Interobserver Variability in Diagnosing STIC With Frequent Underdiagnosis of Early High-grade Serous Carcinoma: Insights From a Multicenter Review From the CANSTIC Study

It has become widely accepted that most cases of tubo-ovarian high-grade serous carcinoma (HGSC) arise from precursor lesions in the fallopian tube called serous tubal intraepithelial carcinoma (STIC). This study evaluated the interobserver reproducibility of STIC diagnosis and its diagnostic boundaries from a multicentre Canadian cohort study. The multicentre CANSTIC study included patients with isolated STIC diagnosed between 1998 and 2020 across 15 Canadian centres. A blinded panel of 4 subspecialized gynecologic pathologists independently reviewed scanned slides (hematoxylin and eosin, p53, and Ki67) from 57 cases and met for a consensus meeting. Diagnoses were grouped as HGSC, STIC, and serous tubal intraepithelial lesion/other. The initial interobserver agreement for the diagnostic categories was moderate (Fleiss’s kappa =0.495, 95% CI: 0.269–0.675; Cohen’s kappa =0.492, 95% CI: 0.226–0.758, percent agreement 83.3%). A concordant diagnosis among all 4 reviewers was initially reached in 68% (39/57; 95% CI: 55.5%–79.0%), which increased to 91% (52/57; 95% CI: 81.1%–96.2%) after the consensus meeting. Consensus or majority (3 of 4 reviewers) was considered the final diagnosis; this resulted in STIC in 79% (45/57; 95% CI: 66.7–87.5), in agreement with the original diagnosis. Nine percent of cases were downgraded to serous tubal intraepithelial lesion/other (5/57; 95% CI: 1.8%–14.4%), while 12% were upgraded to HGSC (7/57; 95% CI: 6.1%–23.2%). Six patients developed HGSC clinically (11%; 95% CI: 4.9%–21.1%), of whom 3 had a review diagnosis of HGSC and 3 of STIC. This study shows suboptimal interobserver reproducibility for a diagnosis of STIC, despite consideration of ancillary p53/Ki67 testing. This study offers guidance to increase interobserver reproducibility in this challenging area, particularly in avoiding an underdiagnosis of HGSC.