Axial computed tomographic image with reference curve in a patient with invasive squamous cell carcinoma. Cross-sectional locations are numbered.
Reconstructed panoramic view.
Reconstructed mandibular cross sections. Numbers correspond with axial image in Figure 1.
Algorithm for management of patients with squamous cell carcinoma of the oral cavity suspicious for mandibular invasion.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Brockenbrough JM, Petruzzelli GJ, Lomasney L. DentaScan as an Accurate Method of Predicting Mandibular Invasion in Patients With Squamous Cell Carcinoma of the Oral Cavity. Arch Otolaryngol Head Neck Surg. 2003;129(1):113–117. doi:10.1001/archotol.129.1.113
Copyright 2003 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2003
To determine the diagnostic accuracy of a dental computed tomographic software program, DentaScan, in assessing mandibular bone invasion in patients with squamous cell carcinoma (SCC) of the oral cavity clinically fixed to the mandible.
Retrospective chart review.
Academic tertiary care medical center.
Thirty-six consecutive patients underwent DentaScan imaging of the mandible prior to composite resection for tumor ablation. All patients included in this review had SCC of the oral cavity clinically fixed to the mandible. A final histopathology report specifically addressing the presence or absence of bone invasion was available for each patient.
The diagnostic accuracy for DentaScan in this study was as follows: sensitivity, 95%; specificity, 79%; positive predictive value, 87%; and negative predictive value, 92%.
DentaScan is an accurate method of preoperative evaluation for mandibular invasion in patients with SCC of the oral cavity.
ACCORDING TO American Cancer Society statistics, there were 29 000 estimated new cases of oral cavity cancer diagnosed in 2002. Approximately 90% of these were squamous cell carcinoma (SCC). A total of 7400 deaths occurred as a result of oral cavity cancer in 2002.1 Therefore, this disease represents a significant problem for both patients and surgeons who care for them.
Given the anatomic constraints of the oral cavity, many of these tumors lie in close proximity to the mandible. Lesions encroaching upon the adherent gingiva of the mandible are often fixed to it, and therefore a dilemma exists: how much, if any, of the mandible will need to be resected to obtain oncologically clear margins. Surgeons of the Halstead era favored segmental mandibulectomy for oral cavity tumors. The basis for the "jaw-neck" or "commando" procedures was the long-held belief that oral cavity lymphatics passed through the mandible en route to the cervical lymphatics. Thus, removal of the mandible was considered oncologically necessary. In 1964, Marchetta et al2 disproved this notion. Through careful anatomic and histologic analyses of segmental mandibulectomy specimens, they demonstrated that oral cavity tumors invade the mandible by direct extension, not lymphatic channels. Therefore, segmental resection of bone is not always necessary in patients with oral cavity cancer, even in tumors seemingly close to the mandible on physical examination.3
Segmental resection of mandibular bone often presents significant reconstructive challenges. In addition to cosmetic considerations, mastication and deglutition can be significantly altered depending on the location and amount of bone taken. While microvascular free tissue transfer techniques significantly assist in rehabilitation, not all patients are candidates for these lengthy, complicated, and costly procedures.
Marginal mandibulectomy, in preserving mandibular continuity, is associated with less functional and cosmetic morbidity, and can provide satisfactory oncologic margins if tumor has not invaded through cortical bone. Preoperative knowledge of bone invasion is, therefore, critical in planning the appropriate procedure.
Determining mandibular invasion preoperatively, however, has proven difficult and controversy surrounding this subject has existed for decades. Several modalities including clinical examination, plain radiographs, nuclear medicine studies, magnetic resonance imaging (MRI), and computed tomography (CT) have all been studied. Each method has its advantages and disadvantages, but CT imaging has, in recent years, demonstrated some of the best results.
DentaScan, a dental computed tomographic software program, is an extension of CT technology. Developed in the 1980s to assist oromaxillofacial surgeons in planning for endosseus implantation, DentaScan reformats standard axial CT scans into 2 unique views: panelliptical and parasagittal (Figure 1, Figure 2, and Figure 3). Reformatting images allows for close inspection of buccal and lingual cortices, and in theory should improve specificity and sensitivity over standard CT imaging. While a few investigators have described its use for imaging patients with oral cavity SCC, there are no existing studies large enough to determine its diagnostic accuracy for this purpose. This article presents a series of 36 consecutive patients who underwent DentaScan imaging prior to either segmental or marginal mandibulectomy for ablation of oral SCC. Histopathologic evaluation describing the presence or absence of bone invasion was available for each patient. From this information the diagnostic accuracy of DentaScan was calculated.
Between October 1997 and October 2001, 85 composite resections were performed by a single surgeon (G.J.P.). Of those, 38 consecutive patients underwent DentaScan as part of their preoperative workup. Patients were selected for DentaScan if on physical examination the tumor was fixed to the mandible without obvious bone involvement. Thirty-six of these patients had available final histopathologic confirmation of the presence of SCC. The following information was retrospectively gathered for each patient: sex, age, procedure performed, DentaScan results, tumor histologic type, presence or absence of bone invasion, and final pathologic staging.
Scans were completed with axial orientation parallel to the mandibular occlusal plane. All scans were performed on 1 of 3 CT scanners (Highlight, LightSpeed, or CTI Systems; General Electric Medical Systems, Milwaukee, Wis). Axial images were 1 × 1-mm slice thickness and table increment with the exception of images obtained on the multislice scanner, where thickness and table increment were 1.25 mm. Images were obtained with bone algorithm only, without intravenous contrast.
Computed tomographic examinations were postprocessed from image data using the DentaScan software package (General Electric Medical Systems). A curve was designated from an axial image at the roots of the teeth, irrespective of the tumor location, with derivation of direct cross-sectional images at 2- or 3-mm increments and panellipse images at 2-mm increments. Hardcopy images were reviewed for evidence of cortical destruction or erosion, and for trabecular disruption from tumoral invasion.
Following composite resection (either segmental or marginal mandibulectomy) tumor specimens were prepared for histopathologic review in the following fashion. Mandibulectomy specimens were fixed in 10% formalin solution and allowed to decalcify for 2 days. Specimens were then sectioned in bread-loaf fashion and grossly examined for tumor invasion. Microscopic evaluation of the tumor-bone relationship was then performed to definitively determine the presence of tumor invasion into bone. Pathologists were unaware of preoperative imaging and intraoperative findings.
Diagnostic accuracy of DentaScan was then determined by comparing preoperative imaging findings with final histopathologic findings. Statistical analysis was performed by determining sensitivity, specificity, positive predictive value, and negative predictive value.
Of our group of 36 patients, 13 were women and 23 were men. Ages ranged from 41 to 84 years. Twenty-one patients underwent segmental mandibulectomy, and 15 underwent marginal mandibulectomy. All patients had final histopathologic confirmation of SCC. Data are summarized in Table 1.
DentaScan results were as follows: 12 studies were read as negative and 24 as positive. Final pathologic findings showed 14 specimens as negative for bone invasion and 22 as positive.
Statistical analysis revealed a sensitivity of 95%, specificity, 79%; positive predictive value, 87%; and negative predictive value, 92%.
The ideal test to determine mandibular invasion in patients with SCC of the oral cavity would be highly sensitive and specific, noninvasive, inexpensive, and widely available. While several different diagnostic tests have been advocated, each has its own particular set of advantages and disadvantages. The modalities that have been studied to date include clinical examination, Panorex radiographs, MRI, radionuclide scanning, CT, and now DentaScan.
Clinical examination has been shown by several authors to be an accurate method of preoperative mandibular assessment. Leipzig4 compared clinical examination with plain film and bone scan in a series of 31 patients and found clinical examination to be more accurate than the other 2 modalities. Shaha5 determined that clinical examination was more accurate than Panorex radiographs and CT scan. In his series, the overall accuracy of clinical examination was 88% compared with the other modalities, which had accuracies closer to 70%.
Clinical examination is best performed when the patient is under general anesthesia. Patients with oral cancer often have significant pain and trismus, making examination in the office difficult. Shaha5 did not state whether preoperative assessment took place in the office or under general anesthesia. We routinely examine these patients under general anesthesia at the time of definitive resection to confirm our preoperative examination and imaging findings. However, we believe that mandibular invasion is best determined prior to taking a patient to the operating room for definitive resection. If segmental resection of bone is necessary, reconstruction will likely be more complicated, particularly if large amounts of bone or anterior mandible are involved.
Magnetic resonance imaging has also been investigated as a tool to assess mandibular invasion. Ator et al,6 in a series of 11 patients with a variety of malignant and benign tumors, suggested that MRI may be superior to CT and other modalities for this purpose. They stress the superior resolution of tumor and soft tissue interface and ability to better evaluate the mandibular medullary space using MRI. Other authors have compared this modality with CT and found it to be less accurate. The most frequent problems stated are the lack of signal generated by bone on MRI, and high cost. In their multivariate analysis, Tsue et al7 found CT and physical examination to be the best predictors of mandibular invasion. They state that MRI was not as accurate as other modalities.
Panorex is the best plain film for determining mandibular involvement because of its view of the body, ramus, and condyle. Spine artifact, however, limits its usefulness for the parasymphysis and symphaseal regions. The major limitation of Panorex radiography is its low sensitivity (50%), because 30% to 50% demineralization must occur before change is visible to the naked eye.8 For this reason, it should only be used in conjunction with other more sensitive tests for determining bone involvement.
Bone scanning for detection of mandibular invasion has been plagued with low specificity. Introduced in the 1970s, this technique has been advocated as a useful adjunct study by several authors. Weisman and Kimmelman9 did a retrospective study of 40 patients who underwent bone scanning prior to mandibulectomy and found a false-positive rate of 53%. While bone scanning was able to detect mandibular invasion in 8 of 9 cases, they point out that this technique is sensitive to any osteogenic focus including periodontal disease, fractures, osteoradionecrosis, osteomyelitis, and neoplastic invasion. Single positron emission computed tomography has been recently used with slightly better results. Imola at al10 recently showed this diagnostic test to be 95% sensitive and 72% specific for mandibular invasion in patients with oral cancer. A 72% specificity is still relatively low compared with other available technologies, reflecting the inherent limitations of this sort of study.
Computed tomography has been used with varying results. In a series of 43 patients, Close et al11 determined its sensitivity to be 100% with a false-positive rate of 8.3%. These impressive results, however, have not been duplicated in more recent studies. Lane et al12 determined that CT was a useful but potentially inaccurate predictor of bone invasion in tumors of the retromolar trigone. In their study, bone invasion was missed in 27% of patients, but their positive predictive value of 91% suggested good specificity. Recent studies have shown better results. Mukherji et al,13 in a series of 49 patients, calculated the diagnostic accuracy of CT as follows: sensitivity, 96%; specificity, 87%; positive predictive value, 89%; and negative predictive value, 95%. They attribute their excellent results to superior imaging technique, using 3-mm-thick sections and reconstructing with both bone and soft tissue algorithms. Coronal images were not obtained in their study.
DentaScan has been proposed as a means to improve the diagnostic accuracy of standard CT. Talmi et al14 described using DentaScan in 17 patients with varying intraoral pathological conditions. Although the authors found it to be useful, they did not determine its diagnostic accuracy. In addition, their series consisted of a variety of tumor types. King et al15 analyzed a series of 26 patients who underwent DentaScan for various intraoral conditions but the diagnostic accuracy was not quantified in this study either. Similarly, Yanagisawa et al16 provided several case reports demonstrating the usefulness of DentaScan, but their series consisted of a variety of pathological conditions and no statistical analysis was provided.
To date, to our knowledge, ours is the first study to quantify the diagnostic accuracy of DentaScan in preoperative determination of mandibular invasion in patients with SCC of the oral cavity. DentaScan was able to predict mandibular invasion in 21 of 22 patients. The single patient with a false-negative finding had a primary lip cancer extending to the mandible. This patient received a marginal mandibulectomy, followed by postoperative radiation therapy; unfortunately, he had local recurrence. His final pathology report demonstrated bony involvement by tumor.
We had 3 false-positive results, for a specificity of 79%. Our specificity was relatively low partly because our patient population without bone invasion was small (14 of 36 patients). In all 3 cases, the radiologist was unable to determine mandibular invasion with certainty. Because the final radiology reports suggested bone invasion, they were listed as positive for the purposes of this study.
These cases illustrate the importance of correlating physical examination results with DentaScan results, especially when the radiologist's certainty is low. In 2 of these cases (patients 2 and 5; Table 1), our clinical examination did not support mandible invasion, and marginal mandibulectomy was performed. The final pathology report in both of these patients showed no bone invasion, and thus the correct operation was performed, despite incorrect DentaScan results. Although long-term follow-up for these patients was not available at the time of this study, both patients were free of disease at 3 months and 12 months postoperatively.
The third false-positive study occurred in a patient who presented with a recurrence following primary radiation therapy (patient 14; Table 1). This patient also underwent marginal mandibulectomy since the physical examination did not suggest bone invasion and the DentaScan was equivocal. Despite a final pathology report that showed no evidence of bone invasion, this patient had local recurrence 10 months after surgery.
One of the drawbacks of DentaScan is the difficulty in resolving the difference between cortical irregularities and true tumor invasion. In addition, highly curved areas such as the parasymphysis are slightly more difficult to evaluate using this technique. Despite its imperfections, however, DentaScan provides a detailed anatomic map of the mandible and is therefore useful in planning the extent of surgery, even when cortical erosion is equivocal.
The importance of preoperative identification of mandibular invasion cannot be overemphasized. Optimal management of these unfortunate patients requires much planning, and often reconstruction must be coordinated with a microvascular surgical team for segmental bony defects. The primary surgical goals in managing these patients are complete resection, preservation of physiologic function, and maximization of cosmesis. Achieving complete resection often comes at the expense of the latter two. If marginal mandibulectomy can be performed without sacrificing oncologically sound margins, the patient will most likely have a superior physiologic and functional outcome and require a less complicated reconstruction. A simple algorithm for management of these patients is shown in Figure 4.
This article does not compare DentaScan with other techniques, including standard CT. The image processing software costs approximately $16 000, and the additional cost of DentaScan to the patient is $650 at our institution. We believe this cost is justified by the accuracy of information provided and that preoperative planning is greatly enhanced by this technique. Future studies will be aimed at determining the additional benefits DentaScan provides over other techniques, such as standard axial CT.
Corresponding author and reprints: John M. Brockenbrough, MD, Loyola University Medical Center, Department of Otolaryngology–Head and Neck Surgery, 2160 S First Ave, Maywood, IL 60153 (e-mail: firstname.lastname@example.org).
Accepted for publication August 19, 2002.
This study was presented at the annual meeting of the American Head and Neck Society, Boca Raton, Fla, May 13, 2002.