The presence of an adnexal mass is a frequent reason for

Transcription

1 GYNAECOLOGY GYNAECOLOGY Risk of Malignancy Index in the Evaluation of Patients With Adnexal Masses Sharon E. Clarke, MD, PhD, 1 Robert Grimshaw, MD, FRCSC, 2 Paula Rittenberg, MD, FRCSC, 2 Katharina Kieser, MD, FRCSC, MSc, 2 James Bentley, MBChB, FRCSC 2 1 Department of Diagnostic Radiology, Dalhousie University, Halifax NS 2 Department of Obstetrics and Gynaecology, Division of Gynaecologic Oncology, Dalhousie University, Halifax NS Abstract Objective: To determine if the risk of malignancy index (RMI) can distinguish between benign and malignant adnexal masses in a population of women referred to a department of gynaecologic oncology for surgical resection of an adnexal mass. Methods: We performed a retrospective review of the medical of charts of 259 consecutive patients. Ninety-six charts did not have data available to calculate the RMI, leaving a total of 163 for review. Three definitions of RMI were compared; each incorporated menopausal status of the patient, ultrasound characteristics of the adnexal mass, and serum CA-125 level. Results: Of the masses resected, 105 were benign and 58 were malignant. The area under the ROC curve for all three definitions of RMI was Using a cut-off of 120, the first RMI definition (RMI 1) had a sensitivity of 72% and a specificity of 87%; the second (RMI 2) had a sensitivity of 76% and a specificity of 81%; and the third (RMI 3) had a sensitivity of 74% and a specificity of 84%. These results are generally in agreement with published values. Conclusion: We have validated the use of RMI to predict the risk of malignancy in a Nova Scotia population of women with adnexal masses. This will aid in more selective referral of patients to specialized oncology centres for cancer surgery, allowing for appropriate management of health care resources and optimization of treatment for women with gynaecological malignancies. Résumé Objectif : Déterminer si l indice de risque de malignité (IRM) permet de distinguer entre les masses annexielles bénignes et malignes, au sein d une population de femmes orientées vers un service d oncologie gynécologique en vue de la résection chirurgicale d une masse annexielle. Méthodes : Nous avons mené une analyse rétrospective des dossiers médicaux de 259 patientes consécutives. Quatre-vingt-seize dossiers ne contenaient pas les données nécessaires au calcul de l IRM, ce qui nous laissait un total de 163 dossiers à analyser. Trois définitions d IRM ont été comparées; chacune d entre elles incorporait l état ménopausique Key Words: Adnexal mass, ovarian cancer, risk of malignancy, ultrasound, CA-125 Competing Interests: None declared. Received on November 21, 2008 Accepted on January 26, 2009 de la patiente, les caractéristiques échographiques de la masse annexielle et le taux sérique de CA-125. Résultats : Cent cinq des masses réséquées étaient bénignes et 58 étaient malignes. La surface sous la courbe FER pour les trois définitions de l IRM était de 0,87. En présence d un seuil de 120, la première définition de l IRM (IRM 1) comptait une sensibilité de 72 % et une spécificité de 87 %; la deuxième (IRM 2) comptait une sensibilité de 76 % et une spécificité de 81 %; et la troisième (IRM 3) comptait une sensibilité de 74 % et une spécificité de 84 %. D ordre général, ces résultats concordent avec les valeurs publiées. Conclusion : Nous avons validé l utilisation de l IRM pour prédire le risque de malignité au sein d une population de femmes de Nouvelle-Écosse présentant des masses annexielles. Cela contribuera à accroître la sélectivité de l orientation des patientes vers des centres oncologiques spécialisés en vue d une chirurgie visant le cancer, permettant ainsi une gestion appropriée des ressources de soins de santé et l optimisation de la prise en charge des femmes qui présentent des tumeurs malignes gynécologiques. J Obstet Gynaecol Can 2009;31(5): INTRODUCTION The presence of an adnexal mass is a frequent reason for a woman to be referred to a gynaecologist. The discrimination between benign and malignant adnexal masses is central to decisions regarding clinical management and surgical planning in such patients. Patients with malignant tumours should be referred to a gynaecological oncologist, as the quality of cytoreductive surgery and surgical staging/lymph node dissection are important prognostic factors in ovarian cancer. 1,2 These specialized surgical procedures require the specific skills and experience provided by gynaecologic oncology surgeons. Furthermore, appropriate and timely referral to a gynaecologic oncologist has been proven to increase survival in patients with ovarian cancer. 3 Conversely, patients believed to have a benign mass requiring surgery are able to have this performed by a general gynaecologist. A standardized method for preoperative identification of probable malignant masses would allow 440 MAY JOGC MAI 2009

2 Risk of Malignancy Index in the Evaluation of Patients With Adnexal Masses Table 1. Definitions of RMI that were compared in this study (see text for explanation) RMI 1 4 RMI 2 9 RMI 3 11 M (menopausal status) M = 1 if premenopausal M = 3 if postmenopausal M = 1 if premenopausal M = 4 if postmenopausal M = 1 if premenopausal M = 3 if postmenopausal US (ultrasound score) U = 0 if no features U = 1 if one feature U=3if 2features U=1if 1feature U=4if 2features U=1if 1feature U=3if 2features optimization of first-line treatment for women with ovarian cancer. A risk of malignancy index would be valuable for the selective referral of relevant patients to specialized oncology centres. Currently, clinical examination, ultrasound assessment, and assays of tumour markers are part of the standard work-up for an adnexal mass. Although none of these indicators alone is very sensitive or specific for detecting malignancy, an index developed by Jacobs et al. 4 incorporates information about the patient s menopausal status and serum CA-125 levels, and ultrasound characteristics of the mass to predict the risk of malignancy with greater sensitivity and specificity than any one factor alone. 5 7 Some of the potential advantages of RMI include rapid triage of patients through the referral system and fewer operations for benign masses being performed by gynaecologic oncologists. Furthermore, there will be less need for re-operation on women not fully staged for early ovarian cancer. If more women are operated on early in the course of cancer, this may lead to increased survival. The RMI is easy to use and has been associated with a high rate of compliance in peripheral centres However, it has not been routinely implemented in North America, despite its ease of use. The purpose of this study was to determine if the RMI can distinguish between benign and malignant adnexal masses in the population of women referred to the Department of Gynaecologic Oncology at the QE II Medical Centre in Halifax (a tertiary level referral centre) for surgical resection RMI ROC US ABBREVIATIONS risk of malignancy index receiver operating characteristic ultrasound score of an adnexal mass. The retrospective validation of the RMI is the first step towards the long-term aim of this study, which is to implement a risk scoring system in Nova Scotia. METHODS Using a database that collects information prospectively about all women seen in the Department of Gynaecologic Oncology at the QE II Medical Centre, we identified the medical records of women referred for management of an adnexal mass over a one-year period, beginning January 2004 and ending December A total of 259 charts were found in the database. Sixty-one of these charts were either incomplete or unavailable for review. In 30 of the remaining charts, the patient did not have an ultrasound examination, and in five the serum CA-125 value was not recorded. Therefore, a total of 163 charts were included in our retrospective review. Three definitions of RMI were compared; each incorporated menopausal status, ultrasound characteristics of the adnexal mass, and serum CA-125 level. The original definition of RMI (RMI 1) was published by Jacob et al. in This was modified by Tingulstad et al. in 1996, and we refer to this modified definition as RMI 2. 9 RMI 2 has been reported to be more sensitive than RMI 1 6,9 ; therefore, it was included in our study to facilitate a direct comparison of these two definitions. Tingulstad et al. 11 described a third definition of RMI in 1999, which we refer to as RMI 3. The RMI is calculated using the formula RMI = M US serum CA 125. M refers to the patient s menopausal status, US refers to the ultrasound score, and serum CA 125 is the assayed level expressed in ku/l. For calculation of RMI 1 and RMI 3, M = 1 for premenopausal and M = 3 for postmenopausal; for RMI 2, M = 1 for premenopausal and M = 4 for postmenopausal. Postmenopausal status was assigned if the woman had had more than one year of amenorrhea or was over 50 years of age if she had undergone hysterectomy. The ultrasound score is calculated according to the presence or absence of the following features: multilocularity of the ovarian mass, solid areas, bilaterality, ascites, and extra-ovarian tumours (metastases). For calculation of RMI 1, the US is 0 if none of these features is present, 1 if one feature is present, and 3 if two or more features are present. In the RMI 2 definition, the ultrasound score is 1 if none or one of these features is present, and 4 if two or more features are present. For calculation of RMI 3, the ultrasound score is 1 if none or one of the features is present, and 3 if two or more features are present. The definition of each RMI is shown in Table 1. At our institution, a normal CA-125 level is defined as < 21 ku/l. More commonly, 35 ku/l is used as a cut-off point. 12 Since MAY JOGC MAI

3 GYNAECOLOGY Table 2. Breakdown of histological diagnoses in 163 patients with adnexal masses Histology Benign 105 Cystadenofibroma 16 Serous cystadenoma 16 Teratoma/dermoid 15 Mucinous cystadenoma 13 Ovarian cyst 10 Fibroid 8 Endometrioma 6 Fibrothecoma 6 Peritoneal pseudocyst 3 Paratubal cyst 1 Hydrosalpinx 2 Ovarian torsion 2 Brenner tumour 3 Smooth muscle cell tumour 1 Necrotizing granulomatous salpingitis 1 Normal ovaries 2 Malignant 58 Serous cystadenocarcinoma 27 Mucinous cystadenocarcinoma 3 Endometriod carcinoma 3 Primary peritoneal carcinoma 3 Metastases 2 Clear cell carcinoma 1 Borderline Brenner tumour 1 Anaplastic 1 Serous low malignant potential tumour 9 Mucinous low malignant potential tumour 8 Table 3. Cancer staging for the 56 patients with primary peritoneal or ovarian cancer Stage (56 primary ovarian cancers) Stage Ia 12 Stage Ic 8 Stage IIa 2 Stage IIb 1 Stage IIc 2 Stage IIIa 3 Stage IIIb 6 Stage IIIc 18 Stage IV 4 n n there are no international reference preparations for the CA-125 marker, different assays will produce different cut-off values. A value of 21 ku/l in our assay is roughly equivalent to 35 ku/l in commonly used assays. 12 For analysis purposes, tumours of low malignant potential were classified as malignant because it was considered ideal for these tumours to be surgically managed by a gynaecologic oncologist. The histopathologic diagnosis was considered to be the true outcome. When the tumour was malignant, the surgical stage was recorded. The study protocol was approved by the Capital District Health Authority s research ethics review board. RESULTS Of the 163 adnexal masses, 105 were benign, and 58 were malignant. The average age of the study participants was 55 years (range 14 91). The average age of women with benign masses was 54 ± 16 years, and the average age of women with malignant masses was 57 ± 14 years. Sixty-one patients were premenopausal, and of these, 46 had benign lesions and 15 had malignant lesions. The number of postmenopausal women was 102, of whom 59 had benign lesions, and 43 had malignant adnexal masses. The average serum CA-125 level in the benign and malignant groups was 32 ku/l (range ) and 1471 ku/l (range: ) respectively. In the benign group, 25 masses had no ultrasound features, 52 masses had one ultrasound feature, and 28 masses had two ultrasound features. No benign masses had three, four, or five ultrasound features. In the malignant group, two masses had no ultrasound features, 25 masses had one ultrasound feature, 24 masses had two ultrasound features, six masses had three ultrasound features, one mass had four features, and none had five ultrasound features. Seventeen adnexal masses were tumours of low malignant potential; as described above, these lesions were classified as malignant for the purposes of ROC analysis. Classifications by histological results and by tumour staging for malignant lesions are shown in Table 2 and Table 3, respectively. The area under the ROC curve for RMI 1, RMI 2, and RMI 3 was 0.87 (Figures 1 3). Using a threshold (cut-off) value of 120, RMI 1 had a sensitivity of 72% and a specificity of 87%; RMI 2 had a sensitivity of 76% and a specificity of 81%; and RMI 3 had a sensitivity of 74% and a specificity of 84%. The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of each definition of RMI obtained with a cut-off value of 120 are listed in Table 4. Using a threshold of 120 for RMI 1, RMI 2, and RMI 3, causes of false positive results included five 442 MAY JOGC MAI 2009

4 Risk of Malignancy Index in the Evaluation of Patients With Adnexal Masses Figure 1. ROC curve for RMI 1 Figure 2. ROC curve for RMI 2 fibrothecomas/fibromas, one fibroma involving periappendiceal adipose tissue, two benign ovarian cysts, one Brenner tumour, one mature teratoma, one serous cystadenoma, one serous cystadenofibroma, one degenerated leiomyoma, and one case of necrotizing granulomatous salpingitis. For RMI 2, additional false positives included two cases of endometriosis, an additional cystadenofibroma, one serous cystadenoma, one case of normal ovaries, and one combined Brenner/mucinous neoplasm. RMI 3 had 17 false positives, and the additional causes were one case of endometriosis, one case of normal ovaries, and one serous cystadenoma. For RMI 1, RMI 2, and RMI 3 using a threshold of 120, causes of false negative results included four mucinous tumours of low malignant potential (one stage Ia and three stage Ic), two serous tumours of low malignant potential (stage Ia, stage IIb), five serous cystadenocarcinomas (one stage Ia, two stage IIa, two stage IIc), one borderline MAY JOGC MAI

5 GYNAECOLOGY Figure 3. ROC curve for RMI 3 Table 4. Results of ROC analysis for the three definitions of RMI described in the text; the values are given for an RMI threshold value of 120 Brenner tumour (stage Ia), one clear cell carcinoma, and one endometrioid carcinoma arising from the right ovary (stage Ic). Additional false negatives for RMI 1 were one serous cystadeno- carcinoma and one mucinous cystadenocarcinoma (stage Ia). The additional false negative for RMI 3 was a mucinous cystadenocarcinoma (stage Ia). DISCUSSION RMI 1 % (n) RMI 2 % (n) RMI 3 % (n) Accuracy 82 (133/163) 79 (129/163) 80 (131/163) Sensitivity 72 (42/58) 76 (44/58) 74 (43/58) Specificity 87 (91/105) 81 (85/105) 84 (88/105) PPV 75 (42/56) 69 (44/64) 72 (43/60) NPV 85 (91/107) 86 (85/99) 85 (88/103) PPV: positive predictive value; NPV: negative predictive value The results of our study are in agreement with previously published findings. Because a serum CA-125 level of 21 KU/L using our assay is equivalent to a level of 35 KU/L in most other assays, a cut-off RMI of 120 in our study is equivalent to a cut-off RMI of 200 in other studies. In two similar retrospective studies, the sensitivity and specificity for RMI 1 using a cut-off of > 200 were found to be 74% and 92% respectively 13 and 73% and 91% respectively. 7 These values correspond well with our values of 72% for sensitivity and 87% for specificity. Similarly, in previous studies using RMI 2, sensitivity and specificity were 74% and 89%, 13 71% and 89%, 10 and 76% and 82%. 7 Our values for RMI 2 were a sensitivity of 76% and a specificity of 81%. Previously reported sensitivity and specificity figures for RMI 3 at a cut-off of 200 are 74% and 91% 7 ; again, these values are similar to our findings of a sensitivity of 74% and a specificity of 84%. In a study describing prospective external validation of several prognostic models, 14 the area under the ROC curve for RMI 1 was 0.86 and for RMI 2 was 0.83, similar to our values of 0.87 for both RMI 1 and RMI 2. RMI 1, RMI 2, and RMI 3 had comparable overall accuracies and the same area under the ROC curve; however, at the cut-off of 120, RMI 2 was slightly more sensitive and less specific than RMI 1 or RMI 3. RMI 2 is likely to be chosen for use in a prospective trial because the benefit to patients of detecting more cancers (higher sensitivity) is felt to outweigh the negative impact on resources as a consequence of gynaecologic oncologists operating on more benign cases (lower specificity). A significant limitation of this study is the retrospective design. Since all of the cases included in our analysis were referred to our centre for consultation with a gynaecologic oncologist, the proportion of malignant lesions (36%) would presumably be higher than in patients seen in 444 MAY JOGC MAI 2009

6 Risk of Malignancy Index in the Evaluation of Patients With Adnexal Masses peripheral centres, where a larger percentage of women would present with benign adnexal masses. The efficacy of using RMI in the Nova Scotia community has yet to be studied; the sensitivity and specificity would not be expected to change, but the negative predictive value and positive predictive value would depend upon the prevalence of disease. In a recent study by Bailey et al., 8 the effectiveness of RMI for identifying cases of ovarian malignancy in peripheral centres for subsequent referral to a specialized cancer centre for treatment was assessed. RMI performed favourably in this situation, and the sensitivity of the algorithm for detecting ovarian malignancy remained high despite the fact that serum CA-125 levels and ultrasound findings were obtained from a diverse range of laboratories and sonographers. We expect that RMI would perform similarly well in our population of women. Although all of the cases included in our study had the serum CA-125 assay performed at our institution, a significant number of the patients had their ultrasound examinations performed at community hospitals. The majority of ultrasound reports contained adequate information about the morphology of adnexal masses for inclusion in the RMI algorithm. Although the ultrasound reports were generally of good quality, standardization of reporting would be useful to ensure inclusion of the criteria required for RMI calculation. This is particularly important in light of the results of a multi-centre prospective trial (International Ovarian Tumor Analysis), which concluded that pattern recognition of ultrasound images by a trained observer was superior to serum CA-125 for discrimination between benign and malignant adnexal masses. 15 CONCLUSION We have validated the use of RMI in a Nova Scotian population of women with adnexal masses. This result will aid in more selective referral of relevant patients to specialized oncology centres for cancer surgery, permitting the appropriate management of health care resources and optimization of treatment for women with gynaecological malignancies. REFERENCES 1. Giede KC, Kieser K, Dodge J, Rosen B. Who should operate on patients with ovarian cancer? An evidence-based review. Gynecol Oncol 2005;99(2): ACOG Committee Opinion. The role of the generalist obstetrician-gynecologist in the early detection of ovarian cancer. Gynecol Oncol 2002;100: Gillis CR, Hole DJ, Still RM, Davis J, Kave SB. Medical audit, cancer registration, and survival in ovarian cancer. Lancet 1991;337: Jacobs I, Oram D, Fairbanks J, Turner J, Frost C, Grudzinskas JG. A risk of malignancy index incorporating CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol 1990;97: Davies AP, Jacobs I, Woolas R, Fish A, Oram D. The adnexal mass: benign or malignant? Evaluation of a risk of malignancy index. Br J Obstet Gynaecol 1993;100: Morgante G, la Marca A, Ditto A, De Leo V. Comparison of two malignancy risk indices based on serum CA125, ultrasound score and menopausal status in the diagnosis of ovarian masses. Br J Obstet Gynaecol 1999;106: Manjunath AP, Pratapkumar Sujatha K, Vani R. Comparison of three risk of malignancy indices in evaluation of pelvic masses. Gynecol Oncol 2001;81: Bailey J, Tailor A, Naik R, Lopes A, Godfrey K, Hatem HM, et al. Risk of malignancy index for referral of ovarian cancer cases to a tertiary center: does it identify the correct cases? Int J Gynecol Cancer 2006;16: Tingulstad S, Hagen B, Skjeldestad FE, Onsrud M, Kiserud T, Halvorsen T, et al. Evaluation of a risk of malignancy index based on serum CA125, ultrasound findings and menopausal status in the pre-operative diagnosis of pelvic masses. Br J Obstet Gynaecol 1996;103(8): Andersen ES, Knudsen A, Rix P, Johansen B. Risk of malignancy index in the preoperative evaluation of patients with adnexal masses. Gynecol Oncol 2003;90: Tingulstad S, Hagen B, Skjeldestad FE, Halvorsen T, Nustad K, Onsrud M. The risk-of-malignancy index to evaluate potential ovarian cancers in local hospitals. Obstet Gynecol 1999;94: Sibley PE. BR-MA, GI-MA and OM-MA: Immunoassays for the tumor markers CA15 3, CA19 9, and CA125. Diagnostic Products Corporation Limited. 13. Aslam N, Tailor A, Lawton F, Carr J, Savvas M, Jurkovic D. Prospective evaluation of three different models for the pre-operative diagnosis of ovarian cancer. BJOG 2000;107: Mol BW, Boll D, De Kanter M, Heintz AP, Sijmons EA, Oei SG, et al. Distinguishing the benign and malignant adnexal mass: an external validation of prognostic models. Gynecol Oncol 2001;80: Van Calster B, Timmerman D, Bourne T, Testa AC, Van Holsbeke C, Domali E, et al. Discrimination between benign and malignant adnexal masses by specialist ultrasound examination versus serum CA-125. J Natl Cancer Inst 2007;99: MAY JOGC MAI