Abstract
New targeted therapies for non-small cell lung cancer (NSCLC) have increased the demand for accurate diagnosis rendered by histology or cytology. We evaluated small specimens value in NSCLC diagnosis against surgical pathology.
Methods
Our database was searched for all resected NSCLCs between 2008-2011 with pre-resection cytology and/or biopsy. Small specimens' sensitivity was determined taking surgical pathology as gold-standard. Pre-resection diagnoses for type of procedure: bronchoscopic (FB) or transthoracic (TT); size; pathology; distance from thoracic wall were correlated with resection pathological diagnosis. SPSS for windows was used to analyze Wilson method estimated 95% confidence intervals, agreement between results by Cohen's Kappa; factors associated with sensitivity and agreement by uni and multivariate regression.
Results
131 (69%) patients had known preoperative diagnosis (171 FB, 85 TT);. Pre-resection pathologic subtypes: adenocarcinoma 105 (55%), squamous 48 (25%), adenosquamous 9 (5%); others 27 (15%). Tumor size: median (min-max) 27 mm (4-110). Distance from thoracic wall median (min-max) 7 mm (0-50).
Sensitivity: all 69%, TT 57%, FB 55%. Agreement: all k=0,54, TT k=0,69, FB k=0,69. Tumor size by additional mm, in TT ODD 5% (p=0,001), in BF ODD 3% (p=0, 004) was associated with diagnosis sensitivity in multivariated analysis. Size of tumor in TT was associated with agreement in multivariated analysis.
Conclusion
Results from our study confirm that tumour size is an influencial predictor for diagnosis accuracy. Distance from chest wall is a relevant additional issue to be considered. Overall small specimens may accurately be used to diagnose NSCLC subtypes.
- © 2014 ERS