Assessment standards: comparing histopathology, digital image analysis, and stereology for early detection of experimental cisplatin-induced kidney injury in rats.

Histopathology generally represents the reference standard for performance evaluation of nonclinical biomarkers used to inform regulatory decision making. This study uses drug-induced nephrotoxicity in rats to evaluate histopathology methods utilized in biomarker performance assessments. Male Sprague-Dawley rats received a single dose of cisplatin (0.5-5.0 mg/kg, intraperitoneally) to produce mild renal injury. Animals were euthanized 72 hr postdose and perfusion fixed. Kidneys were processed for histology and stereology procedures. Kidney injury molecule-1 (Kim-1) was measured in urine and in kidney tissue. Digital slide images were generated and analyzed by pathologists after collaborating on a training set of glass slides and digital images. Image analysis identified immunohistochemistry (IHC)-defined tubular injury. Stereology methods yielded estimations of proximal tubular cell number and volume. Statistical relationships among data sets were determined using correlation coefficients. Receiver operator characteristic (ROC) analyses determined the effect of method on biomarker assessment. Urinary Kim-1 was strongly correlated with digital image analysis and secondarily to histopathology evaluations. Stereology demonstrated weak or no correlation to pathology and urinary Kim-1. In ROC analyses, semiquantitative evaluations determined higher values for urinary Kim-1 performance than did IHC-based qualitative digital analyses. Semiquantitative evaluation as used in this study was most predictive of urinary Kim-1 values.