Introduction

With an estimated annual incidence of 313 959 new cases and 207 252 deaths worldwide, ovarian cancer is the eight leading cause of death from cancer in women.1 When treating patients with ovarian cancer, complete resection to no residual disease in primary cytoreductive surgery has been demonstrated to be the most important prognostic factor.2 3 The complete resection rate recommended by the European Society of Gynecologic Oncology (ESGO) is >65% for primary debulking surgeries.4 However, the success rate for achieving complete resection surgery varies widely, from 19% to 86%, in the literature.5 6

Several studies have tried to establish factors that most accurately predict which patients will achieve complete resection in the upfront surgery.7–12 Various factors have been evaluated, including circulating biomarkers, laparoscopy based scores, and imaging based factors.7–9 Previous works have investigated the value of 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) or CT scan in predicting optimal cytoreduction in upfront debulking surgery.9–12 However, all of these series showed low rates of complete tumor resection (33–35%).

Our objective was to evaluate the value of a preoperative 18F-FDG PET/CT scan in predicting complete resection in primary surgery of advanced ovarian cancer among centers with a high rate of complete resection.

Methods

This multicenter, observational, retrospective study evaluated patients with advanced ovarian cancer who underwent primary cytoreductive surgery in two Spanish institutions between January 2017 and January 2022. The study protocol was approved by the institutional review board and the local ethics committee. In 2017, both centers routinely included 18F-FDG PET/CT scan as part of the preoperative diagnostic work-up in patients with suspected advanced ovarian cancer. In all cases, a previous ultrasound evaluation was performed. If a high tumor burden was suspected on preoperative 18F-FDG PET/CT or CT scan (peritoneal carcinomatosis), diagnostic laparoscopy was performed to assess cytoreducibility.

The peritoneal cancer index score was calculated laparoscopically, and if >24 points, neoadjuvant chemotherapy was administered. If isolated suspicious extra-abdominal lymph nodes were detected on preoperative examination, but a complete abdominal resection was considered feasible, upfront surgery was performed. Whenever possible, extra-abdominal lymph nodes were removed at the time of surgery. All cases were presented at the multidisciplinary tumor board of each institution.

18F-FDG PET/CT Image Acquisition

18F-FDG PET/CT studies were performed on a PET/CT scanner (Biograph mCT Flow, Siemens, Knoxville, USA). Patients fasted for at least 6 hours before 18F-FDG injection. Blood glucose level was determined in capillary blood samples (Accu-Chek Aviva System, Roche). 18F-FDG (335.3 MBq) was intravenously administered, providing a mean serum glucose level of 5.2±0.7 mmol/L. In those patients with a blood glucose level >8.0 mmol/L, a bolus of short acting insulin analog (Lispro, Eli Lilly) was administered. If capillary blood glucose exceeded 13.8 mmol/L, the study was canceled and rescheduled. Before the PET scan, for attenuation correction, a low dose CT scan was obtained without contrast enhancement from the skull base to the thigh. After that, CT images with contrast medium were also obtained. Finally, PET images were acquired 1 hour after 18F-FDG injection, from the skull to the mid-thigh, at a speed of 0.7 mm/seg.

PET data were reconstructed according to the standard protocol consisting of three-dimensional ordered subset expectation maximization iterative reconstruction with both time of flight and point spread function modeling corrections, using three iterations and 21 subsets. A 2.0 mm full width at half maximum Gaussian post-filter was applied. An additional reconstruction was performed according to the European Association of Nuclear Medicine standards13 (ordered subset expectation maximization+time of flight, three iterations 21 subsets, 6.5 mm full width at half maximum Gaussian filter). All resulting PET images had a matrix size of 200×200 with a 4.1×4.1 mm2 pixel size.

18F-FDG PET/CT performed outside of the two institutions was not included in the study. All 18F-FDG PET/CT results were reported by nuclear medicine physicians specialized in gynecologic oncology malignancies.

All reports and 18F-FDG PET/CT images were reviewed. A finding was considered positive only if described as 'suspicious for malignancy' in the report (based both on increased FDG uptake and morphological features). Following ESGO recommendations for ovarian cancer,14 preoperative 18F-FDG PET/CT findings were divided into: right and left adnexa, pouch of Douglas, vagina, uterus, bladder, rectosigmoid, rectovaginal septum, pelvic wall, pelvic lymph nodes, para-aortic lymph nodes (from aortic bifurcation to left renal vein), right and left gutters, small bowel, omentum, large bowel, appendix, bowel mesentery, right and left diaphragm, liver surface, liver parenchyma, lesser omentum, stomach, pancreas, spleen, hepatic hilium lymph nodes, celiac lymph nodes, extra-abdominal lymph nodes (retroperitoneal nodes above left renal vein or inguinal nodes), abdominal wall, skin, and presence of ascites.

Calculation of Peritoneal Cancer Index

Previous studies11 15 have investigated the peritoneal cancer index score on 18F-FDG PET/CT. Regions corresponding to the small bowel (9-12) have the lowest accuracy on 18F-FDG PET/CT. For this reason, most of the studies evaluating the peritoneal cancer index on18F-FDG PET/CT keep the quadrants of the small bowel out of the score, limiting the score to nine quadrants. Regions are usually numbered as follows: 0, central; 1, upper right; 2, epigastrium; 3, upper left; 4, left flank; 5, lower left; 6, pelvis; 7, right lower; and 8, right flank. A score of 0–3 is assigned to each quadrant based on the peritoneal cancer index surgical score. A lesion size score of 0 indicates no visible tumor burden in the peritoneum. Lesion size scores of 1, 2, and 3 describe maximum tumor diameters of up to 0.5, 5.0, and >5 cm, respectively. A maximum of 24 points can be obtained.

After thorough discussion, a modified PET/peritoneal cancer index scale was proposed based on the morphometabolic findings of the 18F-FDG PET/CT. Because patients with advanced ovarian cancer are likely to have a high tumor burden in the pelvis, but this does not usually translate into incomplete cytoreductive surgery, we decided to create a composite score from the three lower quadrant scores of the pelvic area, resulting in a scale of seven quadrants with a maximum score of 18 (online Supplemental Material).

Two gynecologic oncologists (FB and MC) retrospectively calculated modified 18F-FDG PET/CT peritoneal cancer index scores independently by reviewing reports and images. If any discrepancies were observed, both modified peritoneal cancer index scores were calculated to obtain the final peritoneal cancer index score. Surgical peritoneal cancer index scores were obtained from surgical reports, which had been calculated and registered by the surgeons at the time of surgery following the recommendations of Jacquet and Sugarbaker.16 Complete resection surgery was defined as 'no macroscopic disease' at the end of surgery.2 Clinical and pathological parameters, including age, International Federation of Gynecology and Obstetrics (FIGO) stage, histology, tumor grade, preoperative CA 125 level, ascites, Eastern Cooperative Oncology Group Performance Status, surgical time, and surgical procedures were reviewed and retrieved.

Inclusion criteria were: histological confirmation of invasive epithelial ovarian carcinoma, primary peritoneal carcinoma, or fallopian tube carcinoma; preoperative FIGO stage III or IV; upfront debulking surgery; and 18F-FDG PET/CT performed 1 month before surgery. Exclusion criteria were 18F-FDG PET/CT performed at a non-participating center or any concurrent cancer at the time of 18F-FDG PET/CT.

Statistical Analysis

Continuous variables are reported as median (range) and categorical variables as absolute frequencies and percentages. Associations between categorical variables were assessed using Fisher’s exact test, and the Wilcoxon rank sum test was used for continuous variables between groups. Several cut-offs were analyzed for age, PET peritoneal cancer index, CA125, and predictive value score. The most predictive cut-off of residual disease in the receiver operating characteristic was used to group patients. All variables with a p value <0.20 in the univariate analyses were introduced in a forward stepwise procedure. Two significance levels were specified in the process: 5% for predictor addition to the model and 10% for predictor removal; b coefficients were divided by the smallest value and rounded to integers to calculate each variable’s ratio in the predictive index. The area under the receiver operating characteristic curve of the predictive index was calculated.

According to their 'predictive value score' for complete resection surgery, participants were classified into favorable (0–4 points) or unfavorable (5–16 points) risk groups. A p value <0.05 was considered to indicate statistical significance. Statistical analysis was performed using SPSS 27.0 (IBM SPSS Statistics, Armouk, USA)

Results

Among 299 patients diagnosed with advanced ovarian cancer, 45 underwent upfront primary cytoreductive surgery that met the inclusion criteria. Complete resection was achieved in 36 (80%) patients. Patient selection is shown in Figure 1. Patient and tumor characteristics are shown in Table 1. The median operating time was 335 min (range 50–651). The median modified 18F-FDG PET/CT peritoneal cancer index was 5 (range 3–15) and the median surgical peritoneal cancer index score was 11.5 (range 3–28). Diaphragmatic stripping was performed in 25 (55.6%) patients and colorectal resection in 23 (51.1%) patients. Seven (15.6%) patients underwent small bowel resection, 5 (11.1%) patients had a splenectomy, and 6 (13%) patients had an ostomy at completion of surgery. Seven patients (15.6%) had suspicious extra-abdominal lymph nodes on preoperative 18F-FDG PET/CT (one inguinal, four cardiophrenic, and two supraclavicular). In four of these seven cases, complete abdominal resection was not possible. The most frequent site of residual disease after surgery was the mesentery and serosa of the small bowel (four patients, 44%). Location and size of residual disease are presented in the supplementary material (online Supplemental Material).

Figure 1

Receiver operating characteristic (ROC) curve for the predictive score model. AUC, area under the curve.

Table 1

Patient characteristics

With the data from this study, a receiver operating characteristic curve was generated and the most predictive cut-off values for the modified 18F-FDG PET/CT peritoneal cancer index, age, and CA125 were calculated. The cut-off values were: 6 for the peritoneal cancer index, 58 years for age, and 500 U/mL for CA125. On univariate analysis, only the presence of extra-abdominal lymph node involvement was significantly associated with macroscopic residual disease after cytoreduction surgery (Table 2). However, on multivariate analysis, the presence of extra-abdominal lymph node metastasis, modified peritoneal cancer index score≥6, age ≥58 years, and American Society of Anesthesiology (ASA) score ≥3 were significantly associated with incomplete tumor resection (Table 3).

Table 2

Radiologic and clinical criteria (univariate analysis)

Table 3

Multivariate model and predictive score

A predictive value score (Table 3) was assigned to criteria that were significant in the multivariate model, which was based on the odds ratio. For this multivariate predictive model of non-complete cytoreduction, a receiver operating characteristic curve was generated and the area under the curve was 0.881 (Figure 2). A predictive value of 5 or more was the most predictive cut-off for non-complete cytoreduction. The complete resection rate was 91.7% in patients with a score of ≤4 and 33.3% in patients with a score of ≥5 points for the predictive value score.

Figure 2

Selection criteria. Consolidated Standards of Reporting Trials (CONSORT) diagram. PET/CT, positron emission tomography/computed tomography; PCI, peritoneal cancer index.

DiscussionSummary of Main Results

Our study showed that extra-abdominal lymph node involvement was the main predictive factor of gross residual disease at primary cytoreductive surgery. On multivariate analysis, a modified 18F-FDG PET/CT peritoneal cancer index score of ≥6, age ≥58 years, and ASA score ≥3 were also associated with gross residual disease. With these criteria, a predictive score (0–16) was created. A value of ≥5 was the most predictive cut-off point, with a rate of complete resection surgery of 33.3% in this group of patients.

Results in the Context of Published Literature

Proper selection of patients for primary debulking surgery remains a challenge. Currently, in many centers, diagnostic laparoscopy is performed prior to cytoreductive surgery to better select patients for primary surgery. This strategy results in a better selection of patients who will benefit from higher rates of complete cytoreduction.

Previous studies have analyzed different imaging techniques and clinical variables to predict complete resection surgery. The cut-off values in our cohort for age were 58 years and for CA125, 500 U/mL. Our results are consistent with previous reports7 17–19

A previous study by Suidan et al19 included 669 patients and analyzed the clinical characteristics and radiological findings on preoperative CT scan. The authors initially evaluated optimal cytoreduction (<1 cm) and found three clinical (CA125, age, and ASA) and six radiological criteria (suprarenal retroperitoneal lymph nodes>1 cm, diffuse small bowel adhesions/thickening, and lesions>1 cm in the small bowel mesentery, root of the superior mesenteric artery, perisplenic area, and lesser sac) that could predict optimal cytoreduction. Three years later, the authors performed a secondary post hoc analysis9 focusing on patients with complete resection (no macroscopic disease). In this analysis, they found the same three clinical criteria but nine radiologic criteria associated with complete resection and developed a predictive model score with an area under the curve of 0.72. In our study, two of the clinical criteria (ASA ≥3 and age ≥58 years) were also associated with complete resection. In the radiologic findings, we found extra-abdominal lymph node involvement as the most predictive factor of gross residual disease.

In the study by Suidan et al, retroperitoneal lymph nodes above the renal hilium demonstrated an odds ratio of 1.79. However, these studies are not comparable. First, the complete resection rate in the Suidan cohort was 33% (which is the same complete resection rate as our unfavorable prognosis group). Second, some of the radiologic criteria included in the Suidan study were indications in our centers to recommend neoadjuvant chemotherapy (intraparenchymal liver, lung, or pleural metastases). In addition, we also included the modified peritoneal cancer index, trying to better assess the total tumor burden, which was not used in the Suidan study. Another study12 analyzed the Suidan model in PET/CT. The authors prospectively studied eight quantitative criteria based on the Suidan model and constructed a PET score in 31 patients. They found that higher PET scores and tumor burden were associated with gross residual disease. However, the complete resection rate in this cohort was also 35.5%, which would independently impact oncological outcomes.

The presence of distant lymph node metastases is associated with a worse prognosis in advanced ovarian cancer.20 However, suspicious preoperative extra-abdominal lymph nodes should not preclude primary cytoreductive surgery when complete abdominal cytoreduction can be achieved otherwise.21 22 Luger and colleagues investigated the prognostic role of enlarged cardiophrenic lymph nodes on preoperative CT scan. The authors collected 178 patients with stage III/IV ovarian cancer and found that the presence of enlarged cardiophrenic lymph nodes was associated with worse survival and lower rates of complete abdominal resection.22 This may be important, as many reports previously demonstrated better accuracy of 18F-FDG PET/CT for detecting extra-abdominal lymph node metastases.23–26 The importance of 18F-FDG PET/CT in the primary setting remains unclear. Other studies have analyzed its use in predicting complete resection surgery27 28 but evidence comparing the predictive value of complete resection between 18F-FDG PET/CT and CT alone is scarce.

Strengths and Weaknesses

To our knowledge, this is the first study to validate a 18F-FDG PET/CT predictive score for complete resection in primary cytoreductive surgery in a cohort with >50% complete resection rate. In addition, all patients were collected from two qualified centers for ovarian cancer (51.1% colorectal resection and ESGO accredited centers). Both circumstances minimize the risk of surgery dependent bias and allow better interpretation of preoperative variables. The main weaknesses of this study were the retrospective design and the small number of patients. Additionally, the COVID-19 pandemic occurred during this study period, which may have affected scheduled surgeries and treatments.

Implications for Practice and Future Research

In this study, we have proposed a new predictive score that combines preoperative 18F-FDG PET/CT findings, including a modified peritoneal cancer index score, and two clinical variables to predict no gross residual disease at primary cytoreductive surgery. Our results need to be validated in a larger cohort to ascertain external validity. In future studies, investigating the role of 18F-FDG PET/CT before primary cytoreductive surgery in advanced ovarian cancer could be implemented to help predict the best surgical outcomes.