ORIGINAL ARTICLE

FIGO Versus Silverberg Grading Systems in Ovarian

Endometrioid Carcinoma
A Comparative Prognostic Analysis
Carlos Parra-Herran, MD,*† Dina Bassiouny, MD, PhD,*‡ Danielle Vicus, MD, MSc,§∥
Ekaterina Olkhov-Mitsel, PhD,* Matthew Cesari, MD, CM, FRCPC,†
Nadia Ismiil, MBChB, FRCPC,*† and Sharon Nofech-Mozes, MD*†

laterality, lymphovascular space invasion, and stage also corre-

Abstract: The International Federation of Obstetrics and Gyne- lated with outcome. Stage showed prognostication superior to all
cology (FIGO) grading system for endometrial carcinoma is other variables in multivariate analysis. As currently defined, the
Downloaded from [Link] by BhDMf5ePHKbH4TTImqenVJ2toCr/9wZZQx3cb4JDYsokm3ogpi/h/Os0v6nk7FSU on 02/06/2019

currently applied to ovarian endometrioid carcinoma (OEC) in Silverberg grading system is a better predictor of survival than
many practices. However, previous reports claim superior FIGO. Such differences may be explained by the G2 OEC groups,
prognostication by using the Silverberg grading system for with G2 Silverberg clustering with G1 tumors, and having a more
ovarian carcinoma. Thus, a thorough comparison between favorable behavior compared with G2 FIGO. Thus, Silverberg
FIGO and Silverberg in OEC is still warranted. A total of 72 may be preferable in order to stratify patients in low and high-risk
OECs diagnosed at our institution were independently graded categories for prognosis and disease management.
using both systems. Grade (G) following Silverberg was based on
combined scores for architecture, nuclear atypia, and mitotic Key Words: ovarian cancer, endometrioid, ovarian endometrioid
activity. FIGO grading was based on the % of nonsquamous carcinoma, FIGO, Silverberg, grade, grading
solid component; severe atypia warranted upgrade to the archi- (Am J Surg Pathol 2019;43:161–167)
tectural FIGO grade (G1 to G2 or G2 to G3). Case grouping by
grade was correlated with disease-free (DFS), disease-specific
(DSS), and overall (OS) survival. Eleven (15.3%) OECs were
bilateral, 26 (36.1%) had ovarian surface involvement, and 12
(16.7%) had lymphovascular space invasion. Forty-seven OECs
O varian endometrioid carcinoma (OEC) represents the
second most common type of epithelial ovarian carci-
noma. The International Federation of Obstetrics and
were stage I (65%), 16 (22%) stage II, and 9 (13%) stage III.
Gynecology (FIGO) grading system for endometrial endo-
Median follow-up period was 62 months (range: 1 to 179 mo).
metrioid carcinoma is used to grade OEC in many practices
Median DFS was 60.5 months (1 to 179 mo); median OS was
given the morphologic similarities between these 2 tumors.
61 months (1 to 179 mo). Sixteen (22%) OECs recurred and 9
However, the prognostic value of the FIGO grading system
(13%) patiets died of disease. In univariate analysis, both FIGO
has not been addressed using an evidence-based approach.
and Silverberg correlated significantly with DFS, DSS, and OS
Conversely, the universal Silverberg grading system for
(all with P < 0.05). However, when compared in multivariate
ovarian carcinoma has documented association with overall
analysis, only Silverberg retained statistical correlation with
survival (P < 0.05). G1+G2 OEC by Silverberg had significantly
survival (OS) and disease-free survival (DFS).1,2 Fur-
thermore, Ishioka et al3 found that, while both grading sys-
better DFS, DSS, and OS compared with G3; such separation
tems were prognostically significant in a large cohort of
was not seen with FIGO. Survival was similar in Silverberg G1
and G2 tumors even 5 years after diagnosis, whereas FIGO
ovarian carcinomas, the Silverberg system had a stronger
association with lymph node involvement and residual tumor
G2 tumors had survival approaching G1 in the first 5 years, but
compared with FIGO, thus suggesting a superior value for
declined after the 5-year mark approaching G3 tumors. Tumor
the former. When further assessing the 18 patients with OEC,
7 (39%) were found to have an inconsistent grade between the
From the Departments of *Laboratory Medicine; §Gynecologic Oncology, 2 systems, although both systems correlated equally to
Sunnybrook Health Sciences Centre; Departments of †Laboratory prognosis. This further emphasizes the need to address OEC
Medicine and Pathobiology; ∥Obstetrics and Gynaecology, University independently from other histologic subtypes.
of Toronto, Toronto, ON Canada; and ‡Department of Pathology, Even though a universal grading system for ovarian
Mansoura University, Egypt.
Conflicts of Interest and Source of Funding: The authors have disclosed cancer has documented prognostic value, such an approach
that they have no significant relationships with, or financial interest is not widely used. This is mainly because of our current
in, any commercial companies pertaining to this article. understanding of the major ovarian cancer types as bio-
Correspondence: Carlos Parra-Herran, MD, Deparment of Laboratory Med- logically different diseases with their own grading system.
icine and Pathobiology, University of Toronto, Pathologist, Sunnybrook
Health Sciences Centre, 2075 Bayview Ave., Room E4-27a, Toronto ON,
Grading of ovarian serous carcinoma has moved toward a
Canada M4N 3M5 (e-mail: [Link]@[Link]). 2-tier system with proven biological and clinical differences
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved. between low-grade and high-grade tumors, as well as

Am J Surg Pathol  Volume 43, Number 2, February 2019 [Link] | 161

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

Parra-Herran et al Am J Surg Pathol  Volume 43, Number 2, February 2019

acceptable reproducibility between these 2 categories.4,5 mitotic activity in 10 high-power fields (0 to 9 = 1, 10 to

Ovarian clear cell carcinoma and carcinosarcoma are, by 24 = 2, ≥ 25 = 3); added score determined grade (G1: 3 to 5,
definition, high grade. Grading of OEC, in contrast, lacks G2: 6 to 7, G3: 8 to 9). FIGO grading was based on the % of
the same level of consensus among practices, and a thor- solid component (G1: <5%, G2: 5% to 50%, G3: > 50%);
ough comparative analysis of the currently available grad- severe atypia warranted upgrade to the architectural FIGO
ing systems is still necessary. Thus, our aim is to compare grade (1 to 2 or 2 to 3).
the prognostic value of the Silverberg and FIGO grading
systems in a well-annotated cohort of OEC. Statistical Analysis
For analysis purposes, time of recurrence and time
MATERIALS AND METHODS of death were used to estimate DFS, disease-specific sur-
This study was approved by the Research Ethics vival (DSS), and OS. The time to recurrence was defined
Boards at Sunnybrook Health Sciences Centre. from the date of primary surgery to the date of first re-
corded recurrence by imaging or tissue diagnosis. Like-
Case Selection and Review wise, time to death was defined from the date of primary
Surgical pathology reports of in-house ovarian cancer surgery. DSS counted exclusively events of cancer-related
patients treated surgically at Sunnybrook Health Sciences death; death of other causes was counted in OS, but not in
Centre between 2000 and 2013 were retrieved from the DSS. Univariate DFS, DSS, and OS probabilities were
Anatomic Pathology database. After exclusion of meta- estimated and plotted using the Kaplan-Meier method
static tumors, cases with controversial site of origin, and and compared using the log rank test. Cox proportional
cases with minimal microscopic disease after neoadjuvant hazards models were used to generate univariate hazard
chemotherapy, cases of 702 patients with primary epithelial ratios and 95% confidence intervals, to analyze individual
ovarian cancer were available. Tumors diagnosed as en- contributions of each recorded clinical and pathologic
dometrioid type as per 2014 World Health Organization variable to DFS, DSS, and OS probabilities. Multivariate
criteria were further selected. Histologic diagnosis of en- cox proportional hazard regression model analysis was
dometrioid carcinoma was confirmed in all cases by subsequently conducted to weigh clinically meaningful
independent review of all available hematoxylin and eosin– variables that were significant on univariate analysis. The
stained slides by 1 fellow and at least 1 gynecologic patho- correlation between tumor grade and other clinicopatho-
logist; only cases with full consensus on histotype were logic characteristics was calculated using appropriate sta-
included. In addition, the presence of at least one con- tistical tests, that is, χ2 test for categorical variables and
firmatory endometrioid feature was required for inclusion: 1-way analysis of variance for continuous variables. All
background of endometriosis and/or endometrioid ad- statistical analyses were performed using the SPSS soft-
enofibroma or metaplasia (squamous, mucinous, and/or ware 24.0 (IBM Corporation, New York, NY). Two-sided
secretory). Further confirmation of the diagnosis involved P-values <0.05 were considered statistically significant.
the use of immunohistochemistry, which was performed in
tissue microarrays containing two 1 mm cores from each RESULTS
case, and stained with monoclonal antibodies against WT1 A total of 72 OECs were included. Clinical and
(6F-H2; Ventana) and Napsin-A (MRQ-60; Ventana). A pathologic characteristics of the cohort were previously
positive control was included in every run. All stains were described in a separate study.6 Briefly, 11/72 (15.3%)
reviewed independently by at least 2 observers; any staining OECs were bilateral, 26/72 (36.1%) had ovarian surface
for these markers was interpreted as positive and warranted involvement, and 12/72 (16.7%) had lymphovascular space
exclusion of the case. Finally, tumors with synchronous invasion. Distribution by stage was as follows: stage I, 47
myoinvasive endometrial carcinoma or endometrioid (65%); stage II, 16 (22%); and stage III 9 (13%).
metastases to the ovary from the endometrium as per report
6 were also excluded. Comparative Analysis: Tumor Grade
Clinical and pathologic variables were extracted Table 1 shows the distribution of OECs among the
from the institutional database Biomatrix, including pa- FIGO and Silverberg grades. Most tumors showed
tient age, pathology data (tumor size, laterality, lympho- concordance between FIGO and Silverberg grades (54
vascular space invasion, stage), and follow-up data. The tumors, 75%). Grade discordance was observed in 18
latter included time to recurrence in months, last date of
follow-up and status at follow-up.
TABLE 1. Histologic Grade Distribution of Ovarian
Tumor Grade Assessment Endometrioid Carcinomas by FIGO and Silverberg Systems
OECs selected were independently graded by 2 gyne- Silverberg
cologic pathologists, first as per Silverberg and, subsequently 1 2 3 Total (N [%])
( > 12 mo), as per FIGO criteria. A consensus grade through FIGO
a third reviewer was obtained if initial grading by the 2 1 26 7 0 33 (45.8)
observers was discordant. Silverberg grading was based on 2 5 18 1 24 (33.3)
architecture (majority glandular = 1, papillary = 2, solid = 3), 3 1 4 10 15 (20.8)
Total (N [%]) 32 (44.4) 29 (40.3) 11 (15.3) 72 (100)
nuclear atypia (mild = 1, moderate = 2, severe = 3), and

162 | [Link] Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

Am J Surg Pathol  Volume 43, Number 2, February 2019 Ovarian Endometrioid Carcinoma Grading System

FIGURE 1. OEC with discordant FIGO and Silverberg grades. A and B, OEC with a Silverberg grade 1 (score 1+1+1 = 3) and FIGO
grade 2 (20% solid growth). C and D, OEC with a Silverberg grade 2 (score 3+2+1 = 6) and FIGO grade 3 (80% solid growth). In
general, a predominant solid growth pattern will result in both a FIGO grade 3 and an architectural score of 3 by Silverberg;
however, Silverberg also takes nuclear atypia and mitotic activity into consideration.

TABLE 2. Univariate Analysis of Survival in Patients With Ovarian Endometrioid Carcinoma

Disease-free Survival Overall Survival Disease-specific Survival
HR 95% CI P HR 95% CI P HR 95% CI P
Age (y) 1.045 1.010-1.081 0.011 1.025 0.983-1.069 0.240 1.011 0.964-1.061 0.644
Size 0.970 0.896-1.050 0.449 0.982 0.895-1.077 0.703 0.959 0.860-1.070 0.457
Surface 4.785 1.658-13.813 0.004 2.794 0.846-9.228 0.092 4.344 1.081-17.463 0.039
Laterality 3.689 1.267-10.743 0.017 5.532 1.560-19.616 0.008 4.282 1.042-17.607 0.044
LVI 3.905 1.406-10.849 0.009 3.946 1.135-13.726 0.031 5.169 1.366-19.566 0.016
Stage < 0.001 < 0.001 < 0.001
Stage 2 vs. stage 1 4.093 1.095-15.305 0.036 1.684 0.306-9.259 0.549 3.239 0.453-23.166 0.242
Stage 3 vs. stage 1 15.680 4.549-54.042 < 0.001 11.635 3.092-43.781 < 0.001 22.188 4.262-115.513 < 0.001
FIGO 0.017 0.026 0.022
FIGO II vs. I 2.968 0.741-11.881 0.124 2.725 0.499-14.886 0.247 4.097 0.426-39.412 0.222
FIGO III vs. I 7.103 1.809-27.894 0.005 9.087 1.706-48.393 0.010 16.047 1.826-141.002 0.012
FIGO I or II vs. III 3.946 1.445-10.777 0.007 5.268 1.533-18.105 0.008 6.977 1.788-27.226 0.005
Silverberg 0.001 0.003 0.004
Silverberg 2/3 vs. 1/3 2.362 0.589-9.472 0.225 1.182 0.238-5.869 0.838 1.189 0.167-8.483 0.863
Silverberg 3/3 vs. 1/3 12.047 2.924-49.636 0.001 14.093 2.628-75.579 0.002 18.098 2.619-125.054 0.003
Silverberg 1 or 2 vs. 3 7.322 2.549-21.034 < 0.001 12.988 2.995-56.326 0.001 16.613 3.201-86.208 0.001
95% CI indicates confidence interval; HR, hazard ratio; LVI, lymphovascular space invasion.

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved. [Link] | 163

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

Parra-Herran et al Am J Surg Pathol  Volume 43, Number 2, February 2019

FIGURE 2. Kaplan-Meier curves stratifying patients with ovarian endometrioid carcinoma on the basis of FIGO (A, C, E) and
Silverberg (B, D, F) grading systems in terms of DFS (A, B), DSS (C, D), and OS (E, F).

(25%) tumors. Of these, 10 cases had a higher FIGO grade Comparative Analysis: Tumor Grade and Patient
compared with their respective Silverberg grade, and 8 Outcome
cases had a higher Silverberg grade compared with their Median follow-up period for the patient cohort was
respective FIGO grade. Figure 1 depicts examples of OEC 62 months (range: 1 to 179 mo). Tumor recurrence after
with concordant and discordant grading. initial surgery was documented in 16 (22%) patients.

164 | [Link] Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

Am J Surg Pathol  Volume 43, Number 2, February 2019 Ovarian Endometrioid Carcinoma Grading System

TABLE 3. Multivariate Cox Regression Model Including FIGO and Silverberg Grading Systems
Disease-free Survival Overall Survival Disease-specific Survival
HR 95% CI P HR 95% CI P HR 95% CI P
FIGO I or II vs. III 1.124 0.225-5.618 0.887 1.348 0.191-9.538 0.765 1.467 0.156-13.809 0.738
Silverberg 1 or 2 vs. 3 6.685 1.299-34.408 0.023 10.382 1.292- 83.387 0.028 12.268 1.092-137.854 0.042
95% CI indicates confidence interval; HR, hazard ratio.

Death of disease occurred in 9 (13%) subjects. Two pa- carcinoma (serous, endometrioid, mucinous, clear cell) has
tients died of other causes. Median DFS was 60.5 months been actively promoted by some authors, in particular the
(1 to 179 mo), and median OS was 61 months (1 to 179 proponents of the Silverberg grading system. This system,
mo). In univariate analysis, both grading systems corre- first proposed in 1998, has been validated as a useful pre-
lated significantly with DFS (FIGO P = 0.01, Silverberg dictor of patient survival, and has been implemented in
P = 0.001), DSS (FIGO P = 0.02, Silverberg P = 0.004), many practices.7,8 Interestingly, the Silverberg grading sys-
and OS (FIGO P = 0.02, Silverberg P = 0.003, Table 2). tem has also been found to provide superior correlation
Patient outcome was also significantly correlated with with the transcriptomic profile obtained from multiple
tumor laterality (unilateral vs. bilateral), lymphovascular published ovarian cancer data sets.9
space invasion, and stage. Despite the above evidence, the prognostic value of
Kaplan-Meier survival analyses for tumor grade universal grading systems has been questioned by some
(FIGO and Silverberg systems) are shown in Figure 2. authors, who reported lack of correlation between grade
Remarkably, curves for Silverberg G1 and G2 OEC and patient outcome, particularly in those with advanced-
overlapped and showed considerable separation from stage tumors.10,11 Furthermore, over the last 2 decades,
patients with Silverberg G3 tumors. Such clustering was our knowledge of ovarian carcinoma pathogenesis has
not seen using FIGO grading: patients with FIGO G2 expanded, and the major ovarian carcinoma types (high-
OECs had intermediate survival between those with FIGO grade serous, low-grade serous, endometrioid, clear
G1 and G3 tumors. Survival for G1 tumors was similarly cell, mucinous) are now regarded as pathogenically and
favorable in both FIGO and Silverberg systems; likewise, morphologically distinct diseases.12–14 Consequently, the
G3 tumors were associated with similar aggressive course, grading approach has shifted to model each histologic
regardless of the grading system. Importantly, the clustering type. For example, tuboovarian serous carcinoma is now
of G1 and G2 in the Silverberg system persisted after divided using a 2-tier grading classification of low versus
5 years of follow-up, whereas the FIGO G2 curve declines high grade, which is now widely accepted as reproducible
after 5 years, approaching the survival pattern of FIGO G3 and congruent with the biology of serous neoplasia.4,5,15
tumors. As regards clear cell carcinoma, grade proposals have
In multivariate comparative analysis of both Sil- been published16; however, outcome is heavily dictated
verberg and FIGO grading systems, only Silverberg by stage, and superior prognostication based on tumor
retained statistical association with DFS, DSS, and OS grading has not been documented.17
(Table 3). However, when adding stage to the analysis, Given the above shifts in grading for serous and
stage was the only factor found to independently correlate clear cell carcinomas, an optimization of the grading ex-
with patient outcome in (Table 4). Analysis restricted to ercise in other subtypes (endometrioid and mucinous)
patients with early-stage disease (FIGO stage 1 or 2) failed becomes relevant. Grading of ovarian endometrioid neo-
to reveal a statistical association between survival and plasia remains based on traditional grading systems,
tumor grade by either system (Table 5). namely Silverberg and FIGO. The latter, an extrapolation
of the FIGO grading scheme for endometrial carcinoma,
has not been rigorously validated in ovarian endometrioid
DISCUSSION tumors. Conversely, the Silverberg system has been shown
Histologic grade of epithelial ovarian cancer is fre- to provide prognostication in all major subtypes including
quently considered for prognostic and therapeutic purposes. endometrioid tumors but has not been compared with the
A universal approach to grade all major types of ovarian FIGO system exclusively in OEC. Our study addresses this

TABLE 4. Multivariate Cox Regression Model Including FIGO Stage, FIGO Grade, and Silverberg Grade
Disease-free Survival Overall Survival Disease-specific Survival
HR 95% CI P HR 95% CI P HR 95% CI P
Stage 1 or 2 vs. 3 6.410 2.223-18.489 0.001 6.529 1.823-23.389 0.004 8.378 2.018-34.771 0.003
FIGO I or II vs. III 0.906 0.120-6.840 0.923 1.070 0.090-12.786 0.957 1.053 0.049-22.801 0.974
Silverberg 1 or 2 vs. 3 5.344 0.692-41.272 0.108 8.086 0.615-106.378 0.112 8.714 0.354-214.635 0.185
95% CI indicates confidence interval; FIGO, International Federation of Obstetrics and Gynecology; HR, hazard ratio.

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved. [Link] | 165

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

Parra-Herran et al Am J Surg Pathol  Volume 43, Number 2, February 2019

TABLE 5. Multivariate Cox Regression Model Including FIGO Stage, FIGO Grade, and Silverberg Grade in Patients With Early-stage
Disease
Disease-free Survival Overall Survival Disease-specific Survival
HR 95% CI P HR 95% CI P HR 95% CI P
Stage 1 vs. 2 2.715 0.561-13.145 0.215 1.264 0.163-9.834 0.823 2.559 0.255-25.699 0.425
FIGO I or II vs. III 1.198 0.140-10.258 0.869 0.741 0.038-14.360 0.843 0.920 0.024-34.771 0.964
Silverberg 1 or 2 vs. 3 2.472 0.216-28.314 0.467 5.249 0.158-174.729 0.354 4.438 0.063-313.747 0.493

gap by comparing the prognostic value of both FIGO and either FIGO or Silverberg in terms of reproducibility remains
Silverberg grading systems in a well-annotated cohort of to be elucidated. Furthermore, whether a combination of both
OECs with an exceptional long-term follow up. This data grading systems in a single algorithm improves prediction also
set provides an opportunity to examine the robustness of remains to be studied. Likewise, incorporation of other mor-
prognostication beyond 5 years from diagnosis. We found phologic variables such as the pattern of stromal invasion may
significant correlation of both grading approaches with be of value.21 These hypotheses will constitute the aim of
DFS, DSS, and OS in univariate analysis. However, the future investigations.
Silverberg system was superior for 2 reasons: (1) it re- It is also important to note that tumor stage showed
tained statistical correlation with survival in a multivariate prognostic value superior to all variables, including grade.
analysis, and (2) it provided a sharper separation of pa- Even when restricting the analysis to patients with early-
tients in terms of outcome (grade 1 and 2 OECs having a stage disease (FIGO stage 1 and 2), multivariate calcu-
significantly better outcome than grade 3 OEC), especially lations did not reveal a statistical association between
beyond 5 years. grading and survival. This may be explained by the low
As a separate tumor category, the definition of OEC number of adverse events in this subset of early-stage tu-
has been refined in the recent literature. For instance, a mors, and by our cohort size. Larger series may show a
significant subset of tumors previously regarded as endo- more significant relationship between grade and patient
metrioid can now be correctly identified as high-grade serous outcome. More importantly, our findings highlight the need
with solid, endometrioid, and transitional (SET) variant to refine tumor grade categories and their definitions,
morphology, thanks to improvements in the morphologic so that they correlate better with the cancer behavior.
evaluation and the use of WT1 immunohistochemistry.18,19 While Silverberg appears to be preferable for this purpose,
While the existence of high-grade OEC has been questioned we foresee that meaningful grading of OEC will be best
by some authors,18 the high-grade OEC category is well achieved through a combination of morphologic, immuno-
characterized in the literature as an aggressive, albeit in- histochemical, and molecular characteristics.
frequent, subset of ovarian endometrioid neoplasia.6,19,20 In summary, as currently defined, Silverberg is a
This highlights the importance of grading in the clinical superior grading approach of OEC in terms of patient
evaluation of OEC, in particular the identification of high- outcome prediction, compared with FIGO. Grade 1 and
grade tumors with potential for adverse behavior. grade 3 OECs showed similar survival in both systems.
In terms of patient outcome analysis, our study re- However, grade 2 OECs by Silverberg have a more in-
vealed that grade 1 (low grade) OECs have similar dolent outcome in contrast to the grade 2 OECs by FIGO,
behavior, regardless of the grading system used. Using which showed an intermediate behavior. Such differences
FIGO grading, grade 2 OECs had survival curves in be- may lead to a more distinct stratification of OEC in low-
tween grade 1 and grade 3 tumors, and therefore behave as risk and high-risk categories when using Silverberg. This
a real “intermediate” category. Of note, OECs with FIGO may potentially allow for grouping of grades 1 and 2
grade 2 and grade 3 tumors had similar survival in uni- OECs as low grade and grade 3 OEC as high grade. As
variate analysis in one recent study.19 In contrast, strat- management algorithms have shifted toward a 2-tier
ification using the Silverberg system resulted in much grading for other types of ovarian cancer, our data suggest
better separation of patients with worse DFS, DSS, and that Silverberg is a preferable approach in the grading
OS (grade 3) from those with more indolent behavior, exercise of OEC.
which clustered closer together in the survival curves
(grade 1 and grade 2 OEC). Thus, the Silverberg system is REFERENCES
more prone to simplification into a 2-tier classification, in 1. Shimizu Y, Kamoi S, Amada S, et al. Toward the development
which grades 1 and 2 OEC can be regarded as “low grade” of a universal grading system for ovarian epithelial carcinoma.
I. Prognostic significance of histopathologic features—problems
and grade 3 OEC as “high grade,” similar to the 2-tier involved in the architectural grading system. Gynecol Oncol. 1998;70:
approach in the clinical management of endometrial 2–12.
cancer. Moreover, Silverberg G2 cases had DFS, DSS, 2. Shimizu Y, Kamoi S, Amada S, et al. Toward the development of a
and OS comparable to G1 throughout the follow-up pe- universal grading system for ovarian epithelial carcinoma: testing of
riod, whereas FIGO G2 clustered with G1 in the first a proposed system in a series of 461 patients with uniform treatment
and follow-up. Cancer. 1998;82:893–901.
5 years, but approached G3 beyond 5 years. 3. Ishioka SI, Sagae S, Terasawa K, et al. Comparison of the usefulness
In our study, we did not address interobserver agreement between a new universal grading system for epithelial ovarian cancer
of the grading exercise by either system. The superiority of and the FIGO grading system. Gynecol Oncol. 2003;89:447–452.

166 | [Link] Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

Am J Surg Pathol  Volume 43, Number 2, February 2019 Ovarian Endometrioid Carcinoma Grading System

4. Battista MJ, Cotarelo C, Almstedt K, et al. Validation of a two-tier 14. Gilks CB, Ionescu DN, Kalloger SE, et al. Tumor cell type can be
grading system in an unselected, consecutive cohort of serous ovarian reproducibly diagnosed and is of independent prognostic significance
cancer patients. Arch Gynecol Obstet. 2016;294:599–606. in patients with maximally debulked ovarian carcinoma. Hum
5. Bodurka DC, Deavers MT, Tian C, et al. Reclassification of serous Pathol. 2008;39:1239–1251.
ovarian carcinoma by a 2-tier system: a Gynecologic Oncology 15. Hannibal CG, Vang R, Junge J, et al. A binary histologic grading
Group Study. Cancer. 2012;118:3087–3094. system for ovarian serous carcinoma is an independent prognostic
6. Parra-Herran C, Lerner-Ellis J, Xu B, et al. Molecular-based factor: a population-based study of 4317 women diagnosed in
classification algorithm for endometrial carcinoma categorizes Denmark 1978-2006. Gynecol Oncol. 2012;125:655–660.
ovarian endometrioid carcinoma into prognostically significant 16. Yamamoto S, Tsuda H, Shimazaki H, et al. Histological grading of
groups. Mod Pathol. 2017;30:1748–1759. ovarian clear cell adenocarcinoma: proposal for a simple and
7. Mayr D, Diebold J. Grading of ovarian carcinomas. Int J Gynecol reproducible grouping system based on tumor growth architecture.
Pathol. 2000;19:348–353. Int J Gynecol Pathol. 2012;31:116–124.
8. Silverberg SG. Histopathologic grading of ovarian carcinoma: 17. Yamamoto S, Kasajima A, Takano M, et al. Validation of the
a review and proposal. Int J Gynecol Pathol. 2000;19:7–15. histologic grading for ovarian clear cell adenocarcinoma: a retro-
9. Yao CQ, Nguyen F, Haider S, et al. The transcriptomic profile of spective multi-institutional study by the Japan Clear Cell Carcinoma
ovarian cancer grading. Cancer Med. 2015;4:56–64. Study Group. Int J Gynecol Pathol. 2011;30:129–138.
10. Matsuno RK, Sherman ME, Visvanathan K, et al. Agreement for 18. Lim D, Murali R, Murray MP, et al. Morphological and
tumor grade of ovarian carcinoma: analysis of archival tissues from immunohistochemical reevaluation of tumors initially diagnosed as
the surveillance, epidemiology, and end results residual tissue ovarian endometrioid carcinoma with emphasis on high-grade
repository. Cancer Causes Control CCC. 2013;24:749–757. tumors. Am J Surg Pathol. 2016;40:302–312.
11. Kommoss S, Schmidt D, Kommoss F, et al. Histological grading in a 19. Assem H, Rambau PF, Lee S, et al. High-grade endometrioid
large series of advanced stage ovarian carcinomas by three widely carcinoma of the ovary: a clinicopathologic study of 30 cases. Am
used grading systems: consistent lack of prognostic significance. J Surg Pathol. 2018;42:534–544.
A translational research subprotocol of a prospective randomized 20. Soyama H, Miyamoto M, Takano M, et al. A pathological study
phase III study (AGO-OVAR 3 protocol). Virchows Arch Int J Pathol. using 2014 WHO criteria reveals poor prognosis of grade 3 ovarian
2009;454:249–256. endometrioid carcinomas. Vivo Athens Greece. 2018;32:597–602.
12. Köbel M, Kalloger SE, Boyd N, et al. Ovarian carcinoma subtypes 21. Chen S, Leitao MM, Tornos C, et al. Invasion patterns in stage
are different diseases: implications for biomarker studies. PLoS Med. I endometrioid and mucinous ovarian carcinomas: a clinicopathologic
2008;5:e232. analysis emphasizing favorable outcomes in carcinomas without
13. Köbel M, Bak J, Bertelsen BI, et al. Ovarian carcinoma histotype destructive stromal invasion and the occasional malignant course of
determination is highly reproducible, and is improved through the carcinomas with limited destructive stromal invasion. Mod Pathol.
use of immunohistochemistry. Histopathology. 2014;64:1004–1013. 2005;18:903–911.

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved. [Link] | 167

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.