Su1878 Risk Factors for the Development of Esophageal Precancerous Lesions: an Analysis of the Cases Registered for a Multicenter Prospective Cohort Study

Abstract

Introduction: Finding an accurate and convenient biomarker for cancer progression in Barrett's oesophagus (BE) is of high clinical importance. DNA ploidy abnormalities (DNAPA) are a reliable predictor of cancer risk in BE, but measurement is expensive and not routinely available. We have previously shown in a sample of surgical oesophageal resection specimens that replication licensing factors (RLFs), particularly polo-like kinase-1 (PLK1) may act as surrogate markers of DNA-PA. This study aimed to examine the potential of PLK1 in predicting DNA-PA in the metaplasia-dysplasia-OA sequence. Method: 36 paraffin embedded oesophageal tissue specimens were selected from patients with non-dysplastic Barrett's (NDBE, n=5), low grade dysplasia (LGD, n=5), high grade dysplasia (HGD, n=12), intramucosal cancer (IMC, n=5) and invasive OA (IOA, n=9). Sections from these blocks were mounted and immunostained with two PLK-1 antibodies from Millipore (PLK1-M) and Leica (PLK1-L) with the automated BOND-MAX system (Leica Microsystems) for consistency. Staining was reported by 2 independent expert pathologists blinded to patient status with the Allred scoring system, a composite of the percentage (0-no cells, 1= 66%) and intensity of cells staining (0=negative, 1=weak, 2=moderate, 3=strong). Adjacent sections were analysed with image cytometry to identify DNA ploidy abnormalities. Results: Aneuploidy was present in LGD (20%), HGD (75%), IMC (20%) and IOA (56%) samples. Using linear regression analysis, Pearson coefficient was calculated for the correlation between DNA-PA and mean Allred expression scores for each PLK1 antibody. PLK1-M had a higher degree of correlation (r2=0.26, p=0.001) then PLK1-L (r2= 0.22, p=0.004). Inter-observer analysis with linear kappa scores confirmed good correlation of PLK1-M (κ=0.72, 95% CI:0.60-0.83) and PLK1-L (κ=0.53, 95% CI:0.38-0.68), but a Bland-Altman plot found pathologist 2 had a trend to score PLK-L more highly. However, intra-observer analysis confirmed both PLK1-L scores correlated with ploidy status (r2=0.14, p=0.023 and r2=0.27, p=0.0016). Pathologists took 90.3 seconds/slide (mean) to review and score with the Allred method. Using a mean Allred cut off score of 3.5, the sensitivity and specificity for the detection of aneuploidy were 81.3% and 75% for PLK1-M, 93.8% and 45% for PLK1-L. Conclusion: This pilot study demonstrated a statistically significant correlation between Allred reporting of PLK1 staining and DNA-PA. PLK1-L antibody appears more sensitive and PLK1-M more specific. PLK1 immunostaining is relatively inexpensive, less work intensive and more available than current methods to predict DNA-PA. Scoring staining with the Allred method is rapid and reproducible. In future, PLK1 staining may provide the basis of a more durable clinical biomarker panel to predict DNA ploidy.