In this study, we identified two important clinical features of postoperative ovarian cancer. The first is that chest metastasis was never detected alone, and was always preceded by abdominopelvic recurrence. Second, either medical examination findings or tumor marker levels suggested recurrence in 13 of the 15 patients with recurrence (87%), but not in the remaining 2 patients.
Regarding the first clinical feature, 2 of our 3 patients with chest metastases originally had stage I disease, but abdominopelvic recurrence preceded chest recurrence in all 3 of them. Hematogenic spread is a rare route of metastasis in ovarian cancer,5 in contrast to lymphatic metastasis and metastasis to the adjacent organs, which are the most common routes.6 In our study, one patient (Patient No. 1), had peritoneal dissemination (PD) and para-aortic lymph node (PAN) metastasis preceding CPLN and supraclavicular lymph nodes (SNL) metastasis, and another patient (Patient No. 15) also had PD and PAN metastasis preceding CPLN and PSLN metastasis. These facts imply that the cancer spread from abdominopelvic lymph nodes to thoracic lymph nodes through the lymphatic tract. This result indicates that routine chest CT can be eliminated, and it may be reasonable to add chest CT only after detection of abdominopelvic recurrence on abdominopelvic CT. Such a strategy could reduce radiation exposure and radiologists’ reporting workload, and eliminate the additional medical costs for having radiologists create reports for unnecessary chest CT scans.7
There are two previous studies similar to our study. Sella et al. reported the rate of lung metastasis from ovarian cancer was 6% in 82 patients with stage III or IV disease.8 The lower incidence of lung metastasis (2.1%) in our study can be explained by inclusion of patients with stage I and II disease, which is unlikely to metastasize to the lungs. In previous studies, pulmonary metastases tended to be preceded either by abdominopelvic disease9, 10, 11 or a rise in tumor markers.8 This result is consistent with our study. On the other hand, Dachman et al. observed chest metastases without abdominopelvic recurrence in six (2.7%) of 226 follow-up CT scans.12 This is in conflict with our study, but we presume they were unable to detect the abdominopelvic disease in those six patients because their CT scan sensitivity (7–10 mm collimation) was too low to depict small implants. Thin slices and axial and multiplanar reconstructed images have been reported to be useful in detection of small implants.13
Regarding the second clinical feature, 13 of 15 patients with recurrence (87%) had signs of recurrence in their medical examination findings or tumor marker levels, which indicates that clinical examinations such as medical examination findings and tumor marker elevation are useful to identify signs of recurrence. However, the fact that the remaining 2 patients had none of those signs implies the need for routine abdominopelvic CT scan regardless of results of tumor marker levels and medical examination findings.
Although transvaginal ultrasound is a noninvasive examination that is easily performed and effective in detecting ascites and Douglas fossa dissemination,14 it has some inherent limitations. It cannot always detect peritoneal dissemination due to its low sensitivity,15 and cannot detect pelvic or para-aortic lymphadenopathy and inguinal lymphadenopathy because of its limited field of view. In our study, transvaginal ultrasound showed low sensitivity, specificity, and AUC for detection of recurrence of ovarian cancer (sensitivity: 33.3%, specificity: 97.7%, and AUC: 0.69).
In our study, only one patient among the 15 symptomatic patients with recurrence had respiratory distress. In previous studies, very few patients first presented with symptoms,16 and very few recurrences were detected by symptoms alone.17 These results are consistent with our results and demonstrate that detection of recurrence by symptom monitoring may be difficult.
A multicenter European trial concluded that treating recurrences only based on detectable CA-125 levels in patients who are asymptomatic is not associated with increased survival and is associated with decreased quality of life,18 but many other studies19,20 have indicated that CA-125 is useful for the detection of ovarian cancer recurrence, although they used various thresholds. In our study, tumor markers (CA19-9 and CA-125) showed higher sensitivity, equal specificity, and higher AUC for detection of ovarian cancer recurrence (sensitivity: 73.3%, specificity: 97.7%, and AUC: 0.86) than transvaginal ultrasound. This was especially true for stage III tumors: all patients (7/7) in that stage had elevated tumor markers. In patients with stage III ovarian cancer, tumor marker levels might be reliable enough to predict recurrence, and even routine abdominopelvic CT might be omitted until tumor markers become elevated. However, our sample size is too small to draw a conclusion. When tumor markers and transvaginal ultrasound were combined, the sensitivity was 86.7%, the specificity 95.4%, and the AUC 0.91. These values are all better than transvaginal ultrasound alone and tumor markers alone, but they cannot completely predict recurrence. As a result, routine abdominal CT is unavoidable to detect recurrence during follow-up for ovarian cancer.
Our study has two limitations. First, most of the positive findings of recurrence and metastasis were based on our criteria, not pathological examination. However, in clinical practice, chemotherapy is started when CT findings indicate distant metastasis as biopsy to obtain a histological specimen of a recurrent lesion is invasive. Second, our study is a single-center retrospective study in Japan. The distribution of histological types of ovarian cancer within our patient group is similar to that in Asia, but serous carcinoma is the most common type in Europe and North America.21 Most patients with ovarian cancer present with stage III disease per FIGO staging,22 but in our study over half of patients presented with stage Ⅰ disease. This may be because pelvic examinations using transvaginal ultrasound even for screening have become routine and more cases of asymptomatic ovarian cancer were detected in an earlier stage.23 Therefore larger scale multicenter studies are needed to support our findings.