Evaluation of curve-fitting methods for the simultaneous determination of isoenzymes from inhibition kinetic responses
Abstract
This thesis describes the evaluation of nonlinear regression methods for simultaneous determinations of isoenzymes from combined, inhibition kinetic data. Simulated data are used to determine best fitting methods and optimal values for enzyme activities, rate constants and ratios of rate constants. Then the method is evaluated for the quantitation of the liver- and bone-type isoenzymes of alkaline phosphatase in human serum. Because a first-order model does not fit the response for bone-type alkaline phosphatase well, an alternative computational method is developed. Absorbance- and fluorescence-based methods are developed and characterized using the alternative method and the feasibility of determining a third component, either intestinal- or placental-type alkaline phosphatase is evaluated. The results for simulated data show that values for rate constants, ratios of rate constants and fitting ranges which provide results with desired relability are restricted. Small ratios of rate constants and short fitting ranges produced the largest useful activity ranges and shortest measurement times but also gave the worst accuracy and most dependence on initial estimates of fitting parameters. Either the extended Kalman filter or Marquardt algorithm can be used to obtain fits of the data and there are no significant differences in parameters estimated by these methods. Inhibition responses for the alkaline phosphatase data are optimized with respect to the criteria determined from the evaluation of simulated data. For the absorbance-based method, best conditions are for inhibition by urea with 2-(ethylamino)ethanol as buffer. The ratio of rate constants is about 2, fitting range is 8 half-lives, and measurement times are 46 min. The range of activities which can be determined is 10-160 U/L (Bowers and McComb units), within-run standard deviations for total activity and isoenzyme fraction are 4 and 6%, respectively, and results correlated well with expected values. All characteristics of a fluorescence-based method are similar except that measurement times are shorter (8 min) and values for ranges of activities which can be measured and for imprecision are 1.5-fold smaller and 2-fold larger, respectively. No suitable method was found for the determination of three components, but placental and intestinal-type isoenzymes can be detected when present in sufficiently large concentrations.
Degree
Ph.D.
Advisors
Pardue, Purdue University.
Subject Area
Analytical chemistry
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.