Immunostaining boosts pathologists’ accuracy in Barrett’s esophagus



SAN DIEGO – Years of experience and an academic medical center affiliation predicted the accuracy of pathologists reviewing biopsies from patients with Barrett’s esophagus, according to the results of a multinational study.

Those with 5 or more years of experience were less likely to make major diagnostic errors in reviewing Barrett’s esophagus biopsies (odds ratio [OR], 0.48, 95% confidence interval, 0.31-0.74). Pathologists who worked in nonacademic settings were more likely to make a major diagnostic error (OR, 1.76; 95% CI, 1.15-2.69) when reviewing hematoxylin and eosin-stained slides alone, but the addition of p53 immunostaining greatly improved accuracy.

Current guidelines recommend expert evaluation of Barrett’s esophagus biopsies that show dysplasia, but exact determination of expert review status had been lacking, according to Marnix Jansen, MD, a pathologist at University College London.

“The guidelines say that biopsies with dysplasia need to be reviewed by an expert pathologist, but don’t define what makes an expert pathologist,” Dr. Jansen said in an interview at the annual Digestive Disease Week.

“We wanted to advance the field by for the first time creating objective and quantitative standards” to delineate the characteristics of an expert pathologist in reviewing Barrett’s esophagus tissue samples, said Dr. Jansen. The study’s first author is Myrtle J. van der Wel, MD, of Amsterdam University Medical Center, the Netherlands.

More than 6,000 individual case diagnoses were used in the study, which included pathologists from more than 20 countries. Before the pathologists began reviewing the case set, they answered a questionnaire about training, practice context, years of experience, case volume, and other demographic characteristics.

“We then sent those biopsies around the world to ... 55 pathologists in the U.S., in Europe, Japan, Australia, even some in South America – so really around the whole globe,” explained Dr. Jansen. Biopsies were assessed by each pathologist before and after p53 immunostaining.

“Once we had the final dataset – which is massive, because we had 6,000 case diagnoses within our dataset – we could then regress those variables back onto the consensus data,” providing a first-ever look at “clear predictors of what the pathologist looks like that will score on a par with where the experts are,” said Dr. Jansen.

The results? “You need at least 5 years of experience. On top of that, if you are a pathologist working in a [nonacademic center], you are at a slightly increased risk of making major diagnostic errors,” said Dr. Jansen. However, the analysis convincingly showed that the addition of p53 immunostaining neutralized the risk for these pathologists – a strength of having such a large dataset, he said.

The study also affirmed the safety of digital pathology for expert review, said Dr. Jansen: “One of the reassuring points of our study was that we found that the best concordance was for nondysplastic Barrett’s, and high-grade dysplasia, which really replicates known glass slide characteristics. So we can really say that digital pathology is safe for this application – which is very relevant for pathologists that are taking in cases from outside for expert review.”

Concordance rates for nondysplastic Barrett’s esophagus and high-grade dysplasia were over 70%; for low-grade dysplasia, rates were intermediate at 42%.

Going forward, the study can inform the next iteration of guidelines for pathologist review of Barrett’s dysplasia, said Dr. Jansen. Rather than just recommending expert review, the guidelines can include a quantitative assessment of what’s needed. “You need to have to have at least 5 years of experience, and if you work in a [community hospital], to use a p53, and that is collectively what amounts to expertise in Barrett’s pathology.”

A follow-up study with a similar design is planned within the United Kingdom, the Netherlands, and the United States. This study, which Dr. Jansen said would enroll hundreds of pathologists, will include an intervention arm that administers a tutorial with the aim of improving concordance scoring.

Dr. Jansen reported no relevant conflicts of interest.

Next Article: