Novel staining pattern of p53 in Barrett’s dysplasia – the absent pattern

Sir: Barrett's oesophagus (BO) is conversion of oesophageal squamous mucosa to a glandular phenotype, and is a consequence of gastro-oesophageal reflux. This is a precursor to oesophageal adenocarcinoma (OA), which is rising rapidly in western countries and carries a poor prognosis.1 This pathway is characterized by intestinal metaplasia and increasing grades of dysplasia before cancer supervenes. Recognizing dysplasia early allows close monitoring as well as treatment preventing OA or cure at an early stage. The recog`nition of dysplasia by pathologists is critical, and while pathologists can recognize dysplasia reproducibly2,3 this may sometimes be difficult. Therefore additional prognostic markers would be helpful. Several studies have shown that p53 overexpression is associated closely with dysplasia and predicts progression in its own right.3–6 Overexpression is a consequence of mutations which stabilize the inactivated protein. However, a subset of unequivocal Barrett's dysplasia cases are negative for p53. In some cases p53 may not be involved in the dysplastic progression, but in other cases truncating TP53 mutations or epigenetic silencing may cause protein inactivation. In these cases it is lack of expression rather than overexpression which would be expected. With this in mind, we re-analysed the dysplastic and indefinite p53-negative cases from our previously published study4 to see if this pattern might have accounted for some p53-negative dysplasias and to assess its significance prognostically. As described in detail previously,3 cases of oesophageal glandular dysplasia were identified from pathology databases at Queens Medical Centre and City Hospital, Nottingham between 1987 and 2004. Thirty-two random cases of BO without dysplasia were also included. Five pathologists blinded to the original diagnosis classified each case independently using the revised Vienna classification. One pathologist scored each p53-stained section (D07 antibody with microwave antigen retrieval; Dako, Ely, UK) as positive, negative or not representative. A consensus Vienna score for each case was determined. Cases negative for p53 but with a study diagnosis of dysplasia or indefinite for dysplasia were reviewed again by two pathologists (PVK and MI). Two distinct patterns were identified: (i) wild-type (p53-wt) with weak immunoreactivity in dysplastic/atypical epithelium similar to the background non-dysplastic epithelium and (ii) absent staining (p53-abs) with complete lack of staining in dysplastic/atypical epithelium relative to weak positivity in surrounding non-dysplastic epithelium (Figure 1). This contrasted with positive staining described above, where the dysplastic epithelium showed much stronger staining than the surrounding non-dysplastic epithelium (p53-pos). Patient records were examined to determine length of follow-up and progression. Three patterns of p53 immunoreactivity. A, p53-positive. Dysplastic epithelium strongly positive with weak background positivity in non-dysplastic epithelium. B, p53-wild-type. Weak staining in atypical villiform epithelium similar to background. C,D, p53-absent. Complete lack of p53 expression in dysplastic epithelium contrasting sharply with weak positivity in surrounding non-dysplastic epithelium. A total of 175 cases were identified. In 33 cases p53 staining could not be assessed due to lack of representative tissue, 43 cases were p53-positive (42 dysplasia, one indefinite) and 99 p53-negative (11 dysplasia, 10 indefinite). Re-analysis of categories 2–4 p53-negatives are shown in Table 1. The 10 indefinite for dysplasia cases all had staining in equivocal areas indistinguishable from background non-dysplastic Barrett's epithelium (p53-wt). Of the 11 cases with dysplasia, five showed the absent pattern of staining in dysplastic areas (p53-abs), while six showed weak staining indistinguishable from background Barrett's epithelium (p53-wt) (Table 1). Of the 40 patients with positive p53 with follow-up data available, 28 progressed within 10 years. Of the 10 patients with indefinite for dysplasia and p53-wt, two progressed within 10 years. Of the 11 dysplastic cases, all five with p53-abs progressed clinically or histologically but only two of six with p53-wt progressed; four remained alive with regression without intervention. We and others have shown that in addition to dysplasia, immunohistochemical analysis of p53 overexpression is a powerful predictor of progression in BO.3–6 Herein we have identified a novel pattern of complete absence of staining (p53-abs) which has a similar predictive value to the classical pattern of overexpression (p53-pos). This should add power to the technique of p53 immunohistochemistry by including cases with abnormal expression of the protein which was formally regarded as negative. p53-abs may correlate with truncating mutations in TP53 which cannot be recognized by the commonly used D07 antibody. Alternatively, a variety of other non-mutational mechanisms for TP53 inactivation have been described. The proportion of oesophageal cancers harbouring mutant p53 ranges up to 80%.7 Of our assessable definite dysplasia cases, 47 of 53 (89%) showed an abnormal p53 immunophenotype. Abnormalities of p53 may be detected by direct mutation detection, inferred by finding loss of heterozygosity at locus 17p or chromosome 17 aneuploidy by fluorescence in situ hybridization (FISH).8 These techniques are of use in the research environment, but immunohistochemistry is more easily achievable in routine pathology and has the advantage of detecting non-mutational p53 abnormalities. Interpretation of p53 immunohistochemistry requires training and experience, and may vary between laboratories. However, if careful attention is paid to discriminating between staining in non-dysplastic areas relative to the area of interest these variations can usually be accounted for. The evidence base is now accumulating to consider taking p53 immunohistochemistry from the research domain into quality-assured routine practice.