To what degree is there interobserver variability among pathologists for T1 penile cancer?

Abstract

381 Background: Molecular and genetic markers have yet to be developed to predict those patients that are at risk for lymph node metastasis. Currently, the grading of penile cancer plays a critical role in the determination of which patients receive an inguinal lymph node dissection (ILND) along with other treatment modalities. We sought out to determine the variance among genitourinary (GU) pathologists at a tertiary cancer center for penile cancer based on a European model. Methods: Nine patients that were diagnosed with stage pT1 primary penile squamous cell carcinoma were selected who underwent either a partial (8) or radical penectomy (1) from 10/2000 to 09/2009. All slides from each case were reviewed by each of the 3 reviewing pathologists, independently, who diagnosed the subtype of squamous cell carcinoma according to WHO criteria, assigned a grade, noted whether lymphovascular invasion was present or not and finally staged the tumor according to the AJCC Cancer Staging Manual, 7th edition. No access to the original, final pathological diagnosis was allowed. Interobserver variance between the 3 GU pathologists and each variable was calculated using Cohen’s kappa coefficient. Results: Complete agreement was reached in 3 cases for tumor grade and 4 cases for tumor stage out of 9. Overall, the 3 GU pathologists only displayed fair agreement at 30% for tumor grade (ê = 0.30, p = 0.018) and trended towards fair agreement at 24% (ê = 0.24, p = 0.077) and 25% (ê = 0.250, p = 0.097) for tumor stage and LVI respectively. Conclusions: The variance displayed herein demonstrates the difficulty in identifying individuals that would benefit from a diagnostic/therapeutic lymph node dissection based on pathological staging. This corroborates that of the European model and calls for novel methods to determine reproducible prognostic markers.