Dysplasia in Barrett's oesophagus: p53 immunostaining is more reproducible than haematoxylin and eosin diagnosis and improves overall reliability, while grading is poorly reproducible.

ABSTRACT

AIMS: p53 immunostaining in Barrett's oesophagus (BO) has been shown to be predictive of progression, but data regarding its generalizability to routine practice are lacking. This study compared the reliability of p53 and dysplasia interpretation and grading. METHODS AND RESULTS: Seventy-two cases encompassing the full spectrum of BO were circulated to 10 pathologists from four institutions after a brief training session in p53 interpretation. Each pathologist classified cases on haematoxylin and eosin (H&E) alone using the Vienna classification and assessed the p53 staining using a qualitative system. Agreement was assessed using kappa statistics. For the four-tier Vienna system, the average unweighted kappa was 0.30. Weighted kappa values varied from 0.27 to 0.69 with an average of 0.47. When grouped into definite dysplasia versus no definite dysplasia the average kappa was 0.55, but the kappa for low-grade dysplasia (LGD) versus high-grade dysplasia (HGD) was only 0.31. For p53, using the three recognized patterns, the unweighted kappa was 0.6 (confidence interval 0.58-0.63). When cases were evaluated with both H&E and p53 the average kappa was 0.61 for definite dysplasia versus the rest. CONCLUSIONS: p53 immunohistochemistry interpretation is more reliable than dysplasia diagnosis, even with limited training. As it is predictive of prognosis and improves diagnostic reproducibility, it is suitable for routine use by pathologists as an adjunct to dysplasia diagnosis. The distinction of LGD versus HGD was poor. This study supports simplifying dysplasia diagnosis into 'present', 'indefinite' or 'absent', and the use of p53 as an ancillary marker in difficult cases. This should help to prevent overdiagnosis of dysplasia and inappropriate treatment.