Directional Association Measurement in Contingency Tables: Genomic Case.

Analysis of large data sets is currently a major challenge. Strong efforts are being undertaken to tackle this problem by developing new methods or modifying existing ones. The Z association method is a new method for describing directional association in contingency tables. It allows to arbitrarily group categories for each of the two variables, for which the contingency table is analyzed. The Z coefficient was calculated on a sample data set with gene mutations in different cancer types. Results showed some association with both gene mutations and annotation groups. Detailed results obtained for particular cancer types versus particular genes and annotation groups were in line with well-known facts in cancer genomics. The "MEUSassociation" R library allows to analyze the directional association between two categorical variables, and the mutual relationship is summarized in a contingency table, by means of the Z association coefficient. The method implemented in the library allows to compute the standard Z coefficient and to apply it in a case, where all possible singular coefficients Z(A:B) are computed at the same time, giving information of association between particular rows and columns. Investigating the ranked list of the highest singular coefficients allows to reduce the complexity of a large-scale data set. Both the Z coefficient and its R implementation are important tools in categorical data analysis.

Reproducibility of grayscale and radiofrequency IVUS data acquisition in stented coronary arteries.

OBJECTIVES: Variability in data acquisition from intervened coronary arteries could represent a source of error that has implications for the design of serial stent studies. We assessed inter-pullback reproducibility of volumetric grayscale and radiofrequency intravascular ultrasound (IVUS) data in stented coronary arteries. DESIGN: Fifteen patients with coronary artery lesions treated with stent implantation were included and examined with two separate pullbacks using the Eagle Eye Gold-phased array 20 MHz IVUS catheter (Volcano). The arteries were divided into five segments, giving a total of 150 sub-segments for analyses. Matching of frames was performed using landmarks that were clearly visible in coronary angiography and intravascular pullbacks. Data were analyzed off-line at an independent Corelab. RESULTS: The inter-pullback reproducibility of geometrical data was very good for non-stented segments with relative differences less than 5% between pullbacks for lumen-, vessel-, and plaque volumes. For stented segments reproducibility was poorer with relative differences between pullbacks in the range of 5-10%. The inter-pullback reproducibility of compositional data demonstrated large standard deviations of relative differences, indicating a weaker agreement. CONCLUSIONS: Agreements between pullbacks were weaker in stented than those in non-stented segments. Based on our data, future longitudinal IVUS studies in intervened vessels should account for a variability of 5-10% attributed to the acquisition of images.

Borderline trend towards long-term mortality benefit from drug eluting stents implantation in ST-elevation myocardial infarction patients in Poland-data from NRDES registry.

OBJECTIVES: To analyze long-term outcome of ST-Elevation Myocardial Infarction (STEMI) and non-STEMI (NSTEMI) patients treated with Drug Eluting Stents (DES) stents with regard to mortality and stent thrombosis rates based on the national Polish PCI registry database. BACKGROUND: Only a few studies suggested potential trend towards lower mortality in STEMI patients treated with DES. Current European Society of Cardiology STEMI guidelines recommend DES use only with one restriction to suspected patient poor compliance or contraindication to dual antiplatelet therapy. METHODS: Thirteen high-volume interventional cardiology centers in Poland enrolled patients in National Registry of Drug Eluting Stents (NRDES) Registry from October 2010 till October 2011. RESULTS: There were 2686 patients enrolled in NRDES Registry. Eighty five patients (3%) had both DES and BMS implanted at index PCI procedure and were excluded from further analysis. A subpopulation of STEMI (1709; 66%) and NSTEMI (892; 34%) patients was selected for BMS vs DES comparison. A significant difference in favor of DES group for 1-year mortality was found in STEMI subgroup (P < 0.0001-unadjusted and P = 0.0497 after propensity score adjustment). No such differences were noticed for NSTEMI subgroup or stent thrombosis comparisons. CONCLUSIONS: A strong selection bias for DES stents was observed with regard to demographic and angiographic characteristics in both STEMI and NSTEMI. DES implantation was associated with similar ischemic outcome to BMS at 1-year follow-up. STEMI patients with DES presented a trend towards lower long-term mortality at 1 year in comparison to BMS.