Diagnostic value of integrated ¹⁸F-fluoro-2-deoxyglucose positron emission tomography/computed tomography in recurrent epithelial ovarian cancer: accuracy of patient selection for secondary cytoreduction in 134 patients

Ovarian cancer is the sixth most common cancer in women and the leading cause of death in gynecologic malignancies [1]. Most women with advanced ovarian cancer experience recurrence within 5 years of diagnosis, despite previously achieving complete remission with cytoreductive surgery and adjuvant chemotherapy [2]. To detect cancer recurrence, various surveillance methods after the completion of primary treatment have been proposed, including periodic evaluation of tumor marker levels and the use of imaging modalities, such as ultrasonography, computed tomography (CT), and integrated ¹⁸F-fluoro-2-deoxyglucose-positron emission tomography/computed tomography (¹⁸F-FDG-PET/CT) [3]. In previous studies, serial monitoring of the cancer antigen 125 (CA-125) level showed a sensitivity of 79% to 95% and a positive predictive value (PPV) close to 100% [4, 5, 6]. However, unlike tumor marker levels, imaging modalities can assess the site of recurrence, disease volume, and spread extent, which can help gynecology oncologists to decide on a treatment plan [7]. Rapid identification of recurrence and its range through these imaging modalities can improve survival in patients who undergo complete resection through secondary cytoreduction, but there is no benefit in survival when residual tumor remains [8, 9].

Recently, ¹⁸F-FDG-PET/CT has become widely used to determine whether secondary cytoreduction in patients with suspected recurrence of epithelial ovarian cancer (EOC) [10, 11]. Compared with CT, positron emission tomography/computed tomography (PET/CT) may be able to detect recurrence earlier through detect hypermetabolic uptake of ¹⁸F-fluoro-2-deoxyglucose (¹⁸F-FDG) when localized recurrent tumors smaller than 1 cm occur. However, limited data are available on the diagnostic accuracy of ¹⁸F-FDG-PET/CT in suspected recurrent EOC with non-disseminated lesions, which can be an indication for secondary cytoreduction. In addition, several previous studies were derived from patient-based analysis rather than lesion-based analysis, and only a few of the study patients actually underwent secondary cytoreduction.

Therefore, we investigated the diagnostic value and accuracy of patient selection for secondary cytoreduction of ¹⁸F-FDG-PET/CT and compared them with those of contrast-enhanced CT (CECT) in patients with suspected recurrence of EOC with non-disseminated lesions by correlating imaging findings with pathologic diagnosis of secondary cytoreduction.

The patients' baseline characteristics are described in Table 1. The median interval between primary treatment and secondary cytoreduction was 38.2 months (range, 7.1–136.6). The median overall survival after secondary cytoreduction was 39.1 months (range, 2.1–126.1). At the end of follow-up, 23 patients (17.2%) had achieved complete remission, 23 patients (17.2%) were alive with disease, and 88 patients (65.7%) had died of disease.

Table 1
Baseline characteristics of the study cohort (n=134)

• Click for larger image
• Click for full table
• Download as Excel file

1. Patient-based analysis

Among the 134 patients, 73 underwent ¹⁸F-FDG-PET/CT, 129 underwent CECT, and 68 underwent both imaging tools before secondary cytoreduction (Table 2). Overall, 124 patients (92.5%) were confirmed to have recurrence by pathologic diagnosis after secondary cytoreduction. In the 73 patients who underwent preoperative ¹⁸F-FDG-PET/CT, 65 (89.0%) were confirmed to be positive for malignancy, giving 98.5% sensitivity (95% confidence interval [CI]=91.8%–99.7%), 87.7% accuracy (95% CI=80.1%–95.2%), and 88.9% PPV (95% CI=79.6%–94.3%). In the 129 patients who underwent CECT, 119 (92.2%) were diagnosed with recurrence after secondary cytoreduction, giving 96.6% sensitivity (95% CI=91.7%–98.7%), 89.1% accuracy (95% CI=83.8%–94.5%), and 92.0% PPV (95% CI=85.9%–95.6%). In the 68 patients who underwent both imaging tools, 60 (88.2%) were confirmed to have recurrence. Among these, 59 patients were accurately detected by ¹⁸F-FDG-PET/CT, with 98.3% sensitivity (95% CI=91.1%–99.7%), 86.8% accuracy (95% CI=78.7%–94.8%), and 88.1% PPV (95% CI=78.2%–93.8%), whereas 60 patients were detected by CECT, with 100% sensitivity (95% CI=94.0%–100.0%), 88.2% accuracy (95% CI=80.6%–95.9%), and 88.2% PPV (95% CI=78.5%–93.9%).

Table 2
Patient-based analysis (n=134)

• Click for larger image
• Click for full table
• Download as Excel file

The mean preoperative CA-125 level in the 10 false-positive patients was very low, at 28.8 U/mL (range, 3.2–81.8). Seven patients had less than 35 U/mL and only 3 patients had a 10 U/mL or more increase compared with the previous CA-125. Secondary cytoreduction according to ¹⁸F-FDG-PET/CT readings revealed false positives in 8 patients and residual tumor in 14 patients, giving an accuracy of patient selection for secondary cytoreduction of 69.4% (50 of 72). Secondary cytoreduction according to CECT readings revealed false positives in 10 patients and residual tumor in 35 patients, giving an accuracy of patient selection for secondary cytoreduction of 64.0% (80 of 125). Secondary cytoreduction according to ¹⁸F-FDG-PET/CT and CECT readings revealed false positives in 8 patients and residual tumor in 14 patients, giving an accuracy of patient selection for secondary cytoreduction of 67.2% (45 of 67). Most patients with residual tumor were found to have peritoneal disseminated recurrence smaller than 1 cm during surgery. The average treatment-free interval of these patients was 24.9 months.

2. Lesion-based analysis

A total of 313 macroscopic localized lesions with suspected recurrence by preoperative imaging tools or intraoperative findings were resected during the secondary cytoreduction and submitted for pathologic examination (Table 3); 169 lesions were removed from patients who underwent preoperative ¹⁸F-FDG-PET/CT, 304 lesions were removed from patients who underwent preoperative CECT, and 160 lesions were removed from patients who preoperatively underwent both ¹⁸F-FDG-PET/CT and CECT. Overall, 254 of the 313 lesions (81.2%) were confirmed to be recurrence. In the 169 lesions removed from patients who underwent preoperative ¹⁸F-FDG-PET/CT, 135 lesions (79.9%) were confirmed to be recurrence and 124 lesions were accurately detected by ¹⁸F-FDG-PET/CT, giving 91.9% sensitivity (95% CI=86.0%–95.4%), 81.1% accuracy (95% CI=75.2%–87.0%), and 85.5% PPV (95% CI=78.9%–90.3%). The mean of the maximum standardized uptake value (maxSUV) of the hypermetabolic lesions was 6.92 (range, 1.6–26.7). The mean maxSUV of false-positive sites and recurrence sites were 5.60 and 6.85, respectively. The p-value in logistic regression analysis of the maxSUV was 0.216. In the 304 lesions removed from patients who underwent preoperative CECT, 247 lesions (81.2%) were positive for malignancy and 212 lesions were accurately detected by preoperative CECT, giving 85.8% sensitivity (95% CI=80.9%–89.6%), 78.6% accuracy (95% CI=74.0%–83.2%), and 87.6% PPV (95% CI=82.9%–91.2%). In the 160 lesions removed from patients who preoperatively underwent both imaging tools, 128 lesions (80.0%) were diagnosed to be recurrence after secondary cytoreduction. Of these, 119 lesions were accurately detected by ¹⁸F-FDG-PET/CT, giving 93.0% sensitivity (95% CI=87.2%–96.3%), and 109 lesions were accurately detected by CECT, giving 85.2% sensitivity (95% CI=78.0%–90.3%). The characteristics of the 21 false-positive lesions and the 11 false-negative lesions on ¹⁸F-FDG-PET/CT are summarized in Tables 4 and 5.

Table 3
Lesion-based analysis (n=313)

• Click for larger image
• Click for full table
• Download as Excel file

Table 4
Characteristics of 21 lesions with false-positive ¹⁸F-FDG-PET/CT findings

• Click for larger image
• Click for full table
• Download as Excel file

Table 5
Characteristics of 11 lesions with false-negative ¹⁸F-FDG-PET/CT findings

• Click for larger image
• Click for full table
• Download as Excel file

In the current study, we evaluated the diagnostic value of ¹⁸F-FDG-PET/CT and CECT for detecting the recurrence of EOC with non-disseminated lesions. In patient-based analysis, we found no significant difference in the diagnostic ability of the 2 imaging tools. In lesion-based analysis, ¹⁸F-FDG-PET/CT tended to have higher sensitivity than CECT, suggesting the possibility of early diagnosis and treatment before lesion dissemination. However, in terms of accuracy and PPV, there was no significant difference in ability compared with CECT, and the combination of both imaging tools showed no superiority in diagnostic value compared with ¹⁸F-FDG-PET/CT alone.

There have been several reports on the diagnostic value of ¹⁸F-FDG-PET/CT in patients with suspected recurrent EOC [12, 13, 14, 15, 16]. In these studies, ¹⁸F-FDG-PET/CT showed a sensitivity of 82%–100%, a PPV of 93.8%–100%, and an accuracy of 80%–97.7%. However, most previous studies involved patient-based analysis rather than lesion-based analysis. In patient-based analysis of patients with multiple lesions, even if the predictions of individual lesions are somewhat incorrect, any malignant lesions are considered to indicate an accurate prediction. Accurate detection of each lesion is essential for precise surgical range determination and prevention of unprofitable surgery. In one lesion-based analysis in 2012, the approach showed a sensitivity of 80.8% [17]. A prospective lesion-based study of 22 patients published in 2009 determined a sensitivity of 60.5% and accuracy of 72.4% [18]. Although these studies were performed on a small number of patients, the overall diagnostic values were much lower than those of patient-based studies. We performed both patient-based and lesion-based analyses. The diagnostic value of the lesion-based analysis was lower than that of the patient-based analysis, but the difference was not as marked as in previous studies.

There have been 2 meta-analyses, with sensitivities of 88.6% and 91% and specificities of 90.3% and 88%, respectively [19, 20]. Gu et al. [19] demonstrates that AUC of an ROC curve is greater for PET/CT than that of CT or magnetic resonance imaging (MRI) and that this difference is statistically significant. However, they had many limitations, such as considerable variations between the included studies in terms of patient population, data presentation, and insufficient numbers of high-quality and large-scale prospective studies, and a mixture of patient-based and lesion-based analyses. Above all, only a few studies were focused on the diagnostic accuracy of ¹⁸F-FDG-PET/CT in suspected recurrent EOC with “non-disseminated lesions.” The most important goal of the rapid detection of recurrence using ¹⁸F-FDG-PET/CT is to promptly detect a non-disseminated status, which can be expected to improve the survival rate by performing complete resection through secondary cytoreduction. In recurrent EOC cases, local recurrence of cancer enough to perform secondary cytoreduction is very rare. Our study focused on non-disseminated recurrent EOC. And this is the only study we know of to only include patients who underwent secondary cytoreduction rather than ascites analysis or interventional pathologic confirmations such as fine needle biopsy.

Most previous studies were performed on less than 50 patients and there was no study included more than 100 patients. Our study is based on long-term data from a relatively large number of patients. We also attempted to increase the reliability of the data by excluding patients who were lost to follow-up.

In this study, the mean preoperative CA-125 level was as low as 28.8 U/mL (range, 3.2–81.8) in 10 patients with false-positive ¹⁸F-FDG-PET/CT findings and the preoperative increase was not clear. Based on this finding, patients with a CA-125 of less than 35 U/mL and no significant increase compared with previous tests should consider short-term follow-up without surgery if the imaging findings are uncertain because they are more likely to undergo unnecessary surgery.

Among 21 lesions with false-positive ¹⁸F-FDG-PET/CT findings, foreign body granuloma was found in 33.3% (7 of 21). Foreign body granuloma is a rare benign tumor caused by an inflammatory response of biological tissue to any exogenous substance in the tissue. If there is an implant, the healing reaction is different and it is called foreign body reaction. Foreign body reaction involves protein adsorption, fibroblasts, macrophages, multinucleated foreign body giant cells (macrophage fusion), and angiogenesis [21]. In women who have undergone surgery for ovarian cancer, foreign body granulomas mainly formed around the suture materials or clips. Silk sutures cause more allergic reactions and infections than monofilament sutures, and foreign body granulomas occur more frequently in Asia where silk sutures are used more frequently [22]. Foreign body granuloma is not well detected on CT or MRI because it is radiopaque. However, it has an inflammatory nature that can be detected as a local hypermetabolic lesion in ¹⁸F-FDG-PET/CT and thus increases its false-positive rate (Fig. 1). In our results, the mean maxSUV of foreign body granuloma was 4.41 (range, 2.2–7.2). A number of case reports have been published on this entity in various carcinomas [23, 24, 25]. However, only 1 or 2 cases were reported in each study and no large numbers of cases were identified, as in this study.

Fig. 1
¹⁸F-FDG-PET/CT images of a 47-year-old women with stage IV papillary serous carcinoma of the ovary after a staging operation followed by chemotherapy with paclitaxel plus carboplatin for 12 cycles and radiation therapy to the left supraclavicular area. The preoperative CA-125 level before secondary cytoreduction was 8.4 U/mL. These images show a hypermetabolic nodule (maxSUV=7.2) in the right pelvic region (white or black arrows) suggestive of malignancy. After secondary cytoreduction, pathology confirmed foreign body granuloma in the lesion. This patient achieved complete remission with no recurrence for more than 5 years.

¹⁸F-FDG-PET/CT, ¹⁸F-fluoro-2-deoxyglucose-positron emission tomography/computed tomography; CA-125, cancer antigen 125; CT, computed tomography; maxSUV, maximum standardized uptake value; PET, positron emission tomography.

• Click for larger image
• Download as PowerPoint slide

Of the 11 lesions with false-negative ¹⁸F-FDG-PET/CT findings, 9 were less than 1.5 cm in size. This result reconfirms the difficulty of detecting disseminated recurrence lesions smaller than 1 cm on PET/CT or CT [18].

This is the first study to compare the accuracy of patient selection for secondary cytoreduction according to imaging modality. In our present analysis, the accuracy of patient selection for secondary cytoreduction of ¹⁸F-FDG-PET/CT was 69.4%, which was higher than that of CECT at 64.0%. This accuracy makes it easy to judge whether or not to perform the surgery and will help clinicians to present accurate percentages of the possibility of complete resection in secondary cytoreduction to patients who considering secondary cytoreduction.

This study had some limitations of note. It was a retrospective study and there could be a selection bias by its nature. In our hospital, ¹⁸F-FDG-PET/CT is performed in most patients with suspected local recurrence based on CECT findings. Therefore, PET-CT interpretation may not be independent of CECT interpretation. Regarding the accuracy of patient selection for secondary cytoreduction, operators can make incorrect decisions about resectability by not referring to the results of imaging studies and this can act as a confounding factor. We also could not determine the specificity because the study population comprised patients who underwent secondary cytoreduction due to the positive results of an imaging study.

In conclusion, this study shows that ¹⁸F-FDG-PET/CT have higher diagnostic utility and the accuracy of patient selection for secondary cytoreduction than CECT. Early identification of recurrence and its range by imaging modalities can lead to higher survival in patients with recurrent EOC by complete resection through secondary cytoreduction. Patients with a CA-125 of less than 35 U/mL with no significant increase compared with previous tests show high risk of false-positive findings. Thus, there is still a need for integration of the results of ¹⁸F-FDG-PET/CT, CECT, and CA-125 levels to aid treatment planning and short-term follow-up should be considered instead of an operation if the imaging findings are uncertain.