Identification of Malignant Pleural Effusions (MPEs) by Tumour Markers: a National Multi-centre Trial

MPEs are a common and important cause of cancer-related mortality and morbidity. Prompt diagnosis using minimally invasive procedures is a key point in the evolution of disease since the overall median survival after diagnosis is only 4-9 months. Tumour markers analysis has been proposed as a less invasive alternative for categorizing malignant and non-malignant pleural effusions. This multi-centre study aimed at establishing diagnostic cut-offs for a panel of markers in pleural fluid and serum to identify patients with malignancy. Methods Pleural fluid specimens and serum samples from 162 patients (102 malignant, 60 non-malignant) consecutively admitted in three Italian university hospitals were analyzed for Ca15.3, Ca 19.9, Ca72, Ca 125, Cyfra 21.1, NSE, CEA, M2PK and ADA. The diagnosis of malignant or non-malignant effusion was based on cytology, pleural biopsy, thoracoscopy, video-assisted thoracic surgery (VATS). Statistical evaluation included Kolmogorov-Smirnov, Mann-Whitney, Chi-square and Fisher’s exact tests. Non parametric (Spearman) correlations were determined. ROC curve analysis was performed to determine analyte cut-offs, sensitivity, specificity and AUC values of each marker. A p-value <0,05 was considered statistically significant. Results Cytological negative samples were analyzed to ensure that they truly represented non-malignant effusions related to other diseases such as congestive heart failure (CHF), renal imbalance, pneumonia, tuberculosis or post-traumatic. Patients with a history of malignancy or subsequent diagnosis of malignancy with a cytological negative pleural sample were excluded from the study. All examined concentrations were significantly higher in malignant effusions compared to non-malignant effusions. The value of AUC of pleural samples was always higher than in serum for all malignant cases. The best AUC value in pleural samples (Fig.1) was detected for Ca72 (0,800), Cyfra 21.1 (0,776) and CEA (0,773) while in serum samples Ca72 (0,761), Ca15.3 (0,653) and CEA (0,648) were the best performers (Fig.2). Conclusions Tumour markers assay in pleural fluid complements cytology and other classifying tests. Our study shows a significant performance of pleural fluid vs serum samples when comparing AUC values. A few tumour markers over-expressed in pleural fluid of patients with known malignancy have been identified. Therefore the use of a panel including the best performing markers may prevent patients with suspected malignancy from undergoing invasive procedures such as thoracoscopy or VATS. While routine analysis of markers is not recommended in all pleural effusions, tailoring a specific marker assay in pleural fluid for a specific malignancy is highly advisable especially in patients with relevant comorbidity.