UICC World Cancer Congress 2006

Objective:Theoretically early detection (ED) of malignant neoplasms allows increasing 5-year survival (5YS) of cancer patients (CP) by several times while combined/complex treatment - only by 5-30% with incomparable financial expenses. Therefore, we examined 29 blood factors of CP with different localization, patients with non-malignant pathology (PNMP) and old healthy donors (HD). Methods:In trial (1985-2005) input level of blood factors of consecutive radical operated and monitored 283 CP with stage IA (T1N0M0) (80 non-small cell lung CP after lobectomies, 123 gastric CP after gastrectomies, 80 endometrial CP after hysterectomies+bilateral salpingo-oophorectomies; tumor size=0.5-2 cm; 5YS=100%), 403 PNMP, operated with suspicion on lung, gastric or endometrial cancer and 120 HD was analyzed (n=806, age=52.2±12.4 years, male=371, female=435). Logistic regression, multi-factor clustering, discriminant analysis, structural equation modeling, Monte Carlo, bootstrap simulation and neural networks computing were used to determine any significant overall regularity. Results:Logistic regression analysis displayed that ED of CP from PNMP and HD significantly depended on: blood lymphocytes, ESS, color index, prothrombin index, bilirubin. Neural networks computing, genetic algorithm selection and bootstrap simulation revealed relationships between ED of CP from PNMP and HD and lymphocytes (rank=1), eosinophils (2), segmented neutrophils (3), leucocytes (4), stick nuclear neutrophils (5), color index (6), weight (7), thrombocytes (8), monocytes (9), protein (10), glucose (11), Hb (12), bilirubin (13), ESS (14), prothrombin index (15). Correct recognition of CP from PNMP and HD was 73.1% by logistic regression (odds ratio=6.14), 74.2% by discriminant analysis and 96.5% by neural networks computing (area under ROC curve=0.942; errors=0.186).