Receiver Operating Characteristics (ROC) Application to Immnuhistochemistry for Determining Optimal Antibody Concentration: A Mathematical and Novel Application for Quality Control
Teresa D Samulski, Timothy Baradet, Paul D Zhang, Priti Lal, Michael S Feldman, Karen S Gustafson, Emma E Furth. Hospital of the University of Pennsylvania, Philadelphia, PA; Fox Chase Cancer Center, Philadelphia, PA
Background: No standards exist by which working antibody concentrations are determined in clinical immunohistochemistry (IHC) laboratories. Such decisions are made by microscopic examination of the staining intensity of differing antibody concentrations using selected tissue sections, and choosing an “optimal” concentration (that which yields the stongest true staining and weakest false staining) “by eye." We utilized a mathematical approach derived from signal detection theory, receiver operator characteristics, coupled with computer image analysis to create a quantitative algorithm for IHC optimization.
Design: Immunohistochemical staining of formalin-fixed paraffin embedded normal kidney tissue using antibodies against Multi-cytokeratin AE1/AE3 (Novocastra; Wetzlar, Germany; #NCL-AE1/AE3) was performed at concentrations 1:1, 1:2, 1:4, 1:40, 1:400, 1:4,000, 1:40,000, and 1:400,000 using standard methods on a Leica Bond instrument using the Novocastra Bond Polymer Refine Detection System. For each antibody concentration, the signal intensity of the tubules (biologic positive) and interstitium (biologic negative) per cell was measured by InForm® data collection software. The area under the curve (AUC), as metric of test performance and standard deviation were calculated. Considering each dilution as an individual “test”, we compared the AUC among each “test” and chose the “test” with the highest AUC as the optimal antibody concentration.
Results: The performance of the AE1/AE3 antibody was high across most antibody dilutions (1:1, 1:2, 1:4, 1:40, 1:400, 1:4,000, 1:40,000, and 1:400,000) despite dropping signal intensity. For each respective antibody dilution the AUC was 0.97, 0.91, 0.94, 0.99, 0.97, 0.98, 0.5, and 0.5 (S.D. <.0001)and the average tubule signal was 0.28, 0.22, 0.22, 0.34, 0.19, 0.13, 0, and 0.
Conclusions: Our application of ROC analysis to IHC-based staining is a quantitative, rigorous, mathematical method to determine an optimal antibody concentration for reliable performance. When the performance characteristics (AUC) of a particular antibody are high across multiple dilutions, other factors, such as signal intensity and reagent costs, may be used to determine an optimal antibody concentration. Our methodology will enhance quality control and optimization of IHC staining and also enable comparison of antibody performance over time and among different laboratories and platforms.
Category: Quality Assurance
Monday, March 19, 2012 9:30 AM
Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 301, Monday Morning