Communication between the surgeon and the radiation oncologist is improved with the use of virtual models of the final tumour resection, combining three-dimensional imaging and conventional clip marking with computer-aided navigation. This investigation was designed to determine the deviation of virtual marking procedures compared to conventional marking by titanium ligature clips in oral cancer with different localizations. Seventeen patients with surgically placed clips and virtual landmarks on the resection margin after complete tumour ablation were evaluated. To determine whether the virtual landmarks remain predictive of the resection margin, the deviation of the virtual points from their corresponding clips was analyzed by measuring the distance between their centres of gravity. In total, 189 clips were evaluated. Metric analyses of the deviation between the virtual points and clips showed a deviation of 2.3 ± 0.6 mm for tumours with a maxilla localization, 7.2 ± 2.5 mm for tumours with a mandible localization, and 12.6 ± 3.8 mm for tumours with a tongue localization. A significant statistical relationship was demonstrated in the virtual point–clip deviation as a function of tumour localization. Virtual marking of maxillary tumour resection margins allows accurate definition of the former tumour bed and could lead to novel adjuvant treatment strategies.
Up to 90% of patients with stage I or II oral squamous cell carcinoma (SCC) are cured with surgery, however outcomes for patients with locally advanced stage III or IV head and neck cancer have been less promising. The primary pattern of failure is loco-regional. As a consequence, adjuvant radiotherapy is often recommended for locally advanced SCC of the head and neck. Although adjuvant radiotherapy is associated with significantly improved overall and cancer-specific survival, the outcomes for lymph node-positive patients remain suboptimal. Even with combined surgery and radiotherapy, the 5-year survival of patients with lymph node-positive head and neck cancer has been shown to be only 43%. These data highlight the importance of investigating novel treatment strategies, such as increasing the radiation dose intensity or integrating new treatment modalities for improved visualization of the tumour bed.
The optimal therapy of advanced SCC consists of complete excision of the tumour, followed by external beam radiotherapy to the primary tumour site and the dissected neck.
A strategy that aims to improve the therapeutic ration for patients with SCC at a relatively low risk of local tumour relapse, involves limiting high radiation doses to the tumour bed and the involved lymph nodes and reducing the dose to the head and neck tissue remote from the former tumour bed.
In most centres, a tumour bed boost is delivered. Usually the electron boost is marked clinically using information such as the area of surgical induration, data on the preoperative tumour location, surgical and histopathological annotations, and postoperative three-dimensional (3D) imaging such as computed tomography (CT) or magnetic resonance imaging (MRI) datasets. During recent years, in view of indications suggesting that the tumour bed boost may contribute to increased local control, attention has focused on strategies to increase the accuracy of radiation planning. A key requirement of adjuvant radiotherapy is the precise localization of the tumour bed after surgery, for which many oncologists are using 3D imaging.
A number of investigations have been reported in which surgical clips left at the excision cavity margins have been used to delineate the boost treatment field in breast-conserving tumour therapy. The accurate determination of surgical cavity volume is also important for adjuvant irradiation after head and neck tumour resection.
Therefore the precise orientation of the boundaries between resection borders and native tissue becomes important. The delineation of reconstructed tissue from resection margins on postoperative CT scans is often difficult. In this situation, marking of the margins with titanium clips could be useful. Markers implanted in surgical cavity walls provide additional localization information compared to CT imaging alone.
Several improvements in the area of computer-assisted planning have been implemented to improve the labelling of resection margins. Surgical navigation systems have become an established technique in the field of head and neck surgery. They allow simultaneous visualization of an operative site as well as surgical instruments, and relate them to the patient’s preoperative treatment. 3D virtual marking and mapping of resection margins is a new technique that enables the precise delineation of skull base and midface tumour margins. However, virtual marking of tumour resection margins could be insufficient apart from hard tissue, due to the effects of swelling or gravity on the tissue, whether native or reconstructed.
The aim of the current investigation was to determine the deviation of virtual marking by navigation probe compared to conventional marking by titanium ligature clips in oral SCC with different localizations.
Materials and methods
Seventeen patients with SCC had surgical clips inserted and additional virtual marking at the margins of the excision cavity during surgery. Ethical approval was obtained before undertaking this study. Written informed consent was obtained from all study patients.
All patients underwent tumour resection, neck dissection, defect reconstruction, and marking of the tumour resection margin in one stage, performed by the same oral and maxillofacial surgery team.
The standard protocol was to insert clips at the margins of the excision cavity. All patients had radio-opaque titanium ligature clips placed after tumour ablation and before tissue reconstruction (Ligaclip Extra Titanium Medium; Ethicon Endo-Surgery, Cincinnati, Ohio, United States). Each excision cavity boundary was defined by clips positioned at the deep, superficial, medial, lateral, inferior, and superior boundaries. Additional markers were placed at 15–20-mm intervals on the soft tissue margin if tumour resection margins were irregularly configured.
In addition, the precise localization of the resection margin was marked and mapped by the navigation probe. Computer-assisted surgery (CAS) systems require a headset transmitter to be secured to the patient with an additional transmitter on a pointer. Based on preoperative imaging and the identification of fiducial markers on the CAS monitor, the optical detector allows the computer to integrate spatial information from the actual patient with a 3D virtual treatment plan (iPlan 3.0.3; BrainLab, Feldkirchen, Germany). This allows precise localization of the patient and surgical instruments at the patient. The navigation pointer was then used to mark each titanium clip placed at the resection margin and these precise localization data were stored. The virtual marking procedure was performed precisely in the centre of each clip placed ( Fig. 1 ).
In the case of a tumour of the lower jaw, a constant fixed positioning of the mandible was ensured and additional virtual landmarks on bony and on dental fiducial markers of the mandible were determined. All virtual landmarks were incorporated into and stored in the 3D virtual treatment plan.
The numbers of clips inserted and virtual landmarks were recorded for each patient. Successful clip insertion was defined as at least six clips inserted in the correct position around the tumour resection margin.
A postoperative CT scan was done at least 3 weeks after surgery to avoid postoperative changes due to swelling. All patients underwent a postoperative CT scan with slices of no greater than 3 mm in thickness, as per standard practice. The postoperative CT scan was performed using the same head position as in the first CT scan.
The preoperative virtual treatment plan and postoperative images were fused using a bone fusion registration based on mutual information. All landmarks and ligature clips were detected on axial view using the planning software iPlan 3.0.3. In the case of a tumour of the lower jaw, the mandible was chosen as the region of interest for bone fusion registration, taking into consideration the additional virtual landmarks placed on bony and or dental fiducial markers on the mandible.
All titanium clips placed were then outlined manually using the brushing tool on the software and all corresponding virtual markings were identified. Data related to the clips and the corresponding landmarks were exported as STL files and imported into a similar coordinate system, and the distance between the centres of gravity were measured (Rapidform XOR; Inus Technology Inc., Seoul, South Korea) ( Fig. 2 ).
Descriptive statistics for quantitative variables are given as the mean ± standard deviation. For statistical analysis, a mixed model for fixed and random effects and pair-wise comparisons of marginal linear predictions and an overall test were performed. The data were analyzed with IBM SPSS Statistics version 21.0 (IBM Corp., Armonk, NY, USA). Figures were generated with SPSS and Microsoft Office Excel (Microsoft Excel for Windows 7; Microsoft Corporation, Redmond, WA, USA).
The clinical marking of the resection margin after complete tumour ablation was successful in 17 patients. One hundred and eighty-nine clips were placed intraoperatively. The mean number of inserted clips was 11 (range 6–25). For each clip placed, a corresponding virtual marker was assigned. For all patients, the clips placed intraoperatively and the virtual landmarks could be identified in the postoperative CT scan and allowed delineation of the former tumour bed ( Fig. 3 ).