Long-term Behavior of Serous Borderline Tumors Subdivided Into Atypical Proliferative Tumors and Noninvasive Low-grade Carcinomas: A Population-based Clinicopathologic Study of 942 Cases.

Ovarian serous borderline tumors (SBTs) have been the subject of considerable controversy, particularly with regard to terminology and behavior. It has been proposed that they constitute a heterogenous group of tumors composed, for the most part, of typical SBTs that are benign and designated "atypical proliferative serous tumor (APST)" and a small subset of SBTs with micropapillary architecture that have a poor outcome and are designated "noninvasive low-grade serous carcinoma (niLGSC)". It also has been argued that the difference in behavior between the 2 groups is not due to the subtype of the primary tumor but rather the presence of extraovarian disease, specifically invasive implants. According to the terminology of the 2014 WHO Classification, typical SBTs are equivalent to APSTs and SBTs displaying micropapillary architecture are synonymous with niLGSC. In addition, "invasive implants" were renamed "low-grade serous carcinoma" (LGSC). The argument as to whether it is the appearance of the primary tumor or the presence of extraovarian LGSC that determines outcome remains unsettled. The current study was initiated in 2004 and was designed to determine what factors were predictive of outcome, with special attention to the appearance of the primary tumor (APST vs. niLGSC) and that of the extraovarian disease (noninvasive vs. invasive implants). Our study is population based, involving the entire female population of Denmark. None of the women in the study were lost to follow-up, which ranged up to 36 years (median, 15 y). All the microscopic slides from the contributing hospitals were rereviewed by a panel of 2 pathologists (R.V. and R.J.K.) who were blinded to the follow-up. After excluding those that were not SBTs by the pathology panel, as well as cases with a prior or concurrent cancer or undefined stage, 942 women remained, of which 867 were APSTs and 75 were niLGSCs. The median patient age was 50 years (range, 16 to 97 y). Eight hundred nine women (86%) presented with FIGO stage I disease, whereas 133 (14%) had advanced stage disease. Compared with APSTs, niLGSC exhibited a significantly greater frequency of bilaterality, residual gross disease after surgery, microinvasion/microinvasive carcinoma, advanced stage disease, and invasive implants at presentation (P-values <0.003). Because the cause of death is difficult to accurately ascertain from death certificates, we used development of invasive serous carcinoma as the primary endpoint as following development of carcinoma, the mortality is very high. In the entire cohort, subsequent development of carcinoma occurred in 4%, of which 93% were low grade and 7% high grade (median time, 10 y; range, up to 25 y). After adjusting for age at and time since diagnosis of APST or niLGSC, occurrence of subsequent carcinoma was significantly higher with niLGSC than APST among all stages combined (hazard ratio [HR]=3.8; 95% confidence interval [CI], 1.7-8.2). This difference was still significant for stage I but not advanced stage cases. Moreover, all-cause mortality was not statistically significantly different between APST and niLGSC. Of all women with advanced stage disease, 114 (86%) had noninvasive implants, whereas 19 (14%) were invasive. Noninvasive implants were significantly associated with subsequent development of carcinoma (HR=7.7; 95% CI, 3.9-15.0), but the risk with invasive implants was significantly higher (HR=42.3; 95% CI, 16.1-111.1). In conclusion, although invasive implants are the most important feature in predicting an adverse outcome, subclassification into APST and niLGSC is important as it stratifies women with respect to risk for advanced stage disease and invasive implants for all women and development of serous carcinoma for stage I cases.