1. Epidemiology
Uterine cervical cancer is the fourth most common cancer worldwide, with roughly 604,127 cases diagnosed and an annual mortality of 341,831
[1][2]. About 80–90% of the cases described are encountered in developing countries due to the lack of proper screening practices
[3]. On the other hand, the incidence has drastically reduced in the United States due to robust screening with pap smear exams and Human Papilloma Virus (HPV) DNA testing and cases hae remained stable during the recent decade (2009–2018). It is estimated that 14,100 new cases and 4280 deaths of invasive cervical cancer will be observed in the United States in 2022
[4]. The survival rate, in general, has been reported as 66%. However, it is lower (39%) in African American women of age ≥65 years
[4].
2. Classification
Most cervical cancers arise from the junctional zone between the cervix’s outer squamous and inner columnar epithelial lining. According to World Health Organization (WHO) classification, cervical cancer can be of various histologic subtypes: (i) squamous cell carcinoma (SCC), (ii) adenocarcinoma, (iii) clear cell adenocarcinoma, (iv) adenosquamous carcinoma, (v) serous carcinoma, (vi) glassy cell carcinoma, (vii) adenoid basal carcinoma, (viii) adenoid cystic carcinoma, (ix) undifferentiated carcinoma, and (x) adenocarcinoma
[5]. The SCC constitutes 75% of cervical cancer encounters, while the adenocarcinoma comprises 10–25%, adenosquamous 20%, and the rest of the histologies <5% of cases
[3][5]. The dysplastic lesions of SCC can be divided into high-grade squamous intraepithelial lesions (HGSIL) and low-grade squamous intraepithelial lesions (LGSIL).
3. Imaging
Although the previous FIGO classifications did not include the imaging criteria for tumor staging, the 2018 classification has permitted the utility of imaging, which enabled better tumor assessment and staging
[2][6][7]. The FIGO staging was based on clinical evaluation due to a limited access to imaging in low-income countries with high cervical cancer prevalence
[8]. However, clinical staging is suboptimal for certain tumor characteristics such as size, parametrial invasion, and lymph node involvement. In patients with early-stage cervical cancer (IA and part of IB1), the microinvasion is only detectable using tissue evaluation
[9]. The rest of the tumor stages, including local extension, can be assessed using reliable imaging modalities such as CT, MRI, and PET/CT, which have higher sensitivity and comparable specificity to the clinical evaluation
[10][11]. The National Comprehensive Cancer Network 2022 Practice Guidelines in Oncology recommended CT or PET/CT for tumor surveillance and follow-up, and MRI for the local assessment of the stage ≥ IB1
[8][12]. Staging is essential to predict survival, and surgical planning is considered standard management for early-stage (≤IIA) cervical cancers
[13]. Nguyen et al. compared PET/CT and PET/MRI and found that both modalities could identify all the primary and metastatic lesions and could strongly correlate standardized uptake value (SUV) (
p = 0.03)
[12] (
Figure 1).
Figure 1. A 51-year-old woman with squamous cell carcinoma of the cervix. (A). Sagittal T2-weighted imaging (T2WI), (B). axial T2WI, and (C). axial fused T2WI positron emission tomography/MRI showing a large (18)F-fluorodeoxyglucose (FDG) avid cervical mass (arrow). (D). An axial positron emission tomography/computed tomography image showed FDG avidity cervical tumor (arrow).
In general, diffusion weighted imaging (DWI) is considered sensitive to assessing parametrial involvement. It has high false-positive rates if the patients have large tumor sizes or a superimposed infections. Imaging must be highly specific to demonstrate the local tumor invasion since the curative surgery can be performed based on the parametrial invasion
[13]. Moreover, identifying stromal, ovarian, or corpus invasion is crucial as they are risk factors for lymphovascular space invasion (LVSI) and para-aortic lymph nodal metastases
[14][15].
The successful integration of PET and MRI enabled tumor evaluation and staging in a “one-stop” approach. In a study by Steiner et al., PET/MRI has proven to have a benefit over MRI with an Area Under Curve (AUC) of 0.85 vs. 0.74 for vaginal invasion and 0.89 vs. 0.73 for parametrial invasion
[13]. Similar findings were observed in the study by Sarabhai et al., who reported that PET/MRI and MRI are similar in characterizing the T-stage of the tumor (85% vs. 87%)
[16]. Wang et al. reported that PET/MRI has a sensitivity, specificity, and negative predictive value (NPV) of 78.5%, 64.9%, and 74.5%, respectively, compared to MRI
[17]. PET/MRI characterizes the parametrial invasion with a sensitivity and specificity of 90% and 94%
[17]. Kitajima et al. reported a diagnostic accuracy of 83% compared to MRI alone in a study comprising 30 patients
[18].
All the studies above were based on morphological observations on PET/MRI. Instead, Wang et al. quantified the gray level values to evaluate the parametrial invasion. They reported that high gray values corresponded to the higher FIGO stages (
p < 0.05); hence, this quantification technique is practical to implement in clinical practice
[17]. Wang et al. described the sensitivity, specificity, and NPV of combined PET/MRI+ gray values as 87%, 84%, and 86%, respectively, compared to MRI or PET/MRI alone, in assessing parametrial invasion (
p < 0.05)
[17].
The tumor cells drain from the cervix, through the lymphatic vessels, into parametrial lymph nodes, pelvic sidewall nodes, external and internal iliac nodes, and para-aortic nodes
[19]. Around 10–30% of patients with cervical cancer demonstrate pelvic lymph node metastases (LNM) during an early stage. This reduces the 5-year survival rate from 94.1% (negative LNM) to 64.1% (positive LNM)
[20]. Accurate lymph nodal assessment is essential for developing the individualized treatment algorithm, enhancing the prognosis, and reducing mortality. According to FIGO 2018 classification, micro- or macro-metastases to the lymph nodes are staged as IIIC regardless of tumor size or extent
[21]. CT and MRI are less sensitive and specific in detecting metastatic lymph nodes, as they cannot differentiate metastatic from non-metastatic lymph nodes
[22][23]. The combined PET/CT was studied, which showed high sensitivity (91% vs. 37.3%) and diagnostic accuracy (98% vs. 95%) compared to MRI (
p < 0.034), and hence is recommended by the National Comprehensive Cancer Network clinical guidelines
[23][24]. However, the PET is limited by identifying small lymph nodal metastases of size < 5 mm
[25]. Later, PET/MRI was found to have improved diagnostic confidence over PET/CT with the advantage of a reduced radiation dose
[26] (
Figure 2). PET/MRI has a sensitivity, specificity, and diagnostic accuracy of 91%, 94%, and 93% in detecting nodal metastases
[12]. Compared to PET/CT, PET/MRI identifies nodal metastases with a sensitivity, specificity, and accuracy of 92.3%, 88.2%, and 90%, respectively
[18].
Figure 2. A 55-year-old woman with squamous cell carcinoma of the cervix, status post hysterectomy. (A). Sagittal T2-weighted imaging (T2WI), (B). post-contrast sagittal T1-weighted imaging (T1WI), (C). coronal diffusion-weighted image (DWI), (D). coronal apparent diffusion coefficient (ADC), (E). coronal fused T2WI, and (F). axial T2WI. (G). Axial fused T2WI positron emission tomography/MRI showed an enhancing (18)F-fluorodeoxyglucose (FDG) avid plaque-like thickening at the left cervix (arrow) with restricted diffusion. (H). An axial positron emission tomography/computed tomography image showed ill-defined FDG avidity (arrow). b: urinary bladder.
Cervical cancer is one of the tumors that demonstrate heterogeneity to hypoxia. Narva et al. studied the association between hypoxia and increased resistance to chemotherapy and radiotherapy in patients with SCC of the cervix
[27][28]. In addition, the cancerous cells adapt to the hypoxic microenvironment, leading to genetic instability, DNA damage, and mutagenesis. This results in a rapid tumor invasion to the adjacent and distant organs.
18Fluorine-labeled 2-(2-nitro-1-
H-imidazol-1-y)-
N-(2,2,3,3,3-pentafluoropropyl)-acetamide (
18F-EF5) is a hypoxia radiotracer that can be used in PET imaging. Increased uptake of
18F-EF5 is strongly associated with poor prognosis compared to (18)F-fluorodeoxyglucose (
18F-FDG) uptake. Narva et al. reported that an increased
18F-EF5 uptake on
18F-EF5-PET/MRI correlates with hypoxia intensity, which is proportional to the tumor stage
[27].
Radiotherapy is the cornerstone in the management of patients with cervical cancer. Around 25% of cervical cancer cases recur, and 24% among those are observed in already-treated patients, which points to the importance of identifying the radio-resistant tumor areas that may be managed with radiation dose escalation. A new PET tracer
68Ga-NODAGA-E[c(RGDyK)]2 ([
68Ga] (Ga-RGD)) identifies the α
vβ
3, an integrin that is found on the newly formed vasculature. Pelvic insufficiency fractures (PIF) are a late complication of radiotherapy, and Sapienza et al. studied the incidence of PIF in patients who underwent radiotherapy for various gynecologic cancers
[29]. They found that 10–18% of patients are affected by PIF, with the sacrum as the most common fracture site
[29]. Azumi et al. noticed PIF in 20% of patients with cervical cancer treated with radiotherapy
[30]. They also demonstrated that PET/MRI discovers PIF earlier than PET/CT (
p <0.05), with the added advantage of reduced radiation exposure
[30]. The earliest sign of PIF is medullary edema, which can be observed as T1 hypointense and T2 hyperintense on MRI as early as 18 days after the symptom onset
[30].
The maximum standardized uptake value (SUV
max) derived from [
18F] FDG-PET and diffusion metrics such as the apparent diffusion coefficient (ADC) from the MRI are studied as the prognostic indicators in patients with cervical cancer
[28][31][32][33][34][35][36][37]. Many studies reported that SUVmax and ADC minimum (ADC
min) values of cervical cancer are inversely related. Olsen et al. also described reduced ADC value in intense SUV
max [38]. In addition, the SUVmax was seen to vary based on the histology and degree of differentiation of cervical cancer, and this feature aided in the prognostication
[39]. SCC of the cervix is found to have higher SUV
max values than the non-squamous tumors (
p = 0.153), and poorly differentiated ones have higher SUVmax than do the well-differentiated tumors (
p = 0.0474)
[39]. The underlying reason for the difference in SUVmax is secondary to the degree of Glucose Transporter (Glut) expression that aids in FDG uptake; however, it still needs to be validated through further studies
[40][41].
The simultaneous acquisition of PET/MRI provides precise spatial correlation and a more appropriate insight into the imaging biomarkers on the voxel level. The inverse correlation between SUV
mean, SUV
max, and ADC
min was also supported by Brandmaier et al. on hybrid PET/MRI. The correlations between SUV
mean and ADC
min (
r = −0.403) and SUV
max and ADC
min (
r = −0.532) were significant in primary cervical tumors
[42]. The authors demonstrated a stronger correlation between SUV
mean and ADC
min (
r = 0.773) and SUV
max and ADC
min (
r = −0.747) in the case of recurrent cervical tumors
[42]. Grueneisen et al. reported significant SUVmax and ADC
min in primary tumors but not the recurrent cervical tumors
[43]. Later, Ho et al. described no correlation among SUV
max, SUV
mean, ADC
min, or ADC
mean. However, they found that the ratio of ADC
min/ADC
mean (relative admin) and the ratio of SUV
max and SUV
mean (relative SUV
max) correlated well with the adeno- and adenosquamous carcinoma of the cervix (
r = −0.685) and with the well- to moderately differentiated tumors (
r = −0.631)
[44]. No significant correlation between relative SUVmax and relative ADC
min was found in squamous cell carcinoma and poorly differentiated tumors
[44]. Surov et al. studied the SUV and ADC parameters and their relation with the KI 67 proliferation index
[45]. They found that SUVmax (
r = 0.59), SUV
mean (
r = 0.45), SUV
max/ADC
min (
r = 0.71), SUV
max/ADC
mean (
r = 0.75), and ADC
min (
r = −0.48) correlated significantly with the KI 67 proliferative index, thereby reflecting the tumor proliferation rate
[45]. Additionally, SUV
mean (
r = 0.71) and SUV
max (
r = −0.71) strongly correlate with epithelial and stromal areas and locate the metabolically active areas
[45]. In addition to SUV and ADC, the other parameters include metabolic tumor volume (MTV) and total lesion glycolysis (TLG). It has been studied that these parameters conventionally correlate with the SCC antigen levels, FIGO staging, tumor size, and depth of stromal invasion
[46][47].
Table 1 and
Table 2 summarize the essential characteristics of PET/MRI studies in cervical and pelvic malignancies.
Table 1. Characteristics of PET/MRI studies in cervical cancer.
Table 2. Characteristics of PET/MRI studies in pelvic malignancies.