BACKGROUND AND OBJECTIVE: We have carried out a comparative prediction analysis in sporadic colorectal carcinoma of plasmatic lipids and currently tumor markers used in this neoplasia (carcinoembrionary antigen - CEA, CA 19.9 and sialic acid -SA). PATIENTS AND METHODS: Transversal hospital-based case-control study. Sample was composed by 53 sporadic colorectal neoplasia patients just before surgery and 40 non matched controls. A 90 per cent of cases were at Dukes A and B stages. A multivariable model was fitted with non-conditional logistic regression. Confidence intervals were calculated at 95 per cent of confidence. Model sensibility and specificity were performed at 50 per cent cutting point. We also explored possible interactions. RESULTS: All three tumor markers (CEA, CA 19.9 and SA) were elevated in cases (p<0.01). Multivariable model included: Total cholesterol (TC) (OR= 0.69; CI: 0.50-0.96) high density lipoprotein fraction (HDL) (OR = 0.30; CI: 0.11-0.83) very low density lipoprotein fraction (VLDL) (OR = 2.21; CI = 1.07-4.55) years of age (OR = 1.11; CI = 0.98-1.26) CA 19.9 (OR = 1.20; CI = 1.02-1.42) and alkaline phosphatase (OR = 1.09; CI = 1.01 - 1.19). No interactions were found out. Model sensibility reached 96.23% and a specificity of 92.50%. CONCLUSIONS: HDL showed a similar diagnostic strength than CA 19.9 in this sporadic colorectal carcinoma sample but with an inverse OR. This multivariable model is going to be validated.
BACKGROUND AND OBJECTIVE: We have carried out a comparative prediction analysis in sporadic colorectal carcinoma of plasmatic lipids and currently tumor markers used in this neoplasia (carcinoembrionary antigen - CEA, CA 19.9 and sialic acid -SA). PATIENTS AND METHODS: Transversal hospital-based case-control study. Sample was composed by 53 sporadic colorectal neoplasiapatients just before surgery and 40 non matched controls. A 90 per cent of cases were at Dukes A and B stages. A multivariable model was fitted with non-conditional logistic regression. Confidence intervals were calculated at 95 per cent of confidence. Model sensibility and specificity were performed at 50 per cent cutting point. We also explored possible interactions. RESULTS: All three tumor markers (CEA, CA 19.9 and SA) were elevated in cases (p<0.01). Multivariable model included: Total cholesterol (TC) (OR= 0.69; CI: 0.50-0.96) high density lipoprotein fraction (HDL) (OR = 0.30; CI: 0.11-0.83) very low density lipoprotein fraction (VLDL) (OR = 2.21; CI = 1.07-4.55) years of age (OR = 1.11; CI = 0.98-1.26) CA 19.9 (OR = 1.20; CI = 1.02-1.42) and alkaline phosphatase (OR = 1.09; CI = 1.01 - 1.19). No interactions were found out. Model sensibility reached 96.23% and a specificity of 92.50%. CONCLUSIONS: HDL showed a similar diagnostic strength than CA 19.9 in this sporadic colorectal carcinoma sample but with an inverse OR. This multivariable model is going to be validated.
Authors: J M Villadiego-Sánchez; M Ortega-Calvo; R Pino-Mejías; A Cayuela; P Iglesias-Bonilla; F García-de la Corte; J M Santos-Lozano; José Lapetra-Peralta Journal: Int J Med Sci Date: 2009-01-30 Impact factor: 3.738