Tumoral Morphologic Features From Cervical Biopsies That Are Predictive of a Negligible Risk for Nodal Metastasis and...

The metastatic or recurrent potential of localized human papillomavirus-associated endocervical adenocarcinoma (HPVA EAC) is difficult to predict, especially based upon biopsy alone. Recent analyses of small cohorts indicate that high tumor nuclear grade (TNG) and the presence of necrotic tumor debris (NTD) from HPVA EACs in cervical biopsy specimens are highly predictive of nodal metastasis (NM). In the present study, we aimed to investigate how reliably tumoral morphologic features from cervical biopsy specimens predict NM or tumor recurrence (TR) and patient outcomes in a large cohort of endocervical adenocarcinoma patients. A cohort comprised of 397 patients with HPVA EAC treated at 18 institutions was identified, and cervical biopsies were paired with their associated complete tumor resections for a total of 794 specimens. A variety of tumoral histologic features were examined for each paired specimen, including TNG (assessed on a 3-tiered scale of increasing abnormalities-TNG1, TNG2, TNG3) and NTD (defined by the presence of necrotic and apoptotic tumor cells within tumor glandular lumens admixed with granular and eosinophilic amorphous material and inflammatory cells), which were correlated with outcomes. The distribution of TNG in biopsies was as follows: 86 (21.7%) TNG1, 223 (56.2%) TNG2, and 88 (22.2%) TNG3. NTD was identified in 176 (44%) of the biopsy specimens. The sensitivity, specificity, positive predictive value, and negative predictive value of a TNG1 assignment in the biopsy being predictive of the same assignment in the full resection were 0.82 (95% confidence interval [CI]: 0.7-0.9), 0.895 (0.86-0.93), 0.593 (0.48-0.696), and 0.96 (0.94-0.98), respectively. Respective values for an NTD-negative status were 0.89 (95% CI: 0.83-0.92), 0.715 (0.64-0.77), 0.72 (0.65-0.77), and 0.89 (0.83-0.93), respectively. Compared with the other cases in each category, both TNG1 and an NTD-negative status were each significantly associated with lower rates of NM (odds ratio for TNG1=0.245, 95% CI: 0.070-0.857, P=0.0277; for NTD=0.199, 95% CI: 0.094-0.421, P<0.0001) and TR (odds ratio for TNG1=0.225, 95% CI: 0.051-0.987, P=0.0479; for NTD=0.367, 95% CI: 0.171-0.786, P=0.0099) independent of depth of stromal invasion, lymphovascular invasion, tumor size, FIGO stage, and Silva pattern. Overall, 73/379 (19%) cases were both TNG1 and NTD-negative on the biopsy, and none of these 73 cases showed NM (0%), but a single case (1.4%) showed TR. In contrast, among the 324 biopsies with TNG2/3 and/or presence of NTD, 62 (19.1%) had NM, and 41 (12.9%) had TR. In summary, 2 variables in combination (ie, TNG1 and NTD-negative) identified a subset of HPVA EAC patients-∼19%-with a 0% frequency of nodal metastases and only 1.4% frequency of recurrence. Biopsies highly but imperfectly predicted these features. Nonetheless, these findings may potentially be of clinical utility in the risk stratification of patients with HPVA EACs. This may allow some patients with a minimal risk of nodal metastases and TR to be identified at the biopsy phase, thereby facilitating more personalized, possibly less aggressive treatment.