eFigure. Ultrasound Images of a Normal Ovary (A), a Simple Ovarian Cyst (B), and a Complex Cystic Mass (C)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Smith-Bindman R, Poder L, Johnson E, Miglioretti DL. Risk of Malignant Ovarian Cancer Based on Ultrasonography Findings in a Large Unselected Population. JAMA Intern Med. 2019;179(1):71–77. doi:10.1001/jamainternmed.2018.5113
What is the risk of malignant ovarian cancer associated with simple cysts?
Among 72 093 women in this nested case-control study of patients enrolled in Kaiser Permanente Washington, simple cysts were common in both premenopausal and postmenopausal women undergoing pelvic ultrasonography, but they were not associated with an elevated risk of being diagnosed as having malignant ovarian cancer. In contrast, complex cysts or solid masses are far less common but are associated with a significantly increased risk of developing malignant cancer.
Simple cysts should be considered normal findings and do not need surveillance.
The risk of malignant ovarian cancer associated with simple cysts is unknown.
To quantify the risk of ovarian cancer based on ultrasonographic characteristics of ovarian masses, including simple cysts, in a large unselected population.
Design, Setting, and Participants
This was a nested case-control study of patients enrolled in Kaiser Permanente Washington, a large integrated health care system in Washington State. Participants were 72 093 women who underwent pelvic ultrasonography between January 1, 1997, and December 31, 2008. Analysis was completed in April 2017.
Ultrasonographic characteristics of ovarian masses measured in 1043 women, and also, using weights derived from the sampling strategy, estimated frequencies for the entire cohort.
Main Outcomes and Measures
Malignant ovarian cancer, identified through December 31, 2011, by cancer registry linkage.
Among 210 women who were diagnosed as having ovarian cancer, 49 were younger than 50 years, and 161 were 50 years or older. Ultrasonography findings were predictive of cancer (C statistic, 0.89). The risk of cancer was significantly elevated in women with complex cysts or solid masses, with likelihood ratios relative to women with normal ovaries ranging from 8 to 74 and the 3-year risk of cancer ranging from 9 to 430 cases per 1000 women based on patient age and ultrasonography findings. In contrast, the 23.8% of women younger than 50 years and the 13.4% of women 50 years or older with simple cysts were not at a significantly increased risk of ovarian cancer compared with women with normal ovaries. Likelihood ratios associated with the detection of a simple cyst were 0.00 in women younger than 50 years (no cancers were identified) and 0.10 (95% CI, 0.01-0.48) in women 50 years or older, and the absolute 3-year risk of cancer ranged from 0 to 0.5 cases per 1000 women.
Conclusions and Relevance
According to this study, the ultrasonographic appearance of ovarian masses is strongly associated with a woman’s risk of ovarian cancer. Simple cysts are not associated with an increased risk of ovarian cancer, whereas complex cysts or solid masses are associated with a significantly increased risk of ovarian cancer.
Increased use of transvaginal pelvic ultrasonography has led to the frequent identification of ovarian masses.1 While most ovarian masses are benign and reflect normal physiological processes, a small number are malignant. Historically, surgical exploration and resection of all identified ovarian masses represented the standard of care, and as many as 5% to 10% of women have undergone a surgical procedure for a suspected ovarian neoplasm; most of these procedures did not diagnose cancer.1,2 With the growing realization that benign masses are frequent even in postmenopausal women1,3-12 as well as more widespread use of risk prediction models to estimate the likelihood that a given mass is malignant based on its ultrasonographic appearance combined with clinical and laboratory data,4 surgical resection has largely been replaced by conservative management, including surveillance for masses thought to be benign.1
The most commonly identified ovarian mass is a simple cyst.3 While a large number of studies have concluded that simple cysts are almost certainly benign and not cancer precursors, researchers3,5-11 (and professional guidelines that reference these studies1,13) nonetheless recommend ongoing surveillance of these lesions. The recommendations for ongoing surveillance of simple cysts, despite widespread belief that they are almost certainly benign, in part reflects the poor prognosis of malignant ovarian cancer and concern that there is a small but unknown risk of cancer even in masses with the most benign appearances.
To our knowledge, no prior study has described the prevalence of ovarian masses (and their appearance) and cancer outcomes among a well-defined, unselected population with respect to ovarian cancer that would permit quantification of absolute risks of cancer based on imaging characteristics or has compared these risks with those among women without such findings. Previous studies limited their analysis to women participating in an ovarian cancer screening program,3,6,12 restricted the analysis to patients who underwent resection for whom the decision to resect may have been influenced by ultrasonography,4,7,10,14,15 or did not have an adequate comparison group so that appropriate relative risk could be quantified.6,7,10,14 The objectives of this study were to investigate the ultrasonographic characteristics that are associated with ovarian cancer and to quantify the risk by specific imaging findings in a large unselected population. We sought to identify features predicting with a high level of certainty that an ovarian mass is benign and thus does not require surveillance.
We conducted a nested case-control study of patients enrolled in Kaiser Permanente Washington (previously Group Health Cooperative), a large integrated health care system in Washington State. Kaiser Permanente Washington enrolls between 250 000 and 350 000 female members annually. Included were patients 18 years or older who underwent pelvic ultrasonography, including transvaginal imaging, between January 1, 1997, and December 31, 2008, and who were previously cancer free. Analysis was completed in April 2017. During the study period, 72 093 women underwent 118 778 pelvic ultrasonography examinations. Institutional review board approval of the study was obtained from the University of California, San Francisco and from Kaiser Permanente Washington, with a waiver of individual informed consent. The results were stratified by age (<50 vs ≥50 years) to approximate menopausal status.
Cancers were identified through linkage of the Kaiser Permanente Washington enrollment files with the Kaiser Permanente Washington pathology records and with the Seattle–Puget Sound Surveillance Epidemiology and End Results Cancer Registry, a population-based registry collecting cancer incidence and mortality data for western Washington State. By cancer registry linkage, we identified ovarian cancer diagnosed through December 31, 2011, providing a minimum of 3 years of follow-up (range, 3-14 years) after the last ultrasonography during which a cancer could be diagnosed. Patients diagnosed as having nonovarian malignant neoplasms (other than nonmelanomatous skin cancer) were excluded to prevent inclusion of metastatic cancer to the ovary, which would not be captured by the cancer registry. Cancer registry linkages were performed twice to ensure complete identification of all cancers in the cohort, including cases with delayed reporting to the registry. The linkages resulted in the identification of 210 women diagnosed as having ovarian cancer within the cohort of women who had undergone previous pelvic ultrasonography.
For each woman diagnosed as having ovarian cancer whose ultrasonographic images could be retrieved (n = 142) as well as a matched sample of ultrasonography examinations in women not diagnosed as having cancer, we reviewed all pelvic ultrasonography examinations within the 3-year period before diagnosis or the matched date of ultrasonography. For each woman diagnosed as having cancer, we matched to ultrasonography examinations in 7 women without cancer on age, body mass index (calculated as weight in kilograms divided by height in meters squared), and year of ultrasonography. For each matched ultrasonography examination, we selected the first ultrasonography in the same woman within 3 years of the matched ultrasonography to ensure that we had the initial ultrasonography. To protect against bias, each control was sampled independently over time, meaning that the same individual could be matched to more than 1 case.16 We reviewed 901 ultrasonography examinations in women without cancer.
A board-certified radiologist (L.P.) masked to cancer status reviewed the ultrasonographic images. For each examination, the radiologist recorded the presence and features of ovarian masses, including size and whether cystic or solid; if cystic, it was recorded whether the mass was simple, contained internal echoes, had a thin septation, or had complex features. Simple cysts were defined as thin, smooth walled, round or oval, entirely anechoic masses 1 cm or larger, with posterior enhancement and no internal architecture. Cysts smaller than 1 cm were considered normal (tiny cysts), and their presence was not recorded. If internal echoes were identified (diffuse low-level echoes, fibrinous strands, or retracting clot) within a cyst that did not have any other features of a complex mass, the cyst was characterized as having low-level echoes. Septated cysts were defined as having a single, thin septation. Complex cysts were defined as masses with any of the following: multiple (>1) loculations or septations, thick septations, mural nodules, wall excrescences, papillary projections, or solid components with vascularity (eFigure in the Supplement). Patients with masses involving both ovaries or with more than 1 mass within the same ovary were characterized using the features of the most abnormal finding. Septated cysts were uncommon (n = 16), with none subsequently diagnosed as cancer; these masses were combined for analysis with simple cysts. We did not review the ultrasonography examinations in women identified through the second cancer registry linkage and assumed that their ultrasonography examinations had a similar distribution in the findings as women identified through the first cancer registry linkage.
All analysis is at the level of the woman, and the results are weighted to the full study cohort. We assessed the frequency of ultrasonographic findings in the reviewed images; using weights derived from the sampling strategy, ultrasonographic characteristics of ovarian masses were measured in 1043 women, and frequencies were estimated for the entire cohort of women who underwent pelvic ultrasonography (n = 72 093), adjusting for the distribution of matching variables in the cohort. We calculated the frequency distribution of ultrasonographic findings and calculated the odds ratio and likelihood ratio for each finding type by age. We assessed the predictive accuracy of ultrasonographic findings for discriminating between women with and without malignant ovarian neoplasm using the C statistic. We assessed the risk of developing cancer within 3 years per 1000 women using unconditional logistic regression with empirical standard errors, controlling for variables used in the matching scheme. Risk predictions and 95% CIs were calculated via marginal standardization using inverse probability weights to match to the prevalence of women among the general population. All analyses were performed using statistical software (Stata, version 13.1; StataCorp LP).
Among the 72 093 women who underwent pelvic ultrasonography, 210 were subsequently diagnosed as having ovarian cancer (Table 1). Most of the women in the cohort (75.5%) were younger than 50 years, whereas most of the women with cancer (76.7%) were 50 years or older. Quiz Ref IDThe following 6 histological subtypes of ovarian cancer accounted for most cases: papillary serous cystadenocarcinoma (36.6%), endometrioid carcinoma (12.0%), serous cystadenocarcinoma (10.6%), clear cell adenocarcinoma (8.5%), adenocarcinoma not otherwise specified (8.0%), and mucinous adenocarcinoma (4.9%). The remaining histological subtypes each accounted for 2% or less of cases. Ovarian cancer was diagnosed a mean of 3.4 months (range 0-35.5 months) after the first ultrasonography; among patients with ovarian cancer who underwent more than 1 ultrasonography, the mean (SD) time between examinations was 11.0 (10.5) months. Among women without cancer, the mean (SD) time between ultrasonography examinations was 17.2 (12.5) months.
Most women in the cohort had normal ovaries (Table 2). Quiz Ref IDThe most common ovarian finding among women in the cohort other than a normal ovary was a simple cyst, estimated to occur in 12 957 women (23.8%) younger than 50 years and 2347 women (13.4%) 50 years or older (Table 2). The estimated number (proportion) of women with simple cysts decreased with increasing cyst size (Table 3). Among the estimated 15 306 women with a simple cyst, only a single woman was subsequently diagnosed as having cancer within 3 years (Tables 2 and 3). Quiz Ref IDNo simple cysts were subsequently diagnosed as cancer in women younger than 50 years, and the sole simple cyst in women 50 years or older subsequently diagnosed as cancer results in a likelihood ratio of 0.06 (95% CI, 0.01-0.48) (Table 2).
The most common ovarian finding in patients diagnosed as having ovarian cancer was a complex cystic mass, found in 31 of 49 women (63.3%) younger than 50 years and in 90 of 161 women (55.9%) 50 years or older (Table 2). Approximately one-third of the women with a complex cystic mass and ovarian cancer also had ascites. In contrast, a complex cystic mass was estimated to occur in only 4150 women (7.6%) younger than 50 years without ovarian cancer and 884 women (5.1%) 50 years or older without ovarian cancer (Table 2). The identification of a complex cystic mass increased the likelihood of cancer approximately 8-fold in women younger than 50 years and 50 years or older. In women 50 years or older with a complex cyst with ascites, the likelihood of cancer was increased substantially (likelihood ratio, 74.17; 95% CI, 15.80-349.12).
Solid masses were estimated to occur in approximately 10% of women with ovarian cancer and in approximately 1% of women without ovarian cancer. The identification of a solid mass significantly increased the likelihood of cancer approximately 10-fold in women younger than 50 years (likelihood ratio, 8.08; 95% CI, 1.86-36.12) and in women 50 years or older (likelihood ratio, 10.08; 95% CI, 3.25-31.21).
In the multivariable analysis, ultrasonography findings predicted the probability of cancer (C statistic, 0.89). The odds of cancer were significantly elevated in women with complex cysts or solid masses (odds ratio range, 23.06-249.28 based on patient age and finding) (Table 4). Women whose ovaries were not identified, women with simple cysts, and women having cysts with low-level echoes were not at a significantly increased risk of ovarian cancer compared with women in whom normal ovaries were seen.
Ovarian cancer prevalence is much higher among women 50 years or older,17 and the absolute risk of cancer associated with the identification of these findings was also higher herein in women 50 years or older. In the present study, among 1000 women diagnosed as having complex cysts or solid masses, the 3-year risk of ovarian cancer was estimated as 9.4 to 11.0 cases per 1000 women younger than 50 years and as 65.2 to 429.8 cases per 1000 women 50 years or older. The corresponding estimated 3-year risk of ovarian cancer in both age groups was less than 1 case per 1000 women in whom a simple cyst is identified.
If ultrasonography technique is limited, it may be possible to identify a cystic mass in the ovary but not to confidently characterize it. Likelihood ratios associated with cystic masses based on size alone are listed in Table 5. In both age groups, there was no statistical increase in cancer risk until cysts were 7 cm or larger.
We found a single case of cancer after an ultrasonography obtained because of severe pain that demonstrated a 1-cm simple cyst. This patient had persistent severe pain, and computed tomography performed 30 days after the initial ultrasonography diagnosed extensive peritoneal metastatic disease, presumed to be ovarian in origin.
Ultrasonographic characteristics of ovarian masses are predictive of 3-year ovarian cancer risk, with an accuracy similar to that of the other frequently used diagnostic tests, such as the Breast Imaging–Reporting and Data System (BI-RADS) used for mammography.18 Women with complex cysts or solid masses have elevated relative risk of ovarian cancer, whereas women with simple cysts have low relative risk of ovarian cancer (absolute 3-year risk of ovarian cancer, 0-0.5 cases per 1000 women, which is not different from that of women with normal ovaries) (Table 4). Our results add to the existing literature because we were able to document the high prevalence of simple cysts among a large unselected population; we could determine the occurrence of cancer irrespective of whether women were referred for follow-up imaging, surgical consultation, or ovarian resection; we could quantify absolute cancer risks; and we could statistically compare risks in women with simple cysts vs in women with normal ovaries. Despite some of these methodological advantages of our study, the results are similar to those of prior reports. For example, we found 13.4% of women 50 years or older have simple cysts, similar to the 14.1% previously reported by Greenlee et al,3 and we found that the risk of cancer associated with their identification is low, as has been demonstrated by others.1,3-11
We observed that the most frequent appearance of ovarian cancer was a complex cystic mass and estimate that 6.5% of women 50 years or older with such masses will be diagnosed as having ovarian cancer within 3 years based on the data in Table 4. Sharma et al,12 reporting on the risk of ovarian cancer in postmenopausal women enrolled in the United Kingdom Collaborative Trial of Ovarian Cancer Screening, described 4 categories of masses, which together would correspond to our complex cyst categories, and the absolute risk of cancer in these groups ranged from 0.4% to 6.6%. Sharma et al did not present data on the prevalence of simple cysts or their associated risks.
In our study, the absolute cancer risks associated with the identification of complex cysts or solid masses were significantly higher among women 50 years or older. This outcome occurs because of the higher incidence of ovarian cancer among older women17 and the greater likelihood of physiological hemorrhagic cysts that can mimic complex cystic masses in younger women. We found a single case of cancer after an ultrasonography demonstrating a 1-cm simple cyst. While not proven, the identification of a simple cyst on ultrasonography was likely an incidental finding because the patient already had symptoms suggestive of her metastatic cancer diagnosed 1 month later. However, even with inclusion of this case, the statistical risk of cancer was significantly reduced, and not elevated, when simple cysts were identified.
One of the justifications for the surveillance of simple cysts is that imaging may be inaccurate and might miss complex features.13 This supposition was not supported by our data: cysts interpreted as being simple, including large ones, were not associated with cancer. Furthermore, even if the detailed characterization of a cyst was ignored (Table 5), there was no statistical elevation of cancer risk until cysts were 7 cm or larger.
Ovarian cancer is the fifth leading cause of cancer deaths among women, with 22 000 cases and 14 000 deaths annually.17 Because of its high mortality rate and because ovarian cancer is typically a cystic disease on ultrasonography (albeit complex), radiologists and gynecologists who perform imaging, as well as primary care physicians who are told of these findings in reports, are often reluctant to ignore the frequently encountered simple cysts, although it is widely known that simple cysts are almost universally benign, and this has resulted in frequent surveillance and referral to gynecologists and oncologists even in the absence of data demonstrating that malignant lesions develop from simple cystic precursors and without data demonstrating value from the ongoing surveillance of these lesions. For example, the Society of Radiologists in Ultrasound advises that, in premenopausal women, simple cysts as small as 3 cm could be reimaged annually if it makes the physicians more comfortable, suggests annual surveillance of simple cysts larger than 5 cm, and recommends magnetic resonance imaging of simple cysts 7 cm or larger.13 The Society of Radiologists in Ultrasound guidelines suggest even greater surveillance of simple cysts in postmenopausal women, recommending annual surveillance of lesions as small as 1 cm. The American College of Obstetricians and Gynecologists likewise recommends ongoing surveillance of simple cysts, although admitting that the optimum frequency and duration of surveillance is not known.1 Several publications suggest ongoing surveillance of cysts but without data to support this recommendation.19-21 While surveillance may not seem harmful, there is a growing realization across all areas of medicine that unnecessary imaging is associated with morbidity, including wasted time, false-positive results, overdiagnosis, unnecessary surgery, and anxiety.22-25Quiz Ref ID The best way to minimize the harms of unnecessary surveillance of incidental benign lesions is to avoid surveillance. Given the high prevalence of simple cysts, their lack of association with ovarian cancer, and no elevated risk compared with women with normal ovaries, incidental and asymptomatic simple cysts should be considered normal and frequent findings in women of any age and ignored.
A large number of prediction models for characterizing the risk of ovarian cancer have been developed that combine the ultrasonographic appearance of masses with demographics and clinical risk factors.1,4,26 Women with simple cysts are typically included in these predictive models. Given that simple cysts have no association with cancer, the prediction models should not be applied to women with these cysts because this imaging finding alone predicts that such masses are not at risk of developing into cancer.
The main strength of our study is the large sample size and linkage of the cohort with a comprehensive cancer registry, which will identify more than 98% of cancers in cohort members.27,28 This protocol allowed determination of malignant cancer outcomes even in women who did not undergo surveillance or immediate surgery. We used an efficient nested case-control study design; by matching and selecting controls from relevant risk sets within the entire cohort, we were able to estimate the prevalence of ovarian masses and the specific findings within the larger cohort. Therefore, our study provides the largest and most precise estimate to date of the prevalence and absolute risk of ovarian masses.
The analysis had several limitations. We did not include outcomes not captured in the cancer registry, such as benign ovarian masses or endometriomas. Ovarian masses should be managed based on controlling relevant symptoms rather than future cancer risk or future potential symptoms. We did not have information on patient symptoms, and future cancer risk may vary by symptoms or other potential risks such as potential future symptoms that could derive from a cyst. We did not include women with a history of cancer, and our results do not apply to these women. Quiz Ref IDWe did not have menopausal status, and the stratification by age was inexact for characterizing menopausal status. We were unable to obtain all of the imaging examinations in patients with cancer, but this factor is unlikely to have systematically biased the results. Finally, our findings reflect imaging performed through 2008, allowing cancer diagnosis through 2011, with confirmation or ascertainment of cancer as of 2014. There is no reason to suspect that this date range would have biased the results because it is unlikely that the prevalence of cancer among women with simple cysts has changed since that time.
According to the results of this study, increased ultrasonography use,29 with concomitant improved technical quality, has led to the identification of imaging findings without clear criteria about what images requires further evaluation.25 The result has been tissue sampling and surgery when little value comes from this assessment. Simple cysts are frequently encountered incidental and normal findings on pelvic imaging, and additional evaluation of these findings is not warranted.
Accepted for Publication: August 4, 2018.
Corresponding Author: Rebecca Smith-Bindman, MD, Department of Radiology and Biomedical Imaging, University of California, San Francisco, 350 Parnassus Ave, Ste 307, San Francisco, CA 94143 (firstname.lastname@example.org).
Published Online: November 12, 2018. doi:10.1001/jamainternmed.2018.5113
Author Contributions: Dr Smith-Bindman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Smith-Bindman, Miglioretti.
Acquisition, analysis, or interpretation of data: Smith-Bindman, Poder, Johnson.
Drafting of the manuscript: Smith-Bindman, Poder, Johnson.
Critical revision of the manuscript for important intellectual content: Johnson, Miglioretti.
Statistical analysis: Smith-Bindman, Johnson, Miglioretti.
Obtained funding: Smith-Bindman, Miglioretti.
Supervision: Smith-Bindman, Poder, Miglioretti.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by grants R21CA131698 and K24CA125036 from the National Cancer Institute (Dr Smith-Bindman).
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not represent the official views of the National Cancer Institute or the National Institutes of Health.
Additional Contributions: Deborah Seger (Kaiser Permanente Washington) provided valuable assistance in gathering the electronic data for this study. No compensation was received.
Create a personal account or sign in to: