Search

In context of Evidence Based Medicine concept, Good Clinical Practice rules specify that "data generated should be reliable and robust". Reliability and robustness are further translated in requirements concerning statistical and clinical significance of results. Paper presents main aspects connected with comparison of evolutions of endpoints as function of different parameters like time, administered dose, proportion of active components etc., leading to problems of comparison of curves, with direct application to comparison of platelets aggregation curves in presence of different concentrations of ionic calcium. Theoretical part presents comparison of curves in biopharmacy using f2 metric and area under curve metric, and comparison of survival curves in clinical studies. Platelet aggregation test was performed using Born turbidimetric light transmission method using Helena PACKS-4 Aggregometer. Blood samples were collected from patients in internal medicine ward of Colentina Clinical Hospital. Platelet rich plasma (PRP) was obtained by centrifugation at 200G. Washed platelets where extracted by centrifugation of PRP at 2700 G. The supernatant was replaced with sodium chloride 0.9%. Platelets aggregation was induced by adding different concentrations of calcium gluconate into cuvettes which contained washed platelets. After digitalization, curves were compared using similarity factor f2 and areas under curves. Paper puts in evidence that both type of comparison, after mathematical and statistical evaluation, have to define a clinical threshold for clinical significance. In case of f2, in dissolution studies the threshold is 10%, in case of bioequivalence based on area under curves threshold is 20%. Establishment of the threshold for significant clinical difference in comparison of aggregation curves is not only a problem of statistics. Graphical representation of data suggested significant differences between curves obtained with different concentrations of calcium ion. Application of both f2 method and log-rank test let to conclusion that differences were statistical significant. Representation of aria under curves as function of calcium concentration put in evidence an approximate linear dependence. In spite of apparently objective character of mathematical approach, the problem of comparison of aggregation curves remains practically unsolved since we do not know the threshold between clinical significant and non-significant results. [ABSTRACT FROM AUTHOR]