Until recently, the clinical staging relied on clinical evaluation, preferably by an experienced examiner
[1]. Conventional procedures, such as cystoscopy, proctoscopy, intravenous urography, and X-ray of the lungs, were performed
[1]. Examination under anaesthesia was recommended, especially when the examination was difficult, or there was uncertainty regarding the involvement of the vagina, parametrium, or pelvic wall
[2]. However, physical examination alone is known to have low accuracy for assessing tumour size and parametrial infiltration. Historically, Innocenti et al. reported sensitivities of clinical evaluation and transrectal ultrasound for diagnosing parametrial involvement, using surgico-pathology as the reference standard, of 52% and 78%, respectively
[3]. Twenty years ago, the multicentric clinical trial of the American College of Radiology Imaging Network/Gynecologic Oncology Group (ACRIN/GOG183) enrolled 208 patients (from 25 centres, with invasive cervical cancer proved by biopsy results) who underwent pelvic magnetic resonance imaging (MRI) and contrast-enhanced computed tomography (CT) before definitive radical hysterectomy
[4]. Correlation between maximum histopathologic tumour size and tumour size from clinical assessment (rs = 0.37, low correlation), CT (rs = 0.45, moderate correlation), and MRI (rs = 0.54, moderate correlation) were reported with the highest figures for MRI
[4]. It is obvious that clinical assessment alone is insufficient to accurately assess the size of smaller tumours that do not cause cervical enlargement. In the same ACRIN/GOG 183 study, MRI yielded higher sensitivity (53%) than clinical assessment (29%) for diagnosing parametrial invasion
[5]. A systematic review including 3254 patients also found that MRI had much higher sensitivity than clinical examination for diagnosing parametrial invasion (sensitivity: 84% vs. 40%) and locally advanced disease (79 vs. 53%)
[6]. Clinical understaging can be caused by the inability to detect incipient parametrial invasion by clinical examination, especially ventrally and/or positive (metastatic) lymph nodes, while clinical overstaging can be due to subjective clinical assessments falsely interpreted as parametrial spread. Although findings from modern imaging examinations (i.e., ultrasound, MRI, CT, and PET-CT) were not considered for staging purposes in the guidelines until 2018, they had already largely replaced staging results based on conventional procedures in many gynaecologic oncology centres in high-income countries. A multicentric clinical trial by the American College of Radiology Imaging Network/Gynecologic Oncology Group (ACRIN/GOG183) from 2005 showed only sporadic use of conventional procedures recommended for cervical cancer staging
[5][7]. Cystoscopy was performed in 8.1%, sigmoidoscopy or proctoscopy in 8.6%, intravenous urography in 1%, and examination under anaesthesia in 27%
[5][7]. Similar trends were reported in the European Society of Gynaecological Oncology (ESGO) survey published in 2018; cystoscopy was used during preoperative workup in 17%, rectoscopy in 10%, and examination under anaesthesia in 26%
[8]. Therefore, in 2018, the initiative to update the guidelines was undertaken under the joint umbrella of ESGO, the European Society for Radiotherapy and Oncology (ESTRO), and the European Society for Pathology (ESP) and new guidelines for the staging, treatment, and follow-up of cervical cancer patients were established, implementing imaging into the staging and treatment decision-making process
[9]. Of note, the updated recommendations after five years (2023) are unchanged in terms of recommended imaging modalities for local, nodal, and distant staging of this disease
[10].
2. Local (Pelvic) Workup for Different Stages
The role of modern imaging in local staging is to delineate the cervical tumour to determine if fertility-sparing surgery can be offered, tailor the radicality of parametrial resection based on the minimum thickness of uninvolved cervical stroma, and to assess parametrial infiltration and tumour invasion into the pelvic side wall or adjacent organs (bladder, rectum). Following the revised FIGO 2018 staging system for cervical cancer
[11][12], imaging methods include ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), combined PET-CT or PET-MRI, etc., based on local resources
[13]. In the ESGO survey published in 2018, CT, PET-CT, MRI, and ultrasound were frequently used in the pre-treatment diagnostic workup. The survey showed that in early-stage disease, MRI was the most frequently used imaging method (74%), but more than half of the respondents used CT (54%), and a minority preferred PET-CT (25%). Pelvic ultrasound was reportedly considered in 23%
[8]. Real-life data on gynecologic oncologists’ preferred primary staging modality and their diagnostic performance in early-stage cervical cancer was published in the prospective, international SENTIX study
[14]. Each participating site was instructed to choose their preferred method based on their routine clinical practice. Among 690 prospectively enrolled patients with early-stage cervical cancer, 46.7% and 43.2% of patients underwent MRI and pelvic ultrasound, respectively, whereas 10.1% underwent both modalities. Pelvic MRI and ultrasound yielded similar diagnostic performance for predicting histological tumour size, parametrial involvement, and macrometastatic nodal involvement. CT is a well-established imaging method being widely used in cancer staging. Improvements in CT technologies using helical and multi-detector CT (MDCT) yield higher spatial resolution with shorter acquisition times, albeit with reported slightly higher radiation exposure
[15]. However, CT is still inferior to MRI in assessing tumour size and local tumour extension due to its lower soft-tissue contrast resolution, even when using contrast-enhanced (CE) CT
[4][5][16][17][18]. Iodinated contrast agents are routinely used for CT examinations to visualise contrast-enhancing neoplastic lesions or metastases and diagnose various non-malignant conditions, e.g., infectious- and vascular disease (
Figure 1)
[4][16]. PET-CT provides a unique combination of anatomic information provided by CT and tissue-specific metabolic characteristics provided by PET using the glucose analogue (18)F-fluorodeoxyglucose (FDG). The fused images acquired during a single examination facilitate localising malignant lesions, typically exhibiting increased FDG-avidity and depiction of physiologic FDG uptake in non-malignant tissue. However, PET-CT is not optimal for local staging due to the low soft-tissue contrast on CT and the low spatial resolution of PET. Furthermore, the partial volume effect from FDG-avid urine in the bladder (physiologic renal FDG excretion) may preclude an accurate definition of the tumour volume or parametrial invasion (
Figure 1)
[19]. PET-CT involves slightly higher radiation exposure than diagnostic CT alone
[20]. MRI thoroughly assesses the pelvic anatomy with a wide field of view and no radiation risk. MRI has for years been considered the modality of choice for detecting local tumour spread
[21] and has been shown to yield high interobserver reproducibility for tumour size measurements with high concordance between maximum primary tumour size from MRI and from hysterectomy specimens
[22]. However, MRI is relatively expensive and time-consuming and may be contraindicated in some patients (e.g., in the presence of MRI-incompatible implants). Furthermore, limitations in access to MRI scanners are particularly common in low-income countries. As for all imaging modalities, their diagnostic accuracy depends on the radiologist’s experience in gynaecologic oncologic imaging. The MRI protocol, traditionally based on morphological sequences, has recently been supplemented by functional sequences, including diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) MRI (
Figure 1)
[23][24]. DWI depicts the free water motion of the tissue. The free water movement is restricted in malignant lesions, normally being highly cellular. The tumour appears hyperintense on high b-value (e.g., b = 1000 s/mm
2) images and correspondingly hypointense on the apparent diffusion coefficient (ADC) maps. The ADC map depicts true restricted diffusion and allows measurements of tumour ADC value. DWI combined with conventional MRI sequences enable assessments of both morphologic and physiologic features in a single examination. In DCE-MRI, dynamic image acquisition is accomplished after the administration of an intravenous bolus of gadolinium-based contrast agent. Typically, cervical tumours enhance rapidly, followed by an early washout of contrast. In the early arterial phase (30 sec post-contrast), the tumour is hyperintense. In contrast, in the late venous phase (2 min post-contrast), it is hypointense relative to the more gradually enhancing normal cervical epithelium and stroma
[17][23]. Using a contrast agent could increase the reader’s confidence in identifying stromal and parametrial invasion. On the other hand, no significant improvement in staging accuracy has been demonstrated, and therefore, its use is not considered essential
[25]. The addition of DWI to T2-weighted MRI demonstrated a promising role in improving the detection of parametrial invasion and increasing reader confidence
[26][27], allowing better tumour delineation for less-experienced radiologists. Importantly, the measured maximum tumour dimensions are reportedly virtually identical based on DWI and conventional series. Functional magnetic resonance, including DWI and DCE-MRI imaging, have also been recently addressed and studied as a tool supplementing conventional MRI in brachytherapy settings for patients with locally advanced cervical cancer. Their complementary use resulted in lower interobserver variability in target delineation (Gross tumour volume)
[28]. Nevertheless, validation through robust prospective data, before extensive adoption of DWI and DCE-MRI in cervical cancer, is essential. Particular attention must be paid to the use of uniform protocols, standardised nomenclature and correlation of imaging findings with histopathology
[29][30]. PET-MRI integration has not yet been shown to significantly improve local staging performance compared to MRI alone
[31]. Examination from the renal hila to the pubic symphysis is recommended to assess the presence of hydronephrosis in case of pelvic wall and lymph node infiltration (see below)
[24].
Figure 1. Bulky cervical squamous cell carcinoma in a 48-year-old woman with FIGO stage IVA (T4 N1 M0).Upper row: a 7 cm large (maximum diameter), mostly necrotic tumour depicted in the sagittal plane by (a) transrectal US (hypoechogenic tumour marked with dotted line having central irregular necrotic cystic areas (ellipse); (b) T2-weighted MRI depicting a hyperintense cervical lesion and contrast-enhanced (CE) CT (c) depicting a hypodense tumour. The infiltration of the rectal wall (R) and sigmoid colon (S) are marked with arrows and single letters of corresponding adjacent organ infiltration. Middle row: the same tumour depicted in the transverse plane by transrectal three-dimensional ultrasound (d); T2-weighted MRI (e) and CE CT (f); exhibits infiltration of lateral parametria (P) and rectum (R) marked with arrows and single letters of corresponding adjacent organ infiltration. Lower row: three-dimensional colour Doppler US (g) depicts high perfusion within the same tumour except in the central necrotic part (ellipse). Paraaxial DWI (h) (high b-value image: b = 1000) depicts restricted diffusion within the tumour and contrast-enhanced MRI (DCE-MRI) (i) enhancement rim in the periphery of the lesion with central necrosis. Axial FDG- PET-CT (j) depicts FDG-avid tumour except for the central necrotic part (ellipse). B, bladder; FDG, fluorodeoxyglucose; FIGO, the International Federation of Gynaecology and Obstetrics; CT, computed tomography; MRI, magnetic resonance imaging; P, lateral parametria; PET-CT, positron emission tomography fused with computed tomography; R, rectum; S, sigmoid colon.
The diagnostic potential of ultrasound is likely to have been underestimated in gynaecologic oncology until recently. In cervical cancer, ultrasound was formerly reserved for the screening of hydronephrosis. The reported limitations of ultrasound were low-contrast resolution, which may have limited the differentiation of the tumour from the adjacent tissue, small field of view disallowing the evaluation of the pelvic side wall, dependence on operator skill, subjectivity when interpreting the image, and challenging technical storage and retrieval of high-quality images on demand. However, ultrasound has undergone major technological developments over the past decades, especially the development of endovaginal high-resolution probes with a wide field of view, allowing the depiction of detailed pelvic anatomy comparable to that from MRI (
Figure 1).
Ultrasound can be performed by gynaecological oncologists who benefit from the meticulous knowledge of the disease. The same endoluminal probe can be introduced transvaginally or transrectally. The transrectal approach, performed without any patient preparation, such as enema or fasting, is preferred for cervical cancer due to the diminished risk of bleeding from the tumour. Additionally, the transrectal approach allows a better acoustic setting to show the distal part of the cervix
[32]. The combination of transvaginal/transrectal and transabdominal ultrasound allows a complete assessment of the abdomen and pelvis for abdominal staging (
Figure 2)
[33].
Figure 2. Ultrasound for cervical cancer staging.Transvaginally inserted probe (a). Transrectally inserted probe (b). Transabdominal scanning (c).
Nowadays, expert sonographers dedicated to gynaecologic oncology scanning can perform local staging by evaluating the maximum tumour size, depth of stromal invasion, infiltration of pericervical fascia and parametrial involvement up to the pelvic side wall, including the assessment of iliac vessels, pelvic muscles and nerves, and the depth of bladder and rectal wall infiltration
[34][35][36][37][38]. Moreover, the dynamic aspect of an ultrasound scan can be applied to reliably exclude the infiltration of adjacent organs (bladder and rectosigmoid colon)
[39][40]. In addition to two-dimensional (2D) ultrasound, three-dimensional (3D) ultrasound obtains the third (coronal) plane, similar to MRI. It enables easy storage of measured volumes and their retrieval for second readings, planning radiotherapy, and evaluating the treatment effect (
Figure 3). Moreover, Doppler ultrasound allows highly accurate, non-invasive, in vivo assessment of vascular features reflecting tumour angiogenesis, which has been shown to predict clinical response to neoadjuvant or definitive chemoradiation in patients with locally advanced cervical cancer
[41][42]. The transabdominal ultrasound not only evaluates hydronephrosis
[43] but may also detect abdominal intraparenchymatous metastases, lymph node metastases, and peritoneal spread during one examination (see below)
[33]. The ultrasound examination is well tolerated by patients and does not require any patient preparation, such as fasting or contrast agent application. Contrast-enhanced ultrasound is not established in cervical cancer staging
[44][45]. In addition, ultrasound is widely available and cheaper than MRI and does not have any known contraindications (
Table 1 Comparison of Different Imaging Methods for Application in Gynecologic Oncology).
Figure 3. Three imaging planes of three-dimensional (3D) ultrasound and MRI. Cervical squamous cell carcinoma in a 57-year-old woman with FIGO stage IIIC1 (T2b N1 M0). Sagittal (a), transverse (b), and coronal plane (c) on 3D ultrasound depict a large tumour of 5 cm (max diameter) as a hypoechogenic mass. T2-weighted MRI images in sagittal (d), transverse (e), and coronal (f) planes visualise the tumour as a slightly hyperintense cervical mass. Both imaging methods showed in the sagittal view an infiltration of the ventral vaginal wall (V) and in the transverse and coronal plane a bilateral invasion of lateral parametria (P) marked with arrows and single letters of corresponding adjacent soft tissue infiltration. B, bladder; MRI, magnetic resonance imaging; V, vagina.
Table 1. Comparison of Different Imaging Methods for Application in Gynecologic Oncology.