Ovarian cancer remains the leading cause of death from gynecologic cancers in the United States. It was estimated that in 2009 there would be 21,554 new cases of ovarian cancer, and that 14,600 women would die as a result of disease.1 Case–control studies have indicated that women with ovarian cancer commonly experience a pattern of symptoms that include bloating, pelvic/abdominal pain, difficulty eating/feeling full quickly, and urinary urgency or frequency.2,3 These symptoms were found to be more commonly associated with ovarian cancer, when they were newly experienced, and occurred more that 12 times per month.4 Recently, consensus groups have recommended that women who experience symptoms suggestive of ovarian cancer should undergo a complete physical examination, and in certain cases, transvaginal ultrasonography and CA 125 testing. Although the majority of patients with these symptoms will not have ovarian cancer, those who do will require complete surgical staging and aggressive tumor debulking to maximize their chances of survival.5–7 In this regard, it is important to establish risk profiles of patients with ultrasonographically confirmed adnexal tumors so that they can receive appropriate treatment and, when necessary, referral for specialty cancer care. The authors postulate that the addition of ultrasonographically generated tumor morphology to patient demographics and serum biomarker profiles could improve prediction of malignancy in a clinically detectable adnexal mass. This investigation was undertaken to estimate the accuracy of a combination of patient demographics, ultrasonographically generated tumor morphology, and serum CA 125 values in predicting risk of malignancy in adnexal masses.
MATERIALS AND METHODS
This investigation was undertaken after approval from the University of Kentucky Human Subjects Institutional Review Board. Study participants were women referred to the University of Kentucky–Markey Women's Cancer Center with a diagnosis of an adnexal mass on pelvic examination who underwent surgery at this institution from 2001 to 2008.
The following demographic data were obtained for all study patients: age, height, weight, gravidity, family history of breast or ovarian cancer, personal history of cancer, and menopausal status. Postmenopausal was defined as the absence of menses for a minimum of 12 months. Family history was considered positive if the patient had a first-degree relative (ie, mother, sister, or daughter) or a second-degree relative (ie, grandmother, granddaughter, aunt, or niece) with documented ovarian or breast cancer. All women underwent pelvic examination, transvaginal ultrasonography, and serum CA 125 determination within 2 weeks before surgery. Transvaginal ultrasonography was performed using General Electric (Milwaukee, WI) Logic 400 or Voluson ProV ultrasound units with a 5-mHz vaginal probe as described previously.8 Tumor dimensions from ultrasonographic images were recorded, and tumor morphology was classified as cystic, solid, or complex (containing both solid and cystic components). All ultrasonograms were reviewed by at least one of the authors. All tumors classified as cystic were unilocular, whereas cystic tumors with septations were included in the complex group. Ascites was defined as free fluid more than 60 mL in the abdomen/pelvis confirmed by ultrasonography. Serum CA 125 determinations were performed using a two-site sandwich paramagnetic particle chemiluminescent immunoenzymatic assay with a normal value less than 35 units/mL. Serum samples with values exceeding 5,000 units/mL were diluted to end point for a final result.
After surgical removal, the dimensions of each tumor were recorded, and frozen section histologic evaluation was performed. Tumors were classified histologically according to the World Health Organization system. Patients with a histologic diagnosis of ovarian malignancy underwent immediate tumor debulking and surgical staging according to the International Federation of Gynecology and Obstetrics system. Data were entered into a MEDLOG database (MEDLOG Systems, Crystal Bay, NV) and exported into an Excel (Microsoft Corp., Redmond, WA) spreadsheet for analysis using WinSTAT (R. Fitch Software, Bad Krozingen, Germany), SPSS (SPSS, Inc., Chicago, IL), and PC-SAS with the Enterprise Miner statistical software (SAS Institute, Inc., Cary, NC).
Proportions were compared using χ2 statistics or Fisher exact tests. Means were compared using two-sample t tests. Statistical significance was determined at the .05 level. Multivariable analyses used the classification and regression tree procedure, a nonparametric method that defines or accepts cut points and uses training and validation sets to optimize rules for the analysis.9 The training set is formed by splitting the entire sample in half after stratification for malignant cases and benign controls.
Between July 2001 and December 2008, 399 patients referred to the outpatient clinic of the University of Kentucky–Markey Women's Cancer Center for evaluation of an adnexal mass on pelvic examination were included in this investigation. Four patients had their first CA 125 determination after surgery and were excluded from further evaluation. Demographic data, biomarker profiles, and tumor characteristics of the patients evaluated are listed in Table 1. Fifty-five percent of patients were postmenopausal and 40% were aged 55 years or older. Sixty-four patients (16%) had a family history of breast cancer, 48 patients (12%) had a family history of ovarian cancer, and one patient was BRCA1 positive. Two hundred thirty-six masses (60%) were ultrasonographically complex (Fig. 1A), 123 (31%) were cystic (Fig. 1B), and 36 (9%) were solid (Fig. 1C). Five of the 236 complex adnexal masses (2.1%) were cystic ovarian tumors with thick septa but no solid areas. Radiologic evidence of ascites was present in 54 patients (14%). Serum CA 125 level was elevated (more than 35 units/mL) in 148 patients (38%) and was more than 120 units/mL in 87 patients (22%).
At the time of surgery, 264 patients (67%) were found to have benign ovarian tumors, 118 patients (30%) had ovarian cancer, and 13 patients (3%) had ovarian tumors of borderline malignancy. The stage distribution of the patients with ovarian tumors of borderline malignancy and ovarian cancer was as follows: stage I, n=38; stage II, n=17; stage III, n=74; and stage IV, n=2. The most common cell types of ovarian malignancy were adenocarcinoma, followed by serous cystadenocarcinoma, mucinous cystadenocarcinoma, and clear-cell carcinoma.
The relationship of demographic, biomarker, and tumor variables to risk of malignancy in patients with an adnexal mass is presented in Table 2. Variables statistically related to risk of malignancy were tumor morphology, ascites, serum CA 125 level, tumor size, tumor bilaterality, menopausal status, and age.
Tumor morphology from ultrasonographically generated images was related directly to risk of malignancy. There were 236 complex adnexal masses, and 120 (51%) were malignant. None of the five complex masses with septal morphology without solid areas were malignant. There were 36 solid adnexal masses, and 11 (32%) were malignant. In contrast, there were 123 cystic adnexal masses, and none were ovarian tumors of borderline malignancy or invasive ovarian cancers (P<.001). The finding of purely cystic morphology in an adnexal mass was associated with a negative predictive value (NPV) for malignancy of 100%.
Fifty-four patients with an adnexal mass had radiologically confirmed ascites, and all of these patients had invasive epithelial ovarian cancer (stage IC, n=2; stage IIC, n=1; stage IIIC, n=49; and stage IV, n=2). Therefore, the finding of documented ascites in a patient with a complex or solid adnexal mass had a positive predictive value (PPV) for malignancy of 100%.
Serum CA 125 values were related directly to risk of malignancy in women with an adnexal mass. Only 19 (7.7%) of 247 patients with a normal serum CA 125 value (less than 35 units/mL) had ovarian cancer. Conversely, 13 of 83 patients (34.2%) with a serum CA 125 value of 35–59 units/mL, 17 of 23 patients (73.9%) with a serum CA 125 value of 60–120 units/mL, and 82 of 87 patients (86.8%) with a CA 125 value more than 120 units/mL had borderline or malignant ovarian tumors (P<.001). Serum CA 125 values related to specific histologic diagnoses in all patients with complex and solid adnexal masses are listed in Table 3. The only benign histologic finding consistently associated with an elevated serum CA 125 value was ovarian endometriosis. Thirteen (48%) of 27 women with an ovarian endometrioma had an elevated serum CA 125 value (more than 35 units/mL), and 5 (18%) had a serum CA 125 value more than 60 units/mL. There were 114 women with other benign complex or solid adnexal masses, and only 1 patient (with an ovarian fibroma) had a CA 125 level more than 60 units/mL. As expected, serum CA 125 values were related directly to stage of disease in patients with ovarian malignancies. Serum CA 125 values were elevated (more than 35 units/mL) in 30.7% of patients with an ovarian tumor of borderline malignancy, 77.2% of patients with stage I and stage II ovarian cancer, and 98.6% of patients with stage IIIC and IV ovarian cancer. All 54 patients with ascites had an elevated (more than 35 units/mL) serum CA 125 level, and 52 (96.2%) had a CA 125 level more than 60 units/mL.
Risk of malignancy in an adnexal mass was significantly higher in women older than 55 years than in younger women (P<.001), in postmenopausal women compared with premenopausal women (P<.001), in women with bilateral compared with unilateral ovarian tumors (P<.01), and in women whose adnexal masses were more than 10 cm in diameter compared with smaller tumors (P<.001). A family history of ovarian cancer or breast cancer in a woman with an adnexal mass was not associated statistically with an increased risk of malignancy in the population studied.
Although many variables were correlated with malignancy, only a small number had acceptable positive or NPVs (Table 4). Classification and regression tree multivariable analysis considered age, gravidity, postmenopausal status, weight, family history of ovarian cancer or breast cancer, tumor morphology, ascites, tumor bilaterality, maximum tumor diameter, and CA 125 value. This analysis found the most accurate significance of interactions to declare a high risk of malignancy if a patient had an adnexal mass with complex or solid morphology and a serum CA 125 value more than 35 units/mL. The statistics for the training and validation sets indicated uniformly high performances for sensitivity, specificity, and predictive values across both the training and validation sets.
Statistical parameters associated with the high-risk definition are listed in Table 5. Using the stated high-risk parameters resulted in a sensitivity of 30.8% for ovarian tumors of borderline malignancy, 77.3% for early stage (I and II) ovarian cancer, and 98.6% for advanced stage (III and IV) ovarian cancer. Increasing the cutoff value of CA 125 from 35 units/mL to 60 units/mL and keeping the other parts of the definition the same increased specificity from 92.4% to 96.6% but lowered sensitivity from 84.7% to 80.9% and missed 13 patients with stage I or stage II ovarian cancer. Therefore, a cutoff value of 35 units/mL for CA 125 was used in the final high-risk definition. This definition correctly identified 34 of 44 patients with stage I and stage II ovarian cancer and 93 of 94 patients with stage III and stage IV ovarian cancer. Also, it correctly excluded 244 of the 264 patients with benign ovarian tumors.
The observation that the occurrence of certain symptoms may precede the clinical diagnosis of ovarian cancer has resulted in the recommendation that women experiencing the recent onset of bloating, pelvic/abdominal pain, feeling full quickly after eating, or urinary urgency/frequency should consult a physician and undergo a complete physical examination. Women having a clinically palpable abnormality in the pelvis or those with persisting symptoms in the presence of a normal pelvic examination are advised to undergo transvaginal ultrasonography and CA 125 testing.
The findings of this investigation indicate that analysis of data concerning patient demographics, tumor morphology obtained from ultrasonographic images, and serum CA 125 levels is useful in estimating the risk of malignancy in women with an adnexal mass. For example, the risk of neoplasia in unilocular cystic ovarian tumors is very low. This morphologic pattern was present in 123 (31%) of 395 adnexal tumors, and no patient had either a borderline or an invasive ovarian malignancy. This confirms the observation by Roman and colleagues10 who, in a summary of the literature, reported a 0.7% rate of malignancy in 569 unilocular cystic ovarian tumors 10 cm or less in diameter. Similarly, Modesitt and colleagues11 followed more than 3,200 women with unilocular cystic ovarian tumors less than 10 cm in diameter for an average of 6 years without operative intervention. No patient developed ovarian cancer, and 69% of these tumors resolved spontaneously over the period of observation. There is no doubt that some unilocular cystic ovarian tumors grow to significant size and require surgical removal. However, the risk of malignancy even in larger cystic lesions is low, particularly in women with a normal CA 125 level. In the present study, there were 27 unilocular cystic ovarian tumors more than 10 cm diameter, and all were benign.
In contrast, adnexal masses with complex or solid morphology are associated with a significant risk of malignancy. There were 272 patients with a complex or solid adnexal mass, and 131 (48%) had an ovarian malignancy. These women form the basis of a high-risk group for ovarian cancer. In the present investigation, documented ascites in a woman with a complex or solid adnexal mass was uniformly predictive of ovarian cancer. There were 54 patients with this finding, and all had epithelial ovarian cancer. These results are consistent with the findings of Im and colleagues12 who, in a multiinstitutional review, reported that 79% of postmenopausal women with a clinically detectable pelvic mass and ascites had an ovarian malignancy.
The use of serum CA 125 as a method of predicting risk of malignancy in patients with a clinically detectable pelvic mass was suggested in 2002 by a joint publication of the American College of Obstetricians and Gynecologists and the Society of Gynecologic Oncologists.13 This report stated that 69.8% of postmenopausal women with an adnexal mass and a serum CA 125 value more than 35 units/mL had an ovarian malignancy and that an elevated serum CA 125 value in a patient with a clinically detectable pelvic mass could be used as one indication for patient referral for subspecialty care. In the present investigation, more than three fourths of women with a complex or solid adnexal mass and a CA 125 value more than 35 units/mL had either borderline or invasive ovarian cancer.
When evaluating a number of variables, including patient demographics, tumor morphology, and CA 125 levels, as predictors of malignancy, multivariable classification and regression tree analysis defined high risk as women with a complex or solid adnexal mass and a serum CA 125 value of more than 35 units/mL. In the population studied, this definition of high risk was associated with a PPV of 84.7% and a NPV of 92.4% and correctly identified 34 (77.3%) of 44 patients with stage I or stage II ovarian cancer as well as 73 of 74 patients (98.6%) with stage III or stage IV ovarian cancer. Thus, including tumor morphology significantly increases predictive values beyond the PPV of 24.3% for stage I or II ovarian cancer and 56.8% for stage III or IV ovarian cancer associated with the original American College of Obstetricians and Gynecologists/Society of Gynecologic Oncologists high-risk criteria.14 In the present study, raising the cutoff value of CA 125 from 35 units/mL to 60 units/mL in the definition of high risk would have increased specificity for identifying ovarian cancer cases but would have lowered sensitivity and resulted in missing a significant number of patients with early-stage disease. This is important because several studies have reported as high as a 24% survival advantage for patients with early-stage ovarian cancer who received comprehensive surgical staging and treatment by a gynecologic oncologist.15,16
Although family history of ovarian cancer or breast cancer was not statistically associated with an increased risk of ovarian cancer in this investigation, a hereditary cancer risk assessment was not routinely performed on every patient enrolled in this study. Clearly, women with a strong family history of ovarian or breast cancer or women who are members of populations with a high prevalence of “founder mutations” should undergo genetic testing,17 and this information should be taken into consideration in determining optimal treatment for a patient with an adnexal mass.
As an increasing number of women who have symptoms suggestive of ovarian cancer are evaluated, clinicians will be asked to determine which patients are at significant risk for ovarian cancer. Data from the present investigation suggest that the combination of ultrasonographic tumor morphology and serum CA 125 value improves the discrimination of women at risk of ovarian cancer from those with benign adnexal lesions. These findings should be helpful in determining which patients can be followed without surgery, which patients are likely to have a benign ovarian tumor, and which patients are at high risk of ovarian malignancy and should be referred for subspecialty care.
1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun M. Cancer statistics 2009. CA Cancer J Clin 2009;59:225–49.
2. Goff BA, Mandel LS, Melancon CH, Muntz HG. Frequency of symptoms of ovarian cancer in women presenting to primary care clinics. JAMA 2004;291:2705–12.
3. Smith LH, Morris CR, Yasmeen S, Parikh-Patel A, Cress RD, Romano PS. Ovarian cancer: can we make the clinical diagnosis earlier? Cancer 2005;104:1398–407.
4. Goff BA, Mandel L, Drescher CW, Urban N, Gough S, Schurman KM, et al. Development of an ovarian cancer symptom index. Cancer 2007;109:221–7.
5. Bristow RE, Tomacruz RS, Armstrong DK, Trimble EL, Montz FJ. Survival effect of maximal cytoreductive surgery for advanced ovarian carcinoma during the platinum era: a meta-analysis. J Clin Oncol 2002;20:1248–59.
6. Eisenhauer EL, Abu-Rustum NR, Sonoda Y, Aghajanian C, Barakat RR, Chi DS. The effect of maximal surgical cytoreduction on sensitivity to platinum-taxane chemotherapy and subsequent survival in patients with advanced ovarian cancer. Gynecol Oncol 2008;108:276–81.
7. Eisenhauer EL, Abu-Rustum NR, Sonoda Y, Levine DA, Poynor EA, Aghajanian C, et al. The addition of extensive upper abdominal surgery to achieve optimal cytoreduction improves survival in patients with stages IIIC-IV epithelial ovarian cancer. Gynecol Oncol 2006;103:1083–90.
8. van Nagell JR, DePriest PD, Ueland FR, DeSimone CP, Cooper AL, McDonald AM, et al. Ovarian cancer screening with annual transvaginal sonography: findings of 25,000 women screened. Cancer 2007;109:1887–96.
9. Breiman L, Friedman J, Olshen R, Stone C. Classification and regression trees. New York (NY): Wadsworth; 1984.
10. Roman LD. Small cystic pelvic masses in older women: is surgical removal necessary? Gynecol Oncol 1998;69:1–2.
11. Modesitt SC, Pavlik EJ, Ueland FR, DePriest PD, van Nagell JR Jr. Risk of malignancy in unilocular ovarian cystic tumors less than 10 cm in diameter: a long-term follow-up study. Obstet Gynecol 2003;102:594–9.
12. Im S, Gordon A, Buttin B, Leath CA III, Gostout BS, Shah C, et al. Validation of referral guidelines for women with pelvic masses. Obstet Gynecol 2005;105:34–41.
13. The role of the generalist obstetrician-gynecologist in the early detection of ovarian cancer. ACOG Committee Opinion No. 280. American College of Obstetricians and Gynecologists. Obstet Gynecol 2002;100:1413–6.
14. Dearking AC, Aletti G, McGree M, Weaver AL, Sommerfield M, Cliby WA. How relevant are ACOG and SGO guidelines for referral of an adnexal mass. Obstet Gynecol 2007;110:841–8.
15. Le T, Adolph A, Krepart G, Lotocki R, Heywood MS. The benefits of comprehensive surgical staging in the management of early-stage epithelial ovarian carcinoma. Gynecol Oncol 1992;47:223–7.
16. Suh-Burgman E. Long-term outcomes following conservative surgery for borderline tumor of the ovary: a large population-based study. Gynecol Oncol 2006;103:841–7.
© 2010 The American College of Obstetricians and Gynecologists
17. Hereditary breast and ovarian cancer syndrome. ACOG Committee on Practice Bulletins. Gynecol Oncol 2009; 113:6–11.