High-precision MRI-guided adaptive brachytherapy for cervical carcinoma

Abstract

Over the past 20 years advances in radiological technology have led to dramatic changes in cervical cancer management. External beam radiotherapy has evolved radically as a result of these changes. More recently, though, three-dimensional (3D) image guidance and volume-based prescribing has been adopted into cervical cancer brachytherapy. This overview covers the science and technology behind magnetic resonance imaging (MRI)-guided adaptive brachytherapy and explains why it is now considered the new international standard of care.

• Clinical
• CT
• imaging
• MRI
• PET
• physics

Introduction

Brachytherapy plays a fundamental role in the treatment of cervical carcinoma. Internationally, concomitant chemoradiation followed by brachytherapy is regarded as the standard of care for patients with tumours staged between IB2 and IVA. The addition of chemotherapy to external beam radiotherapy (EBRT) has been found to improve overall survival by approximately 12% [1–3] but at the cost of increased gastrointestinal (GI) and haematological (only one use) toxicity. In patients unsuitable for concurrent chemoradiotherapy, studies examining the addition of hyperthermia to radiotherapy have shown improved local control rates and overall survival in patients with advanced tumours [4], [5]. The radiobiological and radiophysical principles of brachytherapy though allow a higher dose of radiation to be delivered to the cervical tumour whilst minimising toxicity to the organs at risk (OAR).

The principle of brachytherapy is that the radiation source is placed directly into the area to be treated thus delivering an intense high dose which dissipates rapidly according to the inverse square law. The procedure has been well developed over many years with traditional applicator systems which have subsequently been adapted for use with modern afterloading techniques. Afterloading is now universally employed allowing safe and accurate placement of applicators prior to radiation delivery and minimising the radiation exposure to staff and patients. Conventional applicator systems are based on a central intrauterine tube (the tandem) and a vaginal applicator which may be either a pair of ovoids within each fornix or a ring placed at the top of the vagina. These are usually rigid applicators, fixing together in a defined format which enables accurate and reproducible dose distributions. An alternative method which has been developed in Paris is to use a vaginal mould technique [6], [7]. This method involves the creation of a customized vaginal mould for each patient, which houses one intrauterine and two vaginal catheters. The position and length of the catheters within the mould are based upon the patient's anatomy and the size and extent of their tumour. Following insertion of the mould under general anaesthetic the patient is then scanned in the MRI scanner and the necessary planning volumes delineated.

The radiation can be delivered using either low dose rate (LDR), high dose rate (HDR) or pulsed dose rate brachytherapy (PDR). The dose rate refers to the rate at which radiation energy is release from the source and is expressed in Gray per hour (Gy/h). Although the equivalent prescribed biological dose is the same no matter which dose rate is used, a correction is required with a reduction in total physical dose as the dose rate increases, and for HDR, fractionated delivery of the dose over 3 or 4 fractions. Typically in LDR and PDR the source stays in situ for up to 24–48 h before being removed, while in HDR this time is typically a few minutes [8]. Selection of dose rate depends upon the centre's brachytherapy equipment and staff experience, but LDR machines have now largely been phased out and modern practice is based on either HDR or PDR. Published outcomes using all three dose rates show high local control rates with acceptable levels of toxicity [6], [9], [10]. It has been postulated that PDR brachytherapy allows a wider range of treatment optimisation and facilitates the treatment of a larger volume of tissue without increasing toxicity [10]. Lower rates of local failure when compared with LDR brachytherapy have been reported but will need to be confirmed with larger studies [6].

The historical convention when delivering cervical brachytherapy is to prescribe the radiation dose to a point termed Point A and to use the International Commission on Radiation Units and Measurements (ICRU) Report 38 for OAR point dose reporting [11]. The combination of this prescription convention with radiographic 2D imaging has consistently shown high local control rates for early stage disease, but unfortunately these rates reduce significantly when the disease becomes bulkier and more advanced [12]. Studies have shown that by escalating the brachytherapy dose delivered using conventional non-adaptive 2D X-ray imaging there is an improvement in local control rates but at the cost of significantly higher rates of toxicity [13–16]. Interest has therefore grown in discovering ways to improve local control while maintaining or reducing existing OAR toxicity. This has led to the development of high-precision computed tomography (CT) and MRI-guided volume-based brachytherapy for carcinoma of the cervix.

Search strategies used and sources of information

PubMed was searched for articles published before January 2012 containing the following terms: cervix cancer, cervical cancer, brachytherapy, MRI-guided, Royal College of Radiologists, RCR, CT, PET, and MRI. All relevant and any related articles were then fully reviewed.

Imaging technologies used in treating cervical cancer

Important prognostic factors associated with cervical carcinoma are tumour size, tumour extension and nodal involvement [17–19]. Hence radiological imaging in combination with physical examination plays a vital role in the diagnosis, treatment and follow-up of patients with carcinoma of the cervix. This applies at every step of the management process and embraces ultrasound (US), CT, MRI and positron emission tomography (PET). Over the past 20 years there has been a significant evolution in the application of imaging technology to the planning and delivery of radiotherapy for cervical carcinoma.

Since the early 1990s CT has been adopted into the routine planning and delivery of EBRT. This has led to the creation of what is termed conformal radiotherapy which describes a volume-based approach to EBRT planning whereby the tumour and other clinically important areas are radiologically visualised and delineated on the CT to create 3D volumes. ICRU Reports 50, 62, 71, and 78 define the structures and terminology used to represent and report these volumes [20–23]. The structures normally delineated on the CT planning scan are the gross tumour volume (GTV), clinical target volume (CTV), planning target volume (PTV) and OAR. Once these volumes have been delineated on the planning CT then advanced planning software is used to shape the radiotherapy dose so that it conforms to the PTV.

This volume-based approach has been shown to improve dose homogeneity across the tumour while helping to minimise unwanted normal tissue toxicity by the adoption of volumetric dose constraints for OAR [24], [25]. In one study comparing 3D conformal radiotherapy planning with older conventional radiotherapy planning, the dose received by the critical organs, the rectum and bowel, was approximately 50% lower (96.82% versus 38.23% and 61.37% versus 30.79% respectively) [26].

Despite the benefits of CT leading to significant advances in EBRT planning and delivery it still remains inferior to MRI with regard to soft tissue visualisation. With MRI it is possible to alter the contrast of different tissues to improve tissue plane visualisation and detect subtle abnormalities. In cervical carcinoma this means that the relationship of the tumour to the uterus and vagina can be accurately determined and any early invasion of the parametria identified. Compared with CT, MRI has been evaluated as the more accurate imaging modality for cervical cancer staging, 86% versus 63% respectively [27], with a sensitivity of 93% for early tumour detection and an overall staging accuracy of 86% [28]. MRI staging accuracy has also been assessed by comparing it with findings found at the time of surgery. MRI was found to accurately estimate tumour size to within ±5 mm and correctly identify parametrial invasion in 88% of cases [29–31].

Although MRI offers a distinct advantage over CT in terms of tumour visualisation it has not replaced CT in EBRT planning. MRI relies upon the signals derived from different water densities in tissue so unlike CT it does not measure the electron density of tissues. Also, due to the low water density of cortical bone, MRI cannot accurately detect a signal from it. Because of these differences MRI cannot be used in EBRT planning because it lacks the information required for dose calculations. Hence the data provided by CT is still regarded as vital in EBRT treatment planning.

In brachytherapy planning however, tissue electron density data is of less importance since dosimetry is dependent primarily upon the inverse square law. Consequently MRI can be adopted directly for brachytherapy using its superior tumour visualisation. Dimopoulos et al. demonstrated that, unlike CT, MRI for brachytherapy planning permitted the precise delineation of the GTV, CTV, OAR and applicators [32]. Hence there has been a transition from Point A-based 2D dosimetry to MR-based 3D dosimetry for cervical cancer which has required significant changes to methods of dose prescribing and reporting.

An introduction to MRI-guided adaptive brachytherapy

The first reports on the use of 3D imaging in pelvic brachytherapy were in the late 1980s [33]. At the time clinicians were using CT-based imaging to calculate brachytherapy dose data for the tumour and OAR. With the advent of 3D conformal EBRT, a similar approach was being explored in brachytherapy which would allow dose escalation while minimising unwanted toxicity. Early proponents of MRI-guided brachytherapy showed using small case series that this technique could offer those potential benefits [34–37].

These early studies revealed that the traditional method of using point doses rather than volumes for dose prescription and reporting was unsatisfactory. The studies which had used CT-based imaging to facilitate brachytherapy had discovered that when the dose delivered to point A using the CT technique was compared with the dose delivered to point A using the traditional radiographic technique that the traditional technique could lead to tumour under-dosage [38], [39]. Kim et al. discovered that the traditional radiographic method of prescribing to point A overestimated the dose received by the tumour and that the tumour stage seemed to be inversely proportional to the volume of GTV covered by the prescription dose. By examining the dose volume histograms created using their CT-based brachytherapy data they reported that for FIGO stages IB1, IB2, IIB and IIIB the dose prescribed to point A encompassed on average 98.5%, 89.5%, 79.5% and 59.5% of the GTV volume respectively [39]. These studies also revealed that the traditional method of using ICRU Report 38 to report the OAR doses was inaccurate and underestimated the maximum dose received by the bladder and rectum when compared with the new CT volume-based OAR data [38], [40]. As a result a new system of tumour voluming, dose prescription and dose reporting was developed to replace the traditional point-based system [41], [42].

Defining standards in image-guided brachytherapy

These developments led to the publication of several different recommendations and guidelines. Following the American Brachytherapy Society's 2004 recommendations [43] the Groupe Européen de Curiethérapie-European Society for Therapeutic Radiology and Oncology (GEC-ESTRO) working group published their own recommendations in 2005 and 2006 [41], [42]. Their published recommendations covered all aspects of tumour target and OAR contouring and provided guidance on the reporting of dose volume parameters for image-guided brachytherapy (IGBT). Their recommendations defined an individualised approach to cervical brachytherapy whereby image guidance was used to help conform the dose to the tumour volume and pelvic topography while balancing the organ at risk dose volume constraints [41], [42]. In 2005 it was agreed internationally that the GEC-ESTRO recommendations would form the basis of future work in gynaecological IGBT [44].

GEC-ESTRO sets out clear recommendations with regard to tumour and OAR voluming. They advocate the delineation of three specific volumes relating to the tumour: GTV, high risk (HR) CTV, intermediate risk (IR) CTV and OAR. The organs at risk are defined as the rectum, bladder, sigmoid colon and any adjacent bowel loops which are considered relevant to the treatment target volume. As is the case with conformal EBRT, it is hypothesised that the risk of microscopic disease varies with tumour cell density. Consequently the GTV, HR-CTV and IR-CTV are considered as having different reducing radiotherapy dose requirements based on this hypothesis.

At the time of brachytherapy planning, the patient is scanned in the MRI scanner with the brachytherapy applicators in situ. Then, using the clinical examination findings, initial MRI staging information and the brachytherapy imaging data, the treating clinician delineates these volumes on the planning scan. Any areas of residual palpable disease which are visible as areas of high signal intensity on the T2 weighted planning MRI scan are delineated and defined as the HR-CTV. Any areas at high risk of containing tumour cells are also included in the HR-CTV. These are characteristically described as ‘grey zones’ on the MRI scan and represent areas of disease regression following chemoradiotherapy.

The IR-CTV is normally determined by the original volume of the tumour and any initial pelvic extension. Typically it is defined by a margin of 10–15 mm around the HR-CTV [41]. Once the HR and IR-CTV's have been volumed then the clinician creates volumes for the OARs and the applicators in situ which are tracked and defined (Figure 1).

Figure 1. The image shows an axial slice from a T2 weighted MRI-guided adaptive brachytherapy planning scan. Delineated on the scan are the GTV (green), HR-CTV (red), IR-CTV (blue) and OAR (sigmoid colon (yellow), bladder (purple) and rectum (green)).

Once these planning volumes have been created it is then the role of the brachytherapy physicist to determine an appropriate radiation delivery strategy. The aim of treatment planning is to ensure that the HR-CTV receive an adequate dose to the D90 (dose to 90% of the HRCTV) while observing the OAR dose constraints. Delivery of an equivalent dose in 2 Gy per fraction (EQD2) of ≥87 Gy to the HR-CTV D90 is recommended as this has been shown by Dimopoulos et al. to correlate with a local pelvic control rate of >90% [39], [45]. If it is not possible to cover the HR-CTV D90 volume adequately using a standard delivery technique e.g. using a tandem ring applicator, then it has been shown that the addition of parametrial interstitial needles to the intracavity treatment can improve dose coverage and homogeneity [46], [47] (Figure 2).

Figure 2. Axial and sagittal T2 weighted images from two different patients MRI planning scans. The top two images show the dose wash from a standard tandem ring applicator technique (red, high dose; blue, low dose). The bottom two images demonstrate how the addition of interstitial needles to the standard tandem ring applicator technique can help conform the dose around the IR-CTV (note how the interstitial needles on the right and left help extend the high dose coverage out towards the pelvic side walls).

Adequate coverage of the HR-CTV D90 must be balanced against achieving the OAR dose constraints. It is well recognised from historical case series using 2D non-adaptive treatment techniques that the incidence of late radiation toxicity is determined by the total radiation dose delivered to the pelvic organs, the volume of normal tissue treated and the rate of dose delivery [48–51].

The GEC-ESTRO recommendations changed the way in which OAR doses were reported by recommending the use of dose volume histograms (DVH). Now that OAR structures were volumed it was possible to calculate the maximum dose received by 2 cm³, 1 cm³ and 0.1 cm³ of OAR tissue. GEC-ESTRO termed these volumes as D2_cc, D1_cc and D0.1_cc and defined them as the maximum dose received by that specified volume of OAR tissue [41], [42].

Recently, Georg et al. published work correlating the DVH parameters defined by GEC ESTRO with OAR toxicity following chemoradiotherapy and MRI-guided adaptive brachytherapy. They found that the D2_cc had good predictive value for late rectal and bladder toxicities graded II–IV [52]. They had difficulty finding a correlation between D2_cc and sigmoid colon toxicity owing to a paucity of bowel toxicity events within their data series [53].

Based on the GEC-ESTRO recommendations and more recently published reports, the current dose constraints are 70–75 Gy for the rectum and sigmoid colon and 85–90 Gy for the bladder [12], [42], [52], [53].

The future of MRI-guided adaptive brachytherapy

MRI-guided adaptive brachytherapy is now considered the gold standard technique for treating cervical carcinoma and is recommended in numerous published guidelines [41], [42], [44], [54–56]. Pötter et al. published an update on their MRI-guided brachytherapy patient series (n = 156) in 2011 which included details on local control and toxicity rates. They reported a complete remission rate of 97% (151/156) with local control rates at 3 years of 95% overall and 98% for tumours 2–5 cm and 92% for tumours >5 cm [57]. Notably, the improved local control rates seen for tumours >5 cm (92% versus 67%/64% from the Vienna historical series [9], [58]) seemed to impact on cancer-specific survival (CSS) rates at 3 years. The CSS rate improved from 57% in 1993–1997 and 40% in 1998–2000 to 70% in 2001–2008 [9], [57], [58]. The CSS rate at 3 years for tumours 2–5 cm remained unchanged during the same time period. They advised caution in drawing conclusions from this trend because their length of follow up was only 3 years and might be explained by a number of other important factors and not just adaptive brachytherapy. Nevertheless it is an important finding that will need to be investigated through future prospective studies.

Despite these recently published findings and those of several other smaller international series which have shown improvements in local control rates (∼20%) without any significant increase in toxicity [10], [46], [59], MRI-guided brachytherapy still remains a new and relatively under-utilised approach.

In the UK the Royal College of Radiologists published recommendations on the implementation of image-guided brachytherapy in 2009. Since their publication there has been a significant increase in the number of UK centres using image-guided brachytherapy techniques. Tan et al. surveyed UK oncology centres in both 2008 and 2011 and found a 27% increase in centres using some form of 3D imaging (51% versus 78%) and a 16% increase (4% versus 20%) in centres using MRI-guided brachytherapy [60]. This would suggest that certainly in the UK clinicians are adopting image-guided treatment techniques into the management of cervical carcinoma.

In 2008 an international collaborative study named EMBRACE [61] was launched which aims at improving MRI-guided brachytherapy implementation and quality assurance across the globe. Through prospective data collection on local control rates and toxicity it hopes to further validate the role of MRI-guided brachytherapy in the management of cervical cancer using the GYN-GEC-ESTRO recommendations. Thus far it has collected data on over 600 treated patients and is on track to reach the target recruitment number. It is hoped that the findings when published will reinforce the future of MRI-guided adaptive brachytherapy in gynaecological brachytherapy. Needless to say their final results are eagerly awaited by the radiation oncology gynaecological community.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.