ORIGINAL ARTICLE
Grading Ovarian Serous Carcinoma Using a Two-Tier System Anais Malpica, MD,* Michael T. Deavers, MD,* Karen Lu, MD,† Diane C. Bodurka, MD,† Edward N. Atkinson, MD,‡ David M. Gershenson, MD,† and Elvio G. Silva, MD*
Abstract: In this study, we evaluate a two-tier system for grading ovarian serous carcinoma. This system is based primarily on the assessment of nuclear atypia with the mitotic rate used as a secondary feature. The study included 50 cases of low-grade ovarian serous carcinoma and 50 cases of high-grade ovarian serous carcinoma retrieved from the files of the Department of Pathology at the University of Texas M. D. Anderson Cancer Center from a 28-year period. Cases assigned to the low-grade category were characterized by the presence of mild to moderate nuclear atypia. As a secondary feature, they tended to show up to 12 mitoses per 10 high power fields (HPFs), whereas those in the high-grade category had marked nuclear atypia and as a secondary feature more than 12 mitoses per 10 HPFs. For comparison, the tumors were also graded using the Shimizu/Silverberg and the FIGO grading systems. Patients in the low-grade ovarian serous carcinoma group ranged in age from 19 to 75 years (mean 41.7 years) while patients in the high-grade ovarian serous carcinoma group ranged in age from 27 to 76 years (mean 55 years). All of the cases except one were advanced FIGO stage. Using the Shimizu/Silverberg system, the low-grade ovarian serous carcinoma cases were distributed as follows: grade 1, 47 cases; grade 2, 3 cases. Using the FIGO grading system, 35 cases were grade 1 and 15 cases were grade 2. Regarding the high-grade ovarian serous carcinoma group using the Shimizu/Silverberg system, 14 of the cases were grade 2 and 36 cases were grade 3. Using the FIGO grading system, 1 case was grade 1, 38 cases were grade 2, and 11 cases were grade 3. Most of the patients in both groups were treated with total abdominal hysterectomy and bilateral salpingo-oophorectomy and also received cisplatinum-based chemotherapy. On follow-up, 37 patients in the low-grade ovarian serous carcinoma group had died of disease at a median 4.2 years after diagnosis compared with 46 patients in the high-grade ovarian serous carcinoma group who died of disease at a median of 1.7 years. Eight patients in the low-grade ovarian serous carcinoma group and 4 patients in the high-grade ovarian serous carcinoma group were alive with disease at median follow-ups of 4.3 and 3.85 years, respectively. Four patients with low-grade serous carcinoma were alive without evidence of disease after a followFrom the Departments of *Pathology, †Gynecologic Oncology, and ‡Biomathematics, University of Texas M. D. Anderson Cancer Center, Houston, TX. Presented in part at the 91st Annual Meeting of the United States and Canadian Academy of Pathology, Chicago, IL, February 23 to March 1, 2002. Reprints: Anais Malpica, MD, Department of Pathology, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030 (e-mail:
[email protected]). Copyright © 2004 by Lippincott Williams & Wilkins
496
up that ranged from 4.4 to 22.6 years (median 6.85 years), and one died of other causes 14 years after the diagnosis of her ovarian tumor. On multivariate analysis, residual tumor and tumor grade based on the M. D. Anderson two-tier system for grading ovarian serous carcinoma were found to be significant independent prognostic factors (P = 0.003 and 0.04, respectively). Of interest, 60% of the low-grade ovarian serous carcinomas in this study were associated with a serous neoplasm of low malignant potential, whereas this association was present in only 2% of the high-grade ovarian serous carcinomas. This finding could reflect a difference in the pathogenesis of ovarian serous carcinomas of different grades. In summary, there is usually a good correlation between the two-tier grading system herein presented and the Shimizu/Silverberg and the FIGO grading systems. Because this system is based on defined criteria that are easy to follow and because it involves only two diagnostic categories, it should provide better reproducibility in the grading of ovarian serous carcinoma. However, additional studies are required to validate these statements. Key Words: serous carcinoma, ovary, grading system, high grade, low grade (Am J Surg Pathol 2004;28:496–504)
H
istologic grade has been shown to be an important prognostic factor in cases of ovarian serous carcinoma,1,2,4,5,7,9,11,13,20–24 but there is no single system universally used for grading these tumors. Current grading systems typically analyze architectural pattern, nuclear/cytologic atypia, mitotic index, or a combination of these features.12,19–21 No criteria exist for defining precise thresholds between grade, which could lead to interobserver variability in the assignment of grade, and there is variation between the systems regarding the designation of different categories and the number of strata (ie, well, moderately, poorly differentiated vs. 1, 2, 3 vs. 1, 2, 3, 4 vs. low and high grade).12,15–21 The purpose of this study was to evaluate the use of a two-tier system for grading ovarian serous carcinoma. This system is based primarily on the assessment of nuclear atypia with the mitotic rate used as a secondary feature.
MATERIALS AND METHODS The study group was comprised of 100 cases of ovarian serous carcinoma retrieved from the files of the Department of Pathology at the University of Texas M. D. Anderson Cancer Am J Surg Pathol • Volume 28, Number 4, April 2004
Am J Surg Pathol • Volume 28, Number 4, April 2004
Center from a 28 year period (1972–2000). The diagnosis of ovarian carcinoma showing exclusively a serous morphology was the criterion of the search to retrieve these cases. Only cases with hematoxylin and eosin-stained slides and clinical information available were included in this study. Cases displaying other histologic types (ie, endometrioid, transitional, and undifferentiated) as defined by the WHO classification were excluded.17 Included were 50 cases of high-grade serous carcinoma and 50 cases of low-grade serous carcinoma graded according to the criteria of the system herein presented. The presence of frank destructive invasion of the ovarian stroma was required to differentiate low-grade serous carcinomas from serous tumors of low malignant potential. Hematoxylin and eosin slides were reviewed in all cases (range, 1–50 slides/case; mean 10.5 slides/case). The following clinical and pathologic parameters were evaluated: patient age, tumor size, stage of disease (according to the FIGO staging system), association with a serous neoplasm of low malignant potential and
FIGURE 1. A: Low-grade serous carcinoma characterized by uniform nuclei. B: Higher magnification demonstrating uniformity of chromatin pattern. © 2004 Lippincott Williams & Wilkins
Grading Ovarian Serous Carcinoma
percentage of this component in the ovarian tumor, presence of a micropapillary/cribriform pattern measuring at least 5.0 mm in greatest dimension in the associated serous tumor of low malignant potential (if this was present), nuclear features, mitotic index, surgical treatment, residual disease after surgery, and adjuvant therapy. Regarding the nuclear features, we evaluated the variation in nuclear size and shape (uniformity vs. pleomorphism) and designated “low-grade” nuclear features as mild to moderate nuclear atypia characterized by the presence of mostly uniform round or oval nuclei with evenly distributed chromatin; in some cases, nucleoli were conspicuous (Figs. 1, 2). “High-grade” nuclear features were characterized by variation (ⱖ3:1) in nuclear size and shape with irregular chromatin and the variable presence of macronucleoli (Figs. 3, 4). The nuclear features recorded were those seen in the epithelial areas with the greatest degree of atypicality. The mitotic index (number of mitoses per 10 HPFs) was evaluated
FIGURE 2. A: Low-grade serous carcinoma demonstrating uniformity of nuclei with visible nucleoli. B: Higher magnification shows mild nuclear size variation and uniform chromatin pattern.
497
Malpica et al
FIGURE 3. A: High-grade serous carcinoma displaying marked nuclear pleomorphism and conspicuous mitotic activity. B: Multinucleated tumor giant cells commonly seen in highgrade serous carcinoma.
in the most mitotically active area of the epithelial component of the tumor; we counted 4 sets of 10 consecutive HPFs (400×, Olympus BX40 microscope) and recorded the highest mitotic count per 10 HPFs. For comparison, tumors were also graded using the systems proposed by Shimizu and Silverberg,20–22 and by FIGO (International Federation of Gynecology and Obstetrics).12 In the Shimizu/Silverberg system, ovarian carcinomas are graded 1, 2, or 3 based on the dominant architectural pattern, the degree of cytologic atypia, and the mitotic index (Table 1). The FIGO system is based on architectural features, and the grade of the tumor depends on the ratio of glandular or papillary structures to solid tumor. In this system, grade 1 is equivalent to 50% solid growth. Clinical information was obtained from the patients’ charts, and information about follow-up was obtained in all
498
Am J Surg Pathol • Volume 28, Number 4, April 2004
FIGURE 4. A: High-grade serous carcinoma composed of small (“micro”) papillae. B: Higher magnification demonstrates mitotic activity and nuclear pleomorphism.
cases either from the patients’ charts or from the referring physicians. Follow-up duration ranged from 1 year to 22.6 years in the low-grade serous carcinoma group and from 0.2 year to 11.2 years in the high-grade serous carcinoma group. The survival duration was measured from the date of diagnosis to the date of death; if the patient was still alive, the survival was measured to the date of last contact. Median survival was calculated using the Kaplan-Meier method and the curves were compared using the log-rank and Wilcoxon tests. Multivariate analysis was performed using Cox proportional hazards model. Contingency tables were used to compare the three grading systems. In some cases, epithelial corrected mitotic index was obtained.
RESULTS The clinicopathologic features of the low-grade and high-grade ovarian serous carcinoma cases are summarized in © 2004 Lippincott Williams & Wilkins
Am J Surg Pathol • Volume 28, Number 4, April 2004
TABLE 1. Shimizu-Silverberg Proposed Universal Grading System for Invasive Ovarian Carcinoma
Score
Predominant Architectural Pattern
Cytologic Atypia*
Mitotic Figures per 10 HPF†
1 2 3
Glandular Papillary Solid
Slight Moderate Marked
0–9 10–24 ⱖ25
Total score: 3 to 5 = grade 1; 6 or 7 = grade 2; 8 or 9 = grade 3. *Evaluated in the area of the tumor with the greatest atypicality and comprising at least 50% of a low power (4×, 10×) microscopic field. Slight: relatively uniform vesicular nuclei (variation in diameter ⱕ2:1), a low nuclear: cytoplasmic ratio, with no chromatin clumping or prominent nucleoli. Moderate: intermediate variation in nuclear size (between 2:1 and 4:1) and shape, nucleoli recognizable but small, more chromatin clumping with no bizarre cells. Marked: shape and nuclear size variation (>4:1) are marked, a high nuclear: cytoplasmic ratio, prominent chromatin clumping, thick nuclear membranes, and large eosinophilic nucleoli, bizarre cells may be present. †Counted in the most active area.
Tables 2 and 3. Patient age ranged from 19 to 75 years (mean 41.7 years) in the low-grade serous carcinoma group and from 27 to 76 years (mean 55 years) in the high-grade serous carcinoma group. Except for 1 case of stage I disease in the lowgrade serous carcinoma group, all of the tumors were advanced stage: stage II, 1 case; stage III, 40 cases; and stage IV, 8 cases in the low-grade group, and stage III, 42 cases and stage IV, 8 cases in the high-grade group. The tumors were bilateral in 37 (74%) of 50 cases in the low-grade group and in 42 (84%) of 50 cases in the high-grade group. The tumor size ranged from 2 cm to 18 cm (mean 8 cm) in the low-grade group and from 1.4 cm to 22 cm (mean 7.4 cm) in the high-grade group. An associated serous neoplasm of low malignant potential was present in 30 (60%) of the 50 cases of low-grade serous carcinoma and in only 1 (2%) of the 50 cases of high-grade serous carcinoma. When present in the low-grade serous carcinoma group, the serous neoplasm of low malignant potential component involved 10% to 99% of the ovarian tumor. A micropapillary/cribriform pattern measuring at least 5 mm was noted in most of the cases, 28 (93%) of the 30 cases with a serous neoplasm of low malignant potential component in patients with low-grade serous carcinoma. In contrast, in the single case of high-grade serous carcinoma of the ovary with a serous neoplasm of low-malignant component, the low malignant potential component represented 2 2 2 >2 >2 2 >2 2 2 >2 2 2 2 >2 >2 >2 >2
