Reassessment of p53 immunohistochemistry thresholds in invasive high grade bladder cancer shows a better correlation with TP53 and FGFR3 mutations

Bladder Cancer

Pathol Res Pract. 2020 Aug 23;216(11):153186. doi: 10.1016/j.prp.2020.153186. Online ahead of print.


FGFR3 mutations are frequently mutually exclusive of TP53 mutations in invasive high grade urothelial carcinoma (HGUC) and p53 immunohistochemistry is often used as a surrogate for TP53 mutations. A 10 % staining cut off has been used in HGUC for designation as p53 positive or negative however, a novel contemporary method we have previously proposed (0% or >50 % - abnormal vs. 1-49 % - wild type) has shown significant correlation with oncologic outcome as well. We aimed to compare how a ≥10 % vs. 0 % and ≥ 50 % cut off p53 assessment method correlates with TP53 and FGFR3 mutation status. Tissue microarrays created from three retrospective cohorts (two cystectomy cohorts (cohort A, n = 206 and cohort B, n = 91; one T1 transurethral resection cohort (cohort C, n = 47)) were stained with p53 and scored by two blinded reviewers using both p53 scoring schemes. 50 cases from cohort A were assessed for TP53 and FGFR3 mutation status using next generation sequencing and FGFR3 mutation status was separately assessed in cohorts B and C using SNaPshot methodology. 202 (58.7 %) and 142 (41.3 %) cases showed abnormal and wild type p53 staining, respectively. Using the 10 % cut off, 254 cases were positive (73.8 %) and 90 cases were negative (26.2 %). 27 (14.4 %) and 15 (30 %) assessed cases demonstrated FGFR3 and TP53 mutations, respectively; 19/27 FGFR3 mutated showed a wild type pattern of p53 expression while 15/15 TP53 mutated tumours showed an abnormal pattern of p53 expression. There was a significant correlation between the contemporary p53 scoring scheme and TP53 and FGFR3 mutations (p < 0.0001 and p = 0.002, respectively). Improved sensitivity, specificity, positive predictive value, and negative predictive value for TP53 mutation was also seen compared to the 10 % cut off; specifically, the sensitivity and negative predictive value were 100 %. These findings might be of clinical relevance in the era of precision medicine.