A-088 Establish Laboratory Limit of Blank and Limit of Detection for LH Assay on the Ortho VITROS 3600 System

Abstract

Abstract Background In our lab, Luteinizing hormone (LH) was tested on VITROS 3600 system with manufacture’s declared limit of detection (LoD) 0.216 mIU/mL, limit of Blank (LoB) 0.0367 mIU/mL. It’s determined with NCCLS guideline EP17, false positives (α) less than 5% and false negatives (β) less than 1%, based on more than 700 determinations on several systems. Chinese government launched industry standard WS/T 514-2017, defining how a clinical laboratory establish its detection capability with stated proportions α = β = 5% and some other differences from EP17. Here we set up the LoB, LoD of VITROS 3600 LH assay in our lab based on Chinese industry standard. Methods a) LoB was designed to run 4 blank samples, LoD with 4 low positive samples, respectively with two VITROS LH kit lots, on one VITROS 3600, over two runs per day, for 3–4 days using two different calibrations, therefore we got 120 results (outliers omitted) for LoB and LoD respectively. b) Negative (blank) samples were matrix fetal bovine serum (FBS), low positive samples were in house produced, VITROS 3600 System was well maintained, fully calibrated and quality controlled. c) Data analysis: LoB and LoD sample data are not Gaussian, according to WS/T 514-2017, the LoB and LoD were computed with nonparametric statistics. Results Perform steps described in method for each lot reagents. LoB data distribution were non-Gaussian (as when truncated at zero). therefore the LoB can be determined by ranking the data in order from lowest to highest concentration and using the following equations, LOB = Pct100-α, here α is 0.05, P is 0.95, 60 determinations, each reagent lot arrangement position = 0.5 + 60*0.95 = 57.5, result was not an integer, LOB should be re-calculated = X57 + 0.5(X57-X58) = 0.5(X57 + X58). The final result for LoB is 0.0105.with proportions α = β = 5%, each reagent lot LoD determinations more than 50, thereof, K-S test is used for normality analysis. All results of both reagent lots have P-value less than 0.05, which means that the samples were not been generated from a normal distribution. According to WS/T 514-2017, the LoD is estimated with non-parametric statistical derivation method. All results are greater than LoB 0.0105, 0% results are lower than LoB which is less than stated probability 5%. The median value for lot 1990 is 0.0898, and lot 2010 0.0899, so LoD is 0.0899. Conclusion The limit of detection (LoD) for VITROS LH assay in our lab is 0.0899 mIU/mL, the limit of Blank (LoB) is 0.0105 mIU/mL, determined with WS/T 514-2017, and with proportions of false positives(α) = false negatives (β) = 5%. In our lab the detection capability is surpass to what manufacture claimed in the IFU. It could be as following factors: singleton platform in our lab, with greater stated proportions, less determinations, and consistently assigned operator. So we can conclude that this system is fit for a more sensitive detection scenario, like pediatric patients.