Oral cancer staging established by magnetic resonance imaging

The aim of this study was to compare clinical staging and magnetic resonance imaging (MRI) staging for oral cancer, and to assess inter-observer agreement between oral and medical radiologists. A total of 10 patients diagnosed with oral cancer were assessed before treatment. A head and neck surgeon performed clinical TNM staging. Two medical radiologists and two oral radiologists performed a new staging assessment by interpreting MRI scans, without prior knowledge of the clinical staging. They evaluated the extent of the primary tumor (T), metastasis to regional lymph nodes (N) and grouping by stages. The data were analyzed using the Kappa Index. There was significant agreement (p < 0.05) between the clinical and MRI staging assessments made by one oral radiologist for N stage, and between those made by one medical radiologist for the T and N stages and for the grouping by stages. In the MRI assessment, there was significant agreement among all four observers for both T stage and grouping by stages. For the N stage, there was no significant agreement between one oral radiologist and one medical radiologist or between both medical radiologists. There was significant agreement among the remaining radiologists. There was no agreement between the clinical and MRI staging. These results indicate the importance of using MRI for the diagnosis of oral cancer. Training initiatives and calibration of medical and oral radiologists should be promoted to provide an improved multidisciplinary approach to oral cancer. Descriptors: Magnetic Resonance Imaging; Mouth Neoplasms; Head and Neck Neoplasms. Introduction The prognosis of carcinoma of the maxillofacial region is influenced by a variety of factors, such as the degree of cellular differentiation, size, location, presence of infiltration into the bone tissue, immune response, age, gender, patient’s socio-economic status and the presence of cervical lymph node metastasis, the latter being considered the most significant factor when determining the prognosis.1-3 Incidence and mortality rates vary from one country to another and even within countries, because of differences in customs, especially tobacco use and alcohol consumption, environmental factors and the quality of medical care.4 Oral cancer is diagnosed after clinical examination, biopsy and anatomic pathology examination of the lesion have been carried out. After the diagnosis has been established, an assessment is then needed of the extent and spread Declaration of Interests: The authors certify that they have no commercial or associative interest that represents a conflict of interest in connection with the manuscript. Paiva RR, Figueiredo PTS, Leite AF, Silva MAG, Guerra ENS 513 Braz Oral Res. 2011 Nov-Dec;25(6):512-8 of the disease. Staging, which can be defined as the quantification of the clinical parameters of the disease, helps in making therapeutic decisions and in establishing a prognosis for the patient.1-3,5,6 The TNM system classifies the anatomical extent of the disease in any part of the body, by using clinical observation and histological and surgical complementation, or diagnostic imaging methods. The choice of appropriate treatment for patients with oral cancer depends largely on accurate pretreatment staging and, above all, on the detection of cervical lymph node involvement.2,3,7 In cases of clinically negative necks (N0), the clinical examination may present up to a 40% failure rate in detecting lymph node metastases.2,3,8,9 A combination of clinical and imaging examinations is essential for detecting metastatic lymph nodes and establishing the prognosis.10,11 Of the imaging modalities, computed tomography (CT) and magnetic resonance imaging (MRI) seem to be the most appropriate for the pre-therapeutic staging of head and neck tumors, because they provide information on the extent of the lesion, infiltration of large vessels and metastases in lymph nodes.3,12-14 The major advantage of MRI is that it provides excellent soft tissue detail visualization and does not involve any biological risks for the patient. Research that can facilitate or provide further information on staging means that patients will be adequately treated and consequently have a greater chance of being cured. The aim of this study was to compare the staging (TNM classification) established by clinical and MRI examinations for oral cancer, and to assess inter-observer agreement between medical and oral radiologists when analyzing MRI scans. Methodology Sample A total of 10 patients seen at the Oral Cancer Center at the Division of Dentistry in the University Hospital of Brasília (UnB), from October 2005 to December 2008, with a histologically proven diagnosis of oral squamous cell carcinoma were examined and submitted to MRI before treatment. This study was approved by the Ethics Committee at UnB (no. 025/2007) and informed consent was obtained from all patients. Table 1 presents the clinical characteristics of the patients. After clinical examination and biopsies, the patients were referred to a head and neck surgeon to establish their clinical TNM stage and undergo treatment. They concomitantly underwent the CT and MRI examinations. The CT data were published in a previous study.15 Four observers interpreted the MRI scans. Observers 1 and 2 were dental specialists holding a Master’s in oral radiology, and Observers 3 and 4 were medical radiologists. These four radiologists established staging based on the MRI scans without any prior knowledge of the clinical staging already established by the head and Case Gender Age Tobacco use Alcohol consumption Primary site – Side Clinical TNM Grouping 1 Female 50 No No Gum – R T2N0M0 II 2 Male 72 Yes Yes Base of tongue – L T3N0M0 III 3 Male 55 Yes Yes Retromolar region – L T2N1M0 III 4 Male 51 Yes Yes Floor of mouth – R/L T4N2M0 IVA 5 Male 41 Yes Yes Tongue – R/L T4N2M0 IVA 6 Male 69 Yes Yes Floor of mouth – R/L T4N0M0 IVA 7 Female 48 Yes Yes Side of tongue – R T1N0M0 I 8 Female 50 Yes No Base of tongue – R T1N0M0 I 9 Male 62 Yes Yes Side of tongue – R T1N0M0 I 10 Male 54 Yes Yes Floor of mouth – R T2N2M0 IVA R = right; L = left. Table 1 Demographic and clinical data of patients in the study sample. Oral cancer staging established by magnetic resonance imaging 514 Braz Oral Res. 2011 Nov-Dec;25(6):512-8 neck surgeon. Three parameters were evaluated, namely, the extent of the primary tumor (T), the presence/absence and extent of metastasis in regional lymph nodes (N) and the grouping by stages. The criteria established by Prehn et al.16 were used to ascertain which cervical lymph nodes were affected when cervical metastasis occurred. Magnetic resonance imaging and interpretation MRI was performed with a Signa Excite 1.5 high-field device (1.5 Tesla) (General Electric Healthcare Inc., Milwaukee, USA), following the protocol of the institution. Images of the head and neck region were taken to assess cervical lymph node involvement. The examinations included T1 (TR/ TE, 350/13.1 ms; FOV 24 × 24 mm; slice width/gap, 3,5/1 mm; slice number, 30), T1 with contrast (TR/ TE, 300/4.8 ms; FOV 24 × 24 mm; slice width/gap, 3,5/1 mm; slice number, 30) and T2-weighted (TR/ TE, 4600/99.4 ms; FOV 26 × 26 mm; slice width/ gap, 5/1 mm; slice number, 30) sequences on three anatomical planes (axial, coronal and sagittal). A gadolinium-based contrast agent (Gd/DTPA Diethylene Triamine Pentaacetic Acid) was used. The MRI scans were interpreted on a Toshiba Satellite A65 laptop computer (Toshiba America Information Systems, Inc., Irvine, USA), with a 14inch screen. Printed films were not used. The eFilm 2.0 program (Merge Healthcare Inc., Chicago, USA), which provides a DICOM (Digital Imaging Communication in Medicine) reading, was used to visualize and analyze the images. The MR images were considered the standard parameter in this study. Statistical analysis A descriptive analysis was performed using mean, median, standard deviation, and both maximum and minimum values. The SPSS for Windows program, version 13.0 (SPSS Inc., Chicago, USA) was used for all the statistical tests. Inter-observer agreement as regards MRI staging, T stage, N stage and grouping by stages was analyzed using Cohen’s kappa index. A statistical significance level of 95% (p value < 0.05) and a 5% error level were considered for the analyses. Interpretation criteria of the kappa index recommended by Landis and Koch17 were used to analyze the results. Results Table 2 shows that agreement for the T stage was excellent (k = 0.85) between Observers 1 and 2 (oral radiologists) and moderate (k = 0.47) between Observers 3 and 4 (medical radiologists). All four observers presented different results for the T stage using the clinical and MRI examinations. The highest rate of agreement (k = 0.46 = moderate) was presented by Observer 4. Table 3 shows that agreement for the N stage was substantial (k = 0.69) between Observers 1 and 2 (oral radiologists) and considerable (k = 0.38) between Observers 3 and 4 (medical radiologists). All four observers presented different results for the N Stage using the clinical and MRI examinations. Table 4 shows that the agreement for grouping by Obs. 1 Obs. 2 Obs. 3 Obs. 4 Clinical T k = 0.29 (p = 0.116) k = 0.31 (p = 0.100) k = 0.17 (p = 0.366) k = 0.46 (p = 0.012)* Obs. 1 K= 0.85 (p < 0.000)* k = 0.42 (p = 0.024)* k = 0.58 (p = 0.001)* Obs. 2 k = 0.57 (p = 0.003)* k = 0.73 (p < 0.000)* Obs. 3 k = 0.47 (p = 0.006)* k = kappa value; *p < 0.05 = statistically significant agreement; Obs. = observer; 1 and 2 = oral radiologists; 3 and 4 = medical radiologists. Table 2 Inter-observer and clinical and MRI examination agreement for T Stage. Paiva RR, Figueiredo PTS, Leite AF, Silva MAG, Guerra ENS 515 Braz Oral Res. 2011 Nov-Dec;25(6):512-8 stages was substantial (k = 0.72) between Observers 1 and 2 (oral radiologists) and moderate (k = 0.44) between Observers 3 and 4 (medical radiologists). All four observers presented


Introduction
2][3] Incidence and mortality rates vary from one country to another and even within countries, because of differences in customs, especially tobacco use and alcohol consumption, environmental factors and the quality of medical care. 4Oral cancer is diagnosed after clinical examination, biopsy and anatomic pathology examination of the lesion have been carried out.After the diagnosis has been established, an assessment is then needed of the extent and spread of the disease.]5,6 The TNM system classifies the anatomical extent of the disease in any part of the body, by using clinical observation and histological and surgical complementation, or diagnostic imaging methods.
The choice of appropriate treatment for patients with oral cancer depends largely on accurate pretreatment staging and, above all, on the detection of cervical lymph node involvement. 2,3,7In cases of clinically negative necks (N0), the clinical examination may present up to a 40% failure rate in detecting lymph node metastases. 2,3,8,9A combination of clinical and imaging examinations is essential for detecting metastatic lymph nodes and establishing the prognosis. 10,11][14] The major advantage of MRI is that it provides excellent soft tissue detail visualization and does not involve any biological risks for the patient.Research that can facilitate or provide further information on staging means that patients will be adequately treated and consequently have a greater chance of being cured.The aim of this study was to compare the staging (TNM classification) established by clinical and MRI examinations for oral cancer, and to assess inter-observer agreement between medical and oral radiologists when analyzing MRI scans.

Methodology Sample
A total of 10 patients seen at the Oral Cancer Center at the Division of Dentistry in the University Hospital of Brasília (UnB), from October 2005 to December 2008, with a histologically proven diagnosis of oral squamous cell carcinoma were examined and submitted to MRI before treatment.This study was approved by the Ethics Committee at UnB (no.025/2007) and informed consent was obtained from all patients.Table 1 presents the clinical characteristics of the patients.
After clinical examination and biopsies, the patients were referred to a head and neck surgeon to establish their clinical TNM stage and undergo treatment.They concomitantly underwent the CT and MRI examinations.The CT data were published in a previous study. 15  neck surgeon.Three parameters were evaluated, namely, the extent of the primary tumor (T), the presence/absence and extent of metastasis in regional lymph nodes (N) and the grouping by stages.The criteria established by Prehn et al. 16 were used to ascertain which cervical lymph nodes were affected when cervical metastasis occurred.
The MRI scans were interpreted on a Toshiba Satellite A65 laptop computer (Toshiba America Information Systems, Inc., Irvine, USA), with a 14inch screen.Printed films were not used.The eFilm 2.0 program (Merge Healthcare Inc., Chicago, USA), which provides a DICOM (Digital Imaging Communication in Medicine) reading, was used to visualize and analyze the images.The MR im-ages were considered the standard parameter in this study.

Statistical analysis
A descriptive analysis was performed using mean, median, standard deviation, and both maximum and minimum values.The SPSS for Windows program, version 13.0 Inc., Chicago, USA) was used for all the statistical tests.Inter-observer agreement as regards MRI staging, T stage, N stage and grouping by stages was analyzed using Cohen's kappa index.A statistical significance level of 95% (p value < 0.05) and a 5% error level were considered for the analyses.Interpretation criteria of the kappa index recommended by Landis and Koch 17 were used to analyze the results.

Results
Table 2 shows that agreement for the T stage was excellent (k = 0.85) between Observers 1 and 2 (oral radiologists) and moderate (k = 0.47) between Observers 3 and 4 (medical radiologists).
All four observers presented different results for the T stage using the clinical and MRI examinations.The highest rate of agreement (k = 0.46 = moderate) was presented by Observer 4.
Table 3 shows that agreement for the N stage was substantial (k = 0.69) between Observers 1 and 2 (oral radiologists) and considerable (k = 0.38) between Observers 3 and 4 (medical radiologists).
All four observers presented different results for the N Stage using the clinical and MRI examinations.
Table 4 shows that the agreement for grouping by  stages was substantial (k = 0.72) between Observers 1 and 2 (oral radiologists) and moderate (k = 0.44) between Observers 3 and 4 (medical radiologists).
All four observers presented different results for the grouping by stages using the clinical and MRI examinations.The highest rate of agreement (k = 0.44 = moderate) was presented by Observer 4.
Table 5 shows that there was significant agreement (p < 0.05) between the clinical and MRI staging assessments made by one oral radiologist (Observer 2) for N stage, and significant agreement between those made by one medical radiologist (Observer 4) for T and N stages and for grouping by stages.
As to MRI staging, there was significant agreement (p < 0.05) among all four observers for T stage and for grouping by stages, and there was no significant agreement between Observers 1 and 4, or between 3 and 4, for N stage.In all other comparisons, there was significant agreement (p < 0.05) for N stage (Table 6).

Discussion
The results of the present study contribute to recognizing the importance of MRI examination in establishing the staging of oral cancer.
Carcinoma lesions in the oral cavity are very aggressive and usually infiltrate the surrounding tissue and lymph vessels, producing metastasis in the cervical region. 2,1814]19 In the preoperative treatment of oral cancer, it has been shown that MRI is better for evaluating soft tissue, bone marrow involvement and perineural invasion,    and has been particularly decisive in the diagnosis of small lesions. 7,20,21he discrepancies found in the comparisons between the staging established by clinical and MRI examinations demonstrate the importance of this study.A clinical staging assessment lower than the real staging could result in ineffective treatment and/ or increase the possibility of recurrence in a particular case, whereas a higher clinical staging assessment could lead to more radical treatment, thereby increasing treatment aftereffects.The related literature states that appropriate staging of a lesion is essential for decision-making during surgical and/or radiotherapy planning, for predicting the prognosis and for deciding how to carry out patient follow-up to guarantee greater life expectancy and cure. 1,2,7ifferent results were found between the clinical and MRI examinations for T staging.If a clinical staging assessment establishes a primary tumor as smaller than it really is, this could result in ineffective surgical margins and in an incomplete removal of the lesion. 7Figures 1 through 3 show case number 8 clinically staged as T1, in which the MRI shows the primary tumor measuring 3.4 cm (T2).Surgical and anatomic pathology confirmation would be necessary to determine the actual size of the primary tumor.
Agreement among all four observers was sig-nificant for MRI T staging.This agreement shows that MR images provide greater interpretation standardization.Calibration of the observers may have been decisive in achieving the agreement levels and should be used in joint training programs that prepare medical and oral radiologists to diagnose oral cancer at reference centers providing multidisciplinary care. 15here was substantial and significant agreement between the clinical and MRI staging performed by Observers 2 and 4 for N stage.Nevertheless, case number 3, clinically staged as N1, was staged as N0 by two observers (1 and 3) in the MRI examination.A higher clinical staging assessment, establishing a false-positive for regional metastasis, may lead to more radical treatment and increase morbidity.According to Malard et al. 10 and Scully and Bagan, 11 a combination of both clinical and imaging examinations is essential for the detection of metastatic cervical lymph nodes, and could improve staging and prognosis determination.
Agreement between Observers 1 and 2, 1 and 3, 2 and 3, and 2 and 4 in the MRI staging was significant, indicating that the pre-established criteria and image interpretation guide could also have been crucial for the level of agreement achieved for the N stage.
Different results were observed for the clinical   and MRI staging, and agreement was significant only for Observer 4.Among these results, three cases were staged as belonging to the IVA grouping, which represents lesions which are larger and at a more advanced stage, thereby facilitating diagnosis. 15This confirms the greater importance of the MRI exami-nation for smaller lesions.The definition of grouping by stages is critical in determining the patient's treatment plan, prognosis and survival span.
Agreement among all four observers for stage grouping using MRI was significant, and greater between the oral radiologists.Based on these results and on literature data, it could be stated that clinical examination, anatomic pathology testing and diagnostic imaging modalities (which could include CT and MRI) are necessary to establish the staging of patients with oral cancer.According to Weber et al. 13 and Scully and Bagan, 22 CT and MRI examinations seem to be the most important diagnostic tools when establishing pre-therapeutic staging of head and neck tumors.CT is essential, insofar as it provides a better evaluation of cervical lymph node involvement and invasion of bone cortices adjacent to the primary tumor area.The MRI examination provides a better evaluation of the soft tissues affected by the lesion and allows a more thorough evaluation of small tumors.This examination should be of the treatment protocol of patients with oral cancer, depending on its availability, accessibility and the possibility of carrying out the examination.According to Warnakulasuriya, 4 improvement in the quality of healthcare helps to reduce mortality rates.While the results would indicate the importance of using MRI in the diagnosis of oral cancer, there is also a very obvious need for a combination of research and surgical and pathological information to identify sources of error in pretreatment staging.Joint training initiatives and calibration of medical and oral radiologists should be promoted to provide an improved multidisciplinary approach to oral cancer.

Conclusion
There was no agreement between the staging established by clinical and MRI examinations for oral cancer.MRI examination is useful to provide a better assessment of TNM staging, and the examinations should be analyzed by different professionals (physicians and dentists) in a multidisciplinary approach.

Figure 2 -
Figure 2 -MRI axial slice, T 1 -weighted, with contrast, showing a tumor of hyper-signal intensity (enhanced with contrast) at the base of the tongue (arrowhead), on the right side.

Figure 3 -
Figure 3 -MRI coronal slice, T 2 -weighted, showing a tumor of isosignal intensity at the base of the tongue (arrowhead), on the right side.

Figure 1 -
Figure 1 -MRI axial slice, T 1 -weighted, without contrast, showing a tumor of isosignal intensity at the base of the tongue (arrowhead), on the right side, measuring approximately 3.4 cm.

Table 1 -
Demographic and clinical data of patients in the studysample.

Table 2 -
Inter-observer and clinical and MRI examination agreement for T Stage.

Table 3 -
Inter-observer and clinical and MRI examination agreement for N Stage.

Table 4 -
Inter-observer and clinical and MRI examination agreement for Group Staging.

Table 6 -
Inter-observer agreement for MRI staging.

Table 5 -
Agreement between clinical and MRI staging assessments for all four observers.