Applied Radiology

Dr. Schweitzer is an Instructor in Radiology and Dr. Majid is a Resident in Radiology at the Medical College of Virginia of Virginia Commonwealth University; and Dr. Paredes is a gynecologist with the Department of Obstetrics & Gynecology at St. Mary's Hospital, Richmond, VA.

Ovarian cancer, the most lethal of all gynecologic cancers, is responsible for approximately 14,500 deaths annually in American women, and remains a challenging enigma. Its natural history remains a mystery, and, despite the new therapeutic modalities, the treatment of the disease is yet an unresolved riddle.

These disturbing facts are better understood when the unusual features of this complex disease are considered. Ovarian cancer represents a subtle array of many types and subtypes of malignancies, each with unique histologic features and idiosyncratic biologic behavior. ¹ Because of these facts, no consensus for a standard classification of the disease exists, and the establishment of a standard seems impossible.

Early diagnosis of ovarian cancer is not yet within reach, because of the silent progress of the disease and the lack of specific and sensitive tests at present. The utilization of family history, genetic testing (BRCA-1 and BRCA-2), CA 125 testing, and vaginal and Doppler ultrasound targeted to those patients who are at high risk may be helpful.

Ovarian carcinoma is basically a disease of surfaces, however, in that it spreads not by infiltrating adjacent organs, but by settling on the surfaces of surrounding tissues. Free-floating cancer cells carried by the intra-abdominal peritoneal fluid are responsible for the occult and silent dissemination and for the inaccurate diagnosis of a localized disease. ¹ Theoretically, for the diagnosis of ovarian cancer to be effective at a very early stage, biochemical markers would be used to detect the expression of the disease. Since no such markers are available, prophylactic oophorectomy at the time of unrelated pelvic surgery is a reasonable option to discuss with selected patients.

Pathology of ovarian cancer

More than 85% of ovarian cancers are of coelomic epithelial origin. ² The remaining are derived from germ cells, specialized gonadal stroma, nonspecific mesenchyma, or are metastatic lesions to the ovary. The epithelial ovarian cancers consist of the following subtypes: 50% serous, 17% undifferentiated, 15% endometrioid, 12% mucinous, and 6% clear cell carcinomas.

Serous tumors occur at any age, but their peak incidence is in the fourth to fifth decade. Serous carcinomas are the most common bilateral ovarian neoplasms and the most often widely disseminated ³ at the time of diagnosis, with widespread intraperitoneal deposits.

Endometrioid carcinomas are less likely to be bilateral or widely disseminated in the abdominal cavity. These tumors are more likely to be associated with coexisting adenocarcinoma of the uterus ⁴ and with non-neoplastic endometriosis. The main age at presentation is 55 to 60 years. Common symptoms include abnormal vaginal bleeding, abdominal pain, and/or mass.

Mucinous carcinomas occur mostly in the fourth to seventh decades, are bilateral in 15% to 20% of cases ³ and are mostly (95%) confined to the ovaries at laparotomy. About 5% of mucinous cancers are associated with dermoid cysts and range in size up to 50 cm in diameter. Rare gigantic lesions have been documented as weighing over 100 kg. ³

Clear cell tumors occur most often in the fifth to seventh decades, and one-half to two-thirds of patients are nulliparous. The most common presenting symptom is an enlarging abdominal mass. At laparotomy, 50% are confined to the ovary and are uncommonly associated with paraneoplastic hypercalcemia. ³

Imaging of ovarian malignancy

Once a pelvic mass is found, the role of the diagnostic radiologic modalities is important to clarify its origin and characteristics. Radiology plays an important role in the diagnosis, preoperative staging, surgical treatment, and evaluation of tumor recurrence of ovarian carcinomas. Ovarian carcinoma has characteristic tumor appearances and modes of tumor spread within the peritoneal cavity. By recognizing these features, the radiologist can assist the clinicians in treatment planning.

Ultrasound

Ultrasound is usually used for the evaluation of an asymptomatic or a symptomatic pelvic mass found on clinical exam. A combination of transvaginal and transabdominal ultrasound (US) is 80% accurate in detecting ovarian masses. ⁵ When an ovarian mass is detected, US is performed to evaluate the other ovary and uterus, as ovarian cancer has a high propensity for uterine cavity and contralateral ovarian involvement. This pattern of involvement is found in endometrioid ovarian cancers and in 66% of malignant serous neoplasms. ⁶ US is an inexpensive means of evaluating ovarian masses. It has been shown that ultrasound is 90% sensitive in detecting pelvic masses and their origins. ⁶ However, considerable overlap is present in the imaging features of benign and malignant lesions and, therefore, US cannot be used to differentiate benign from malignant ovarian masses. ⁷ US is also limited in evaluating the extent of disease; specifically, US is not accurate in evaluating the spread of metastatic ovarian disease to the omentum, mesentery, bowel serosa, and retroperitoneum. ⁶

Transvaginal ultrasound and color Doppler sonography have improved characterization of ovarian pathology. ⁷ Ultrasound characteristics that suggest malignancy are multiloculated masses that are >10 cm in diameter with thick septations and solid components ⁸ (figure 1).

Color Doppler sonography is used to aid in differentiating malignant blood flow patterns from benign or physiologic vasculature. ⁹ Malignant blood flow is characterized by an increase in diastolic flow, which is secondary to neovascularization. ⁷ The ovarian arterial blood flow is evaluated by calculating the resistive index (RI) or the pulsatility index (PI). ⁷ The RI is calculated from the difference between the systolic peak and diastolic trough. The values are found between 0 and 1. The PI is calculated by evaluating the difference over the mean velocity of the cardiac cycle. ⁷ The PI is considered more accurate than the RI, and values range between 0 and 10. Regardless of which index is used, the lower the value, the more diastolic flow. Doppler waveforms with RI <0.4 and PI <1 suggest malignancy, but there are many other nonmalignant causes of low RI or PI. ^7,8

Once ultrasound confirms a mass found on physical examination, other imaging studies are often used to further differentiate the mass and evaluate the extent of disease. The role of cross-sectional imaging in the preoperative evaluation of ovarian cancer is still not universally accepted and is somewhat controversial. ¹⁰

If a clinical diagnosis of ovarian cancer is of little doubt, preoperative performance of cross-sectional imaging will not change patient management. ⁶ Surgery remains at the central point in the management of ovarian cancer. ^6,11 Laparotomy, surgicopathologic staging, and tumor debulking are performed. The gold standard is the performance of a total abdominal hysterectomy, bilateral salpingo-oopherectomy, and omentectomy in addition to aspiration of ascites or peritoneal lavage for cytologic examination. In addition, random peritoneal biopsies are taken from the paracolic gutters, undersurfaces of the hemidiaphragms, and lymph nodes. ¹¹ For patients with advanced disease, chemotherapy is initiated following laparotomy. The routine preoperative use of CT or MRI has not been advocated, as staging and debulking occur at the time of surgery. ^6,11 However, cross-sectional imaging has a role in preoperative surgical planning, as well as in determining patients who may benefit from preoperative adjuvant chemotherapy. ^6,11

Cross-sectional imaging

Cross-sectional imaging is used preoperatively to plan the optimal surgery when the diagnosis is uncertain, and as guidance for percutaneous biopsy or paracentesis for symptomatic relief of gross ascites. ⁶ For surgical planning, imaging can direct the surgeon to areas of disease that may be difficult to assess surgically, such as the dome of the diaphragm and retroperitoneum. ¹² In addition, the identification of bowel invasion can alter surgical approach and treatment. ¹² The identification of nonresectable tumors is another important aspect of preoperative imaging. Tumor nonresectability has been defined as tumor larger than 2 cm at the mesenteric root, porta hepatis, omentum of the lesser sac, intersegmental fissure of the liver, gastro-splenic ligament, diaphragm, or liver dome. Lymphadenopathy >1 cm at or above the celiac axis and the presence of presacral and extraperitoneal disease are also criteria for nonresectability. ⁹ Currently, CT is considered the most useful preoperative imaging technique, with a reported accuracy of 70% to 90%. ^6,9,11,12

On CT, malignant ovarian tumors may be entirely solid or cystic, or have mixed solid and cystic components, including septations or papillary fronds. Fine or coarse calcifications may be present. Contrast enhancement is variable and, if present, is seen in cyst walls and solid areas (figures 2 and 3). ^6,12

The early stages of ovarian cancer include direct invasion to local structures. Metastatic disease occurs via three pathways: 1) intraperitoneal seeding, 2) lymphatic invasion, and 3) hematogenous spread. ^11,12,13

Intraperitoneal metastatic disease has a classic appearance. Peritoneal implants frequently occur on the omental and peritoneal surfaces. These include the cul-de-sac, infundibulopelvic ligaments, right paracolic gutter, right subdiaphragmatic surface, liver surface, and mesenteries. ¹¹ Other common sites of organ involvement include the spleen and bowel; bowel obstruction can occur late in the disease secondary to tumor invasion. Less frequently, extra-abdominal and intrahepatic metastases occur late in the disease (figure 4). ¹⁴

Previously reported limitations of CT include the lack of visualization of peritoneal implants <5 mm in diameter. ^9,12 On CT, peritoneal implants appear as soft-tissue or fluid-attenuated masses; if ascites is present, tumor implants are more readily visible (figure 5). Current CT scanners can detect 50% of peritoneal implants as small as 5 mm. ^12,13 Buy et al ¹³ reported identification of peritoneal implants as small as 2 to 3 mm when large amounts of ascites was present. The most common sites of peritoneal implants identified by CT preoperatively were the right subphrenic space, pouch of Douglas, and greater omentum. Involvement of the greater omentum with spread of tumor is known as omental "caking." The omental "cake" on CT is shown by confluent soft-tissue nodules between the anterior abdominal wall and adjacent bowel, with obliteration of the intervening fat. ¹⁵

Lymphatic involvement with ovarian cancer is shown on CT by enlarged retroperitoneal lymph nodes. Specifically, lymph nodes >1 cm may occur in the para-aortic, external iliac, obturator, and hypogastric lymph node chains. ¹¹ One important complication of lymphadenopathy is hydronephrosis due to extrinsic ureteral compression. Metastatic disease occurs in the anterior mediastinum and right thoracic duct via retrosternal lymphatic channels. ^11,13

Sites of hematogenous spread of ovarian cancer are also seen on CT. The most common sites of hematogenous spread are the liver, lung, spleen, pancreas, and kidney (figure 6). ¹¹

Postoperative imaging is used in three situations: 1) to identify disease progression, 2) to define treatment endpoints, and 3) to determine if further surgery or a "second-look" operation is necessary. ⁶ Imaging evaluation is used in conjunction with clinical examination and CA-125 levels in postoperative patient management. ⁶

Like CT, MRI can diagnose ovarian malignancy in the evaluation of the abnormal pelvis approximately 80% of the time and can detect abdominal spread 80% to 90% of the time. ⁵ Advantages of MRI include tissue characterization, identification of local tumor extension, and identification of tumor implants involving the hemi-diaphragm, liver surface, and liver parenchyma. ¹² Because of its multiplanar capability, MRI can distinguish between uterine and ovarian origin of a pelvic mass. ^6,12 MR appearances of ovarian tumors vary and are based upon tissue content.

T1-weighted images enable detection of hemorrhage or fat in an ovarian mass as well as the presence of lymphadenopathy. The high signal intensity of the ovarian follicles helps to differentiate the adjacent bowel and uterus from the ovary on T2-weighted images. ¹² The T2-weighted images are also used for tissue characterization. Contrast-enhanced T1-weighted images provide higher resolution of tissue characteristics and identify peritoneal and omental implants. ¹² On MRI, these appear as nodules of medium signal intensity. ¹⁵ Fat-saturation and gradient-recall echo technique (GRE), are used to further evaluate tissue character. Finally, fast spin echo (FSE) technique decreases image acquisition time, thereby decreasing motion artifact and improving resolution. ¹²

Conclusion

Ovarian carcinoma, the most lethal gynecologic cancer, is a complex grouping of tumors that often have a silent progression, spreading over surfaces of intraperitoneal structures. An initial evaluation of a pelvic mass is done with ultrasound. Transabdominal ultrasound is performed in conjunction with transvaginal ultrasound and color Doppler evaluation. While ultrasound is a useful, noninvasive, and inexpensive modality to confirm an ovarian mass, its inability to differentiate benign from malignant disease and disease extent limits its utility in the assessment of ovarian carcinoma. CT has been established as the primary imaging modality in the evaluation of ovarian tumors, preoperative staging of disease, and posttreatment response. This is due to its easier accessibility, quicker imaging and interpretation time, physician familiarity, and lower cost than MRI. ⁹ MRI offers a variety of techniques and has multiplanar capability. It is usually reserved for patients with contraindications to CT scan, such as pregnancy, contrast allergy, or abnormal renal function. ⁹ However, MRI is becoming more widely used for tissue characterization, and for the evaluation of local and distant spread of disease, especially in evaluation of the liver surface and diaphragm. AR

Acknowledgment

The authors gratefully thank Ms. Louise Logan for her preparation of the manuscript.

References

1. Gershenson D, McGuire WP: Ovarian cancer: Controversies in Management. p 425. Churchill Livingston, New York, 1998.

2. DiSaia P, Crosman WT: Clinical Gynecologic Oncology. p 282. Mosby, Baltimore, 1997.

3. Russell P, Farnsworthy A: Surgical Pathology of the Ovaries. pp 249-338. Churchill Livingston, Singapore, 1997.

4. Scully R, Young R, Clement P: Tumors of the Ovary, Maldeveloped Gonads, Fallopian Tubes, and Broad Ligaments. p 125. AFIP, Washington 1998.

5. Kurtz AB, Tsimikas JV, Tempany CMC, et al: Diagnosis and staging of ovarian cancer: Comparative values of Doppler and conventional US, CT, and MR imaging correlated with surgery and histopathologic analysis--Report of the Radiology Diagnostics Oncology Group. Radiology 212:19-20, 1999.

6. Johnson RJ: Radiology in the management of ovarian cancer. Clin Radiol 48:75-82, 1993.

7. Spencer JA, Kurtz AB: Diagnosing early ovarian cancer with ultrasound--Research goal or clinical reality? Clin Radiol 48:83-88, 1993.

8. Kurtz AB, Middleton WD: Ultrasound: The Requisites. pp 407-413. Mosby-Year Book, Inc., St. Louis. 1996.

9. Karlan BY, Platt LD: Ovarian cancer screening: The role of ultrasound in early detection. Cancer 76:2011-2015, 1995.

10. Forstner R, Hedvig H, Occhipinti KA, et al: Ovarian cancer: Staging with CT and MR Imaging. Radiology 197:619-626, 1995.

11. Kawamoto S, Urban BA, Fishman EK: CT of epithelial ovarian tumors. RadioGraphics 19:S85-S102, 1999.

12. Forstner R, Hricak H, White S: CT and MRI of ovarian cancer. Abdom Imag 20:2-8, 1995.

13. Buy JN, Moss AA, Ghossain MA, et al: Peritoneal implants from ovarian tumors: CT findings. Radiology 169:691-694, 1988.

14. Mishell DR, Stenchever MA, Droegemueller W, Herbst AL: Comprehensive Gynecology. pp 911-924. Mosby-Year Book, Inc., St.Louis, 1997.

15. Lee J, Sagel S, Stanley R, Heiken J: Computed Body Tomography with MRI Correlation, vol 2. pp 1238-1239. Lippincott Raven Publishers, Philadelphia, 1998.