[en] The report consists of computer programs, operating instructions, and sample input-sample output for calculation of the maximum-likelihood linear regression for logit (Y) versus log dose, utilizing an iterative least-squares method with appropriate weighting (four weighting functions are available). The standard curve is graphed in coordinates of Y versus dose, Y versus log dose, and logit (Y) versus log dose. Potency estimates and their standard errors are obtained for unknowns and corrected for sample volume and recovery. Unknowns are plotted consecutively, in groups if desired. The coefficient of variation for any point on the dose response curve is displayed. The Scatchard plot of Bound/Free versus Bound is shown, and both linear and non-linear regression methods are used to estimate the affinity constant(s) and binding capacity or capacities. This program also calculates the 'parallel-line' potency estimates for unknowns analyzed at two or more different dose levels

[en] A model has been developed to describe the kinetics of '2-site immunoradiometric' assays (2-site IRMA). This model provides a basis for optimization of assay conditions. Further, the model can be used to evaluate the effects of 'secondary' reactions which degrade the performance of the 'ideal' assay system. The model permits Monte-Carlo studies to predict the effects of the dose-response curve of random errors in the concentration of reagents, the rate constants, or reaction time. Further, the model provides justification for the use of a four-parameter logistic model for curve fitting and dose interpolation for analysis of data from two-site IRMAs. (author)

[en] New computer programs have been developed for the description of the radioimmunoassay (RIA) dose-response curve in terms of the Mass Action Law involving multiple independent classes of sites. This permits dose interpolation for unknowns, using a rational, theoretically justifiable approach. This method makes it possible to employ an exact correction to compensate for variable concentrations of labeled ligand, e.g., due to carryover from ''tracer'' used to monitor preliminary extraction and chromatography steps. This approach permits an increase in the mass of labeled ligand which can be utilized for this purpose, and may permit a dramatic reduction in counting time and/or counting errors. The approach can be readily generalized to handle many other types of binding isotherms (e.g., involving cooperativity) expressed in a variety of coordinate systems. This approach has been used successfully and validated for RIAs for testosterone, dehydroepiandrosterone, and aldosterone

[en] A method is presented for calculation of the minimal detectable dose (MDD) or minimal detectable concentration (MDC) for a radioimmunoassay or related radioligand assay (e.g., immunoradiometric assays, ''sandwich'' assays, etc.). It is necessary to consider the uncertainty in the estimate of the ''true'' response for zero dose as well as the error in the mean response for replicates of an ''unknown'' sample. The MDC decreases as the degree of replication for the unknown increases, approaching a nonzero finite lower limit as the number of replicates for the unknown gets large. ''Pooling'' of information about the variability of response at zero dose results in improved utilization of data, with increased degrees of freedom and smaller MDC. A one-tailed, rather than two-tailed, Student's t statistic should be used. The considerations reviewed here should be applicable irrespective of the nature of the dose-response curve or the particular model used for curve fitting

[en] The first section of the report consists of listings of computer programs, operating instructions and sample input-sample output for the calculation of the four parameters of a logistic dose response curve, Y ⁺ (A - D)/(1 [(X/C)]]B) [ D utilizing a weighted iterative nonlinear regression. The dose response curve is plotted versus the logarithm of dose. Potency estimates and their weighted standard errors are obtained for the unknowns and corrected for sample volume and recovery factors. Unknowns may be processed in replicate and/or in blocks. The second section of this report consists of listings of computer programs, operating instructions and sample input-sample output for the calculation of quality control, statistics for assays using an analysis of variance to calculate both within assay and between assay variance

[en] The 'sandwich' or noncompetitive reagent-excess, 2-site immunoradiometric assay (2-site IRMA), ELISA, USERIA, and related techniques, have several advantages compared with the traditional or competitive radioimmunoassays. IRMAs can provide improved sensitivity and specificity. However, IRMAs present some practical problems with nonspecific binding, increased consumption of antibody, biphasic dose response curve, (high dose hook effect), and may require special techniques for dose response curve analysis. We anticipate considerable growth in the popularity and importance of 2-site IRMA. (author)

[en] The mathematical and statistical theory of radioimmunoassays (RIAs) has been used to develop a series of computer programs to optimize sensitivity or precision at any desired dose level for either equilibrium or nonequilibrium assays. These computer programs provide for the calculation of the equilibrium constants of association and binding capacities for antisera (parameters of Scatchard plots), the association and dissociation rate constants, and prediction of optimum concentration of labeled ligand and antibody and optimum incubation times for the assay. This paper presents an experimental evaluation of the use of these computer programs applied to RIAs for human chorionic gonadotropin (hCG) and estradiol. The experimental results are in reasonable semiquantitative agreement with the predictions of the computer simulations (usually within a factor of two) and thus partially validate the use of computer techniques to optimize RIAs that are reasonably well behaved, as in the case of the hCG and estradiol RIAs. Further, these programs can provide insights into the nature of the RIA system, e.g., the general nature of the sensitivity and precision surfaces. This facilitates empirical optimization of conditions

[en] Displacement studies of [³H]-[D-Ala²-MePhe⁴-Gly-ol⁵]-enkephalin ([³H]-DAGO) and [³H]-[D-Ala²-D-Leu⁵]-enkephalin ([³H]-DADL) by the corresponding unlabeled ligands show that there are at least three classes of sites which bind these enkephalin analogs with high affinity. Using computer modeling, the introduction of the third site significantly improved the goodness of fit in ten consecutive experiments. These sites appear to correspond to the μ, delta, and μ₁ sites, with mean dissociation constants of 11, 1.3 and 0.9 nM for DADL and 2.5, 300 and 0.3 nM for DAGO, respectively. 15 reference, 3 figures, 1 table

[en] The authors have developed a new, general approach to analysis of dose-response curves from bioassay, immunoassay (including radioimmunoassay, immunnoradiouretic assay, enzyme-linked immunosorbent assay), and other experimental procedures. It provides a test for parallelism, a similarity of shape, and a measure of relative potency for any set of two or more curves. The methods uses a constrained smoothing spline function to estimate the curve shape, together with a nonlinear least-squares fitting technique to estimate parameters for relative potency and slope. The use of constrained splines permits the analysis of nonlinear dose-response curves that cannot be described by a simple model or equation such as the symmetric four-parameter logistic. A microcomputer program is used for the analysis, providing relative potencies and their SE and evaluation of goodness of fit

[en] The paper focuses on the problem of estimation of within-assay variance, either directly or indirectly. If this deteriorates, it is likely that all other levels of precision will also deteriorate. Also, in view of the well-known observation that within-assay precision is two-to threefold better than between-assay precision, the experimentalist should always design his studies such that all relevant comparisons can be made within an assay - and, if possible, within a small region of a given assay. Likewise, in multi-centre studies, the experimental design should permit all important comparisons to be made within laboratories and, if possible, within assays. In addition to the measurements of within-assay precision, one must also estimate the component of between-assay variance. One should consider the local 'contrast' between today's result and the result on recent assays (local between-assay variance), and compare this with the previous cumulative between-assay variance. The same kind of analyses can be used to evaluate variability between laboratories, or between groups of laboratories employing different reagents or methods