Pathologists at Yale University School of Medicine tested that assumption with five commercial antibodies against one protein, Met, on tissue arrays holding 640 archived breast tumor samples, scored by automated quantitative analysis.
One monoclonal antibody, 3D4, stained cell nuclei with one lot and cell membranes with the next, on samples from the same tumors. The two lots correlated at R = 0.194. On a Western blot, both lots had looked alike.
Four of the five antibodies failed to reproduce at a quantitative level, even though most identified the right protein on a blot. Lots of the one reliable antibody lost specificity after four to five months of proper storage. The authors concluded that validation has to be repeated lot by lot.
Read the original study
Pozner-Moulis et al., Laboratory Investigation, 2007 · doi.org/10.1038/labinvest.3700515
Same catalog number, different answer.
Source
- Pozner-Moulis, S., Cregger, M., Camp, R. L., & Rimm, D. L. (2007). Antibody validation by quantitative analysis of protein expression using expression of Met in breast cancer as a model. Laboratory Investigation; a Journal of Technical Methods and Pathology, 87(3), 251–60. https://doi.org/10.1038/labinvest.3700515