Analytic variability in immunohistochemistry biomarker studies

Abstract

Background: Despite the widespread use of immunohistochemistry (IHC), there are no standardization guidelines that control for antibody probe variability. Here we describe the effect of variable antibody reagents in the assessment of cancer-related biomarkers by IHC.

Methods: Estrogen receptor (ER), epidermal growth factor receptor (EGFR) 1, and human epidermal growth factor receptor 3 (HER3) were evaluated by quantitative immunofluorescence. Correlations between ER clones 1D5, SP1, F10, and ER60c, and EGFR monoclonal 31G7, 2-18C9, H11, and 15F8, and polyclonal 2232 antibodies were assessed in 642 breast cancer patients. HER3 was measured by RTJ1, RTJ2, SGP1, M7297, RB-9211, and C-17 antibodies in 42 lung cancer patients. Survival analysis was done with the use of multiple cutoff points to reveal any prognostic classification.

Results: All ER antibodies were tightly correlated (Pearson's r(2) = 0.94-0.96; P < 0.0001) and western blotting confirmed their specificity in MCF-7 and BT474 cells. All EGFR antibodies but 2232 yielded specific results in western blotting; however, only 31G7 and 2-18C9 were strongly associated (Pearson's r(2) = 0.61; P < 0.0001). HER3 staining was nonspecific and nonreproducible. High EGFR-expressing patients had a worse prognosis when EGFR was measured with H11 or 31G7 (log rank P = 0.015 and P = 0.06). There was no statistically significant correlation between survival and EGFR detected by 2-18C9, 15F8, or polyclonal 2232 antibodies.

Conclusions: Antibody validation is a critical analytic factor that regulates IHC readings in biomarker studies. Evaluation of IHC proficiency and quality control are key components toward IHC standardization.

Impact: This work highlights the importance of IHC standardization and could result in the improvement of clinically relevant IHC protocols.

Figure 1

Representative AQUA output for 31G7 after progressive proteinase K digestion for 5, 10, 15, and 45 min at room temperature. Analysis of pan-cytokeratin used as a control for tumor integrity revealed similar results.

Figure 2

Parametric correlations between ER 1D5, SP1, F10, and ER60c clones (A-F) mapping epitopes in the N terminus (1D5 and ER60c) and C terminus (SP1 and F10) of the 67kD ER protein (G).

Figure 3

Correlations between EGFR primary antibodies (A-K). EGFR 34G7, 2-18C9, and H11 clones target the N terminal of the wild-type receptor, polyclonal 2232 binds an epitope around tyrosine 1068, and 15F8 clone binds to the C terminus (I).

Figure 4

Correlations between different HER3 antibodies measured by AQUA in lung cancer specimens. No association was identified for HER3 measurements by clones RTJ1, RTJ2, SGP1, and M7297, and polyclonal C17 and RB_9211, except for C17 and SGP1 (Pearson’s r² = 0.5155). Specific cytoplasmic and membranous staining was observed only for C17. Inset, representative expression pattern.

Figure 5

Survival curves based on cohort division by the 1st tertile, and median EGFR AQUA score detected by H11 (A and B), 31G7 (C and D), 15F8 (E and F), and 2-18C9 (G and H). High-EGFR expressers have a worse prognosis compared with the low-expressing group when EGFR is detected by H11, and the 1st tertile is used as a cutoff point (log rank P = 0.015). The same trend is observed for 31G7 when the median (log rank P = 0.06) and 1st tertile (P = 0.12) are used for stratification of patients into high/low expressers.