Portal de Periódicos da CAPES

Cyclooxygenase-2 overexpression in ovarian endometriomas is associated with higher risk of recurrence

2008; Elsevier BV; Volume: 91; Issue: 4 Linguagem: Inglês

10.1016/j.fertnstert.2008.01.070

1556-5653

Lei Yuan, Fanghua Shen, Yuan Lu, Xishi Liu, Sun‐Wei Guo,

Inflammatory mediators and NSAID effects

Cyclooxegenase-2 expression was evaluated by immunohistochemistry in endometrioma tissue samples taken from 109 patients, of whom 53 had recurrence by 30 months after surgery and 56 did not. Cyclooxegenase-2 overexpression, along with previous medical treatment of endometriosis and the presence of adhesion, is predictive of recurrence of ovarian endometrioma within 30 months after surgery, with a resultant sensitivity of 72.5% and a specificity of 72.4%, suggesting that there are intrinsic and identifiable biomarkers that confer recurrence risk. Cyclooxegenase-2 expression was evaluated by immunohistochemistry in endometrioma tissue samples taken from 109 patients, of whom 53 had recurrence by 30 months after surgery and 56 did not. Cyclooxegenase-2 overexpression, along with previous medical treatment of endometriosis and the presence of adhesion, is predictive of recurrence of ovarian endometrioma within 30 months after surgery, with a resultant sensitivity of 72.5% and a specificity of 72.4%, suggesting that there are intrinsic and identifiable biomarkers that confer recurrence risk. Although surgery is the treatment of choice for the management of endometriosis, recurrence poses a formidable challenge: 40%–45% of patients have relapse of the disease 5 years after the primary surgery and require further surgery (1Garry R. The effectiveness of laparoscopic excision of endometriosis.Curr Opin Obstet Gynecol. 2004; 16: 299-303Crossref PubMed Scopus (123) Google Scholar, 2Evers J.L. Dunselman G.A. Land J.A. Bouckaert P.X. Management of recurrent endometriosis.in: Couinho E. Spinola P. DeMoura L.H. Progress in the management of endometriosis. Partheon, London1995: 291-297Google Scholar). The causes for recurrence are unclear, and there is no biomarker with any prognostic significance. The revised American Fertility Society (rAFS) staging system is not predictive with regard to recurrence risk (3Roberts C.P. Rock J.A. The current staging system for endometriosis: does it help?.Obstet Gynecol Clin North Am. 2003; 30: 115-132Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar). Numerous epidemiologic studies on risk factors for recurrence have been published, but their findings unfortunately are often conflicting.To date, scant attention has been paid to the identification of biomarkers, preferably available around the time of surgery, that are predictive of recurrence. The availability of these biomarkers would accord identification of patients with a high recurrence risk, allowing for possible intervention to postpone or even prevent recurrence altogether.One possible biomarker is the inducible proinflammatory enzyme cyclooxygenase-2 (COX-2), owing to its apparently critical role in perpetuating the positive feedback loop in inducing inflammation and proliferation in endometriosis (4Bulun S.E. Lin Z. Imir G. Amin S. Demura M. Yilmaz B. et al.Regulation of aromatase expression in estrogen-responsive breast and uterine disease: from bench to treatment.Pharmacol Rev. 2005; 57: 359-383Crossref PubMed Scopus (431) Google Scholar, 5Guo S.W. Nuclear factor-kappaB (NF-kappaB): an unsuspected major culprit in the pathogenesis of endometriosis that is still at large?.Gynecol Obstet Invest. 2006; 63: 71-97Crossref PubMed Scopus (124) Google Scholar). Cyclooxygenase-2 overexpression in ectopic endometrium has been noted (6Ota H. Igarashi S. Sasaki M. Tanaka T. Distribution of cyclooxygenase-2 in eutopic and ectopic endometrium in endometriosis and adenomyosis.Hum Reprod. 2001; 16: 561-566Crossref PubMed Scopus (271) Google Scholar, 7Chishima F. Hayakawa S. Sugita K. Kinukawa N. Aleemuzzaman S. Nemoto N. et al.Increased expression of cyclooxygenase-2 in local lesions of endometriosis patients.Am J Reprod Immunol. 2002; 48: 50-56Crossref PubMed Scopus (139) Google Scholar), and it is correlated with severity of dysmenorrhea (8Matsuzaki S. Canis M. Pouly J.L. Wattiez A. Okamura K. Mage G. Cyclooxygenase-2 expression in deep endometriosis and matched eutopic endometrium.Fertil Steril. 2004; 82: 1309-1315Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar) and nonmenstrual chronic pelvic pain in women with endometriosis (9Buchweitz O. Staebler A. Wulfing P. Hauzman E. Greb R. Kiesel L. COX-2 overexpression in peritoneal lesions is correlated with nonmenstrual chronic pelvic pain.Eur J Obstet Gynecol Reprod Biol. 2006; 124: 216-221Abstract Full Text Full Text PDF PubMed Scopus (25) Google Scholar). It has been shown recently that COX-2 regulates survival, migration, and invasion of endometriotic cells through multiple mechanisms (10Banu S.K. Lee J. Speights Jr., V.O. Starzinski-Powitz A. Arosh J.A. Cyclooxygenase-2 regulates survival, migration and invasion of human endometriotic cells through multiple mechanisms.Endocrinology. 2008; 149: 1180-1189Crossref PubMed Scopus (117) Google Scholar).Not surprisingly, COX-2 selective inhibitors have been shown to prevent implantation of eutopic endometrium to ectopic sites (11Matsuzaki S. Canis M. Darcha C. Dallel R. Okamura K. Mage G. Cyclooxygenase-2 selective inhibitor prevents implantation of eutopic endometrium to ectopic sites in rats.Fertil Steril. 2004; 82: 1609-1615Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar), suppress the growth of endometriosis xenografts (12Ozawa Y. Murakami T. Tamura M. Terada Y. Yaegashi N. Okamura K. A selective cyclooxygenase-2 inhibitor suppresses the growth of endometriosis xenografts via antiangiogenic activity in severe combined immunodeficiency mice.Fertil Steril. 2006; 86: 1146-1151Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar), induce regression of autologous endometrial grafts (13Laschke M.W. Elitzsch A. Scheuer C. Vollmar B. Menger M.D. Selective cyclo-oxygenase-2 inhibition induces regression of autologous endometrial grafts by down-regulation of vascular endothelial growth factor-mediated angiogenesis and stimulation of caspase-3-dependent apoptosis.Fertil Steril. 2007; 87: 163-171Abstract Full Text Full Text PDF PubMed Scopus (92) Google Scholar), and to be effective in the management of pain related to endometriosis (14Cobellis L. Razzi S. De Simone S. Sartini A. Fava A. Danero S. et al.The treatment with a COX-2 specific inhibitor is effective in the management of pain related to endometriosis.Eur J Obstet Gynecol Reprod Biol. 2004; 116: 100-102Abstract Full Text Full Text PDF PubMed Scopus (119) Google Scholar). However, Fagotti et al. (15Fagotti A. Ferrandina G. Fanfani F. Legge F. Lauriola L. Gessi M. et al.Analysis of cyclooxygenase-2 (COX-2) expression in different sites of endometriosis and correlation with clinico-pathological parameters.Hum Reprod. 2004; 19: 393-397Crossref PubMed Scopus (53) Google Scholar) reported that COX-2 expression does not correlate with clinicopathologic characteristics and symptoms of patients with endometriosis (15Fagotti A. Ferrandina G. Fanfani F. Legge F. Lauriola L. Gessi M. et al.Analysis of cyclooxygenase-2 (COX-2) expression in different sites of endometriosis and correlation with clinico-pathological parameters.Hum Reprod. 2004; 19: 393-397Crossref PubMed Scopus (53) Google Scholar). The same team also reported that the increased COX-2 expression is associated with lower recurrence risk in patients receiving complete ablation of severe endometriosis (16Fanfani F. Fagotti A. Ferrandina G. Bifulco G. Legge F. Lorusso D. et al.Increased cyclooxygenase-2 expression is associated with better clinical outcome in patients submitted to complete ablation for severe endometriosis.Hum Reprod. 2005; 20: 2964-2968Crossref PubMed Scopus (8) Google Scholar).These contradictory reports prompted us to further investigate the relationship between COX-2 expression and the risk of recurrence of endometriosis. To this end we selected 109 women with ovarian endometriomas, of whom 53 had recurrence and 56 did not, and compared their COX-2 expression by immunohistochemistry.Solely on the basis of recurrence status, we selected 109 patients from a pool of 710 followed-up patients with histologically confirmed ovarian endometriomas undergoing conservative or semi-radical (defined as surgical removal of cysts plus hysterectomy with preservation of bilateral or unilateral ovary) surgery by either laparotomy or laparoscopy at Shanghai OB/GYN Hospital from 2003 to 2004. Information was collected regarding age at surgery, body mass index at surgery, results of pelvic examinations, type of surgery, mode of surgery (conservative or semi-radical), complaint of dysmenorrhea or infertility, presence of adenomyosis or myoma, if any, laterality of endometrioma, size of endometrioma (defined as the diameter in centimeters of the largest cyst), presence of adhesion or not, rAFS scores and stage, postoperative medication use, improvement of symptomatology, and time to recurrence or duration of follow-up if censored. The menstrual phase in which the patient was at the time of surgery also was determined according to days elapsed since the last period.We defined the recurrence of endometrioma, 2 months after surgery, as [1] the presence of ovarian cysts of ≥3 cm in diameter, along with characteristic echoes as detected by ultrasonography for two consecutive menstrual cycles (17Exacoustos C. Zupi E. Carusotti C. Rinaldo D. Marconi D. Lanzi G. et al.Staging of pelvic endometriosis: role of sonographic appearance in determining extension of disease and modulating surgical approach.J Am Assoc Gynecol Laparosc. 2003; 10: 378-382Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar), with or without the recurrence of dysmenorrhea or pelvic pain requiring medical intervention, or as [2] the presence of de novo ovarian endometrioma as confirmed by histologic examination after a second surgery.For the recurrent group, the length of time to recurrence ranged from 2 to 29.3 months (median = 11.7 months). For the nonrecurrent group, the length of follow-up ranged from 32.1 to 54.2 months (median = 37.5 months). The shortest follow-up duration in the latter group exceeded the longest recurrence-free period of the former, thus ensuring that the latter group had enough opportunity for recurrence yet none occurred. With the exception of previous use of endometriosis-related medication and the complaint of dysmenorrhea, the two groups were remarkably similar in terms of age at surgery, menstrual phase at surgery, rAFS stage, and other characteristics considered.We also included endometrial tissue samples from 15 women with tubal infertility or surgically diagnosed benign ovarian cysts, but without endometriosis, adenomyosis, or myoma. The ages of women in this control group ranged from 22 to 48 years (mean [SD] = 33.3 [7.1] years). Eight women (53.5%) were in the proliferative phase, whereas the rest were in the secretory phase. This study was approved by the institutional ethics review board of Shanghai OB/GYN Hospital.We retrieved archived, formalin-fixed, paraffin-embedded tissue blocks from these patients. Serial 4-μm sections were obtained from each block, with the first resultant slide being stained for hematoxylin and eosin to confirm pathologic diagnosis and the subsequent slides stained for COX-2. Routine deparaffinization, rehydration, and immunohistochemistry procedures were performed. For antibodies we used a mouse monoclonal antibody against COX-2 (Shanghai Chang-dao BioTech Company, Shanghai, China), diluted to 1:50, and a secondary antibody, Supervision™ Universal (Anti-Mouse/Rabbit) Detection Reagent, horseradish peroxidase (HRP) (Shanghai Chang-dao BioTech Company).Cytoplasmic immunoreactivity to COX-2, resulting in a brownish color, was detected in both endometrial epithelial and stromal cells but mostly in glandular epithelial cells. Staining was evaluated with a nomogram as reported by Ota et al. (6Ota H. Igarashi S. Sasaki M. Tanaka T. Distribution of cyclooxygenase-2 in eutopic and ectopic endometrium in endometriosis and adenomyosis.Hum Reprod. 2001; 16: 561-566Crossref PubMed Scopus (271) Google Scholar) that incorporates both the proportion of positive-staining cells and the staining intensity.To evaluate the effect of COX-2 expression level and other possible factors on the risk of recurrence, a logistic regression model was used. Probability (P) values of <.05 were considered statistically significant. All computations were made with R 2.5.1 (18Inhaka R. Gentleman R.R. R: a language for data analysis and graphics.J Comput Graph Stat. 1996; 5: 1923-1927Google Scholar). Because the sensitivity and specificity were calculated from the same data when the classification rule was derived, we used leave-one-out cross-validation to recompute these parameters.The mean of the COX-2 staining scores in the recurrent group was significantly higher than that in the nonrecurrent group (P=.00004; Fig. 1). The mean staining score in the control group was significantly lower than in either the recurrent or the nonrecurrent group (P=.0000004 and P=.002, respectively).The area under the curve of the receiver operating characteristic (ROC) curve for classifying recurrent and nonrecurrent patients using the COX-2 staining score was 0.73, indicating that the performance of the classification system is fair. On the basis of the ROC plot, we found that classifying patients with a COX-2 staining score of ≥4 as recurrent seems to be a reasonable trade-off between sensitivity (62.3%) and specificity (76.8%). Hence we recoded the COX-2 staining score as 0 or 1, depending on whether it was ≤3 or >3, and carried out a logistic regression analysis of the dichotomous COX-2 score along with 13 other variables, including the rAFS score, and the interaction term of the menstrual phase and COX-2 staining score (to account for possible difference in COX-2 expression between the two menstrual phases).The analysis identified higher COX-2 expression level (P=.00006), previous use of medication (P=.04), and the presence of adhesion (P=.03) as the three significant risk factor for recurrence. Under this model, the sensitivity and specificity were 72.5% and 72.4%, respectively, indicating some improvement over the classification based solely on the COX-2 staining cutoff point. Leave-one-out cross-validation gave the average sensitivity and specificity as identical to the estimated parameters. Removing the COX-2 staining score would drop the sensitivity and specificity to 56.7% and 64.3%, respectively.Endometriosis is an inflammatory condition characterized by elevated cytokine levels in peritoneal fluid (19Wu M.H. Sun H.S. Lin C.C. Hsiao K.Y. Chuang P.C. Pan H.A. et al.Distinct mechanisms regulate cyclooxygenase-1 and -2 in peritoneal macrophages of women with and without endometriosis.Mol Hum Reprod. 2002; 8: 1103-1110Crossref PubMed Scopus (95) Google Scholar, 20Keenan J.A. Chen T.T. Chadwell N.L. Torry D.S. Caudle M.R. IL-1 beta, TNF-alpha, and IL-2 in peritoneal fluid and macrophage-conditioned media of women with endometriosis.Am J Reprod Immunol. 1995; 34: 381-385Crossref PubMed Scopus (154) Google Scholar, 21Cheong Y.C. Shelton J.B. Laird S.M. Richmond M. Kudesia G. Li T.C. et al.IL-1, IL-6 and TNF-alpha concentrations in the peritoneal fluid of women with pelvic adhesions.Hum Reprod. 2002; 17: 69-75Crossref PubMed Google Scholar), all of them being inducers of COX-2 expression in peritoneal macrophages (22Wu M.H. Wang C.A. Lin C.C. Chen L.C. Chang W.C. Tsai S.J. Distinct regulation of cyclooxygenase-2 by interleukin-1beta in normal and endometriotic stromal cells.J Clin Endocrinol Metab. 2005; 90: 286-295Crossref PubMed Scopus (116) Google Scholar). Therefore, COX-2 overexpression in ectopic endometrium may perpetuate and even amplify the positive feedback loop in endometriosis, leading to increased proliferation and inflammation. Cyclooxygenase-2 overexpression may also enhance cellular invasiveness (23Tsujii M. DuBois R.N. Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase 2.Cell. 1995; 83: 493-501Abstract Full Text PDF PubMed Scopus (2131) Google Scholar, 24Tsujii M. Kawano S. DuBois R.N. Cyclooxygenase-2 expression in human colon cancer cells increases metastatic potential.Proc Natl Acad Sci U S A. 1997; 94: 3336-3340Crossref PubMed Scopus (1325) Google Scholar). Because endometriotic cells are known to be invasive (25Gaetje R. Kotzian S. Herrmann G. Baumann R. Starzinski-Powitz A. Invasiveness of endometriotic cells in vitro.Lancet. 1995; 346: 1463-1464Abstract PubMed Scopus (140) Google Scholar), COX-2 overexpression may cause residual endometriotic cells to migrate and invade new sites and establish new colonies. As a result of COX-2 overexpression, the enhanced susceptibility to proliferation, coupled with enhanced angiogenesis and invasiveness, may render patients with COX-2 overexpression at higher risk of recurrence.This scenario suggests that COX-2 may be a therapeutic target for intervention to reduce the risk of recurrence of ovarian endometrioma. Nonsteroidal anti-inflammatory drugs that inhibit COX-2 expression and specific COX-2 inhibitors could potentially be used after surgery for patients with COX-2 overexpression to postpone or even prevent recurrence. Other compounds, such as histone deacetylase inhibitors, which have been shown recently to suppress COX-2 expression in endometrial cells (26Wu Y. Guo S.W. Suppression of IL-1beta-induced COX-2 expression by trichostatin A (TSA) in human endometrial stromal cells.Eur J Obstet Gynecol Reprod Biol. 2007; 135: 88-93Abstract Full Text Full Text PDF PubMed Scopus (51) Google Scholar), may also be viable candidates for intervention.Because the tissue samples used for COX-2 staining were obtained at the time of surgery and the follow-up was carried out afterward, our finding that COX-2 overexpression is associated with increased risk of recurrence of ovarian endometrioma strongly suggests that there are identifiable molecular differences intrinsic to the lesion tissues and that these differences confer different risks of recurrence.How can our results reconcile with those of Fanfani et al. (16Fanfani F. Fagotti A. Ferrandina G. Bifulco G. Legge F. Lorusso D. et al.Increased cyclooxygenase-2 expression is associated with better clinical outcome in patients submitted to complete ablation for severe endometriosis.Hum Reprod. 2005; 20: 2964-2968Crossref PubMed Scopus (8) Google Scholar), who found that increased COX-2 expression is associated with lower recurrence risk? Differences in patient population aside, we first note that in the Fanfani study the difference in recurrence-free survival between COX-2 positive and negative groups in patients with ovarian endometrioma alone actually was statistically not significant. Second, our study has a larger sample size and a more powerful design that contrast the difference, if any, in COX-2 expression between the two groups. Third, whereas Fanfani et al. (16Fanfani F. Fagotti A. Ferrandina G. Bifulco G. Legge F. Lorusso D. et al.Increased cyclooxygenase-2 expression is associated with better clinical outcome in patients submitted to complete ablation for severe endometriosis.Hum Reprod. 2005; 20: 2964-2968Crossref PubMed Scopus (8) Google Scholar) only considered COX-2 expression, we also considered and controlled 13 other possible confounding variables. Failure to control for these confounders may lead to spurious results. Finally, all biological plausibility and published reports strongly suggest that increased COX-2 expression should increase, rather than decrease, the risk of recurrence.Although COX-2 positivity seems to be the first easy-to-use biomarker for recurrence of ovarian endometrioma, there should still be further independent validation and evaluation of its performance. There is still ample room for improvement in sensitivity and specificity. Although surgery is the treatment of choice for the management of endometriosis, recurrence poses a formidable challenge: 40%–45% of patients have relapse of the disease 5 years after the primary surgery and require further surgery (1Garry R. The effectiveness of laparoscopic excision of endometriosis.Curr Opin Obstet Gynecol. 2004; 16: 299-303Crossref PubMed Scopus (123) Google Scholar, 2Evers J.L. Dunselman G.A. Land J.A. Bouckaert P.X. Management of recurrent endometriosis.in: Couinho E. Spinola P. DeMoura L.H. Progress in the management of endometriosis. Partheon, London1995: 291-297Google Scholar). The causes for recurrence are unclear, and there is no biomarker with any prognostic significance. The revised American Fertility Society (rAFS) staging system is not predictive with regard to recurrence risk (3Roberts C.P. Rock J.A. The current staging system for endometriosis: does it help?.Obstet Gynecol Clin North Am. 2003; 30: 115-132Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar). Numerous epidemiologic studies on risk factors for recurrence have been published, but their findings unfortunately are often conflicting. To date, scant attention has been paid to the identification of biomarkers, preferably available around the time of surgery, that are predictive of recurrence. The availability of these biomarkers would accord identification of patients with a high recurrence risk, allowing for possible intervention to postpone or even prevent recurrence altogether. One possible biomarker is the inducible proinflammatory enzyme cyclooxygenase-2 (COX-2), owing to its apparently critical role in perpetuating the positive feedback loop in inducing inflammation and proliferation in endometriosis (4Bulun S.E. Lin Z. Imir G. Amin S. Demura M. Yilmaz B. et al.Regulation of aromatase expression in estrogen-responsive breast and uterine disease: from bench to treatment.Pharmacol Rev. 2005; 57: 359-383Crossref PubMed Scopus (431) Google Scholar, 5Guo S.W. Nuclear factor-kappaB (NF-kappaB): an unsuspected major culprit in the pathogenesis of endometriosis that is still at large?.Gynecol Obstet Invest. 2006; 63: 71-97Crossref PubMed Scopus (124) Google Scholar). Cyclooxygenase-2 overexpression in ectopic endometrium has been noted (6Ota H. Igarashi S. Sasaki M. Tanaka T. Distribution of cyclooxygenase-2 in eutopic and ectopic endometrium in endometriosis and adenomyosis.Hum Reprod. 2001; 16: 561-566Crossref PubMed Scopus (271) Google Scholar, 7Chishima F. Hayakawa S. Sugita K. Kinukawa N. Aleemuzzaman S. Nemoto N. et al.Increased expression of cyclooxygenase-2 in local lesions of endometriosis patients.Am J Reprod Immunol. 2002; 48: 50-56Crossref PubMed Scopus (139) Google Scholar), and it is correlated with severity of dysmenorrhea (8Matsuzaki S. Canis M. Pouly J.L. Wattiez A. Okamura K. Mage G. Cyclooxygenase-2 expression in deep endometriosis and matched eutopic endometrium.Fertil Steril. 2004; 82: 1309-1315Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar) and nonmenstrual chronic pelvic pain in women with endometriosis (9Buchweitz O. Staebler A. Wulfing P. Hauzman E. Greb R. Kiesel L. COX-2 overexpression in peritoneal lesions is correlated with nonmenstrual chronic pelvic pain.Eur J Obstet Gynecol Reprod Biol. 2006; 124: 216-221Abstract Full Text Full Text PDF PubMed Scopus (25) Google Scholar). It has been shown recently that COX-2 regulates survival, migration, and invasion of endometriotic cells through multiple mechanisms (10Banu S.K. Lee J. Speights Jr., V.O. Starzinski-Powitz A. Arosh J.A. Cyclooxygenase-2 regulates survival, migration and invasion of human endometriotic cells through multiple mechanisms.Endocrinology. 2008; 149: 1180-1189Crossref PubMed Scopus (117) Google Scholar). Not surprisingly, COX-2 selective inhibitors have been shown to prevent implantation of eutopic endometrium to ectopic sites (11Matsuzaki S. Canis M. Darcha C. Dallel R. Okamura K. Mage G. Cyclooxygenase-2 selective inhibitor prevents implantation of eutopic endometrium to ectopic sites in rats.Fertil Steril. 2004; 82: 1609-1615Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar), suppress the growth of endometriosis xenografts (12Ozawa Y. Murakami T. Tamura M. Terada Y. Yaegashi N. Okamura K. A selective cyclooxygenase-2 inhibitor suppresses the growth of endometriosis xenografts via antiangiogenic activity in severe combined immunodeficiency mice.Fertil Steril. 2006; 86: 1146-1151Abstract Full Text Full Text PDF PubMed Scopus (65) Google Scholar), induce regression of autologous endometrial grafts (13Laschke M.W. Elitzsch A. Scheuer C. Vollmar B. Menger M.D. Selective cyclo-oxygenase-2 inhibition induces regression of autologous endometrial grafts by down-regulation of vascular endothelial growth factor-mediated angiogenesis and stimulation of caspase-3-dependent apoptosis.Fertil Steril. 2007; 87: 163-171Abstract Full Text Full Text PDF PubMed Scopus (92) Google Scholar), and to be effective in the management of pain related to endometriosis (14Cobellis L. Razzi S. De Simone S. Sartini A. Fava A. Danero S. et al.The treatment with a COX-2 specific inhibitor is effective in the management of pain related to endometriosis.Eur J Obstet Gynecol Reprod Biol. 2004; 116: 100-102Abstract Full Text Full Text PDF PubMed Scopus (119) Google Scholar). However, Fagotti et al. (15Fagotti A. Ferrandina G. Fanfani F. Legge F. Lauriola L. Gessi M. et al.Analysis of cyclooxygenase-2 (COX-2) expression in different sites of endometriosis and correlation with clinico-pathological parameters.Hum Reprod. 2004; 19: 393-397Crossref PubMed Scopus (53) Google Scholar) reported that COX-2 expression does not correlate with clinicopathologic characteristics and symptoms of patients with endometriosis (15Fagotti A. Ferrandina G. Fanfani F. Legge F. Lauriola L. Gessi M. et al.Analysis of cyclooxygenase-2 (COX-2) expression in different sites of endometriosis and correlation with clinico-pathological parameters.Hum Reprod. 2004; 19: 393-397Crossref PubMed Scopus (53) Google Scholar). The same team also reported that the increased COX-2 expression is associated with lower recurrence risk in patients receiving complete ablation of severe endometriosis (16Fanfani F. Fagotti A. Ferrandina G. Bifulco G. Legge F. Lorusso D. et al.Increased cyclooxygenase-2 expression is associated with better clinical outcome in patients submitted to complete ablation for severe endometriosis.Hum Reprod. 2005; 20: 2964-2968Crossref PubMed Scopus (8) Google Scholar). These contradictory reports prompted us to further investigate the relationship between COX-2 expression and the risk of recurrence of endometriosis. To this end we selected 109 women with ovarian endometriomas, of whom 53 had recurrence and 56 did not, and compared their COX-2 expression by immunohistochemistry. Solely on the basis of recurrence status, we selected 109 patients from a pool of 710 followed-up patients with histologically confirmed ovarian endometriomas undergoing conservative or semi-radical (defined as surgical removal of cysts plus hysterectomy with preservation of bilateral or unilateral ovary) surgery by either laparotomy or laparoscopy at Shanghai OB/GYN Hospital from 2003 to 2004. Information was collected regarding age at surgery, body mass index at surgery, results of pelvic examinations, type of surgery, mode of surgery (conservative or semi-radical), complaint of dysmenorrhea or infertility, presence of adenomyosis or myoma, if any, laterality of endometrioma, size of endometrioma (defined as the diameter in centimeters of the largest cyst), presence of adhesion or not, rAFS scores and stage, postoperative medication use, improvement of symptomatology, and time to recurrence or duration of follow-up if censored. The menstrual phase in which the patient was at the time of surgery also was determined according to days elapsed since the last period. We defined the recurrence of endometrioma, 2 months after surgery, as [1] the presence of ovarian cysts of ≥3 cm in diameter, along with characteristic echoes as detected by ultrasonography for two consecutive menstrual cycles (17Exacoustos C. Zupi E. Carusotti C. Rinaldo D. Marconi D. Lanzi G. et al.Staging of pelvic endometriosis: role of sonographic appearance in determining extension of disease and modulating surgical approach.J Am Assoc Gynecol Laparosc. 2003; 10: 378-382Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar), with or without the recurrence of dysmenorrhea or pelvic pain requiring medical intervention, or as [2] the presence of de novo ovarian endometrioma as confirmed by histologic examination after a second surgery. For the recurrent group, the length of time to recurrence ranged from 2 to 29.3 months (median = 11.7 months). For the nonrecurrent group, the length of follow-up ranged from 32.1 to 54.2 months (median = 37.5 months). The shortest follow-up duration in the latter group exceeded the longest recurrence-free period of the former, thus ensuring that the latter group had enough opportunity for recurrence yet none occurred. With the exception of previous use of endometriosis-related medication and the complaint of dysmenorrhea, the two groups were remarkably similar in terms of age at surgery, menstrual phase at surgery, rAFS stage, and other characteristics considered. We also included endometrial tissue samples from 15 women with tubal infertility or surgically diagnosed benign ovarian cysts, but without endometriosis, adenomyosis, or myoma. The ages of women in this control group ranged from 22 to 48 years (mean [SD] = 33.3 [7.1] years). Eight women (53.5%) were in the proliferative phase, whereas the rest were in the secretory phase. This study was approved by the institutional ethics review board of Shanghai OB/GYN Hospital. We retrieved archived, formalin-fixed, paraffin-embedded tissue blocks from these patients. Serial 4-μm sections were obtained from each block, with the first resultant slide being stained for hematoxylin and eosin to confirm pathologic diagnosis and the subsequent slides stained for COX-2. Routine deparaffinization, rehydration, and immunohistochemistry procedures were performed. For antibodies we used a mouse monoclonal antibody against COX-2 (Shanghai Chang-dao BioTech Company, Shanghai, China), diluted to 1:50, and a secondary antibody, Supervision™ Universal (Anti-Mouse/Rabbit) Detection Reagent, horseradish peroxidase (HRP) (Shanghai Chang-dao BioTech Company). Cytoplasmic immunoreactivity to COX-2, resulting in a brownish color, was detected in both endometrial epithelial and stromal cells but mostly in glandular epithelial cells. Staining was evaluated with a nomogram as reported by Ota et al. (6Ota H. Igarashi S. Sasaki M. Tanaka T. Distribution of cyclooxygenase-2 in eutopic and ectopic endometrium in endometriosis and adenomyosis.Hum Reprod. 2001; 16: 561-566Crossref PubMed Scopus (271) Google Scholar) that incorporates both the proportion of positive-staining cells and the staining intensity. To evaluate the effect of COX-2 expression level and other possible factors on the risk of recurrence, a logistic regression model was used. Probability (P) values of <.05 were considered statistically significant. All computations were made with R 2.5.1 (18Inhaka R. Gentleman R.R. R: a language for data analysis and graphics.J Comput Graph Stat. 1996; 5: 1923-1927Google Scholar). Because the sensitivity and specificity were calculated from the same data when the classification rule was derived, we used leave-one-out cross-validation to recompute these parameters. The mean of the COX-2 staining scores in the recurrent group was significantly higher than that in the nonrecurrent group (P=.00004; Fig. 1). The mean staining score in the control group was significantly lower than in either the recurrent or the nonrecurrent group (P=.0000004 and P=.002, respectively). The area under the curve of the receiver operating characteristic (ROC) curve for classifying recurrent and nonrecurrent patients using the COX-2 staining score was 0.73, indicating that the performance of the classification system is fair. On the basis of the ROC plot, we found that classifying patients with a COX-2 staining score of ≥4 as recurrent seems to be a reasonable trade-off between sensitivity (62.3%) and specificity (76.8%). Hence we recoded the COX-2 staining score as 0 or 1, depending on whether it was ≤3 or >3, and carried out a logistic regression analysis of the dichotomous COX-2 score along with 13 other variables, including the rAFS score, and the interaction term of the menstrual phase and COX-2 staining score (to account for possible difference in COX-2 expression between the two menstrual phases). The analysis identified higher COX-2 expression level (P=.00006), previous use of medication (P=.04), and the presence of adhesion (P=.03) as the three significant risk factor for recurrence. Under this model, the sensitivity and specificity were 72.5% and 72.4%, respectively, indicating some improvement over the classification based solely on the COX-2 staining cutoff point. Leave-one-out cross-validation gave the average sensitivity and specificity as identical to the estimated parameters. Removing the COX-2 staining score would drop the sensitivity and specificity to 56.7% and 64.3%, respectively. Endometriosis is an inflammatory condition characterized by elevated cytokine levels in peritoneal fluid (19Wu M.H. Sun H.S. Lin C.C. Hsiao K.Y. Chuang P.C. Pan H.A. et al.Distinct mechanisms regulate cyclooxygenase-1 and -2 in peritoneal macrophages of women with and without endometriosis.Mol Hum Reprod. 2002; 8: 1103-1110Crossref PubMed Scopus (95) Google Scholar, 20Keenan J.A. Chen T.T. Chadwell N.L. Torry D.S. Caudle M.R. IL-1 beta, TNF-alpha, and IL-2 in peritoneal fluid and macrophage-conditioned media of women with endometriosis.Am J Reprod Immunol. 1995; 34: 381-385Crossref PubMed Scopus (154) Google Scholar, 21Cheong Y.C. Shelton J.B. Laird S.M. Richmond M. Kudesia G. Li T.C. et al.IL-1, IL-6 and TNF-alpha concentrations in the peritoneal fluid of women with pelvic adhesions.Hum Reprod. 2002; 17: 69-75Crossref PubMed Google Scholar), all of them being inducers of COX-2 expression in peritoneal macrophages (22Wu M.H. Wang C.A. Lin C.C. Chen L.C. Chang W.C. Tsai S.J. Distinct regulation of cyclooxygenase-2 by interleukin-1beta in normal and endometriotic stromal cells.J Clin Endocrinol Metab. 2005; 90: 286-295Crossref PubMed Scopus (116) Google Scholar). Therefore, COX-2 overexpression in ectopic endometrium may perpetuate and even amplify the positive feedback loop in endometriosis, leading to increased proliferation and inflammation. Cyclooxygenase-2 overexpression may also enhance cellular invasiveness (23Tsujii M. DuBois R.N. Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase 2.Cell. 1995; 83: 493-501Abstract Full Text PDF PubMed Scopus (2131) Google Scholar, 24Tsujii M. Kawano S. DuBois R.N. Cyclooxygenase-2 expression in human colon cancer cells increases metastatic potential.Proc Natl Acad Sci U S A. 1997; 94: 3336-3340Crossref PubMed Scopus (1325) Google Scholar). Because endometriotic cells are known to be invasive (25Gaetje R. Kotzian S. Herrmann G. Baumann R. Starzinski-Powitz A. Invasiveness of endometriotic cells in vitro.Lancet. 1995; 346: 1463-1464Abstract PubMed Scopus (140) Google Scholar), COX-2 overexpression may cause residual endometriotic cells to migrate and invade new sites and establish new colonies. As a result of COX-2 overexpression, the enhanced susceptibility to proliferation, coupled with enhanced angiogenesis and invasiveness, may render patients with COX-2 overexpression at higher risk of recurrence. This scenario suggests that COX-2 may be a therapeutic target for intervention to reduce the risk of recurrence of ovarian endometrioma. Nonsteroidal anti-inflammatory drugs that inhibit COX-2 expression and specific COX-2 inhibitors could potentially be used after surgery for patients with COX-2 overexpression to postpone or even prevent recurrence. Other compounds, such as histone deacetylase inhibitors, which have been shown recently to suppress COX-2 expression in endometrial cells (26Wu Y. Guo S.W. Suppression of IL-1beta-induced COX-2 expression by trichostatin A (TSA) in human endometrial stromal cells.Eur J Obstet Gynecol Reprod Biol. 2007; 135: 88-93Abstract Full Text Full Text PDF PubMed Scopus (51) Google Scholar), may also be viable candidates for intervention. Because the tissue samples used for COX-2 staining were obtained at the time of surgery and the follow-up was carried out afterward, our finding that COX-2 overexpression is associated with increased risk of recurrence of ovarian endometrioma strongly suggests that there are identifiable molecular differences intrinsic to the lesion tissues and that these differences confer different risks of recurrence. How can our results reconcile with those of Fanfani et al. (16Fanfani F. Fagotti A. Ferrandina G. Bifulco G. Legge F. Lorusso D. et al.Increased cyclooxygenase-2 expression is associated with better clinical outcome in patients submitted to complete ablation for severe endometriosis.Hum Reprod. 2005; 20: 2964-2968Crossref PubMed Scopus (8) Google Scholar), who found that increased COX-2 expression is associated with lower recurrence risk? Differences in patient population aside, we first note that in the Fanfani study the difference in recurrence-free survival between COX-2 positive and negative groups in patients with ovarian endometrioma alone actually was statistically not significant. Second, our study has a larger sample size and a more powerful design that contrast the difference, if any, in COX-2 expression between the two groups. Third, whereas Fanfani et al. (16Fanfani F. Fagotti A. Ferrandina G. Bifulco G. Legge F. Lorusso D. et al.Increased cyclooxygenase-2 expression is associated with better clinical outcome in patients submitted to complete ablation for severe endometriosis.Hum Reprod. 2005; 20: 2964-2968Crossref PubMed Scopus (8) Google Scholar) only considered COX-2 expression, we also considered and controlled 13 other possible confounding variables. Failure to control for these confounders may lead to spurious results. Finally, all biological plausibility and published reports strongly suggest that increased COX-2 expression should increase, rather than decrease, the risk of recurrence. Although COX-2 positivity seems to be the first easy-to-use biomarker for recurrence of ovarian endometrioma, there should still be further independent validation and evaluation of its performance. There is still ample room for improvement in sensitivity and specificity. The authors thank Drs. Yuan-hua Li and Xue-juan Cao of Shanghai Cancer Hospital for their technical assistance with the immunohistochemical staining; Dr. Yuan Ji of Shanghai Zhong Shang Hospital, Dr. Qi Che, and Dr. Dongmei Zhu for expert assistance in reading immunohistochemistry results; and Dr. Xianrong Zhou for his assistance.