent with no false positives in NSCLC. On the other hand, none of the markers were able to discriminate limited from extensive SCLC as suggested by the fact that AUC-ROC, constructed when sensitivity was defined as a positive test in extensive disease and specificity as a true negative test in limited disease, did not reach the upper left octant (AUC 0.65, 0.71 and 0.63 for ProGRP, NSE and CGA, respectively). Conclusion: ProGRP yields the best sensitivity–specificity feature in SCLC, a result deserving further studies designed to evaluate the clinical applicability of this marker.