Introduction

Cancer cure is dependent on the type of local tumour growth, regional spread and metastasis. Tumour formation occurs following the proliferation of genetically unstable cells that resembles the product of a Darwinian selection process . Tumour heterogeneity in locoregional growth and metastases is further influenced by Lamarckian epigenetics . The first step in managing cancer is staging. Staging helps determine the extent of cancer in an individual patient enabling oncologists to plan treatment commensurate with the extent and severity of the disease. Staging is also necessary in determining the prognosis, the chance of recovery or recurrence. Staging of cervix cancer was first published in 1929 ( http://www.figo.org/ ). This staging system was adopted by the International Federation of Gynecology and Obstetrics (FIGO) in 1958. Later, it became known as the FIGO staging system. FIGO staging of cervix cancer is simple and practical. It is largely based on the clinical assessment of the extent of cancer growth in anatomical compartments. With a simple physical and pelvic examination, it is possible to ascertain if the cervix cancer is localised and would be amenable to surgical resection. If necessary, the surgico-pathological findings can be used to prescribe tailored adjuvant therapies. Using this approach, several important prognostic factors have come to light over the last five decades that have greatly enhanced our understanding of cervix cancer and enabled us to select appropriate adjuvant treatments and improving cure rates. More recently, advances in soft tissue visualisation through magnetic resonance imaging (MRI) and metabolic structural imaging through positron emission tomography and computerised tomography (PET/CT) have made it possible to examine many of the prognostic factors non-invasively. This had helped select appropriate patients for surgery and radiotherapy (RT) while minimising the need for adjuvant therapies. Such a reduction in the use of multi-modality treatment has resulted in the reduced costs of treatment and lower long-term toxicities.

The aim of this paper is to examine prognostic factors in cervix cancer as related to FIGO staging system , and to discuss the application of histopathological (surgical) and imaging techniques in supplementing staging process enabling treatment optimisation.

Methods

FIGO staging and histopathological prognostic factors will be discussed through selected landmark papers. Where possible, these prognostic factors will be discussed and compared with radiological imaging studies from the published literature.

A cohort of patients with cervix cancer staged using FIGO criteria, with modern radiological imaging and treated with either surgery and adjuvant RT or RT between 1996 and 2010 at the Peter MacCallum Cancer Centre will be used to illustrate the application of imaging in the staging and treatment of patients with cervix cancer.

Methods

FIGO staging and histopathological prognostic factors will be discussed through selected landmark papers. Where possible, these prognostic factors will be discussed and compared with radiological imaging studies from the published literature.

A cohort of patients with cervix cancer staged using FIGO criteria, with modern radiological imaging and treated with either surgery and adjuvant RT or RT between 1996 and 2010 at the Peter MacCallum Cancer Centre will be used to illustrate the application of imaging in the staging and treatment of patients with cervix cancer.

FIGO stage I

Cervix carcinoma confined to the uterus (extension to corpus should be disregarded).

Stage IA: Invasive carcinoma diagnosed only by microscopy. All macroscopically visible lesions – even with superficial invasion – are classified as stage IB.

Stage IA1: Patients with stage-IA1 cervix cancer have invisible lesions that can only be diagnosed by cytology or surgery, and they can be adequately treated by a cone biopsy or a simple hysterectomy. There is no role of radiological imaging either in the staging or in the management of these patients.

Patients with stage IA2 had a risk of 2–8% of lymph-node metastases , and almost a third of these cancers had lymphovascular space invasion (LVSI). Of 94 patients, seven patients had nodal metastases, and of these five developed recurrent disease and four died of disease. The recommended treatment for patients with stage-IA2 cervix cancer is hysterectomy with lymphadenectomy, and node-positive patients should be offered adjuvant pelvic RT. Radiological imaging in these patients is unlikely to assist in improved patient management.

Stage IB: Clinically visible lesion confined to the cervix or to the microscopic lesion > IA2.

IB1: Clinically visible lesion ≤4.0 cm in the greatest dimension.

IB2: Clinically visible lesion >4 cm in the greatest dimension.

FIGO stage IB is the most studied stage of cervix cancer. It can be treated equally effectively either by primary surgery or by RT. Clinical behaviour of stage-IB disease has been extensively studied.

The Gynecologic Oncology Group (GOG) undertook a prospective study of the surgical and pathological data on patients with histologically confirmed stage-I cervical cancer . Surgical and pathological data were collected on 1125 patients. Of 940 evaluable patients, 732 had squalors’ carcinoma, and they were selected in the study group. In 87 (12%) patients following exploratory laparotomy, hysterectomy was abandoned due to the presence of extrauterine disease. We shall return to this group later. Six hundred and forty-five patients who did not have a macroscopic nodal involvement were treated with radical hysterectomy, pelvic and para-aortic (PA) lymphadenectomy. The incidence of nodal metastases in this group was 15%. Four independent risk factors for nodal metastasis were LVSI ( P < 0.0001), depth of invasion (DOI) in the outer third of cervix stroma ( P < 0.0001), parametrial involvement ( P = 0.0005) and age ( P = 0.02). Three-year disease-free survival for node-negative patient was 85% and for the node-positive patients was 74%. Clinical tumour size >3 cm, LVSI and a deep cervical stromal invasion were the independent prognostic factors. In this series, lymph-node metastasis did not show any statistically significant adverse prognosis. This was due to the patient selection, for 87 patients although clinically staged as FIGO stage I, they had extrauterine disease at the time of laparotomy and were excluded from the analysis. Five-year survival in these patients was 30% as shown in a later publication . Those patients with grossly involved pelvic nodes had a survival of 33%, which was reduced to 16% when PA nodes were present. Nevertheless, based on these highly significant risk factors, GOG developed a relative score to predict the risk of failure in FIGO stage-I cervix cancer . A GOG score of >120 correlated with a 40% risk of recurrence requiring post-operative RT. The efficacy of adjuvant pelvic RT was investigated in a phase-3 GOG study . Two hundred and seventy-seven patients were accrued. Some of the important risk factors for recurrence were >1/3 stromal invasion, LVSI and >3 cm clinical tumour diameters were accrued in this study. Of 277 patients with at least two of these factors, 15% in the RT group and 28% in the ‘no further treatment’ group had a cancer recurrence. Life table analysis indicated a statistically significant (47%) reduction in the risk of recurrence (relative risk = 0.53, P = 0.008). This improved survival came at a cost of 6% grade 3/4 adverse events.

Generally speaking, primary surgery and primary RT were both equally effective treatments for patients with FIGO stage-IB and stage-IIA cervix cancers. However, long-term toxicity profiles of surgery and RT as a primary modality was different. This was shown in a landmark phase-3 study by Landoni et al. .

Four hundred and sixty-nine patients were randomised to receive either RT or radical hysterectomy and pelvic lymph-node dissection (RHPLND). Overall survival was 83% in both arms. There was a significant difference in toxicity (28% vs. 12%; P = 0.0004), favouring the radiation arm. Fifty-four percent patients in stage IB1 and 84% in stage IB2 in the surgical arm required adjuvant radiation on account of adverse prognostic features. The adverse prognostic features were LVSI (62%), lymph-node metastases (26%), <3 mm safe stromal margin (24%) and positive resection margin (11%).

Surgical treatment has an obvious advantage. The duration of treatment and acute recovery usually takes less than a couple of weeks. Once the patient recovers, pelvic tissues retain the normal pliancy. Very few patients have significant long-term effects of surgery. RT on the other hand typically requires 6–8 weeks of treatment including brachytherapy. A few patients may also develop long-term gastrointestinal and urinary symptoms, and they would also require chronic local application of oestrogen to avoid dry and contracted vagina. Those patients whose tumour cannot be adequately treated with primary surgery end up getting toxicities of both treatments.

Patients with cervix cancer that can be cured by surgery alone should not be treated with primary RT. At the same time, if it is possible to predict the requirement of post-operative adjuvant RT when surgery is contemplated, then these patients should only be treated by primary RT. Of course, the treatment selection would also depend on the available local expertise, resources and patients’ choice and preference. A detailed incremental cost-effectiveness analysis of RHPLND in comparison to concurrent chemo-radiotherapy (CCRT) had been published by Katanyoo et al. For stage-IB patients, RHPLND provided better value for money than CCRT, with a probability of 75%. This advantage was seen only for patients who required no adjuvant treatment. It is therefore desirable to select a patient for either modality appropriately.

Assuming that a centre has excellent expertise in all modalities of treatment, and it is well resourced, following an outpatient clinical assessment of a FIGO stage-I disease, based on her age, fertility and tumour size the patient should be investigated as follows.

Patients wishing for fertility preservation surgery

Fertility can be preserved following a trachelectomy. Radical trachelectomy can be performed vaginally or abdominally as well as endoscopically. Ideally, the lesion should be < 2 cm, should have limited extension in the upper endocervical canal and should be limited to within 3–5 mm clear of parametria. A recent retrospective review of the literature has shown <1% incidence of parametrial infiltration in lesions that were <2 cm and had <50% cervix stromal invasion . In patients with clinically staged <2-cm lesion, about 25% had lesions >2 cm as seen in histology specimen . In properly selected cases following radical trachelectomy and lymphadenectomy, recurrence rate was usually <5%, and death due to recurrent disease was <2% . Eighty percent of these women were able to conceive, albeit with assisted reproductive technologies .

Such precise patient selection was possible through MRI

One of the first MRI reports of uterus and cervix was published in 1983 . The quality of image was such that not only could the uterus be distinguished from the surrounding soft tissues, but the transitional zone, which separates the corpus uteri and the cervix, could also be seen. MRI made it possible to measure tumour dimensions non-invasively, which reflected the corresponding histological measurements . In general, the basic imaging protocol for T1-weighted images of the pelvis in the axial plane and T2-weighted images in the axial and sagittal planes are adequate for the imaging of cervix cancer and measuring the tumour dimensions . T1-weighted sequences are better at delineating small cervix tumours in patients planned for a fertility sparing surgery . Trachelectomy can be performed on women with >2-cm tumours, provided these patients are free from a tumour in the upper endocervix, and they do not have a lymph-node metastasis. Women between 26 and 41 years of age were staged with an MRI scan, and they had undergone laparoscopic lymphadenectomy. Twenty women with >2-cm (2.1–5 cm) primary tumour and negative nodes were then treated with neoadjuvant chemotherapy (NACT) consisting of two or three cycles of paclitaxel/ifosfamide/cisplatin followed by a radical vaginal trachelectomy performed in 18 women . Seven pregnancies occurred in five women. Cancer recurred in one patient.

Generally speaking, patients with FIGO stage IB with tumours <3 cm can be adequately assessed by clinical examination, and they can preferentially be treated by RHPLND. Where possible, prior to hysterectomy, nodal regions should be inspected, and any suspicious nodes should be sent for frozen section. Hysterectomy should be abandoned if lymph-node metastasis is found on a frozen section, to avoid subsequent adjuvant treatment with CCRT. Most of these patients would not require any adjuvant treatment. However, a small number of these patients will remain at a high risk of relapse, and they will require adjuvant RT. High-risk patients in a node-negative group can be identified by scoring their tumour size, stromal invasion and LVSI according to an algorithm based on a prospective study by GOG . Patients with a score of >120 have a 40% risk of relapse. The use of adjuvant pelvic RT in this group resulted in a 47% reduction in the risk of recurrence as shown in a phase-3 study by Sedlis et al. Clearly, there is a subgroup of patients with early cervix cancer where preoperative detection of high-risk features would be an advantage. Presently, LVSI cannot be detected by non-invasive means.

MRI and PET/CT can assist in estimating the accurate tumour size, volume and lymph-node metastases. In 125 patients with FIGO stage-IB1 cervix cancer, the tumour diameter and parametrial invasion (PI) measured preoperatively by MRI correlated accurately with a pathology specimen . Almost half of 127 patients had tumour >2 cm. In this group, the incidence of LVSI and cervical stromal invasion of >1/2 was observed in 68% and nodal metastases were in 19%. Although MRI was superior in detecting pelvic lymphadenopathy in comparison with CT and lymphangiography, the difference was not statistically significant . The MRI detection of metastatic lymph nodes is dependent upon the size of the involved nodes. Nodal metastasis in the early-stage cervix cancer tends to be smaller and fewer. Fifty-nine women with FIGO stage IB1–IIA were treated with pelvic lymphadenectomy, and all except one patient had radical hysterectomy. Pelvic lymph-node metastases were documented histologically in 19 patients. PET correctly detected 10 of the 19 patients . Four PET scans were false positive for pelvic nodal disease.

As the incidence of nodal metastases (6–13%) and PI (<5%) in the early-stage cervix is low, the utility of MRI and PET/CT ²⁷ in identifying patients requiring adjuvant RT prior to surgery is limited.

In this regard, lymphadenectomy and frozen section of suspected lymph nodes prior to radical hysterectomy is more useful in avoiding dual modality treatment in this group of patients. Between 1997 and 2005, 107 patients with stage IB1 and IIA1 had a full laparoscopic pelvic lymph-node dissection. Twenty-one patients were found to have positive nodes, and these were treated with CCRT. The remaining 86 node-negative patients underwent a standard radical abdominal hysterectomy 48 h after the laparoscopic lymphadenectomy. Only 6.4% patients relapsed in hysterectomy arm and 19% in the CCRT arm. Thus, an appropriate patient selection not only avoided the application of two radical treatments in some patients but also improved results for each primary modality employed .

Patients with a clinical tumour diameter >3 cm planned for RHPLND will benefit from MRI. In our centre, 149 patients with a 2–4 cm, stage-IB and stage-IIA tumours, 27 (18%) had cranio-caudal diameter of their tumour between 4.5 and 7 cm, 28 (19%) had parametrial infiltration on MRI and 34 (23%) showed fluorodeoxyglucose (FDG) avid lymph nodes on PET/CT.

FIGO stages IB2, IIA2 and IIB

Patients in these stages often have >4 cm tumours. These patients are usually treated using CCRT (NCCN clinical practice guidelines, 2012) .

A combination of NACT, surgery and systemic chemotherapy is also used in their management, in places where adequate facilities for external beam radiotherapy (EBRT) or brachytherapy is not available. In some countries, NACT and RHPLND are preferred due to strong surgical tradition or higher incidence of severe RT-related complications . Sometimes, all three modalities are added simply in the hope of improving survival . Clinical assessment is considered adequate in staging these patients when the selected treatment is CCRT, although discrepancy in clinical assessment under anaesthesia (EUA) and corresponding anatomical pathology is well known and observed in 25% of tumours designated stage I, and in 50% of tumours designated stage II . Significant prognostic factors in this group of patients include tumour volume, corpus invasion, PI and lymph-node metastasis as well as the level of nodal involvement .

Pathological tumour diameter in FIGO stage IB2 and IIA correlated well with the corresponding MRI diameter. The FIGO measurement of tumour size (EUA diameter) related only poorly to the derived average MRI tumour diameter . In a study consisting of 179 patients who had a pre-treatment MRI prior to their curative RT, only corpus invasion ( P = 0.013) and tumour volume ( P = 0.004) were significantly and independently associated with osteosarcoma (OS) .

Is there a prognostically significant cut-off value of tumour diameter? The cut-off value of 4 cm, used in subdividing FIGO stage IB, has been challenged . Following RHPLND in 109 stage-IB1 and 86 stage-IB2 patients, the tumour size failed to predict a recurrence or a nodal status when stratified by the LVSI, the DOI or the PI . In another study, Soutter et al. found 13 cc as a useful cut-off for patients selected for RDPLND predict survival with a positive-predictive value of 0.93 and a negative-predictive value of 0.75. Similarly, Kodaira et al. found a cut-off value of 50 cc to be significant in separating the good and bad prognostic tumours when treated with RT . In fact, tumour volume is a continuous variable . Dichotomising any continuous variable would result in a loss of information. It is possible to make use of different cut-off values for different purposes. In FIGO stage I, a tumour diameter of <2 cm would be suitable for fertility preserving surgery. A tumour volume of <13 cc would be suitable for RHPLND. Most of these patients would not require post-operative RT. A tumour volume of <50 cc among locoregionally advanced cancer is more likely to do well following current treatment with CCRT. All patients with a tumour volume of >50 cc are likely to fare poorly irrespective of their tumour stage. In our centre, 515 patients had their pre-treatment MRI before receiving curative treatment. The failure rate was 52% in 186 patients who had >50-cc tumour volume, whereas 28% of 329 patients with a tumour volume of <50 cc failed. Clearly, an MRI scan can assist in appropriate patient selection for conservative surgery, RHPLND and for CCRT simply by estimating the tumour volume! The MRI was given a rating of 8–9 out of 10 as a pre-treatment planning investigation of invasive cancer of the cervix by the member of a multidisciplinary expert panel under the aegis of the American College of Radiology (ACR) .