Cervical cancer is the second most common cancer and third most common cause for cancer-related deaths in females worldwide [1], [2], [3]. Endocervical adenocarcinoma accounts for roughly 10-25% of carcinomas of the cervix, with an increasing case frequency despite the widespread implementation of vaccination efforts that have largely decreased the incidence of squamous cell carcinoma [4], [5]. Unfortunately, endocervical adenocarcinoma tends to have a worse overall prognosis than other cervical cancers, with stage of diagnosis being an important prognostic indicator of survival [6].

Beginning in the early 2010s, an international collaboration of gynecologic pathologists devised a pattern-based assessment of invasion for HPV-associated endocervical adenocarcinomas, deemed the “Silva System” [6], [7]. Three distinct patterns of invasion were identified: A, B, and C. Pattern A invasion consists of well demarcated, rounded groups of glands with non-destructive invasion without the presence of lymph-vascular space invasion, reminiscent of adenocarcinoma in situ (AIS). Pattern B tumors tend to be well-demarcated with localized stromal invasion eliciting limited/early desmoplasia and can have the presence of lymph-vascular space invasion. Lastly, Pattern C display diffusely infiltration, angulated glands with high grade morphology, extensive destructive invasion, and frequent lymph-vascular space invasion [7]. Notably, patterns B and C are independent risk factors for presence of nodal disease, conferring a higher stage of diagnosis and poorer survival [8]. However, these patterns of invasion only show risk stratification of lymph node metastases in HPV-associated endocervical adenocarcinoma cases, without any relevance in HPV-independent tumors [9].

Consensus on the utilization of the Silva system in clinical practice is limited given the poor overall reproducibility, especially with distinguishing Pattern A from Pattern B, albeit Pattern C tumors tend to have a worse clinical course [9], [10], [11]. Interobserver variability on Silva patterns of invasion varies greatly from cohort to cohort. In an initial cohort of 49 cases, only moderate agreement by Cohen’s kappa score of 0.51 was achieved [12]. Similarly, a study of 84 cases demonstrated moderate agreement with a kappa score of 0.488, notably showing overall poor agreement with Pattern B cases [10]. A large study of 96 cases revealed a slightly higher kappa agreement of 0.61, but showed poor agreement with demonstrating AIS versus pattern A cases [13].

Given the overall few studies demonstrating moderate concordance rates on pattern recognition among gynecologic pathologists, our objective was to evaluate the reproducibility within our institution. We aimed to identify our concordance utilizing not only gynecologic pathologists, but, in contrast to prior studies, also trainees. As such, this approach allowed us to evaluate the spectrum of training levels and comfortability utilizing the Silva system.