The statistical definitions used in this investigation are presented in Table 3. The 76 women with an abnormal screen and invasive epithelial ovarian cancer (n=47), ovarian tumors of low malignant potential (n=15), nonepithelial ovarian cancer (n=5), or cancer metastatic to the ovary (n=9) were considered true-positives. The 447 women with positive screens who had benign ovarian tumors were considered false-positives. The 36,758 patients with negative screens who did not have ovarian cancer were classified as true-negatives. The 12 women who developed ovarian cancer within 1 year of a negative scan were considered false-negatives. Using these data, the sensitivity of transvaginal ultrasonography screening was 86.4%, and the specificity was 98.8%. The positive predictive value of an abnormal screen was 14.53%, and the negative predictive value of a normal screen was 99.97%. After 2006, the positive predictive value increased to 20.17% when the screening algorithm was changed such that surgery was not recommended in women with unilocular cystic ovarian tumors or septated cystic ovarian tumors less than 10 cm in diameter.
Forty-seven women with invasive epithelial ovarian cancer and 15 women with epithelial ovarian tumors of low malignant potential detected by screening are the subjects of the present investigation. The five women with nonepithelial ovarian cancer and the nine women with cancer metastatic to the ovary are not considered further in this study.
Demographic and histologic variables in women with low malignant potential epithelial ovarian tumors are as follows: the mean age of these women was 61.7 years (median, 63 years; range, 48–78 years), the mean weight was 167 pounds (median, 162 pounds; range, 100–240 pounds), and the mean gravidity was 2.0 (median, 2; range, 0–6). Thirteen of these tumors contained both cystic and solid components on ultrasonography, and two were entirely solid. The mean morphology index score was 5.1 (median, 5; range, 3–7) compared with 3.5 (median, 3; range, 0–10) for patients with benign tumors (P<.001). Forty percent of women with low malignant potential ovarian tumors had a palpable abnormality on clinical examination, and only two had an elevated serum CA 125 (more than 35 units/mL). Twenty percent of women with a low malignant potential tumor (3 of 15) reported a family history of ovarian cancer, and only one was diagnosed under the age of 50 years. The stage distribution of women with low malignant potential ovarian tumors was as follows: stage IA, six; stage IB, two; stage IC, six; and stage IIIA, one. Twelve patients had serous ovarian tumors, one patient had an endometrioid ovarian tumor, and two patients had a mucinous ovarian tumor, all of borderline malignancy. An ovarian abnormality was detected on the first screen in 53.3% of these cases. All women were treated primarily by surgery, and none received postoperative chemotherapy. Patients have been followed 0.2–14.8 years (mean, 6.2 years; median, 6 years) after diagnosis, and none has developed tumor recurrence or died of any cause.
Demographic and histologic variables in the 47 women with invasive epithelial ovarian cancer detected by screening are as follows: the mean age of these women was 64.5 years (median, 66 years; range, 36–86 years), the mean weight was 158 pounds (median, 151 pounds; range, 98–270 pounds), and the mean gravidity was 2.3 (median, 2; range, 0–8). An abnormality in ovarian morphology or volume was detected on the first screen in 16 patients (34%) and on subsequent screens in 31 patients (66%). All ovarian malignancies were solid or complex (both solid and cystic components) except one with cystic morphology and a thickened cyst wall. The mean morphology index score of ovarian malignancies was 6.1 (median, 6; range, 1–10) compared with 3.5 (median, 3; range, 0–10) for benign ovarian tumors and significantly different (P<.001). Twenty-one of these women (45%) had a palpable abnormality on clinical examination (note that a physical examination was performed only on patients in which the ultrasonographic examination was abnormal and was biased by knowledge of the ultrasound result.) Thirteen percent of women with invasive ovarian cancer detected by screening (six of 47) reported a family history of ovarian cancer with two women diagnosed under the age of 50 years. At the time of surgery, 22 of the 47 women (47%) with epithelial ovarian cancer detected by screening had stage I disease and 11 (23%) had stage II disease. In contrast, 701 of 2,560 (27%) unscreened women with clinically detected epithelial ovarian cancer treated at hospitals throughout Kentucky and entered into the Kentucky Tumor Registry had stage I or stage II disease at the time of diagnosis (P<.01). Fourteen women (30%) in the screening group had stage III ovarian cancer at the time of detection as compared with 46% in the unscreened group (P=.028). There was also a substage shift within stage III ovarian cancer in the screening group. Half of the women with stage III ovarian cancers detected by screening had stage IIIA or stage IIIB tumors, whereas 84% of unscreened women with stage III ovarian cancer presented with stage IIIC disease (P<.01). There were no cases of stage IV ovarian cancer detected in women undergoing screening either at the initial prevalence screen or in subsequent incidence screens.
Serum CA 125 was elevated (more than 35 units/mL) in 10 of 31 women (32%) with stage I or stage II ovarian cancer detected by screening and in 11 of 14 women (79%) with stage III disease (note that two women did not have a preoperative CA 125 determination). The cell types of the ovarian cancer were as follows: serous carcinoma, 18; adenocarcinoma, 14; endometrioid carcinoma, 12; clear cell carcinoma, two; and mucinous carcinoma, one. Four tumors were well differentiated, nine were moderately differentiated, and 34 were poorly differentiated. According to the histologic criteria of Kurman and Shih,9 10 women (21%) had type I ovarian cancer and 37 women (79%) had type II ovarian cancer.
Patients with invasive ovarian cancer all underwent total abdominal hysterectomy, bilateral salpingo-oophorectomy, omentectomy, maximal tumor cytoreduction, and staging. Complete tumor debulking (no visible residual disease) was achieved in all patients with stage I–IIIA disease. Eighty-six percent of patients with stages IIIB and IIIC disease were cytoreduced to less than 1 cm residual disease. Three patients with stage IA grade 1 endometrioid carcinoma were treated by surgery alone. The remaining patients all received a minimum of six cycles of platinum and taxane chemotherapy. For the past 10 years of this investigation, patients requiring chemotherapy usually have received six cycles of intravenous carboplatin and paclitaxel every 3 weeks as their initial postsurgery treatment regimen.
After therapy, patients were followed from 0.1 to 20.1 years (mean, 5.8 years; median, 5 years) and none have been lost to follow-up. To date, nine screen-detected women have died of ovarian cancer, and six screen-detected women have died of other causes with no evidence of disease. Thirty-one patients are alive with no evidence of disease, and one patient is alive with disease. The causes of death in patients who died with no evidence of ovarian cancer are as follows: two patients died of a myocardial infarction and one each of multiple myeloma, chronic obstructive airway disease, and colon cancer. One of 22 patients (4.5%) with stage I disease, four of 11 patients (36%) with stage II disease, and four of 14 (29%) with stage III disease have died of ovarian cancer. The 5-year survival rate for invasive epithelial ovarian cancers detected by screening was: stage I, 95%±4.8%; stage II, 77.1%±14.5%; and stage III, 76.2%±12.1%. During this trial, 12 women have developed interval ovarian cancer within 1 year of a normal screen (false-negative), and seven of them have died of disease. These women generally had poorly differentiated cancers with minimal ovarian enlargement on transvaginal ultrasonography, which were diagnosed 6–12 months after a normal scan. The 2-year and 5-year survival rates for these interval cases were 75%±12.5% and 57%±14.6%, respectively.
The 5-year disease-free survival rate for women with stage I and stage II epithelial ovarian cancer detected by screening was 95%±4.8% and 77.1%±14.5%, respectively. The survival rates for women with stage I or stage II ovarian cancer did not vary significantly in the screening or control groups (Fig. 3). However, the survival rate for women with advanced disease was significantly higher in the screening group (Fig. 3). This is attributable to the fact that there were proportionally more cases of stage IIIA and IIIB ovarian cancers detected by screening and that no patient developed stage IV disease while being screened. The 5-year disease-free survival rate for the entire group of women whose invasive epithelial ovarian cancers were detected by screening was 84.6%±5.6% as compared with 53.7%±2.3% in unscreened women with epithelial ovarian cancer treated at the University of Kentucky-Markey Cancer Center from 1995 to 2011 (P<.001). The women in the control group all were surgically staged and treated by the same surgical and chemotherapeutic protocols as those women whose cancers were detected by screening. The 5-year survival rate for women in the screened group was also significantly higher (P<.001) than that of unscreened women treated at multiple centers throughout the state and entered from 1995 to 2011 in the Kentucky Cancer Registry (84.6%±5.6% compared with 48.4%±1.2%). When the interval (false-negative) cases are included, the 5-year disease-free survival rate for women with ovarian cancer in the screening group is 74.8%±6.6%, which remains significantly (P<.001) higher than those women treated at this institution who did not have screening (53.7%±2.3%).
Theoretically, the use of screening to detect ovarian cancer at an earlier stage should result in more curable cases; however, there are limited data to suggest that screening actually results in reduced ovarian cancer mortality.10–15 Screening research involves analysis of large numbers of women and takes considerable planning, discipline, time, and financial support to perform. Ovarian cancer meets the criteria of a disease that should benefit from screening.16 First, ovarian cancer is highly curable with conventional treatment methods when detected at an early stage. Numerous studies have confirmed that the 5-year survival rate for women with stage I epithelial ovarian cancer is as high as 95% in major cancer centers.4 Recent studies have indicated that symptoms such as early satiety, abdominal bloating, and pelvic pain occur more commonly in women with ovarian cancer than in control patients without the disease.2,3 However, these symptoms are nonspecific and are often ignored by many women, resulting in significant delays before seeking medical evaluation. It is estimated that without screening, only 15% of ovarian cancers are localized to the ovary at the time of diagnosis.4 This is understandable because ovaries are palpable in less than 30% of postmenopausal women at the time of examination under anesthesia.17 Therefore, it is extremely difficult to detect subtle changes in ovarian morphology or volume associated with early-stage ovarian cancer by clinical examination alone.
One of the difficulties in designing screening strategies for ovarian cancer is that it is not a common disease. The incidence of ovarian cancer in the general population is 20–30 per 100,000 in women younger than 40 years of age and rises to 40 per 100,000 at the age of 50 years.18–20 Therefore, screening studies must focus specifically on women who are at high risk for the disease. Entry criteria for the present screening trial included only asymptomatic women older than age 50 years or women older than age 25 years with a documented family history of ovarian cancer. The screening algorithm followed in this investigation used transvaginal ultrasonography as the initial screening test with serum CA 125 obtained only in women with a persisting ultrasonographic abnormality. This algorithm detected 47 invasive epithelial ovarian cancers and 15 epithelial ovarian tumors of low malignant potential in 37,293 women screened or 1.6 epithelial ovarian cancer cases detected per 1,000 women screened.
An important factor in screening is the interval between the onset of ovarian cancer and the presence of clinical symptoms (the preclinical phase) during which screening intervention should lead to earlier detection. Recently, Crum and coworkers21 and Kurman and Shih9 have postulated that there are two types of epithelial ovarian cancer. Type I ovarian cancers are typically low grade endometrioid, clear cell, or mucinous cell types and have profiles of gene expression similar to those of low malignant potential tumors. In contrast, type II ovarian cancers are high-grade serous carcinomas, endometrioid carcinomas, or undifferentiated carcinomas, which are biologically aggressive. Type II ovarian cancers typically display p53 mutations and are thought to have a shorter preclinical phase than type I tumors. According to this model, annual screening would be expected to be most effective in detecting type I tumors while missing the more aggressive type II tumors, which are responsible for the majority of ovarian cancer deaths. In the present investigation, 21 of the 33 women (64%) with stage I or II invasive epithelial ovarian cancers detected by screening had poorly differentiated serous carcinomas, poorly differentiated endometrioid carcinomas, or undifferentiated adenocarcinomas (type II cancers). Sixteen of these 21 women (76%) are alive and well with no evidence of disease after conventional treatment with surgery and platinum-based chemotherapy. One can postulate that without screening, the majority of these women would have progressed rapidly to advanced-stage ovarian cancer, and many would have died of disease.
A concern about ultrasonographic screening is a low positive predictive value in differentiating malignant from benign ovarian tumors. In the multicenter Prostate, Lung, Colorectal, and Ovarian Cancer trial, for example, 19.5 surgeries were required to detect one ovarian cancer (positive predictive value 5.1%) when screening was performed using a combination of transvaginal ultrasonography and serum CA 125.13 In the present investigation, 523 women with persisting ovarian tumors underwent surgery. Forty-seven of these women had invasive epithelial ovarian cancer, 15 had epithelial ovarian tumors of borderline malignancy, five had nonepithelial ovarian cancer, and nine had cancer metastatic to the ovary. Therefore, 6.9 surgical procedures were performed per ovarian cancer detected (positive predictive value 14.53%). The difference in positive predictive value noted in this trial and that reported in the Prostate, Lung, Colorectal, and Ovarian Cancer trial can be explained partially by a higher cancer prevalence in the screened population and by two modifications in the University of Kentucky screening algorithm, which were adopted in the past 5 years. First, analysis of more than 3,200 unilocular cystic ovarian tumors less than 10 cm in diameter detected by screening revealed that the risk of malignancy in these tumors was essentially nonexistent. Women with these tumors were followed ultrasonographically for an average of 6.3 years, and none developed ovarian cancer.22 As a result, the screening algorithm was changed, and cystic ovarian tumors less than 10 cm in diameter were no longer removed surgically. In a subsequent investigation,23 2,870 complex septated cystic ovarian tumors identified by ultrasonographic screening were evaluated for risk of malignancy. Surgery was performed in 128 women and no tumor with this morphologic pattern was malignant. The remaining women were followed ultrasonographically every 4–6 months without surgery for an average of 6.4 years. Thirty-eight percent of these tumors resolved spontaneously, and no patient developed ovarian cancer. Therefore, the screening algorithm was further amended in 2009 such that surgery was not recommended in women with septated cystic ovarian tumors without solid areas or papillary projections. Ovarian tumors with these two morphologic patterns made up 55% of the total abnormalities detected by screening. As a result of these changes, the positive predictive value of screening in this trial increased from 14.1% (1987–2006) to 20.2% (2006–2011). Finally, a specific treatment algorithm was followed in this trial for women with a persisting ovarian abnormality on transvaginal ultrasonography. This included careful morphologic evaluation of each tumor, Doppler analysis of tumor blood flow, and serum biomarker analysis before a patient underwent surgery. In the Prostate, Lung, Colorectal, and Ovarian Cancer trial, there was no uniform treatment algorithm, and treatment was at the discretion of the patient's primary care physician.14 Clearly, further research is needed concerning the relationship of tumor morphology, molecular genetic patterns, and biomarker profiles to risk of malignancy so that surgery can be avoided in women with ovarian tumors at minimal risk for neoplasia. This should result in progressively increasing the positive predictive value for multimodality screening.
One of the objective measurements of an effective screening test is its ability to lower stage at detection and disease-specific mortality. In the Prostate, Lung, Colorectal, and Ovarian Cancer trial, both transvaginal ultrasonography and CA 125 were evaluated as screening tests. Although transvaginal ultrasonography was associated with a low positive predictive value, 72% of cases detected by ultrasonography alone were stage I or stage II.13 Screening with serum CA 125 was associated with a higher positive predictive value, but 74% of biomarker-detected ovarian cancers were stage IIIC or IV. Data from the present trial are consistent with these findings in that 33 of the 47 cases (70%) of the women whose ovarian cancers were detected by transvaginal ultrasonography screening had stage I or II disease as opposed to 27% of women with clinically detected ovarian cancer from the same population referred to institutions throughout the state (P<.01). Similarly, in the multi-institutional U.K. Collaborative Trial of Ovarian Cancer Screening trial, 48% of ovarian cancers detected in the prevalence screen by multimodal screening had stage I–II disease compared with 26% in the unscreened population.24
The increase in long-term ovarian cancer survival observed in the screening group in the present trial is directly attributable to lower stage at detection. The 5-year survival rate for women with stage I or II epithelial ovarian cancer was 89%, and 70% of the total screening group had these two stages of disease at the time of detection. Whereas more than two thirds of women in the screening group were detected with stage I–II disease, more than two thirds of unscreened women with clinically detected ovarian cancer presented with stage III–IV ovarian cancer. There was also a substage shift within stage III in the screening group with 50% of women being detected with stage IIIA or IIIB disease as opposed to only 16% in women without screening. Finally, no woman in the screening group had or developed stage IV ovarian cancer during this trial, whereas 19.6% of unscreened women with ovarian cancer had stage IV disease at the time of clinical diagnosis.
Data from the present investigation indicate that annual ultrasonographic screening of women at risk for ovarian cancer caused a decrease in stage at detection and site-specific ovarian cancer mortality when compared with women from the same geographic area who did not have screening. The observed 14.5% positive predictive value associated with transvaginal ultrasonography screening is unacceptably low, and routine screening of the general population cannot be recommended at the present time. However, focused research concerning risk of malignancy in ultrasonographically identified ovarian tumors should result in more selective surgery and an increase in positive predictive value to levels associated with screening for other malignancies.
The importance of early detection in ovarian cancer is evident in that more than two thirds of asymptomatic women with ovarian cancer detected by screening in the present trial had localized disease at the time of detection, and their 5-year survival rate was nearly 90%. In the absence of early detection, most women will continue to present with advanced-stage disease for which the cost of treatment is high and the cure rate is low.
1. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics. 2010. CA Cancer J Clin 2011;61:133–4.
2. Goff BA, Mandel LS, Melancen CH, Munz HG. Frequency of symptoms of ovarian cancer in women presenting to primary care clinics. JAMA 2004;291:2705–12.
3. Goff BA, Mandel LS, Drescher CW, Urban N, Gough S, Schurman KM, et al.. Development of an ovarian cancer symptom index. Cancer 2007;109:221–7.
4. Jelovac D, Armstrong DK. Recent progress in the diagnosis and treatment of ovarian cancer. CA Cancer J Clin 2011;61:183–203.
5. van Nagell JR, Higgins RV, Donaldson ES, Gallion HH, Powell DE, Pavlik EJ, et al.. Transvaginal sonography as a screening method for ovarian cancer. A report of the first 1000 cases screened. Cancer 1990;65:573–7.
6. van Nagell JR, DePriest PD, Ueland FR, DeSimone CP, Cooper AL, Pavlik EJ, et al.. Ovarian cancer screening with annual transvaginal sonography: findings of 25,000 women screened. Cancer 2007;109:1887–96.
7. Pavlik EJ, DePriest PD, Gallion HH, Ueland FR, Reedy MB, Kryscio RJ, et al.. Ovarian volume related to age. Gynecol Oncol 2000;77:410–2.
8. Ueland F, DePriest P, Pavlik E, Kryscio R, van Nagell JR. Preoperative differentiation of malignant from benign ovarian tumors: the efficacy of morphology indexing and Doppler flow sonography. Gynecol Oncol 2003;91:46–50.
9. Kurman RJ, Shih IeM. The origin and pathogenesis of epithelial ovarian cancer: a proposed unifying theory. Am J Surg Pathol 2010;34:433–43.
10. Jacobs I, Skates SJ, MacDonald N, Menon U, Rosenthal A, Prys Davies A. Outcome of a pilot randomised controlled trial of ovarian cancer screening. Lancet 1999;253:1207–10.
11. Sato S, Yokoyama Y, Sakamoto T, Futagami M, Saito Y. Usefulness of mass screening for ovarian carcinoma using transvaginal ultrasonography. Cancer 2000;89:582–8.
12. Buys S, Partridge E, Greene MH, Prorok PC, Reding D, Riley TL, et al.. Ovarian cancer screening in the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial: findings from the initial screen of a randomized trial. Am J Obstet Gynecol 2005;193:1630–9.
13. Partridge E, Kreimer AR, Greenlee RT, Williams C, Xu JL, Church TR, et al.. Results from four rounds of ovarian cancer screening in a randomized trial. Obstet Gynecol 2009;113:775–82.
14. Buys S, Partridge E, Black A, Johnson CC, Lamerato L, Isaacs C, et al.. Effect of screening on ovarian cancer mortality. The Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Randomized Controlled Trial. JAMA 2011;305:2295–303.
15. Havrilesky LJ, Sanders GD, Kulasingam S, Chino JP, Berchuck A, Marks JR, et al.. Development of an ovarian cancer screening decision model that incorporates disease heterogeneity: implications for potential mortality reduction. Cancer 2011;117:545–53.
16. Prorok PC. Evaluation of screening program for the early detection of cancer. Natl Cancer Inst Stat Textbk Monogr 1984;51:267–328.
17. Ueland FR, DePriest P, DeSimone C, Pavlik EJ, Kryscio RJ, van Nagell JR Jr. The accuracy of examination under anesthesia and transvaginal sonography in evaluating ovarian size. Gynecol Oncol 2005;99:400–3.
18. Kosary CL. Cancers of the ovary. SEER survival monograph: cancer survival among adults: US SEER program 1988–2001. National Institutes of Health Publication 7-6215. Bethesda (MD): National Institutes of Health; 2007. p. 133–44.
19. Yancik R. Ovarian cancer. Age contrasts in incidence, histology disease stage at diagnosis, mortality. Cancer 1993;71(suppl):517–23.
20. Quirk JT, Natarajan N, Mettlin CJ. Age-specific ovarian cancer incidence rate patterns in the United States. Gynecol Oncol 2005;99:248–50.
21. Crum CR, Drapkin R, Miron A, Ince T, Muto M, Kindelberger DW, et al.. The distal fallopian tube: a new model for pelvic serous carcinogenesis. Curr Opin Obstet Gynecol 2007;19:3–9.
22. Modesitt SC, Pavlik EJ, Ueland FR, DePriest PD, Kryscio RJ, van Nagell JR Jr. Risk of malignancy in unilocular ovarian cystic tumors less than 10 centimeters in diameter. Obstet Gynecol 2003;102:594–9.
23. Saunders BA, Podzielinski I, Ware R, Goodrich S, DeSimone C, Ueland FR, et al.. Risk of malignancy in sonographically confirmed septated cystic ovarian tumors. Gynecol Oncol 2010;118:278–82.
© 2011 by The American College of Obstetricians and Gynecologists.
24. Menon U, Gentry-Maharj A, Hallet R, Ryan A, Burnell M, Jacobs I, et al.. Sensitivity and specificity of a multimodal and ultrasound screening for ovarian cancer, and stage distribution of detected cancers: results of the prevalence screen of the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS). Lancet Oncol 2009;10:327–40.