Chemistry: molecular biology and microbiology – Measuring or testing process involving enzymes or... – Involving antigen-antibody binding – specific binding protein...
Reexamination Certificate
1998-12-16
2001-08-21
Chin, Christopher L. (Department: 1641)
Chemistry: molecular biology and microbiology
Measuring or testing process involving enzymes or...
Involving antigen-antibody binding, specific binding protein...
C435S007900, C435S007920, C435S967000, C436S513000, C436S811000, C436S815000
Reexamination Certificate
active
06277584
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates an improved calibration method to provide increased accuracy for in vitro chemical diagnostic tests where analyte concentrations of clinical relevance result in low assay signal values. In particular, the invention provides for improved accuracy in the region of a calibration curve corresponding to low measured signal values without utilization of a mathematical weighting routine.
2. Description of the Related Art
Generally, biochemical analyzers employ a combination of analyte specific chemical reagents and reaction monitoring means to assay or determine the presence or concentration of a specific substance or analyte within a liquid sample suspected of containing that particular analyte. Such analyzers are well known and almost universally employ some sort of a calibration curve that relates analyte concentration within a sample having a known analyte concentration against the signal generated by the reaction monitoring means in response to the presence of the analyte. Such samples are frequently called “calibrators” or “calibration solutions” or “standard solutions”. For greatest accuracy, calibration curves are established at regular intervals, to compensate for reagent particulars, on individual analyzers, to compensate for equipment performance. The range of analyte concentrations used in establishing a full calibration curve is typically chosen to extend below and beyond the range of analyte concentrations expected to be found within biological samples like blood, serum, plasma, urine and the like.
It is regular practice within the biochemical analytical industry to establish a full calibration curve for a chemical analyzer by using multiple calibration solutions or calibrators which have been carefully prepared with known, predetermined concentrations of analyte. These calibration or standard solutions are assayed one or more times and the mean resulting reaction signals are plotted versus their respective known analyte concentrations. A continuous calibration curve is then produced using any of several mathematical techniques chosen to produce an accurate replication of the relationship between a reaction signal and the analyte concentration. The shape of the calibration curve is affected by a complex interaction between reagents, analyte and the analyzer's electromechanical design. Thus, even if the theoretical analyte-reagent reaction is known, it is generally necessary to employ mathematical techniques to obtain an acceptable calibration curve.
One of the most widely used techniques to establish a calibration curve is the use of regression analysis, either linear or nonlinear depending on the curve shape. This is particularly true with competitive or sandwich immunoassays which frequently use nonlinear regression analysis to fit calibration data with a general nonlinear model known as the logit or Rodbard function.
A well known drawback in regression analysis, however, is that it can give biased results in analytical systems where the measurement signal variability changes with analyte concentration, such as typically found with immunoassays. This is a particular problem for systems which have greater inherent variability at the high end of the calibration curve while the most clinically important region is at the low end of the calibration curve (e.g., TSH, HCG, CKMB, etc.). In these cases standard regression analysis will preferentially fit the high end calibrators at the expense of the low end and potentially introduce significant bias in the low end region.
Since the accuracy with which the calibration curve is established directly affects the accuracy of an assay made on clinical samples, a solution to the problem is to use weighted regression analysis. In this case, the fit is artificially forced or “weighted” to focus on the portion of the calibration curve that corresponds to the range of greatest clinical significance for the analyte of interest. This is particularly critical for high sensitivity immunoassays, which are capable of detecting very low levels of analyte. In these instances, a high weighting factor is applied to the calibrator solutions that have the lowest analyte concentrations, which are frequently the most precise and clinically relevant, so that the resultant full calibration curve is forced to most accurately model the lower range of reaction signals, albeit at the expense of potentially allowing some bias at higher analyte concentrations.
A popular practice is to incorporate a weighting scheme into the software that controls the operation of the analyzer so that technicians performing the calibration and operating the analyzer are not required to also perform the mathematical calculations. However, this convenience does not exist within all analyzers, in particular within older analyzers designed before such high sensitivity immunoassays were commercially available. For this reason, a simple method for calibrating analyzers without resorting to complex weighting techniques is needed to achieve increased calibration curve accuracy with high sensitivity immunoassays.
U.S. Pat. No. 3,960,497 discloses the basic concepts of calibrating and verifying the calibration of a chemical analyzer using standard solutions having known values of the particular characteristic being measured.
U.S. Pat. No. 4,043,756 discloses a method to provide in an automatic chemical testing apparatus means for providing calibration values to processing circuitry, and selecting one of a plurality of calibration signal values for use as the calibration value based on comparison with suitable values resulting from use of the signal as a calibration value.
U.S. Pat. No. 4,169,125 is a method for calculating calibration curve fit values for a polynomial regression curve equation followed by a Newton-Raphson inversion on the equation. This is illustrative of the sophistication often employed in achieving calibration of a chemical analysis system.
U.S. Pat. No. 5,083,283 discloses a method for obtaining a calibration curve using a least-squares approach in which a portion of the measured data are weighted near a limit value used in deciding the components of a living organism.
U.S. Pat. No. 5,281,540 discloses a conventional least-squares regression technique to obtain the parameters and type of curve fit for different lots of analytical reagents.
U.S. Pat. No. 5,348,889 provides a calibration curve plotting the interrelation between the calibration values of a small number of different known standard solutions and their measured reaction signals, the calibration curve being modified by extrapolating an imaginal point. The calibration value of the imaginal point includes both the upper limit of the measurement range and a zero value point.
U.S. Pat. No. 5,554,539 discloses a method for recalibrating a calibration curve by determining ratios between actual and expected reaction signals at lower and higher concentrations. The ratios are then used in combination with pseudo-signals correspondingly lower and higher than the calibration range to obtain an extended calibration curve range.
U.S. Pat. No. 5,795,791 discloses a calibration curve obtained by splitting an original logistic calibration curve into three parts, a low concentration region represented by a multi-degree function, an intermediate concentration region represented by an exponential degree function, and a high concentration region represented by another multi-degree function to produce a three-part calibration curve that has identical slopes at the boundary between them.
Accordingly, from a study of the different approaches taken in the prior art to the problems presented by the necessity for providing calibration curves in high sensitivity immunoassays, there remains a need for a method to produce a calibration curve having increased accuracy at low analyte concentrations without introducing complex weighting calculations and without unduly adding to the number of calibration standard solutions in the low end region of i
Chu Victor Pichai
Pierson-Perry James Floyd
Sanders Connie Mary
Wei Tie Quan
Chin Christopher L.
Dade Behring Inc.
Jordan Leland K.
LandOfFree
Method for calibrating a chemical analyzer with improved... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for calibrating a chemical analyzer with improved..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for calibrating a chemical analyzer with improved... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2516518