Histotype classification of ovarian carcinoma: A comparison of approaches
Introduction
Ovarian carcinoma is heterogeneous, consisting of distinct histotypes with unique epidemiologic characteristics, molecular features, clinical presentations, and prognostic outcomes [1,2]. In the past twenty years, there has been a considerable evolution in how histotypes are defined. Older classification systems, including the 1973 and 2003 World Health Organization (WHO) Classification of Tumors of Female Reproductive Organs [3,4], identified eight principal histotypes: serous, mucinous, endometrioid, clear cell, transitional cell, undifferentiated, unclassified, and mixed ovarian surface epithelial malignant tumors. However, these classification systems showed only moderate reproducibility among pathologists [[5], [6], [7], [8]], highlighting the need for refinement of diagnostic criteria to make them clinically useful. Recent molecular evidence demonstrates that serous carcinomas are two separate histotypes, high- and low-grade [9]. Further, many high-grade endometrioid and undifferentiated carcinomas diagnosed using morphology alone are high-grade serous carcinomas based on protein expression [10], and transitional cell carcinomas are indistinguishable from high-grade serous carcinomas [11]. It has also become clear that true mixed carcinomas are exceedingly rare [12]. In 2014, the new WHO criteria [13] incorporated these histopathological insights, recognizing five principal ovarian carcinoma histotypes: high-grade serous carcinoma (HGSC), low-grade serous carcinoma (LGSC), endometrioid carcinoma (EC), clear cell carcinoma (CCC), and mucinous carcinoma (MC).
A recent study by Kommoss et al. [14] used data from a clinical trial to evaluate the extent to which histotype diagnosis assigned by a gynecologic pathologist in 2002 changed when the same slides were re-reviewed by the same pathologist using the 2014 WHO criteria. Upon re-review, the histotype diagnoses were confirmed for only 54% of patients. However, when two pathologists independently assessed the same diagnostic slides using the 2014 WHO guidelines, an identical histotype was assigned by the pathologists for 98% of patients, which suggests high reproducibility of the 2014 WHO criteria. The low concordance between historical histotype diagnoses and the pathologic re-review using the 2014 criteria prompted Kommoss et al. [14] to conclude: “it is completely unacceptable to use historical histotype diagnosis for research purposes.” Given that the majority of epidemiologic studies of ovarian cancer were conducted prior to publication of the WHO 2014 criteria and include histotype diagnoses determined by the earlier WHO classification schemes [15], this assertion by Kommoss et al. [14] has major implications for any histotype-specific analyses or research using existing data sources.
In the present study, we evaluated the above statement by Kommoss et al. [14] using data from two population-based case-control studies of ovarian carcinoma which were conducted prior to the publication of the 2014 WHO guidelines. We compared the agreement of histotypes assigned according to three different classification approaches: 1. pre-2014 WHO (“historic”) histotype; 2. Standardized pathology review of H&E slides applying the 2014 WHO criteria alone; and 3. An integrated immunohistochemical (IHC) assessment along with the 2014 WHO criteria. We also evaluated the extent to which histotype-specific survival patterns differed across histotype assignment approaches.
Section snippets
Study population
This study comprised data from two population-based case-control studies, the Diseases of the Ovary and their Evaluation (DOVE) Study and the North Carolina Ovarian Cancer Study (NCOCS), described in detail elsewhere [[16], [17], [18], [19]]. Briefly, the DOVE study was conducted in 13 counties of Washington State. Cases were identified through the local Surveillance, Epidemiology, and End Results (SEER) cancer registry, the Cancer Surveillance System (CSS), and eligible cases, aged
Results
After excluding 32 women with unknown stage disease and 2 women with non-epithelial histology, a total of 2361 women with ovarian carcinoma were identified from DOVE and NCOCS (Supplementary Fig. 1). Supplementary Table 2 shows participant characteristics by age, race, study, and tumor stage. For historic histotype, 403 were classified as other epithelial, and 1958 tumors were assigned one of the five principal histotypes: 1273 HGSC (65%), 66 LGSC (3%), 92 MC (5%), 334 EC (17%), 193 CCC (10%).
Discussion
Our ability to utilize data from historic epidemiologic studies of ovarian carcinoma to answer histotype-specific questions hinges on whether there is a reliable approach to re-classify historic histotype diagnoses to approximate the 2014 WHO criteria, which better reflect the current knowledge of ovarian carcinoma pathogenesis. The comparison of three classification approaches showed overall high agreement for the five principal histotypes, with the largest discrepancies in histotype
Funding
This work was supported by the National Cancer Institute of the National Institutes of Health (K99 CA218681 to L.C.P., R01 CA168758 to M.A.R. and J.A.D., R01 CA076016 to J.M.S., K22 CA193860 to H.R.H., and P30 CA042014 to M. Beckerle); Calgary Laboratory Services internal research support (RS11-508 to M.K.); and the Huntsman Cancer Foundation (J.A.D.).
Conflict of interest
The authors have no conflicts of interest to disclose.
Author contributions
Mary Anne Rossing, Jennifer A. Doherty, and Joellen M. Schildkraut designed the study. Mary Anne Rossing, Jennifer A. Doherty, Joellen M. Schildkraut, Kristine Wicklund, and Andrew Berchuck contributed to data collection. Michael Anglesio, Rex Bentley, C. Blake Gilks, Tayyebeh Nazeran, David G. Huntsman, and Martin Köbel contributed to the pathologic and immunohistochemical assessment of tumors. Kara L. Cushing-Haugen and Lauren C. Peres completed all data analysis, and Lauren C. Peres drafted
References (40)
- et al.
Interobserver variability in the interpretation of epithelial ovarian cancer
Gynecol. Oncol.
(1984) - et al.
Calculator for ovarian carcinoma subtype prediction
Mod. Pathol.
(2011) - et al.
The prognostic significance of residual disease, FIGO substage, tumor histology, and grade in patients with FIGO stage III ovarian cancer
Gynecol. Oncol.
(1995) - et al.
Prognostic factors of stage IV epithelial ovarian cancer: a multicenter retrospective study
Gynecol. Oncol.
(2001) - et al.
Do clear cell ovarian carcinomas have poorer prognosis compared to other epithelial cell types? A study of 1411 clear cell ovarian cancers
Gynecol. Oncol.
(2008) - et al.
Natural history and outcome of mucinous carcinoma of the ovary
Am. J. Obstet. Gynecol.
(2011) Histologic subtypes of ovarian carcinoma
Int. J. Gynecol. Pathol.
(2008)- et al.
Ovarian carcinoma subtypes are different diseases: implications for biomarker studies
PLoS Med.
(2008) - et al.
International Histological Classification of Tumors: Histological Typing of Ovarian Tumors
(1973)