Abstract
Background: Reliable diagnosis plays an important role in the early detection of head and neck tumors. The objective of this study was to compare the value and clinical relevance of magnetic resonance imaging (MRI) and positron emission tomography (PET) regarding the assessment and the identification of head and neck tumors. Patients and Methods: From January 1, 2005, to January 1, 2007, 120 patients suffering from carcinomas of the oropharynx and larynx were examined by means of MRI and PET. Results: The difference between sensitivity and specificity of MRI and PET was not significant with regard to the diagnosis of primary tumors, recurrence, or cancer of unknown primary. A statistically significant difference between both methods only occurred for detectron of malignant lymph nodes of size >10 mm. Conclusion: In particular, the assessment of small tumors by MRI and PET is characterized by a high number of false-negative findings. The future of diagnostic imaging is likely to be a combination of both techniques, as a hybrid technique.
- PET
- MRI
- head and neck cancer
Modern diagnostic methods for head and neck tumors should allow for complete tumor assessment and the determination of tumor identity. The routinely performed imaging procedures such as magnetic resonance imaging (MRI), computed tomograhy (CT), positron-emission tomography (PET), and ultrasound, however, only partly meet these requirements.
The objective of the present study was to assess the value of morphological MRI compared to functional PET with regard to the detection and determination of head and neck tumors and to compare the results to those reported in the literature. Thus, the reliability and clinical relevance of both methods and their combination was evaluated in patients suffering from head and neck tumors.
Patients and Methods
The data of 120 patients presenting at the Department of Otorhinolaryngology in Frankfurt/ Oder, Germany, from January 1, 2005, to January 1, 2007, with suspected carcinoma or confirmed carcinoma of the oropharynx and larynx were retrospectively evaluated (staging or re-staging). Furthermore 46 patients who suffered from carcinomas of other locations in the head and neck area were additionally assessed.
Each patient, hospitalized for staging or re-staging, underwent diagnostic imaging, as well as endoscopy of the upper aerodigestive tract, under general anesthesia. As standard imaging methods, MRI of the head and neck, radiography or CT scan of the thorax, abdominal ultrasound, and PET of the whole body were performed. The choice between conventional radiography or CT of the thorax depended on the size of the tumor. In cases of T1 or T2 tumors, radiography of the thorax was performed; in cases of T3 or T4 carcinomas, CT scan of the thorax was performed. If necessary, additional endoscopic examinations (flexible bronchoscopy, gastroscopy) and additional imaging procedures such as scintigraphy, angiography, barium swallow examination, and video cinematography were performed. The imaging procedures were always performed before obtaining histological specimens, in order to avoid falsification of the results after endoscopy, excisional biopsy, or surgical exploration.
MRI and PET findings were assessed separately and recorded as “positive” or “negative”, and then correlated with the histological result. Depending on the histological results, the results of the imaging procedures were evaluated as “true-positive”, “false-negative”, “true-negative” and “false-negative”.
Lymph nodes were extirpated in 38 out of 120 patients by means of neck dissection. Neck dissection was performed unilaterally in 17 patients and bilaterally in 21 patients. Assessing the lymph nodes by MRI, the lymph nodes were classified into three groups: ≤10 mm, >10 mm and <15 mm, and >15 mm. In this context, two different size limits of the lymph nodes were analyzed as possible criteria of malignancy: >10 mm and >15 mm (1). By correlating these changes with the histological findings, the variations of the statistical parameters and thus, also the real changes of the value of MRI could be assessed with regard to these limits.
For MRI, specific criteria were defined, and based on these criteria, the findings were evaluated as positive (malignant) or negative (benign). All tumoral masses that were inhomogenously structured, poorly-differentiated and that revealed infiltrating growth with destruction of neighbouring structures were assessed as there positive. In addition, all changes that showed inhomogenous or enhanced uptake of contrast agents were also evaluated as positive. For the lymph nodes, inhomogeneity, poor differentiation, and a size of more than 10 mm was considered suspicious (1-3).
An Ecat Exact 922 Gantry PETscanner (Siemens, Germany) was used. Before examination, all patients fasted for at least 12 hours. Regarding diabetics, the precondition of the PET examination was euglycaemia. Each patient underwent imaging from the head down to the middle of the thigh. The radiopharmaceutical agent 18F-(FDG) was applied intravenously, usually at a dose of 5 MBq/kg body weight.
The standardized uptake value (SUV) represents a relative value and refers to the mean activity value of the tracer contained in the tissue volume of the region of interest with regard to the injected active quantity and the body weight of the patient. It is relatively higher when the tissue activity is increased (4, 5). The SUV limit for the indication of a malignant process was defined as 2.5. All findings equal to 2.5 were evaluated as negative, all findings higher than 2.5 were considered as positive (4, 5).
Results
From January 1, 2005, to January 1, 2007, a total of 120 patients with suspected carcinoma or already confirmed tumor of the head and neck were retrospectively evaluated. Ninety-five of the patients were male (79%), 25 were female (21%) (ratio: 3.8:1). The age of the examined patients ranged from 41 to 85 years. In the majority of the patients (69 patients) imaging and endoscopy was performed for diagnosis of a primary tumor. The group of patients undergoing diagnostic methods for recurrences amounted to 44 patients. Further 7 patients were examined for CUP.
In 62/69 patients, the presence of a carcinoma was confirmed in the context of primary tumor diagnostics. In seven patients, a malignant process was excluded. In those patients, the suspecious lesions were identified as chronic inflammation or dysplasia. In three out of 69 patients, a secondary carcinoma was suspected at the time of diagnosis. As though the presence of secondary malignancy was not confirmed in these three cases, they were not excluded from the evaluation. Therefore, 72 cases were evaluated from the group of patients undergoing diagnostic imaging for the detection of a primary tumor. The surgical extirpation of lymph nodes was performed in 24 patients in this group.
A total of 48/62 histologically-positive tumors and 8/10 histologically-negative findings in this group were correctly identified by means of PET (sensitivity of 77.4%; specificity of 80%). The carcinomas were evaluated as malignant tumors in 47/62 patients by means of MRI (sensitivity of 75.8%), the specificity of MRI was 50% (Table I).
In the context of recurrence diagnosis, 7/10 histologically-positive tumors and 26/34 histologically-negative findings were correctly identified by means of PET (sensitivity of 70%, specificity of 76.5%). MRI correctly revealed 8/10 carcinomas and 30/34 histologically-negative findings (sensitivity of 80%, specificity of 88.2%) (Table II). From seven patients with CUP, a primary tumor was only found in one case, which was a thyroid carcinoma. Because of the low number of patients in this group, this statement is only informative (Table III). In all, the difference between sensitivity and specificity of MRI and PET was not significant with regard to diagnosis of primary tumors, recurrence, or CUP.
For assessment of the lymph nodes by means of MRI, two different size limits of the lymph nodes were defined as criteria for diagnosing malignancy (>10 mm and >15 mm). MRI proved to a have sensitivity of 92% and specificity of 23% for a defined malignancy level of >10 mm. The change of the malignancy level from >10 mm to >15 mm led to a reduced sensitivity of 80% and an increased specificity of 82%. The sensitivity of PET was lower with 76% and the specificity was higher with 91.1% compared to MRI (Table IV).
Finally, the findings of MRI and PET diagnosis were combined In the primary tumor group after summation of the results, the sensitivity was 82.2% and the specificity 80%. The diagnosis of recurrences revealed a sensitivity of 90% and a specificity of 94.1%. Regarding the diagnosis of CUP, the combination of PET and MRI did not show any change of sensitivity; however, the specificity increased to 85.7%. The difference between the specificity of MRI and the specificity of the combined procedures was not statistically significant with a p value of 0.5.
For the whole patient population, by means of PET, 80/103 malignant lesions were correctly identified as positive and 71/85 of the non-malignant lesions as negative (sensitivity of 77.6%, specificity of 83.5%). Under the same conditions, the results of MRI depended on the examined lymph node size. With a malignancy level defined as >10 mm, 84 findings were correctly identified as positive and 47 as negative by MRI (sensitivity of 81.5%, specificity of 55.2%). At a malignancy level of >15 mm, 81 findings were identified as true-positive and 67 as true-negative (sensitivity of 78.6%, specificity of 78.8%). A statistically significant difference between these methods only occurred for a lymph node malignancy level of >10 mm (p-value of 0.001).
Discussion
The therapy of head and neck carcinomas is an interdisciplinary challenge. Despite modern surgical and diagnostic techniques, the survival rate of patients suffering from these malignancies has not significantly improved (6-8).
MRI as a morphological procedure allows for identification of a tumor due to its morphological and anatomical characteristics. However, this type of disease presents some problems with regard to identification. PET as functional procedure allows for the assessment and a possible definition of the identity of neoplasms based on their functional characteristics. In the context of primary diagnosis, the results of PET were better than the results of MRI. The superiority of PET was shown regarding the sensitivity as well as the specificity. However, the difference was not statistically significant. These findings correspond to data in the literature (9-17). Only in few published studies (14), was MRI superior to PET.
In the literature, the sensitivity rates of the morphological procedures are 50-95%, those of PET 70%-100% (18-23). However, mainly in the assessment of small tumors, both MRI and PET are characterized by a high number of false-negative findings.
Regarding the diagnosis of recurrences, numerous studies showed the superiority of PET in comparison to MRI. The sensitivity of PET is often higher than 80%, while the sensitivity of MRI is 55-80% (11, 18, 20-22). The presented patient population with recurrences revealed differences in comparison to published studies. MRI achieved a sensitivity of 80.0% and a specificity of 88.2% and was thus superior to PET. The difference between both examination methods, however, was not statistically significant.
In 6/44 cases with suspected tumor recurrences the MRI led to false results since it could differentiate between post-operative or post-radiogenous and tumorous lesions. PET led to a total of 11 false findings. In three false-negative cases, either no or a too low enhancement of the radiopharmaceutical agent was detected. In eight false-positive cases, mainly inflammatory processes were histologically identified. Those cases certainly had an influence on the increase of the SUV. The SUV of those findings were between 3.0 and 4.4. Many publications describe malignant processes with higher SUV and benign lesions with lower SUV (3). The increase of the SUV as a malignancy criterion bears the risk of increasing the specificity and at the same time reducing that of sensitivity (5). However, there are no universal guidelines for the interpretation of FDG uptake (24).
In the present study, there were seven patients with CUP but a primary tumor was only found in one case. Because of the low number of patients in this group, no conclusions could be drawn. However, previous studies have shown that PET may be a valuable diagnostic tool in the detection of the primary tumor in patients with cervical lymph node metastases (25, 26).
The lymph node status plays an important role in the diagnosis of head and neck tumors (27, 28). In the literature, sensitivity rates of PET between 71% and 91% and specificity rates between 82% and 100% are described for the determination of the native of lymph nodes. The corresponding parameters for MRI vary between 67% and 90% and 40% and 90%, respectively (1, 23, 29-31). Generally, PET is superior for the assessment of small metastatic lymph nodes (4-6 mm) (9, 32, 33), however, it may lead to false-negative results in cases of necrotic lymph nodes that do not take up enough of the tracer (34-37).
The reduction of sensitivity leads to a lower number of true-positive findings. In our patient population, there were three lymph node findings that became negative (false negatives) because of the change of the malignancy limit. From a clinical point of view, this situation is adverse. The lymph nodes that were not identified and, in consequence, remained untreated may lead to disastrous results. The increase of the specificity, however, leads to more true-negative and thus fewer false-positive findings. The findings that were false-positives with a malignancy limit of >10 mm became true-negatives with a malignancy limit of >15 mm. In our patients, this situation occurred 20 times. This situation is favourable from a clinical point of view.
However, the presented lymph node parameters of the morphological diagnosis do not justify the assumption of these malignancy limits as a secure indication for the benignity of the lymph nodes. The sensitivity of 92% at a 10 mm limit also means that smaller lymph nodes (≤10 mm) may be malignant. According to the literature, metastatic lymph nodes are smaller than 6 mm in about 30% of the cases (23, 38-40).
Comprising of the findings of the diagnostic procedures, PET was characterized by a sensitivity of 77.6% and a specificity of 83.5%. The corresponding values for MRI were dependent on the defined malignancy criteria for the lymph nodes. With a malignancy level of >10 mm, MRI had a sensitivity of 81.5% and a specificity of 55.2%. With a malignancy level of >15 mm, the sensitivity was 78.6% and the specificity 78.8%. The above-mentioned findings generally demonstrated a better sensitivity of the morphological procedure and a better specificity of the functional procedure. Based on the results of the presented study, carcinomas seem to be better-assessed with MRI and better-excluded with PET.
In addition, the results of MRI and PET were combined in this study. Their combination in the context of primary diagnosis led to an improved sensitivity of 4.8% in comparison to PET-alone and of 6.4% in comparison to MRI-alone. The specificity improved by 30% in contrast to MRI-alone. In the diagnosis of recurrent tumors, there was an improvement of the sensitivity of 10% and of the specificity of 5.9% in comparison to MRI-alone, and of 20% and 17.6% compared to PET-alone. None of the cases showed a statistically significant difference between the single and combined procedures.
Conclusion
Correct staging or re-staging gives important information on the extent of the tumor, the lymph node status, and distant metastasis. The therapy based on such results has a significant impact on the prognosis. In summary, MRI and PET are characterized by a high sensitivity and specificity and represent important tools in current diagnosis of head and neck tumors. However, the data of the present study do not justify defining the tumor identity by imaging methods only. The results of imaging should be verified by histological examinations. The comparison of both procedures showed generally better results of PET for diagnosis of primary tumors and better results of MRI for diagnosis of recurrences (41-44). Sufficient diagnosis of small lymph node metastases less than 1 cm still remains a challenge. The present study shows that both techniques are not only competitive but are also complementary to each other for the identification of primary tumors. Therefore, in the future, hybrid techniques will be increasingly discussed.
- Received January 8, 2013.
- Revision received February 15, 2013.
- Accepted February 18, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved