Analysis and validation of tissue biomarkers for renal cell carcinoma using automated high-throughput evaluation of protein expression

Abstract

The objective of this study was to compare the predictive ability of potential tissue biomarkers to known prognostic factors that predict renal cell carcinoma (RCC) recurrence using an automated system of immunohistochemical analysis. After institutional review board approval, a tissue microarray was constructed using tissue from patients who had partial or radical nephrectomy for RCC. Patients with metastatic disease were excluded. Immunohistochemical staining of the tissue microarray for Ki-67, C-reactive protein, carbonic anhydrase 9, and hypoxia-inducible factors 1α and 2α was analyzed using automated image analysis. Univariable and multivariable analyses were performed to evaluate the association of putative biomarkers and known prognostic factors. Of 216 patients who met the entrance criteria, 34 (16%) patients developed metastatic recurrence within a median follow-up interval of 60.9 (interquartile range, 13.9-87.1) months. RCC morphotypes analyzed in this study include clear cell (n = 156), papillary (n = 38), chromophobe (n = 16), and collecting duct/unclassified (n = 6). Univariate analysis identified that only increased Ki-67 was predictive of RCC recurrence among the proteins evaluated, in addition to other known clinicopathological prognostic factors. After multivariate analysis, Ki-67 was identified as an independently predictive risk factor for RCC recurrence (hazard ratio [HR], 3.73 [confidence interval {CI}, 1.60-8.68]). Other independent predictors of RCC recurrence included tumor diameter (HR, 1.20 [CI, 1.02-1.41]) and perinephric fat invasion (HR, 4.49 [CI, 1.11-18.20]). We conclude that Ki-67 positivity is independently predictive of RCC recurrence after surgery in nonmetastatic patients. Automated analysis of tissue protein expression can facilitate a more objective and expedient investigation of tissue biomarkers for RCC.

Keywords: Biomarkers; Immunohistochemistry; Multispectral imaging; Recurrence; Renal cell carcinoma.

Fig. 1

The spectral curve is defined and enables separation and quantification of different chromogens.

Fig. 2

Separation from the composite image (A) of 3,3′-diaminobenzidine (B), Warp Red (C), and hematoxylin (D) chromogens using Nuance software. Composite image was acquired using the ×20 objective lens on the Nuance microscope.

Fig. 3

Manual training of software from the composite image (A) by tissue segmentation (B) and cell segmentation (C).

Fig. 4

Benign (A–E) and malignant (F–J) tissue samples from patients with clear cell RCC demonstrating the expression of CRP, Ki-67, HIF1α and HIF2α, and CAIX. Images were acquired using the ×20 objective lens on the Nuance microscope.

Fig. 5

Kaplan-Meier estimate of disease recurrence based on ISUP grade (A), perinephric fat invasion (B; P = .01), and Ki-67 expression (C; P = .01).