Abstract
Background: Cyclooxygenase 2 (COX2) expression is up-regulated and associated with adverse prognosis in select types of carcinoma. Although not extensively studied in skeletal or soft tissue sarcoma, expression of COX2 has been described in a variable number of gynecological and non-gynecological leiomyosarcomas. In this study, the prevalence and prognostic implications of COX2 expression in leiomyosarcoma were evaluated further. Materials and Methods: Immunohistochemical stains for COX2 were performed on 33 samples of soft tissue leiomyosarcoma and tested for their association with clinicopathological parameters and patient outcome. Results: COX2 staining was limited to tumor cells surrounding areas of tumor necrosis in 6 cases. There were no statistically significant correlations with the clinicopathological parameters studied, including local recurrence, distant metastasis, or disease-specific death. Conclusion: The low frequency, restricted distribution and absence of prognostic implications of COX2 expression soft tissue leiomyosarcoma suggest that this enzyme may not be a useful pharmacological target in this clinical setting.
Non-gynecological leiomyosarcomas (LMS) arising in the retroperitoneum and deep soft tissues of the extremities generally portend an unfavorable outcome (1-4). Pathological features typically associated with poor prognosis include histological grade, tumor size, and anatomic location (5). Complete surgical resection combined with perioperative adjuvant radiation therapy is currently considered the optimal therapy for LMS (4). However, disease recurrence is common. Although the Scandinavian Sarcoma Group has recently reported 10-year disease-free and overall survival rates of 66% and 49% for patients with localized disease at presentation (5), most other groups have shown much lower rates of disease control (4).
Whereas cyclooxygenase 1 shows ubiquitous and constitutive expression in many human tissues, cyclooxygenase 2 (COX2) expression in non-neoplastic tissue is extremely limited and is typically induced by proinflammatory or mitogenic stimuli such as cytokines, growth factors, oncogenes and endotoxins (6). Since COX2 is up-regulated in a number of different types of carcinoma and correlates with adverse clinical outcome for some tumor types, it has been implicated as being important during carcinogenesis or disease progression, although the mechanisms involved are unknown (7, 8). COX2 may promote neoplasia through inhibition of tumor cell apoptosis, augmentation of angiogenic potential, or suppression of the host immune response via down-regulation of natural killer T-cell function (9). Given the availability of specific inhibitors of COX2 activity, much work has studied this enzyme as a target for chemopreventative therapy (7).
Other than Kaposi's sarcoma and some pediatric sarcomas, COX2 expression has not been studied extensively in mesenchymal neoplasms (10-13). Previous reports of COX2 expression in some cases of uterine and extrauterine LMS prompted us to determine the prevalence of COX2 expression in a larger cohort of LMS of soft tissue (12, 13).
Correlation of clinical features with COX2 staining in soft tissue leiomyosarcoma.
Correlation of pathological findings with COX2 staining in soft tissue leiomyosarcoma.
Materials and Methods
Surgical pathology archives were searched for cases of primary LMS of deep soft tissue. Cases arising in subcutaneous tissue or involving viscera, including the female genital tract, were specifically excluded. Cases that showed nuclear staining for estrogen receptor α or progesterone receptor were considered LMS of gynecological type and excluded from the study. Clinicopathological parameters, including patient age, sex, anatomic site of primary tumor, tumor size, histologic grade, American Joint Committee on Cancer (AJCC) TNM stage (14), presence of and site of metastatic disease, and adjuvant therapy administered were recorded through a retrospective chart review. In addition, intervals from surgical resection to adverse events (defined as local recurrence, metastasis, or tumor-related death) or last clinical follow-up were also recorded.
Association of COX2 staining in soft tissue leiomyosarcoma with patient outcomes.
Thirty-three cases were retrieved. The original hematoxylin and eosin-stained slides were reviewed and blocks were selected for immunohistochemistry. Tumors were graded using the French Federation of Cancer Centers System (15). Formalin-fixed, paraffin-embedded tissue sections were pretreated in EDTA buffer in a steamer for 25 min. Following a 15 min cooling period, slides were incubated with a 1:75 dilution of rabbit polyclonal anti-COX2 antibody (PGHS-2; Oxford Biomedical Research Inc, Oxford, MI, USA) for 1 hour at room temperature. Specific antibody binding was detected using an LSAB2 kit according to the manufacturer's recommendations (Dako, Carpinteria, CA, USA). COX2 staining was scored using the Allred system (16). An Allred staining index ≥2 was considered positive for statistical analysis. Statistical analyses were performed using standard univariate methods (17). Differences in Kaplan-Meier disease-free and overall survival curves were evaluated using the log-rank test statistic (Prism v5; GraphPad Software, Inc. La Jolla, CA, USA). This study was approved by the Institutional Review Board at Vanderbilt University.
Results
Clinicopathological data recorded for 33 cases of primary soft tissue LMS are summarized in Tables I and II. Median age at tumor resection was 58 years (range 22-86 years). Neoadjuvant radiation therapy was administered to 7 patients; one additional patient received preoperative chemotherapy. Median length of follow-up was 23.4 months (range 6 months to 13 years). During this follow-up interval, 14 patients (42%) died of disease and 20 patients (61%) developed recurrent disease. Distant metastasis occurred in 14 patients (42%) a median of 9.6 months (range 4.5 months to 2.5 years) after surgical resection; metastatic disease was detected during initial staging studies in 3 additional patients. Six patients (18%) developed local tumor recurrence (median interval 19.2 months, range 10.4 months to 5.0 years).
Soft tissue leiomyosarcoma is typically negative for COX2 staining by immunohistochemistry (A, H&E, bar=1 mm; B, COX2). In a minority of cases, immunoreactivity is strictly limited to tumor cells surrounding areas of tumor necrosis (C, H&E; D, COX2).
COX2 immunoreactivity was detected in 6 out of 33 cases (18%) and was limited to cells adjacent to areas of ischemic or coagulative tumor necrosis (Table II), corresponding to Allred proportion scores of 1 (4 cases) and 2 (2 cases). Within these foci, however, Allred intensity scores for COX2 were intermediate (5 cases) or strong (1 case) (Figure 1). Accordingly, COX2 staining was seen exclusively in tumors of intermediate to high histological grade and AJCC TNM stage; these associations did not reach statistical significance, however (Table II). Additionally, no significant associations between COX2 staining and adverse events (Table III), tumor size, patient age, sex, anatomic site, or exposure to neoadjuvant therapy were observed in this relatively small study sample. Median overall and disease-free survival for patients with COX2-immunopositive tumors was 34.0 months and 24.3 months, respectively; the corresponding survival intervals for patients with COX2-negative tumors were 37.2 and 22.8 months. Log rank analysis of Kaplan-Meier curves disclosed no significant differences in disease-free or overall survival according to COX2 status, neither for the entire study cohort nor after substratification by presence or absence of tumor necrosis, AJCC TNM stage, or histological grade (data not shown).
Discussion
COX2 expression is associated with negative clinical outcomes in carcinomas from a variety of anatomic sites (13, 18-22). However, its expression has not been studied extensively in mesenchymal neoplasms. Preliminary evidence suggests that while COX2 is expressed in a majority of osteosarcomas, rhabdomyosarcomas, Ewing sarcoma/primitive neuroectodermal tumors, and gastrointestinal stromal tumors, there is correlation to clinicopathological features or prognosis (10, 11, 23). COX2 expression has also been demonstrated in a subset of uterine LMS (13), which led us to evaluate further whether it is detectable in non-gynecological LMS arising in soft tissue.
No soft tissue LMS was diffusely positive for COX2 in our study population. Instead, COX2 immunostaining in LMS was strictly localized to neoplastic cells surrounding areas of tumor necrosis. In a study of 103 soft tissue sarcomas of various histological type, Lassus et al. found that approximately 50% stained for COX2 (12). Similar to our findings, immunoreactivity was limited to areas adjacent to tumor necrosis in over a third of these COX2-positive cases. Of the 19 LMS included in this prior study, 8 (42%) demonstrated COX2 expression (12). However, only one case exhibited strong staining throughout the tumor and two other cases displayed weak but diffuse staining. Thus, 5 out of the 19 LMS in this previous study showed the same spatial expression pattern seen in our samples. Comparisons of COX2-positive and -negative tumors in the current study and that of Lassus et al. disclosed no significant differences in the clinicopathological parameters evaluated, including disease-free or overall survival.
The spatial relationship between COX2 immunoreactivity and coagulative tumor necrosis has been demonstrated previously in some other tumor types. Wolff et al. noted a rim of COX2-immunopositive tumor cells surrounding necrotic carcinomas of the lung (24). A similar pattern of COX2 expression has also been demonstrated in glioma (25-27). One possible explanation for this phenomenon is induction of COX2 expression in tumor cells by regional hypoxia or an inflammatory response (28, 29).
Given the lack of prognostic correlation and limited distribution of expression of COX2 in soft tissue LMS, it is unlikely that such a finding indicates COX2 as a useful pharmacological target in this clinical setting.
Acknowledgements
This study was funded by an intradepartmental clinical translational research award (ACT). The authors thank Jean McClure and Lesley Albert for technical and editorial assistance.
- Received March 26, 2009.
- Revision received May 26, 2009.
- Accepted May 27, 2009.
- Copyright© 2009 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved