Introduction

Ovarian cancer (OC) is one of the leading causes of death from gynecologic cancer in the United States [1]. OCs are often metastatic at the time of presentation, and are associated with a high rate of recurrence and poor prognosis [2]. Ovarian neoplasms are classified histogenetically by cell subtype: epithelial, stromal, or germ cell. Epithelial OC comprises 90% of malignant ovarian tumors [3]. OC metastases may occur due to peritoneal, lymphatic, or hematogenous spread. Peritoneal implantation of cancer cells most commonly occurs along the peritoneal surfaces within the pelvis, bowel, liver surface, omentum, or diaphragm. Lymphatic drainage from the ovaries can cause external and common iliac, para-aortic, inguinal, and supraclavicular lymphadenopathy. Hematogenous spread most often affects the liver, lungs, brain, and bones [4,5,6].

Malignant ovarian tumors are usually first seen by imaging—transvaginal ultrasonography (TVUS) or abdominal contrast-enhanced computed tomography (CT)—and the diagnosis is supported by the finding of elevated levels of the serum biomarker CA125 [7,8,9]. While CT is the most common imaging test used for staging and surveillance, other modalities are being increasingly used in the management of ovarian cancer. In this review, we discuss the current clinical role of positron emission tomography (PET)/CT and PET/magnetic resonance imaging (MRI) for the identification, staging, and restaging of OC, as well as evaluation of treatment response. We also discuss our experience with PET/MRI in imaging OC, with emphasis on the advantages and challenges of this new hybrid imaging modality in gynecologic cancers.

PET/CT

Since its introduction in the 1990s, PET/CT has become an extremely useful imaging modality for staging, restaging, and assessment of treatment response in oncology [10]. In this section, we will discuss the role of PET/CT in characterization of adnexal mass, staging of ovarian cancer, evaluating treatment response and prognostication and in the management of recurrent disease.

Role in characterization of an adnexal mass

PET/CT is not commonly used to characterize an adnexal mass, as physiologic uptake may be seen in normal ovaries, limiting its value. However, several authors have reported on the utility of PET/CT in characterization of pelvic masses (Table 1) [11]. Studies show that PET CT has 81–100% sensitivity and 93–95% specificity for diagnosing malignant ovarian tumor [12, 13].

Table 1 Studies evaluating positron emission tomography/computed tomography (PET/CT) for the characterization of benign and malignant adnexal lesions
Full size table

PET/CT can also aid in distinguishing borderline ovarian tumors from OCs. In a retrospective assessment of 13 patients, PET/CT reported good sensitivity (83.3%), specificity (85.7%), and area under curve value (0.893; p = 0.0001) in distinguishing borderline ovarian tumors from stage I malignant ovarian tumors [16]; a cutoff maximum standard uptake value (SUVmax) of 3.7 differentiated borderline ovarian tumors from stage I malignant ovarian tumors. Similarly, in a prospective study of 30 patients the SUVmax in borderline tumors (2.0 ± 0.70) was significantly lower than that of malignant tumors (9.32 ± 4.58), but not significantly different compared with benign tumors (1.74 ± 1.44, p = 0.005) [22]. For detection of malignant or borderline malignant pelvic tumors, the sensitivity was 71.4% and the specificity was 81.3%, and for OC, the sensitivity was 100% and the specificity was 85.0%. In concordance with prior studies, another retrospective study of 171 ovarian cancer patients reported a strong glycolytic phenotype (average SUVmax 7.6) in epithelial OC [20].

Few studies have compared the diagnostic performance of PET/CT with that of TVUS, CT, and MRI. Although TVUS remains the standard initial screening method for suspicious adnexal findings [15, 18], PET/CT can provide additional value to TVUS or CT findings. Castellucci et al. reported that PET/CT, compared with TVUS, had similar sensitivity (87% vs 90%), specificity (100% vs 61%), negative predictive value (81% vs 78%), positive predictive value (100% vs 80%), and accuracy (92% vs 80%) in characterizing ovarian lesions [14]. A prospective study reported that PET/CT, compared with CT, had higher specificity (77% vs 38%) and similar sensitivity (93% vs 96%) for characterizing malignant tumors [21]. Similarly, another prospective study also reported higher accuracy for PET/CT (92.1%) compared with pelvic ultrasonography (83.0%) and abdominopelvic CT or pelvic MRI (74.9%; p = 0.013) in distinguishing malignant or borderline ovarian tumors from benign tumors [19].

While the above studies report the utility of PET CT in diagnosing ovarian tumors, its cost effectiveness for this purpose remains unproven. Currently, pelvic ultrasound and MR are the most commonly used imaging modality for the diagnosis and characterization of ovarian tumors.

Role in staging of OC

PET/CT an effective imaging modality for staging OC, with 75.5–83.3% sensitivity, 68.4–99.4% specificity, 87.5–95.3% positive predictive value, and 96.5–98.6% negative predictive value (Table 2) [23, 24]. PET/CT findings can lead to an alteration in FIGO stage and modifications in the treatment plan. One study reported migration of FIGO stage III to stage IV OC after PET/CT in 26% of patients [25]. Similarly, another prospective study reported upstaging to stage IV in 41% (27/66) of patients with advanced OC [26].

Table 2 Studies evaluating positron emission tomography/computed tomography (PET/CT) for staging of ovarian cancer
Full size table

PET/CT has also shown better performance in staging OC compared with other modalities (CT or MRI). A meta-analysis reported that PET/CT was more accurate (sensitivity 73.2%, specificity 96.7%, odds ratio 90.32) than CT and MRI in detection of lymph node metastasis in OC [31]. In another prospective assessment, PET/CT was superior to CT for the detection of carcinomatosis in subdiaphragmatic peritoneal surfaces (p = 0.020) and in the bowel mesentery (p = 0.001) in advanced epithelial OC [29]. PET/CT also detected higher rates of extra-abdominal disease spread than did CT (78% vs 60%). Similar results were obtained in another retrospective study of 40 patients [27]. PET/CT had higher lesion-based sensitivity (69.4% vs 37.6%), specificity (97.5% vs 97.1%), and accuracy (94.0% vs 89.7%) in preoperative staging of OC, when compared to CT. There were significant differences (p < 0.05) in sensitivity and accuracy between the imaging modalities. In addition, Dauwen et al. reported that PET/CT was better than CT in detecting retroperitoneal lymph node metastases, but not in detecting peritoneal metastases [21]. There was no statistically significant difference between the two modalities for FIGO staging.

Role in OC treatment prognosis and response evaluation

The metabolic parameters of PET/CT, such as SUVmax, total lesion glycolysis (TLG), and metabolic tumor volume (MTV), can provide important prognostic information and assess response to treatment (Fig. 1; Table 3). A prospective study with 33 metastatic ovarian cancer patient reported a significant correlation between PET metabolic response after the first (SUVmax 4.9 ± 2.8, p < 0.008) and third (SUVmax 3.5 ± 2.8, p < 0.005) cycle of chemotherapy with overall survival in advanced-stage OC [32]. Liao et al. reported that in post-surgery OC patients, high whole-body TLG was associated with poor prognosis (hazard ratio 1.043, p = 0.011) [33]. In yet another study, higher TLG was reported as an independent prognostic factor (p = 0.008) for disease progression after cytoreductive surgery in OC [34].

Fig. 1
figure 1

A 77-year-old woman with high-grade serous ovarian cancer with peritoneal carcinomatosis. a Axial contrast-enhanced computed tomography (CT) and b axial positron emission tomography (PET)/CT images showed FDG-avid (maximum standardized uptake value 5.9) nodularity at the left anterior lower abdomen (arrow), consistent with peritoneal carcinomatosis. At follow-up after 3 months of systemic therapy, axial non-contrast-enhanced CT c and axial PET/CT images d showed excellent response to interval therapy, with marked decrease in size and resolution of FDG avidity at peritoneal carcinomatosis. Coronal maximum intensity projection at baseline (e) and after 3 months of systemic therapy (f) showed treatment response

Full size image
Table 3 Studies evaluating positron emission tomography/computed tomography (PET/CT) for prognosis and response to treatment in ovarian cancer
Full size table

PET/CT may also be useful in predicting prognosis in OC. One study reported that larger fractional decrease in TLG after 2 weeks of systemic therapy predicted partial response after 10 weeks (p = 0.037) in ovarian, breast, and endometrial cancers [46]. Also, a rise in SUV between the second and sixth week predicted progression (p = 0.034) was associated with worse progression-free survival (hazard ratio 1.068, p = 0.013). Vallius et al. reported that the median omental SUVmax change during neoadjuvant chemotherapy (NACT) was − 64% (range − 16% to − 84%) and was associated with treatment response (p = 0.004) [36]. The SUVmax decrease < 57% enabled identification of non-responders to NACT. In another study by the same group, MTV reduction < 85% was associated with progressive (PD)/stable disease (SD) (70% sensitivity, 78% specificity, 0.79 AUC) and was able to identify candidates who may benefit by a change in management [35]. Similarly, poor outcome in epithelial OC was associated with higher values for MTV (p = 0.022, hazard ratio 5.571) and TLG (p = 0.037, hazard ratio 2.967) [38]. Gallicchio and colleagues compared all metabolic parameters for patients with peritoneal carcinomatosis from epithelial OC and reported that a quantitative assessment of MTV (p = 0.01), rather than SUVmax (p = 0.48) or TLG (p = 0.06), was helpful for stratifying patients [41]. MTV survival analysis showed significantly better OS in patients presenting with a high tumor burden than in those presenting with less burden (p = 0.01; 26 months vs 14 months); the higher the MTV, the better the response to chemotherapy [41].

Role in recurrent OC

Tumor recurrence is identified in 60–70% of OC patients and is one of the main prognostic factor in OC [47]. Hence early identification of tumor recurrence is critical in restaging and optimal management (Fig. 2; Table 4).

Fig. 2
figure 2

A 73-year-old woman with recurrent high-grade serous ovarian carcinoma, after neoadjuvant chemotherapy followed by surgery and additional chemotherapy. a Axial contrast-enhanced computed tomography (CT) and b axial positron emission tomography (PET)/CT images showed no evidence of FDG-avid recurrent or metastatic ovarian carcinoma. At follow-up after 3 months of systemic therapy, axial non-contrast-enhanced CT (c) and axial PET/CT images d showed interval progression with FDG-avid para-aortic adenopathy (arrow) and pericolonic nodule abutting the ascending colon (arrowhead). Coronal maximum intensity projection at baseline (e) and after 3 months of systemic therapy (f) showed interval progression. Bilateral ureteral stents were placed

Full size image
Table 4 Studies evaluating positron emission tomography/computed tomography (PET/CT) for detection of recurrent ovarian cancer (ROC)
Full size table

PET/CT is reported to have 41–97% sensitivity, 86–100% specificity, and 83–97% accuracy [49,50,51,52, 56, 60, 63,64,65, 67, 71, 73,74,75]. The quantitative metabolic parameters from PET have been associated with optimal surgery outcome and progression-free survival in recurrent OC [66, 69]. A retrospective study of 56 recurrent OC patients showed that whole-body MTV and whole-body TLG (p = 0.008 for both) were significant prognostic factors for post-relapse survival (median PRS duration for surviving patients was 35 months, range 16–90 months).[69]. Another study reported that MTV above 7.52 mL (p = 0.0191) and/or TLG above 35.94 g (p = 0.0069) were associated with significantly shorter progression-free survival (estimates at 3.5 years) in recurrent OC [66].

Fulham et al. reported that PET/CT altered management in 80% of patients, detected new lesions in 77% of patients, and was superior to other modalities in detecting nodal, peritoneal, and subcapsular liver disease in recurrent OC [72]. Similarly, another study reported that PET/CT altered the known disease distribution in 64% patients and had a high impact on the management plan in 57% of patients with recurrent OC [70]. PET/CT altered the apparent disease distribution in 61%, showing lower disease burden in 9% and higher disease burden in 52%, compared to prior findings on CT, clinical examination, and pathology. In a prospective data registry involving 22,975 studies of OC, 83.7% patients underwent PET/CT for initial staging, restaging, and recurrence detection. The researchers reported that physicians changed their intended management plan in 36.5% of patients (95% confidence interval 35.9–37.2%) after reviewing PET findings [61].

PET/CT has also shown better performance in detecting recurrent OC compared to CA125 and CT. In a retrospective evaluation of 121 patients, PET/CT had superior overall sensitivity (82% vs 59% and 69%), specificity (87% vs 80% and 47%), accuracy (83% vs 63% and 66%), positive predictive value (96% vs 93% and 88%), and negative positive value (55% vs 32% and 21%) in detecting recurrent OC, compared with CA125 and CT [75]. Similarly, another study reported that PET/CT significantly outperformed CT in terms of sensitivity (96% vs 84%), specificity (92% vs 59%), negative predictive value (90% vs 59%), positive predictive value (97% vs 84%), and accuracy (95% vs 76%; p < 0.05) in detecting recurrent OC [49].[51]. PET CT is also reported to have better performance in detecting recurrent OC, compared to MR. Sanli et al. reported that, compared with MRI, PET/CT had superior sensitivity (97.5% vs 95%), specificity (100% vs 85.7%), positive predictive value (100% vs 97.4%), negative predictive value (87.5% vs 75%), and diagnostic accuracy (97.8% vs 93.6%) in the detection of < 2-cm peritoneal implants in recurrent OC [54]. A meta-analysis with 34 articles analyzed the diagnostic accuracy of CA125, PET alone, PET/CT, CT, and MRI in recurrent OC [52]. They reported pooled sensitivities of 69% for CA125, 79% for CT, 75% for MRI, and 91% for PET/CT; pooled specificities of 93% for CA125, 84% for CT, 78% for MRI, and 88% for PET/CT; and area under the curve values of 0.92 for CA125, 0.88 for CT, 0.78 for MRI, and 0.95 for PET/CT, and they concluded that PET/CT has a useful supplemental role in surveillance, particularly in patients with increasing CA125 levels and negative CT and MRI findings.

Pitfalls

On cross-sectional imaging of early OC, it may be difficult to detect small implants which is a major limitation [76] (Fig. 3). False-negative results can occur in patients with low-grade tumors and in the early stages of OC [11, 77]. The clear cell or mucinous malignancies have been reported to have low-level FDG uptake [64, 78]. Hence, the sensitivity of FDG PET/CT for the detection of these neoplasms has been reported to be relatively low. Similarly, large cystic or necrotic tumors can yield false-negative results at PET/CT, due to the reduced cellularity and fewer viable cancer cells [11, 13]. Also, postoperative changes and granulation tissue may show FDG uptake and result in false-positive findings [11]. Also, the bowel uptake can be physiologic and may mask serosal involvement with tumor [11, 13]. In addition, false-positive findings can result from physiologic ovarian uptake, which is common during ovulation and the early luteal phase of the menstrual cycle in premenopausal women. Also postoperative abscesses and perforated viscera may demonstrate FDG uptake. Similarly, endometriomas, fibromas, and benign lesions such as corpus luteum cysts, dermoid cysts, serous cysts, and salpingitis may have increased metabolic activity [11, 79] (Fig. 4) [11, 77].

Fig. 3
figure 3

Pitfalls of PET/CT. A 72-year-old female with metastatic esophageal carcinoma. a Axial non-contrast-enhanced CT and b Axial PET/CT images show a 0.5 cm right upper quadrant peritoneal nodule (arrow) with no FDG uptake due to small size. c Axial non-contrast-enhanced CT and d Axial PET/CT images after 4 months show interval increase in the 1.8 cm nodule (arrow) with no FDG uptake. e Axial non-contrast-enhanced CT and f Axial PET/CT images after 2 months show progressive increase in the size of the peritoneal nodule (arrow) with FDG uptake

Full size image
Fig. 4
figure 4

False-positive PET/CT finding. A 65-year-old female with endometrial carcinoma. a Axial PET/CT images show hypermetabolic activity in left adnexa and uterine body (arrow). b Axial T2WI and c post-contrast T1WI show non-enhancing dilated tubular structure in the left adnexa suggestive of a hydrosalpinx (arrow) with some fluid (asterisk)

Full size image

PET/MRI

PET/MRI is a hybrid imaging modality composed of PET and MRI, and can provide combined anatomical and metabolic imaging similar to PET and CT [80, 81]. The aim is to combine the high soft tissue contrast and functional information of MRI with the metabolic activity of whole-body PET. MRI is seen as more effective for local disease evaluation [82] and PET/CT as more effective for identifying distant metastasis and suspected recurrence in gynecologic cancers. There is, thus, an opportunity for “one-stop shopping” and better anatomic localization with integrated PET/MRI [76, 81, 83, 84].

Several recent publications have described initial experiences with PET/MRI in mixed populations of patients with various gynecologic malignancies (Table 5). These studies evaluated the role of PET/MRI in initial characterization and staging [85,86,87], evaluation of advanced disease [88], and detecting recurrence [89,90,91] of gynecologic malignancies (Figs. 5, 6, 7, 8, and 9).

Table 5 Studies evaluating positron emission tomography/magnetic resonance imaging (PET/MRI) for evaluation of gynecologic cancers, including ovarian cancer
Full size table
Fig. 5
figure 5

A 55-year-old woman with high-grade Mullerian carcinoma. a Axial T2-weighted imaging (T2WI), b sagittal T2WI, c coronal fused and d axial post-contrast T1WI, T2WI magnetic resonance imaging (MRI) and e axial fused T2WI, f sagittal fused T2WI, and g coronal fused T2WI positron emission tomography/MRI showed an FDG-avid large complex right adnexal mass (arrow) with an enhancing soft tissue component within the mass and restricted diffusion. h An axial contrast-enhanced computed tomography image showed a complex right adnexal mass (arrow). b: urinary bladder

Full size image
Fig. 6
figure 6

A 71-year-old woman with high-grade serous carcinoma. a Axial T2-weighted imaging (T2WI), b sagittal T2WI, c coronal T2WI, d axial post-contrast T1WI magnetic resonance imaging (MRI), and e axial fused T2WI, f sagittal fused T2WI, and g coronal fused T2WI positron emission tomography/MRI showed an FDG-avid large complex left adnexal mass (arrow) with an enhancing soft tissue component within the mass and restricted diffusion. h An axial contrast-enhanced computed tomography image showed a complex left adnexal mass (arrow). b: urinary bladder, v: vagina

Full size image
Fig. 7
figure 7

A 72-year-old woman with high-grade serous carcinoma with peritoneal carcinomatosis. a Axial T2-weighted imaging (T2WI), b sagittal T2WI, c coronal fused T2WI, d axial post-contrast T1WI, magnetic resonance imaging (MRI), and e axial fused T2WI, f sagittal fused T2WI, and g coronal fused T2WI positron emission tomography/MRI showed an FDG-avid primary adnexal mass (thin arrow) with tumor implants (arrow) and extensive peritoneal disease (asterisk). h An axial contrast-enhanced computed tomography image from 1 month prior to the other images showed a complex left adnexal mass (arrow) with peritoneal disease (asterisk)

Full size image
Fig. 8
figure 8

A 68-year-old woman with high-grade serous carcinoma. a Axial T2-weighted imaging (T2WI), b sagittal T2WI, c coronal T2WI, d axial post-contrast T1WI, magnetic resonance imaging (MRI), and e axial fused T2WI, f sagittal fused T2WI, and g coronal fused T2WI positron emission tomography (PET)/MRI showed an FDG-avid mass in the right adnexa (arrow). h An axial contrast-enhanced computed tomography (CT) image showed a complex right adnexal mass (arrow). i Axial T2WI and j axial post-contrast T1WI MRI, and k axial fused T2WI PET/MRI. A left hip prosthesis was present (asterisk). b: urinary bladder, v: vagina, r: rectum, u: ureter

Full size image
Fig. 9
figure 9

A 29-year-old woman with endometrioma. a Axial T2-weighted imaging (T2WI), b sagittal T2WI, c coronal T2WI, d axial post-contrast T1WI magnetic resonance imaging (MRI), and e axial fused T2WI, f sagittal fused T2WI, and g coronal fused T2WI positron emission tomography/MRI showed a large right adnexal mass (arrow) with no enhancement and with mild restricted diffusion. h An axial contrast-enhanced computed tomography image showed a complex appearing right adnexal cystic mass (arrow). b: urinary bladder, v: vagina

Full size image

In a retrospective assessment of 26 patients for evaluating adnexal lesions, PET/MRI had higher sensitivity (94% vs 74%), specificity (100% vs 80%), positive predictive value (100% vs 93%), and negative predictive value (83% vs 44%) compared with PET/CT [85]. In that study, PET/CT detected only 74% of malignant lesions (14/19), whereas PET/MRI detected 95% of malignant lesions (18/19). Another study used a three-point grading score and reported that PET/T2-weighted imaging (2.72 ± 0.54) localized the lesion in gynecologic malignancies significantly more convincingly than PET/CT (2.23 ± 0.50) or PET/T1-weighted imaging (2.29 ± 0.53; p < 0.01) [86]. Grueneisen et al. assessed the value of diffusion-weighted imaging in PET/MRI for diagnosis of primary and recurrent pelvic malignancies [87]. They reported that adding diffusion-weighting imaging to PET/MRI increased sensitivity (92.9% to 94.9%), NPV (75.0% to 80.0%), and diagnostic accuracy (91.8% to 92.6) [88].

The effectiveness of PET/MRI in detecting recurrent OC has also been evaluated. Grueneisen and group compared the diagnostic performance of PET/MRI with that of PET/CT for detecting recurrence of pelvic malignancies [90]. They found that, compared with PET/MRI, PET/CTs lesion-based sensitivity (82% vs 85%), specificity (91% vs 87%), positive predictive value (97% vs 96%), negative predictive value (58% vs 63%), and diagnostic accuracy (84% vs 86%) did not significantly differ (p > 0.05) in terms of detecting malignant lesions. Other studies have also reported equivalent diagnostic performance of PET/MRI compared to PET/CT [89, 91, 92, 94, 95, 97].

Challenges

Combining PET and MRI into a single acquisition is technically challenging. However, the development of new solid-state PET detectors, which function in the presence of magnetic fields, has made a single acquisition of PET and MRI possible [98].

One challenge is that MRI acquisitions are not based on X-rays and do not provide a direct reference for attenuation correction by the body [99]. To overcome this, Dixon-based technique (where one image is acquired with fat and water signals in phase and another image is acquired with fat and water signal out of phase) has been used as a reference for attenuation correction [100]. Segmenting the body from Dixon-based technique information allows an attenuation map to be generated [101].

Imaging of the lungs using MRI is also challenging. Ultrashort-echo and zero-echo time pulse could potentially be used to detect small nodules, but in general reduced sensitivity to small lung nodules remains a limitation of PET/MRI compared with PET/CT [102, 103]. Continuous respiratory and bowel motion occurs during PET/MRI acquisition, and this motion presents a particular challenge when acquiring images of the abdomen and upper pelvis [104].

Even though there is a benefit of lower radiation exposure reduction with PET/MR than PET/CT, the longer duration of imaging acquisition with PET/MR than to PET/CT and the lack of widespread availability of PET/MR in the radiology non-academic setting have abated the utilization of PET/MR in clinical practice. Finally, the National Comprehensive Cancer Network [105] and American College of Radiology appropriateness guidelines [106] do not have enough evidence to use PET-MR as standard of care imaging. However, since these guidelines do recommend use of PET and MRI individually in the management of OC, it is likely that this promising hybrid imaging technique would soon be included for this indication.

Future perspective

Newer MRI techniques such as dynamic contrast enhancement, diffusion-weighted imaging, intrinsic susceptibility weighting, and spectroscopy could increase the diagnostic ability of MRI to detect and characterize lesions [107,108,109]. In addition, the expanding field of radiomics in OC is also emerging as a promising tool [110, 111]. Numerous novel PET tracers have been introduced for the evaluation of tumors. Some other tracers, such as 18F-fluoromisonidazole [112], copper-labeled diacetyl-bis (N4-methylthiosemicarbazone) [113], 16a-18F-fluoro-17b-estradiol [114], and 18F-3′-fluoro-3′-deoxythymidine [115], have shown promising results in evaluating cancers. Novel MRI tracers, such as ferumoxytol [116] and hyperpolarized 13C [117], are under development as an alternative to FDG. Also, high-resolution delineation of the tumor in PET/MRI permits precise tumor delineation and can also be useful for optimal stereotactic radiosurgery [118].

Conclusion

PET/CT and PET/MRI may help in staging and assessment of recurrent disease in ovarian cancer. The metabolic parameters such as SUV, MTV, and TLG obtained from PET/CT and PET/MR have been shown to be useful surrogate markers for response to therapy, OS, and PFS. The development of novel targeted therapies and PET tracers will further expand the role of these imaging modalities. Nevertheless, more prospective studies with standardized protocols must be conducted before hybrid molecular imaging can be established as an acceptable mainstream imaging modality for OC and to outweigh the added cost and exposure to ionizing radiation.