| Objective:To investigate the roles of Pro-gastrin-releasing peptide(Pro GRP)in the diagnosis of small cell lung cancer(SCLC).Method: We retrospectively analyzed data from 11,206 patients with clinical suspicion of lung cancer from January 1,2015 to May 31,2018.Pro GRP and neuron-specific enolase(NSE)were detected in peripheral blood.The Receiver operator characteristic curve(ROC)of Pro GRP and NSE,the threshold value of Pro GRP and NSE is defined as the value corresponding to the most approximate index,and the sensitivity,specificity,positive predictive value,negative predictive value of the diagnosis of small cell lung cancer are calculated.Delong’s test was used to compare the difference of AUCs.Results:1.The cut-off values of Pro GRP and NSE in the diagnosis of SCLC: ROC showed that an optimum diagnostic cutoff for Pro GRP and NSE was 66 ng/L and 18 μg/L respectively.2.Comparision of diagnosis value of tumor markers: At the above cutoff value,the sensitivity(86.5% vs 78.8%)and specificity(96.5% vs 86.3%)of Pro GRP were higher than NSE for the diagnosis of SCLC.In addition,Pro GRP had a greater AUC value than did NSE in patients with SCLC(0.943 vs 0.894,95%CI: 0.033-0.065,P<0.001).Furthermore,both the positive predictive value(PPV)(74.0% vs 39.0%)and negative predictive value(NPV)(98.2% vs 96.9%)of Pro GRP were higher than NSE.With the assessment of AUC,sensitivity,specificity,PPV,NPV and likelihood ratios,Pro GRP had a greater differential diagnostic accuracy than NSE.3.Diagnosis value of tumor markers in different stages: The median level of Pro GRP and NSE in SCLC increased with the accompanying stages(P <0.001).The median and staging of Pro GRP in other patients with lung cancer Irrespective(P>0.05),the median NSE of patients with large cell neuroendocrine carcinoma(LCNEC)increased with the stage(P <0.05).Further analysis showed that diagnostic efficacy can be improved by using different cutoff values in different stages,but not stage Ⅰ and Ⅱ.The cut-off values of Pro GRP in the diagnosis of SCLC in stage Ⅰ-Ⅱ,III and IV were 56 ng/L,71 ng/L and 99 ng/L respectively.In addition,the sensitivity(96.6% vs 95.8% and 98.3% vs 94.8%)and concordance rate(c2 = 1526.9 and 988.7,both P < 0.001)of detecting SCLC was improved by using different cutoff values compared with the only criteria of Pro GRP being ≥ 66 ng/L in stage III and IV,but not stage Ⅰ-Ⅱ.Additionally,in stage III and IV,the concordance rates of Pro GRP ≥71 ng /L and Pro GRP ≥99 ng /L were also higher than Pro GRP ≥300 ng/L(both P <0.001),which was conventionally indicated for SCLC.Pro GRP had greater AUC,sensitivity,and specificity values than did NSE in SCLC patients with stage Ⅰ-Ⅱ,Ⅲ and Ⅳ.4.The distribution of tumor markers in patients with differently differentiated pulmonary neuroendocrine tumors,PNET): In all patients with lung tumors,the highest level of Pro GRP and NSE was in SCLC(median 675.3 ng/L and 33.4μg/L,respectively),followed by NSCLC(median 35.2 ng/L and 12.9μg/L,respectively)and Typical carcinoid/ Atypical carcinoid(TC/AC)(median 41.3 ng/L and 11.5μg/L,respectively)(P < 0.001).It was interesting to note that Pro GRP and NSE seem to be abnormally higher in poorly-differentiated neuroendocrine carcinoma(PD-NEC containing SCLC and LCNEC)than well-differentiated neuroendocrine tumor(WD-NET containing TC and AC)and NSCLC.5.Diagnostic value of tumor markers for pulmonary neuroendocrine carcinoma: The critical values for the diagnosis of NEC by Pro GRPp and NSE are 66 ng/L and 18 μg/L,respectively.The sensitivity of Pro GRP and NSE for NEC diagnosis were 86.6% and 76.3%,respectively;specificity was 96.7%,85.2%;AUC: 0.941 vs 0.883(95% CI: 0.033-0.065,P < 0.001);positive predictive value respectively: 76.6%,40.4%;negative predictive values were: 97.7%,96.4%.In summary,Pro GRP’s ability to diagnose NEC is superior to NSE.Conclusion: Pro GRP has significantly higher sensitivity and specificity than NSE in the diagnosis of SCLC.Furthermore,special thresholds for every stage may be more reasonable for the diagnosis of SCLC. |
