Imaging of Peritoneal Metastases in Ovarian Cancer Using MDCT, MRI, and FDG PET/CT: A Systematic Review and Meta-Analysis

Simple Summary Ovarian cancer is the leading cause of death due to gynecologic malignancies. Peritoneal metastases represent the most common pathway for the spread of OC, both at the time of initial diagnosis and at recurrence. Accurate mapping of peritoneal metastases helps in planning the appropriate therapeutic strategy, predicting the likelihood of optimal cytoreduction, and identifying potentially unresectable or difficult disease sites that may require surgical technique modifications. Preoperative diagnostic work-up with multidetector CT (MDCT), MRI, including diffusion-weighted imaging (DWI), or FDG PET/CT plays a vital role in the accurate assessment of the extent of peritoneal carcinomatosis. In this article, the aim was to update the role of MDCT, MRI, including DWI, and FDG PET/CT in the detection of peritoneal metastases in ovarian cancer by conducting a systematic review and meta-analysis of the existing literature. Abstract This review aims to compare the diagnostic performance of multidetector CT (MDCT), MRI, including diffusion-weighted imaging, and FDG PET/CT in the detection of peritoneal metastases (PMs) in ovarian cancer (OC). A comprehensive search was performed for articles published from 2000 to February 2023. The inclusion criteria were the following: diagnosis/suspicion of PMs in patients with ovarian/fallopian/primary peritoneal cancer; initial staging or suspicion of recurrence; MDCT, MRI and/or FDG PET/CT performed for the detection of PMs; population of at least 10 patients; surgical results, histopathologic analysis, and/or radiologic follow-up, used as reference standard; and per-patient and per-region data and data for calculating sensitivity and specificity reported. In total, 33 studies were assessed, including 487 women with OC and PMs. On a per-patient basis, MRI (p = 0.03) and FDG PET/CT (p < 0.01) had higher sensitivity compared to MDCT. MRI and PET/CT had comparable sensitivities (p = 0.84). On a per-lesion analysis, no differences in sensitivity estimates were noted between MDCT and MRI (p = 0.25), MDCT and FDG PET/CT (p = 0.68), and MRI and FDG PET/CT (p = 0.35). Based on our results, FDG PET/CT and MRI are the preferred imaging modalities for the detection of PMs in OC. However, the value of FDG PET/CT and MRI compared to MDCT needs to be determined. Future research to address the limitations of the existing studies and the need for standardization and to explore the cost-effectiveness of the three imaging modalities is required.


Introduction
Ovarian cancer (OC) represents the fifth most commonly diagnosed cancer among women, the fifth cause of cancer death, and the commonest cause of death due to gynecologic malignancies [1][2][3][4][5][6][7][8].An estimated number of 19,680 new cases of OC are expected to be diagnosed in the US and 12,740 women are expected to die from the disease in 2024 [2].Most cases (90%) are epithelial ovarian carcinomas and the majority are high-grade serous carcinomas.The most important revision in the last FIGO staging classification is that ovarian, fallopian, and primary peritoneal cancers are considered as one entity [9].
Ovarian cancer has a poor prognosis, with a 5-year relative survival rate of 48%, mainly because most women are diagnosed with advanced-stage disease [2].Moreover, the percentage of recurrence in OC is very high.The standard of care in OC includes either primary debulking surgery (PDS) followed by adjuvant chemotherapy or neoadjuvant chemotherapy (NAC) prior to interval debulking surgery (IDS) and postoperative chemotherapy [10][11][12].

Introduction
Ovarian cancer (OC) represents the fifth most commonly diagnosed cancer among women, the fifth cause of cancer death, and the commonest cause of death due to gynecologic malignancies [1][2][3][4][5][6][7][8].An estimated number of 19,680 new cases of OC are expected to be diagnosed in the US and 12,740 women are expected to die from the disease in 2024 [2].Most cases (90%) are epithelial ovarian carcinomas and the majority are high-grade serous carcinomas.The most important revision in the last FIGO staging classification is that ovarian, fallopian, and primary peritoneal cancers are considered as one entity [9].
Ovarian cancer has a poor prognosis, with a 5-year relative survival rate of 48%, mainly because most women are diagnosed with advanced-stage disease [2].Moreover, the percentage of recurrence in OC is very high.The standard of care in OC includes either primary debulking surgery (PDS) followed by adjuvant chemotherapy or neoadjuvant chemotherapy (NAC) prior to interval debulking surgery (IDS) and postoperative chemotherapy [10][11][12].

ARs
Sugarbaker's PCI AR0 midline abdominal incision, greater omentum, and transverse colon AR1 superior surface of the right lobe of the liver, undersurface of the right hemidiaphragm, and right retrohepatic space AR2 epigastric fat pad, left lobe of the liver, lesser omentum, and falciform ligament

AR3
undersurface of the left hemidiaphragm, spleen, pancreatic tail, and anterior and posterior surfaces of the stomach AR4 descending colon and left paracolic gutter AR5 pelvic side wall lateral to the sigmoid colon and sigmoid colon AR6 female internal genitalia with ovaries, tubes and uterus, urinary bladder, culde-sac of Douglas, and rectosigmoid colon AR7 right pelvic side wall and base of the cecum, including the appendix AR8 right paracolic gutter and ascending colon AR9-12 small bowel (AR9: upper jejunum; AR10: lower jejunum; AR11: upper ileum; and AR12: lower ileum)  Imaging has a fundamental role in the accurate diagnosis of PMs in OC, helping to plan the appropriate therapeutic strategy, predict the likelihood of optimal cytoreduction, and identify potentially unresectable or difficult disease sites, which may require either IDS following chemotherapy or surgical technique modifications during PDS [3,[13][14][15][16][17][18][19][20][21][22].
Multidetector CT (MDCT) is considered the examination of choice for the initial staging of OC and for the evaluation of the extent of the disease in suspected recurrence [1].However, CT has limitations, mainly low soft-tissue resolution, and difficulty in depicting small peritoneal implants or implants at certain anatomic areas, including the root of mesentery, lesser omentum, and serosal surfaces of the small bowel, especially in the absence of ascites [1,3,18,21,[24][25][26][27][28][29][30].
MRI represents another reliable imaging tool for the assessment of PC.The efficacy of the technique in the detection of PMs has been improved by using fat-suppressed delayed contrast-enhanced imaging and diffusion-weighted imaging (DWI) [16,21,[30][31][32][33][34][35].Specifically, DWI improves the detection of small peritoneal hypercellular implants, even in the absence of ascites, due to their high signal against the hypointense background of normal tissues.However, MRI is recommended in specific circumstances, such as women with borderline ovarian tumors or OCs that have been previously staged with fertility preservation and also in cases with inconclusive CT findings [1].
The hypermetabolic activity of PMs increases their conspicuity using FDG PET/CT.CT and FDG PET/CT are considered equivalent alternatives for the detection of recurrent OC [1,17,30,[36][37][38][39][40][41][42][43][44].Based on the results of a recently published meta-analysis, FDG PET/CT had high diagnostic accuracy, with 88% sensitivity and 89% specificity in the detection of recurrent OC [45].Up to now, FDG PET or FDG PET/CT may be used as an adjunct tool in the initial staging of OC, in cases of indeterminate CT findings [1].
Systematic reviews on the role of cross-sectional imaging in the detection of PC in women with OC are lacking.A few recently published meta-analyses assessed the diagnostic accuracy of imaging modalities in the detection of PMs from various primary malignancies, including OC [46][47][48].
The purpose of this systematic review and meta-analysis was to compare the diagnostic performance of MDCT, MRI, including DWI, and FDG PET/CT in the detection of peritoneal metastases in ovarian cancer.

Materials and Methods
This systematic review and meta-analysis was conducted in accordance with the preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines [49].The systematic review has not been registered.

Search Strategy
A systematic and comprehensive literature search was performed for all publications that reported the diagnostic performance of MDCT, MRI, and FDG PET/CT in the detection of PMs in OC.Data extraction was independently performed by two researchers (ACT and MT) from the PubMed/MEDLINE database and included articles published from 2000 to February 2023.
Articles found to be suitable on the basis of their title and abstract were subsequently selected to further determine appropriateness for inclusion in this meta-analysis.Only papers in the English language were assessed.Full-text studies were further evaluated, and exclusion criteria were applied to identify final papers for inclusion.References were manually screened to identify additional studies.

Eligibility Criteria
The inclusion criteria were as follows: diagnosis/suspicion of PMs in patients with ovarian/fallopian/primary peritoneal cancer; initial staging or suspicion of recurrence (primary outcome); MDCT, MRI, and/or FDG PET/CT performed for the detection of PMs; population of at least 10 patients; surgical results, histopathologic analysis, and/or radiologic follow-up, used as a reference standard; per-patient and per-region data included; and data for calculating sensitivity and specificity reported.Discrepancies regarding potential eligibility and inclusion were resolved by consensus.
Studies were excluded if results for different imaging modalities were presented in combination and if data on the performance of each individual technique were unavailable.Studies including patients with the diagnosis of PMs from tumors other than OC were considered eligible only if it was possible to extrapolate results obtained on PC from OC.

Data Extraction
From each study, the following design characteristics were recorded: first author and year of publication; study design (prospective or retrospective); primary outcome; characteristics of study population, including number of patients with ovarian/fallopian/primary peritoneal cancer, age, number of patients with PMs, number and size of PMs, location of PMs; imaging modality, including MDCT, MRI or FDG PET/CT; report of reference test; time interval between imaging modalities; and time interval between imaging and reference standard.
Imaging characteristics included detailed information on the following: imaging equipment (type of scanner for MDCT and FDG PET/CT, magnetic field strength); imaging technique (phases and reformations for MDCT, type of coil, sequences, section thickness, and b-values for MRI); bowel preparation (laxatives and spasmolytic drugs), and use of luminal and/or intravenous contrast medium.
The numbers of true-positive (TP), false-negative (FN), false-positive (FP), and truenegative (TN) results for the detection of PMs were extracted on a per-patient and per-region basis.When cumulative data on the detection of PMs were not reported, the results from the abdominopelvic region with the best diagnostic performance were included in the analysis.
Regarding the per-patient analysis, the diagnostic odds ratios (DORs) were also estimated.A bivariate random effect meta-analytic method was used to estimate pooled sensitivity, specificity, and summary receiver operating characteristic (SROC) curves.Via the percentage of heterogeneity between the studies, computing I 2 values were calculated.I 2 values equal to 25%, 50%, and 75% were assumed to represent low, moderate, and high heterogeneity, respectively.Study heterogeneity was also assessed visually via funnel plots.

Quality Assessment
Analyses were performed using the RevMan software (ReviewManager, version 5.3; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark).To assess the methodological quality of the included primary studies and to detect potential bias, we used the quality assessment of diagnostic accuracy studies-2 (QUADAS-2) tool [50].

Results
The initial search in the electronic database resulted in 848 articles.Following a review of the titles and abstracts, 187 studies were selected as potentially relevant, and their references were cross-checked.Thirty-three publications eventually fulfilled the inclusion criteria and were selected for quantitative synthesis (Table 2) [26,27,30,42,44,.The flow chart of the selection process is shown in Figure 2.  A total of 2025 women with OC were included in the meta-analysis, with a mean age of 57.4 years (range, 19-91 years).The primary outcome included 19 studies with initial diagnosis of ovarian/fallopian/peritoneal cancer, 7 reports with recurrent cancer, and 7 studies with primary or recurrent OC.Advanced OC (FIGO stages III and IV) was reported in 1453 patients (Table 2).The presence of PMs was reported in 487 women and included 4.588 ARs (Tables 2 and 3).A total of 2025 women with OC were included in the meta-analysis, with a mean age of 57.4 years (range, 19-91 years).The primary outcome included 19 studies with initial diagnosis of ovarian/fallopian/peritoneal cancer, 7 reports with recurrent cancer, and 7 studies with primary or recurrent OC.Advanced OC (FIGO stages III and IV) was reported in 1453 patients (Table 2).The presence of PMs was reported in 487 women and included 4.588 ARs (Tables 2 and 3).

Quality Assessment
Study quality grading using QUADAS II scores showed that in terms of risk bias and regarding patient selection, index test, reference standard, flow, and timing, the majority of studies included in the analysis were of high to good quality (Figure 3).In terms of applicability, the quality of reporting on patient selection, index test, and the gold standard was good (Figure 3).

Quality Assessment
Study quality grading using QUADAS II scores showed that in terms of risk bias and regarding patient selection, index test, reference standard, flow, and timing, the majority of studies included in the analysis were of high to good quality (Figure 3).In terms of applicability, the quality of reporting on patient selection, index test, and the gold standard was good (Figure 3).

Discussion
According to our knowledge, this is an up-to-date systematic review and meta-analysis that exclusively compares the diagnostic performance of MDCT, MRI, including DWI, and FDG PET/CT in the detection of peritoneal metastases in women with ovarian cancer.In total, 33 studies, 23 using MDCT, 10 MRI (including three reports with DWI and three studies with WB-DWI), and 16 using FDG PET/CT (including seven studies with CECT), were evaluated.On a per-patient basis, FDG PET/CT had the highest sensitivity (93.7%) when compared to MRI (82.7%) and MDCT (79.7%).Specificity estimates were high for all imaging modalities (92.1%, 90.3%, and 91.5% for MDCT, MRI, and FDG PET/CT, respectively).Both FDG PET/CT and MRI have comparably higher per-patient diagnostic accuracy for the detection of PMs when compared to MDCT.
No differences in the diagnostic performance between MDCT, MRI, and FDG PET/CT were found on a per-lesion basis.MRI had the highest sensitivity (92.6%), when compared to MDCT (70.1%) and FDG PET/CT (58.3%), although our results are limited, due to the small number of MRI datasets (n = 3).Specificity estimates were comparably high for all imaging modalities (90.2%, 90.3%, and 92.6% for MDCT, MRI, and FDG PET/CT, respectively).Based on the results of this meta-analysis, FDG PET/CT and MRI had higher sensitivity compared to MDCT in the detection of PMs in OC.
Similar to our results, a recently published meta-analysis reported comparable diagnostic performance for DWI MRI and FDG PET/CT, higher than that of CT for the detection of PMs in ovarian and gastrointestinal cancer patients [47].This review was based on 28 articles, including 20, 7, and 10 CT, DWI MRI, and FDG PET/CT datasets, respectively.The pooled sensitivity and specificity were 68% and 88% for CT, 92% and 85% for DWI MRI, and 80% and 90% for FDG PET/CT [47].
The main advantages of MDCT include the following: wide availability, rapid scanning, increased volume coverage, excellent spatial resolution, robustness, and reproducibility of image acquisition.CT is devoid of misregistration artifacts and allows the acquisition of thin sections and the creation of high-resolution MPRs, improving the detection of small PMs, especially when a large amount of ascites is present, and the detailed exploration of curved peritoneal surfaces [18][19][20][21]67,69,70,[81][82][83][84][85].Coronal reformations improve the assessment of hemidiaphragms, hepatic and splenic surfaces, and paracolic gutters and the evaluation of the extent of omental disease.Sagittal MPRs improve the assessment of the hemidiaphragms, the Douglas pouch, the vaginal cuff, the peritoneal surface of the bladder, and the rectosigmoid colon [83][84][85].The use of multiplanar reformations was reported in 11 articles in this meta-analysis, although comparative studies on the diagnostic performance of axial images versus MPRs in the detection of PMs were not performed due to limited data [27,52,56,57,59,65,67,69,70,73,78].
MDCT comprised the largest dataset in our meta-analysis, allowing a comprehensive assessment of the diagnostic accuracy of the technique in the detection of PMs in different abdominopelvic regions.The highest MDCT detection rates were noted at the left hypochondrium (AR3), the central abdomen (AR0), the diaphragm, the pelvis (AR5-7 and AR6), and the left lumbar region (AR4).Our observations are primarily related to the advantages of MDCT technology, namely, the acquisition of thin slices and the creation of high-resolution reformations, resulting in an improvement in the evaluation of curved structures, such as the undersurface of the diaphragms, the paracolic gutter, and the pelvis [3,19].Similar to published data, this review confirmed the low diagnostic performance of CT in the colon and the mesentery [3,13,20,25,26,73,81].The detection of early mesenteric involvement or small-sized serosal bowel PMs may be problematic, as CT signs may be subtle, especially in the absence of adequate bowel distention [3].
Based on the analysis of 10 datasets, MRI proved more accurate compared to MDCT for the detection of PMs, on a per-patient basis.The use of fat suppression, delayed contrastenhanced sequences, and DWI contribute to the improvement in the accuracy of MRI in the detection of PMs [3,5,16,20,21,[30][31][32][33][34][35]66,69,[86][87][88][89][90][91].MRI allows better detection of subcentimeter PMs and PC involving certain anatomic areas, such as the bowel serosal surface, the pelvis, the right hypochondrium, and the mesentery.The interobserver agreement of MRI has also been reported to be higher compared to CT in most ARs [31,81].Although the diagnostic performance of MRI was only assessed in three ARs, including the central abdomen (AR0), the diaphragm, and the mesentery, our systematic review found that MRI was more accurate in the bowel mesentery.
Normal peritoneal enhancement is equal to or less than that of the liver.Contrast enhancement greater than the liver is abnormal and may represent the only finding suggestive of PC.This sign is not always detected by MDCT; however, it is readily appreciated by MRI on delayed fat-suppressed contrast-enhanced imaging [3,5,20,81,88].The sensitivity of MRI for PMs has been reported to increase by using DWI in combination with conventional MRI sequences, even in the absence of ascites.The increased contrast between the hyperintense hypercellular implants against the surrounding hypointense normal tissues enhances the detectability of PC by DWI [5,20,21,31,81,[86][87][88][89].
No direct comparison between the accuracy of conventional MRI sequences and DWI was performed in this analysis, due to limited data.
Limitations of MRI are related to the high cost and long examination time, motion artifacts, lack of routine use of intraluminal contrast agents, and need for experience in image acquisition and interpretation [1,20,81,92].MRI is also limited by its ability to detect small, calcified PMs, which are easily detected by MDCT.Disadvantages related to DWI are due to the low spatial resolution; presence of false-positives, attributed to densely cellular tissue, such as fibrosis, bowel mucosa, endometrium, and abscess; and false-negatives, attributed to mucinous carcinomas and well-differentiated malignancies [1,20].
FDG PET/CT disadvantages are related to limited spatial resolution in the detection of small PMs (<5 mm), difficulty in the evaluation of diffuse peritoneal disease, presence of tissues with low FDG avidity, such as mucinous tumors, and possible discrepancy in lesion location between CT and PET/CT caused by respiratory movements and intestinal peristalsis.False positives may be due to inflammation, infection, and benign conditions or the normal physiological activity in the bowel, gallbladder, vessels, ureters, and urinary bladder.Shortcomings of PET/CT also include the limited availability and the high cost [5,67,68,101].
Complete resection of all macroscopic peritoneal implants has been proven to be the single most important independent prognostic factor in OC.Diagnostic laparoscopy can provide a definitive histologic diagnosis and detailed information on the extent of PC.However, up to 40% of women may be understaged surgically as small PMs in areas such as the subdiaphragmatic surfaces, the porta hepatis, or the hepatorenal fossa are not easily accessible.In addition, diagnostic laparoscopy has been associated with a high incidence of port-site metastases, although these do not worsen the patient's prognosis [1,79].Preoperative diagnostic work-up with CT, MRI, or FDG PET/CT is vital in the assessment of the extent of PC in OC [1,79].
Tumor heterogeneity in OC at a cellular and genetic level is a well-known phenomenon that cannot be thoroughly evaluated using conventional imaging data.Quantitative semiautomated and automated methods based on artificial intelligence techniques have been developed, which can be applied to routine medical images to assess tumor heterogeneity.The use of radiomics and radiogenomics may be helpful in the future in predicting OC genotype and biology and in assessing treatment response, clinical outcome, and patient survival [102][103][104][105][106]. Based on preliminary data, MRI and CT-based radiomics have been reported to predict the presence of PMs in OC [107][108][109][110].
This meta-analysis has inherent limitations, mainly related to publication bias and study heterogeneity.Our systematic review was limited to the PubMed database, including published studies reporting a "positive effect" that might overestimate the actual magnitude of an effect.However, study quality grading showed that most of the studies included in the analysis were of high to good quality.
Heterogeneity among included patient groups is another shortcoming, due to differences in primary outcome (primary staging and recurrent disease).Subgroup analysis assessing the differences in the diagnostic performance of MDCT, MRI, and FDG PET/CT in the detection of PMs between primary and recurrent OC was not performed, due to the lack of relevant data.Heterogeneity in study design, imaging methodologies (including scanners, protocols, sequences, and intravenous/oral contrast), reader experience, and reference standards (ranging from histopathologic confirmation to surgical findings and imaging follow-up) is another limitation.The standardization of imaging techniques and consensus on the interpretation criteria for PMs across different centers would facilitate more accurate and reliable assessments.Finally, no data on the cost-effectiveness of MDCT, MRI, and FDG PET/CT were analyzed.Future research should focus on evaluating the cost-effectiveness of these imaging modalities in detecting peritoneal metastases and their impact on treatment decision making.

Conclusions
Peritoneal metastases represent a common finding in women with primary or recurrent OC.Preoperative diagnostic work-up with MDCT, MRI, or FDG PET/CT is mandatory to define the extent of the disease, predict the likelihood of optimal cytoreduction, identify potentially unresectable or difficult disease locations, requiring surgical technique modifications, and select patients who may benefit from adjuvant chemotherapy.
Based on the results of this meta-analysis, FDG PET/CT and MRI had a higher diagnostic performance in the detection of PMs compared to MDCT on a per-patient analysis.No differences between the three imaging modalities were found on a per-lesion basis.
In summary, while FDG PET/CT and MRI can be considered equivalent alternatives for the detection of peritoneal metastases in ovarian cancer, the limitations of the included studies and the need for standardization should be considered.Future research addressing these limitations and exploring cost-effectiveness would contribute to the improvement in clinical practice in the management of ovarian cancer.

Figure 6 .
Figure 6.Funnel plots of the (a) MDCT, (b) MRI, and (c) FDG PET/CT data on a per-patient basis.

Figure 10 .
Figure 10.Summary receiver operating characteristic curves depicting the diagnostic performance of (a) MDCT, (b) MRI, and (c) FDG PET/CT in the detection of peritoneal metastases in ovarian cancer, on a per-region basis.

Table 8 .
Diagnostic accuracy of MDCT in different abdominopelvic regions on a per-patient basis (AR: abdominopelvic region; CI: confidence interval; AUC: area under the curve).

Table 9 .
Diagnostic accuracy of MRI in different abdominopelvic regions on a per-patient basis (AR: abdominopelvic region; CI: confidence interval; AUC: area under the curve).

Table 10 .
Diagnostic accuracy of FDG PET-CT in different abdominopelvic regions on a per-patient basis (AR: abdominopelvic region; CI: confidence interval; AUC: area under the curve).

Table 11 .
Diagnostic accuracy of MDCT in different abdominopelvic regions on a per-region basis (AR: abdominopelvic region; CI: confidence interval; AUC: area under the curve).

Table 12 .
Diagnostic accuracy of FDG PET-CT in different abdominopelvic regions on a per-region basis (AR: abdominopelvic region; CI: confidence interval; AUC: area under the curve).
are cited in the Supplementary Materials.