The aim of this study was to investigate the correlation between tumour thickness (TT) on intraoral ultrasound (US) and magnetic resonance imaging (MRI) with the histologically determined TT of tongue cancers. Secondary objectives included evaluation of potential confounders that affect this association and the predictive value for simultaneous neck dissection. Eighty-eight consecutive patients referred to the study institution between January 2007 and August 2012 with the presumptive diagnosis of invasive squamous cell carcinoma (SCC) of the tongue were analyzed. Seventy-nine patients had preoperative US and 81 had MRI. Correlation between image-determined TT and histological TT was assessed by Bland–Altman plot and Pearson’s correlation coefficient. Potential confounders were assessed by subgroup analysis. Preoperative TT as determined by US demonstrated high correlation and MRI moderate correlation with histological TT. With subgroup analysis, negative associations were biopsy prior to imaging and resection diagnosis other than invasive SCC. Our experience suggests that US could be considered the initial modality of choice for preoperative assessment of TT.
Oral and oropharyngeal squamous cell carcinoma (SCC) constitutes the sixth most common cancer worldwide, with the oral tongue being the most common oral subsite. Patients with oral tongue SCC and clinically/radiologically involved lymph nodes (N+) at diagnosis will generally receive surgical excision of the primary tumour and therapeutic neck dissection (National Comprehensive Cancer Network (NCCN) guidelines ). For patients with early-stage disease and without nodal metastases at diagnosis (T1/2, clinical/radiological N0), there is an approximately 25% risk of loco-regional recurrence. Immediate prophylactic neck dissection for the N0 neck may reduce regional recurrence at the expense of some surgical morbidity. Furthermore, neck dissection may identify other key prognostic factors – perineural and lymphovascular invasion, occult nodal metastases, and extracapsular spread. The identification of extracapsular spread allows the selective use of highly toxic adjuvant chemotherapy with radiotherapy, which has been proven to improve outcomes.
Current imaging is unreliable for direct detection of nodal disease, particularly micrometastasis. This has led previous authors to examine epidemiological, clinical, and histopathological variables to determine if there are parameters that can be identified preoperatively to increase the detection of subclinical disease. Histological primary tumour thickness (TT) is an important independent predictor of occult nodal disease in clinically/radiologically N0 necks. Current consensus is that immediate prophylactic neck dissection be used for tumours >4 mm in depth. Historically, this has been undertaken either at a second surgery after histological depth has been established or immediately based on clinical assessment. Accurate preoperative imaging assessment would allow treatment of the primary tumour and prophylactic neck dissection to be undertaken at a single surgery. A shorter total treatment time has been shown to improve cancer-specific outcomes for head and neck SCC. Patients are generally referred for radiological assessment after a confident diagnosis, usually histological, has been made.
Our study of patients with biopsy-proven or suspected invasive SCC of the oral tongue aimed to compare preoperative TT on high-resolution intraoral ultrasound (US) and magnetic resonance imaging (MRI) with the currently accepted standard of histologically determined TT. Secondary objectives of the study were to assess for factors that impact on the accuracy of preoperative imaging, as well as to assess whether TT measured on the preoperative imaging could be used to guide the need for neck dissection. It is hoped that the results from this study will be useful in generating guidelines for the preoperative imaging of patients with tongue tumours.
Materials and methods
Human research ethics committee approval was obtained for the study. During the period January 2007 to August 2012, patients with clinically suspected or biopsy-proven oral tongue SCC treated by the head and neck oncology tumour stream at the study hospital were prospectively recorded into a database. Standard preoperative workup for all oral tongue tumours included high-resolution intraoral US, MRI of the oral cavity, computed tomography (CT) of the neck/chest, and orthopantomogram (OPG) unless precluded by patient or technical factors.
Patients were staged according to the sixth edition of the American Joint Committee on Cancer system (AJCC 2010).
All patients were discussed at the multidisciplinary head and neck cancer meeting and treatment decisions were based on clinical and imaging findings. Of the 88 patients, 84 had resection of the primary lesion and 59 of these also had a neck dissection. Four patients had non-surgical management.
High-resolution intraoral US images were obtained using a 15–7 MHz L15-7io linear ultrasound transducer ‘hockey stick probe’ on a Philips iU22 machine (Philips Medical, Netherlands) with a 23-mm rectangular field of view and single near-field focal zone. All examinations were performed or supervised by one of two radiologists (PMP, AY). The probe was placed in a sterile cover and sterile gel was applied to the surface. The patient was instructed to protrude the tongue, which was gently held with gauze ( Fig. 1 ). Sterile gel was placed over the lesion to act as ‘stand-off’; the probe was then gently placed on the lesion so as not to distort the mass and moved along the tumour surface until the deepest hypoechoic region, correlating to the greatest TT, was identified. This image was frozen on the screen. Callipers were used to measure the TT from the surface of the tumour to the deepest point of abnormal echogenicity to the nearest 0.1 mm ( Fig. 2 a ). The maximal TT was indicated in the radiological report.
Magnetic resonance imaging
MRI was performed in patients without contraindication on either a Siemens 3.0 T Trio MRI (Siemens Healthcare, Erlangen, Germany) or a General Electric 1.5 T HDx (GE, Milwaukee, USA) with a head and neck coil. The scanning protocol included T2 coronal with fat suppression (FS) (repetition time (TR) 6120 ms, echo time (TE) 167 ms), T2 axial with FS (TR 4760 ms, TE 134 ms), T1 coronal post gadopentetate dimeglumine (Magnevist; Schering, Germany; dose 0.2 mmol/kg) with FS (TR 455 ms, TE 9.4 ms), and axial post-contrast (Magnevist) T1 with FS (TR 766 ms, TE 9.4 ms). The field of view for all sequences was 180 mm, covering from the skull base to the hyoid bone, and the section thickness was 3 mm with 0.3-mm intersection spacing. Due to patient factors (renal impairment in three patients), patient contraindication, or technically suboptimal studies, not all sequences were obtained or were diagnostic for each patient.
One radiologist (PMP), an experienced head and neck radiologist, reviewed de-identified MRI images of all patients. The TT was measured separately on the T2 and T1 post-contrast sequences to assess for any differences between the sequences. The TT was measured as the depth of abnormal signal from the surface of the lesion (for flat, ulcerated, and exophytic lesions) on either coronal or axial sequences determined by the plane that the lesion was most conspicuous on ( Fig. 2 b). No attempt was made to reconstruct the tumour – normal mucosa junction with subsequent measurement of the ‘depth of invasion’, as advocated by some authors. Maximal TT on MRI was determined to 0.1 mm and recorded into a database.
The surgical specimen was placed in formalin for overnight fixation. The specimen was then cut along the long axis into 2–3-mm thick transverse slices and laid out sequentially for examination ( Fig. 3 ). On histological sections, the maximum TT from the surface of the tumour (flat or exophytic) to the deepest point of invasion was determined. If the lesion was ulcerated, the TT was measured from the ulcer base to the deepest point of invasion. Both measurements were determined to an accuracy of 0.1 mm by an ocular micrometre. For patients who underwent a neck dissection, all lymph nodes were examined microscopically for tumour involvement. Tumour size (in millimetres) was recorded in the pathology report, as was nodal involvement for those patients who underwent neck dissection.