A robust method to screen for ovarian cancer has the potential to reduce mortality from this disease, much as has occurred with cervical cancer and the use of cytological staining. To date, a single modality screening method for ovarian cancer, using either the blood test for cancer antigen 125 (CA125) or transvaginal ultrasonography (TVS), has not provided the required accuracy necessary for reliable, cost-effective, early detection of this cancer. Multimodality screening using serum CA125 and pelvic or TVS appears to improve the median survival in the screened group (5,6,7,8). Currently there are at least two large ongoing trials for multimodal screening (CA125 and TVS) in ovarian cancer. The final assessment of the accuracy of such approaches will have to await the analysis of the complete cohort and longer follow-up to assess the impact on mortality (9,10). It is clear that such programs can identify patients with ovarian masses, which then need further work-up.

Most ovarian masses in premenopausal women are benign in etiology and functionally related to the menstrual cycle. When incidentally detected, these masses are typically followed for regression by physical examination or ultrasound. Those that do not regress have a great probability of malignancy. By contrast, ovarian masses in postmenopausal women are more commonly malignant in etiology and are far more concerning for cancer than masses in the premenopausal woman. Certainly, premenopausal women can develop
ovarian cancer and a mass that does not resolve in such a patient is of considerable concern for cancer and requires further work-up.

Morphologic imaging may be used to characterize adnexal masses as benign or malignant and delineate regional or distant spread prior to surgery. In a recent meta-analysis, Liu et al. (11) compared ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI) in differentiation of malignant from benign ovarian tumors. Sensitivity and specificity estimates of all imaging modalities were comparable: 89% (95% confidence interval [CI], 88% to 90%) for US, 85% (95% CI, 83% to 86%) for CT, and 89% (95% CI, 88% to 92%) for MRI (P = .09). Specificity estimates were also comparable: 95% CI, 83% to 85% for US, 76% to 92% for CT, and 84% to 88% for MRI (P = .12). The authors concluded that ultrasound techniques seem to be similar to CT and MRI in differentiation of malignant from benign ovarian tumors, and that ultrasound morphologic assessment still is the most important and common modality in the detection of ovarian cancer. However, since none of these imaging methods achieve total accuracy, persistent ovarian masses in postmenopausal women are commonly surgically excised to avoid delayed diagnosis of ovarian carcinoma.

Ovarian cancers have been shown to be FDG avid in animal models, to express high levels of the glucose transporter GLUT1, and in human xenograft studies to have higher FDG uptake in areas of viable cancer than in normal tissues (including normal lymph nodes) in human xenograft studies. Pilot studies showing the feasibility of imaging ovarian cancer
with PET were reported as early as 1991 (18,19,20). Nearly all PET imaging of ovarian cancer is currently performed with FDG as the radiotracer.

Much like other FDG-avid cancers, patients should be imaged in the fasting state to minimize insulin levels and the targeting of FDG to normal insulin-responsive tissues like skeletal muscle. The approach to imaging may vary slightly depending on the individual PET center but is generally rather consistent. In the authors’ institution, patient preparation for ovarian cancer imaging is similar to that for oncologic imaging in general, with fasting for at least 4 hours (and ideally overnight), with no specific urinary tract preparation. The authors normally use dilute positive oral contrast to achieve definitive positive bowel visualization. In some centers, various interventions are utilized to decrease the amount of FDG in the urinary tract such as bladder catheterization, continuous bladder irrigation, and/or intravenous hydration (1,000 to 1,500 mL of 0.9% or 0.45% saline solution) followed by the intravenous administration of 20 to 40 mg of furosemide. Another way to deal with urinary tract activity can be use of an additional delayed image postvoiding if there is any confusion regarding the urinary tract radiotracer activity levels. Most centers now use iterative reconstruction methods as opposed to filtered back projection methods in order to image the pelvis, as these iterative imaging approaches appear less sensitive to imaging artifacts from the “hot” activity in the bladder or ureters. The proper time from tracer injection until imaging has to be long enough for the FDG to accumulate avidly into the tumors, but not so long that the radiotracer has decayed away substantially. Most imaging in ovarian cancer is started about 60 minutes posttracer injection, but longer delays of 90 minutes to 2 hours will commonly result in superior target/background uptake ratios but lower total counts in the images.

The potential role of [¹⁸F]-FDG PET and/or PET/CT in ovarian carcinoma has been considered and/or investigated to varying extents in the following settings:

The literature regarding PET in ovarian carcinoma detection is somewhat variable and, at first inspection, somewhat confusing. The original literature was generated using PET only, while most recent studies have utilized PET/CT imaging. Given the complexities and variability of locations for the ovaries in the pelvis, it is not surprising that recent results with PET/CT have been generally more accurate than those with PET only.

The detection sensitivity using PET for ovarian cancer is likely dependent on the lesion histology as well as the lesion size. Small lesions, under 1 cm in size, are likely to be far less easily detected than larger lesions due to the resolution limitations of current PET systems. Similarly, mucinous tumors are less likely to be detected than more cellular tumors based on the experience in other mucinous tumors. In interpreting the results of PET imaging in ovarian carcinoma, it is critical to bear in mind the significant variability that may be present in the study populations included for imaging. Detection of microscopic disease is a task beyond the resolution of current PET systems. By contrast, the detection of macroscopic tumors before treatment-induced reductions in glucose metabolism occur is a much more realistic and feasible task for current imaging methodologies.

Ultrasound has been considered the modality of choice in the evaluation of patients with suspected adnexal masses, with MRI utilized in uncertain or problematic cases to further characterize masses as benign or malignant (26).

Hubner et al. (27) correlated the results of FDG PET and CT with pathology results in imaging ovarian masses. The calculated sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were 83%, 80%, 82%, 86%, and 76%, respectively, for PET and 82%, 53%, 72%, 77%, and 62% for CT. However, when combining PET and CT results, the positive and negative predictive values would be 95% and 100%, respectively. This study was performed with an early generation PET-only system, not PET/CT. Further, filtered back projection was used for image reconstructions. This may have led to degradation of image quality due to streak artifacts, but the overall performance of the test remained encouraging.

Rieber et al. (28) prospectively compared the diagnostic performance of transvaginal ultrasound, PET, and MRI in 103 patients with suspicious adnexal findings on ultrasound, in whom subsequent histology revealed 12 malignant and 91 benign ovarian tumors. The calculated sensitivity, specificity, and accuracy were 58%, 78%, and 76%, respectively, for PET; 83%, 84%, and 83% for MRI; 92%, 59%, and 63% for ultrasound; and 92%, 84%, and 85% for all three modalities in consensus.