Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2372041309
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by King, A. D.
Right arrow Articles by van Hasselt, A. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by King, A. D.
Right arrow Articles by van Hasselt, A. C.
(Radiology 2005;237:563-569.)
© RSNA, 2005

Head and Neck Imaging

Salivary Gland Tumors at in Vivo Proton MR Spectroscopy1

Ann D. King, FRCR, David K. W. Yeung, PhD, Anil T. Ahuja, FRCR, Gary M. K. Tse, FRCPC, H. Y. Yuen, FRCR, K. T. Wong, FRCR and Andrew C. van Hasselt, MMed (Otol)

1 From the Department of Diagnostic Radiology and Organ Imaging (A.D.K., D.K.W.Y., A.T.A., H.Y.Y., K.T.W.), Department of Anatomical and Cellular Pathology (G.M.K.T.), and Department of Surgery (A.C.v.H.), Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong SAR, China. Received July 28, 2004; revision requested October 6; revision received October 20; accepted December 10. Address correspondence to A.D.K. (e-mail: b834756{at}mailserv.cuhk.hk).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
PURPOSE: To prospectively evaluate whether proton magnetic resonance (MR) spectroscopy can be used to characterize salivary gland tumors (SGTs).

MATERIALS AND METHODS: Ethics committee approval and informed consent were obtained. Hydrogen 1 (1H) MR spectroscopy was performed with echo times of 136 and 272 msec at 1.5 T in both SGTs and normal parotid glands. Spectra were analyzed in the time domain by using prior knowledge in the fitting procedure to obtain peak amplitudes of choline (Cho), creatine (Cr), and unsuppressed water. Mean Cho/Cr and Cho/water ratios for each subgroup of SGTs were obtained, and results were compared by using a nonparametric t test.

RESULTS: Successful spectra were acquired in 56 patients (35 men, 21 women; mean age, 56 years) with a total of nine malignant tumors and 47 benign SGTs (24 Warthin tumors, 22 pleomorphic adenomas, one oncocytoma). At an echo time of 136 msec, Cho/Cr ratios were obtained in 26 (47%) of 55 spectra, with a mean value (± standard deviation) of 1.73 ± 0.47, 5.49 ± 1.86, 3.46 ± 0.84, and 2.45 for malignant tumors, Warthin tumors, pleomorphic adenomas, and oncocytoma, respectively. Differences were significant between Warthin tumors and pleomorphic adenomas (P = .028) and between benign SGTs and malignant tumors (P < .001). At an echo time of 272 msec, Cho/Cr ratios were obtained in 16 (30%) of 53 spectra, with a mean value of 2.27 ± 0.69, 6.92 ± 1.47, and 3.67 ± 1.23 for malignant tumors, Warthin tumors, and pleomorphic adenomas, respectively. Differences were also significant between Warthin tumors and pleomorphic adenomas (P = .041) and benign SGTs and malignant tumors (P = .004). There was a significant difference in mean Cho/water ratio for Warthin tumors versus pleomorphic adenomas at echo times of 136 msec (P = .003) and 272 msec (P = .002) but not for benign SGTs versus malignant tumors.

CONCLUSION: 1H MR spectroscopy may be used to characterize SGTs, but a larger study is required to validate these initial results.

© RSNA, 2005


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Salivary gland tumors (SGTs) account for less than 3% of all tumors in the body (1). The treatment of choice for most SGTs is surgery, but there are several advantages to obtaining a preoperative diagnosis. These advantages include better preoperative planning and patient counseling in cases of malignant tumors, the identification of those malignant tumors for which surgery is not the primary treatment, and the identification of benign tumors, such as a Warthin tumor, for which conservative treatment may be warranted (2). In many centers, patients who manifest an SGT undergo cross-sectional magnetic resonance (MR) imaging. However, because MR imaging cannot be used to reliably distinguish between a benign and a malignant SGT, an additional investigation is required to improve the diagnostic accuracy. Over the past few years, the role of in vivo proton MR spectroscopy has been evaluated in the imaging of tumors. The results of hydrogen 1 (1H) MR spectroscopy have been reported in a few articles of in vivo cancer series in the head and neck region (3–7), but, to the best of our knowledge, there are no published series results on the spectroscopic findings for SGTs. The purpose of this study was to prospectively evaluate whether 1H MR spectroscopy can be used to characterize SGTs.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
Patient Selection
Patients in this prospective study were recruited from a group of patients undergoing conventional MR imaging between August 2001 and May 2004 for the investigation of a parotid or submandibular lump. Proton MR spectroscopy was performed in those patients with a tumor larger than 1 cm3 at MR imaging. Histologic confirmation of the diagnosis was made either from the surgical specimen or at ultrasonographically (US)-guided biopsy, which was performed after the MR examination. Patients undergoing MR imaging for the investigation of a mass in the opposite salivary gland or for suspicion of other head and neck cancer were recruited to obtain control spectra from a normal salivary gland. In all cases, these control patients had no past or present history of symptoms within the gland and no history of radiation or surgery of the head and neck, and the gland had a normal appearance on MR images. The ethics committee at our institution approved this study, and informed consent was obtained from all patients.

MR Examinations
The examinations were performed by using a 1.5-T whole-body MR imaging system (Gyroscan ACS-NT; Philips, Best, the Netherlands) with a 23-mT/m maximum gradient capability. A quadrature head and neck coil was used for the imaging of the head and neck, and a 20-cm-diameter circular receive-only surface coil (Philips) placed over the parotid or submandibular gland was used to acquire contrast material–enhanced images and to improve sensitivity when 1H MR spectroscopy was performed. The body coil was used to generate a homogeneous B1 excitation field in all MR examinations. Clinical imaging protocol used for examination in all the patients included a coronal T2-weighted fast spin-echo sequence (repetition time msec/echo time msec, 2500/120; section thickness, 4 mm with no intersection gap; matrix, 256 x 256; two signals acquired), a transverse T1-weighted spin-echo sequence (425/18; section thickness, 4 mm with no intersection gap; matrix, 512 x 512; two signals acquired), and a contrast-enhanced transverse T1-weighted spin-echo sequence performed after a bolus injection of 0.1 mmol gadodiamide (Omniscan; Nycomed, Oslo, Norway) per kilogram of body weight. The latter imaging sequence was performed after 1H MR spectroscopy was completed.

With image guidance from the unenhanced MR images obtained in the coronal and transverse planes, the volume of interest was carefully positioned (A.T.A., A.D.K., H.Y.Y., K.T.W.) within the normal gland or the SGT. When lesions were present bilaterally, the largest lesion was selected for study. By using the point-resolved spectroscopic sequence (at 2000/136 and 2000/272) with the circular surface coil selected as the receiver, two water-suppressed spectra were acquired for each volume of interest. Prescanning automated parameter optimization consisted of frequency and receiver gain adjustment, shimming, and gradient tuning. Water suppression was achieved by means of selective inversion recovery by starting the measurement at the zero crossing of the water signal. Data were acquired at a spectral bandwidth of 1000 Hz, and 64 signals were acquired for each water-suppressed spectrum. An unsuppressed water signal with 16 signals acquired at each echo time was used as a reference spectrum. The acquired signals were exported and processed on an off-line computer.

Data Analysis
Spectra were analyzed (D.K.W.Y.) without knowledge of the histopathologic result by using the time-domain fitting routine that uses prior knowledge known as advanced method for accurate, robust, and efficient spectral, or AMARES; this method is implemented in the MR User Interface software package (8) (available at http://sermn02.uab.es/mrui/). After the removal of residual water (4.65 ppm) and lipid peaks in the chemical shift range of 0.90–2.02 ppm from the free induction decay by means of time-domain Hankel-Lanczos singular value decomposition filtering (9), Cho and Cr peak amplitudes were determined. Manually selected resonance frequency and linewidth of Cho and Cr peaks were used as the starting values in the nonlinear least squares fitting algorithm.

Prior knowledge incorporated into the fitting procedure consisted of the following: linewidth of Cr equal to that of Cho; resonance frequencies were constrained to lie in the range of ±0.05 ppm of the known resonance frequencies of Cho (3.2 ppm) and Cr (3.02 ppm); the zero-order phase correction estimated by means of AMARES and the first-order phase were fixed to zero; and the resonances relative phase was also set to zero and a Gaussian model assumed for all the peaks. For a peak to be included for further analysis, it had to have a signal-to-noise ratio of at least two. For the measurement of water peak in reference spectra, manually selected starting values for resonance frequency and linewidth were used. The linewidth of the water peak acquired without water suppression was measured to assess the quality of magnet shim achievable under the examination conditions. The calculated Cho, Cr, and water peak amplitudes were used to determine the Cho/Cr and Cho/water ratios for each parotid lesion.

Statistical Analysis
To test whether there were significant differences in Cho/Cr and Cho/water ratio values between (a) Warthin tumors and pleomorphic adenomas, (b) pleomorphic adenomas and malignant tumors, (c) Warthin tumors and malignant tumors, or (d) the combined group of benign parotid tumors and malignant tumors, we used the Mann-Whitney nonparametric test. A receiver operating characteristic curve was constructed by plotting the sensitivity versus 1 – specificity of 1H MR spectroscopy for enabling the prediction of tumor category identified from logistic regression. An optimal prediction function should yield a high true-positive rate (ie, high sensitivity), as well as a low false-positive rate (ie, high specificity). The choice of the critical cutoff point, however, varies with the perceived trade-offs between sensitivity and specificity. In this study, the cutoff point was chosen to maximize the sum of sensitivity and specificity. Analyses were performed by using SPSS for Windows (release 11.0; SPSS, Chicago, Ill). The level of significance was set at P ≤ .05 in all comparisons, and all statistical testing was two-sided. A power analysis was performed at the end of the study.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Sixty-seven patients (41 men and 26 women; mean age, 53 years; age range, 19–89 years) and nine control subjects (five men and four women; mean age, 56 years) were recruited for study. Six patients were excluded because in three the final diagnosis was inflammatory disease and in three the histologic diagnosis was not confirmed. Data from five of the remaining 61 (8%) patients (parotid gland tumor in four patients and submandibular gland tumor in one patient) were excluded because the spectra acquired at both echo times were not interpretable. Therefore, in the final data analysis of SGTs, there were 56 patients: 35 men and 21 women (mean age, 56 years; age range, 19–84 years). The 1H MR spectra from the 56 SGTs (one in each patient), which comprised 22 pleomorphic adenomas, 24 Warthin tumors, one oncocytoma, and nine malignant tumors (three acinic carcinomas, two adenoid cystic carcinomas, two mucoepidermoid carcinomas, one B cell lymphoma, and one lymphoepithelioma), were analyzed. Of the 56 tissue samples sent for pathologic evaluation, 29 were obtained by means of fine-needle aspiration with US guidance and 27 were obtained by means of surgical excision.

Spectra
Among the 56 SGTs that were examined at 1H MR spectroscopy, there were 55 spectra successfully obtained at an echo time of 136 msec and 53 spectra obtained at an echo time of 272 msec. All SGTs were located in the parotid gland except for two malignant tumors that were located in the submandibular gland. Proton MR spectroscopy was performed in SGTs by using voxel volumes of 1–15 cm3 (mean, 4.1 cm3) and in normal parotid glands by using voxel volumes of 1.8–5.5 cm3 (mean, 2.9 cm3). Spectra were successfully acquired at both echo times from all normal parotid glands that showed presence of lipids but no detectable levels of Cho or Cr. The mean linewidth of the water peak measured on spectra without water suppression was 5.4 Hz ± 1.8 (standard deviation). Broad and intense overlapping peaks derived from fatty acids in the chemical shift range of 0.90–2.02 ppm were present on all spectra at an echo time 136 msec, and these signals were equally present at an echo time of 272 msec. Lactate doublet (1.32 ppm), which is in the same chemical shift range as fatty acids, was not identifiable on any of the spectra. Removal of broad lipid peaks in the range of 0.90–2.02 ppm with the use of Hankel-Lanczos singular value decomposition filtering was successfully performed on all spectra.

Cho was present on the successfully obtained spectra of all benign and malignant lesions acquired with either echo time. With the use of prior knowledge in the fitting procedure, Cr peak amplitude was obtainable in 26 (47%) of 55 spectra at an echo time of 136 msec and in 16 (30%) of 53 spectra at an echo time of 272 msec. Table 1 summarizes the mean value and standard deviation of Cho/Cr and Cho/water ratios measured on spectra acquired at echo times of 136 msec and 272 msec. Examples of the spectra from a Warthin tumor, a pleomorphic adenoma, and a malignant tumor are shown in Figures 1 –3. The P values obtained in the comparison of mean Cho/Cr and Cho/water ratios between different tumor groups are shown in Table 2.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Mean Cho/Cr and Cho/Water Ratios for SGTs at Different Echo Times

 


View larger version (24K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1a. (a) Spectrum acquired at echo time of 136 msec in a 72-year-old woman with a Warthin tumor of the right parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and unresolved lipids (0.90–2.02 ppm). The nominal voxel volume was 3.1 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted (425/18) MR image shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (171K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1b. (a) Spectrum acquired at echo time of 136 msec in a 72-year-old woman with a Warthin tumor of the right parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and unresolved lipids (0.90–2.02 ppm). The nominal voxel volume was 3.1 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted (425/18) MR image shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (114K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1c. (a) Spectrum acquired at echo time of 136 msec in a 72-year-old woman with a Warthin tumor of the right parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and unresolved lipids (0.90–2.02 ppm). The nominal voxel volume was 3.1 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted (425/18) MR image shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (109K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1d. (a) Spectrum acquired at echo time of 136 msec in a 72-year-old woman with a Warthin tumor of the right parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and unresolved lipids (0.90–2.02 ppm). The nominal voxel volume was 3.1 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted (425/18) MR image shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a. (a) Spectrum acquired at echo time of 136 msec in a 39-year-old woman with a pleomorphic adenoma of the left parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and broad lipid signals (0.90–2.02 ppm). The nominal voxel volume was 4.8 cm3. Lipids and residual water were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (193K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b. (a) Spectrum acquired at echo time of 136 msec in a 39-year-old woman with a pleomorphic adenoma of the left parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and broad lipid signals (0.90–2.02 ppm). The nominal voxel volume was 4.8 cm3. Lipids and residual water were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2c. (a) Spectrum acquired at echo time of 136 msec in a 39-year-old woman with a pleomorphic adenoma of the left parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and broad lipid signals (0.90–2.02 ppm). The nominal voxel volume was 4.8 cm3. Lipids and residual water were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (136K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2d. (a) Spectrum acquired at echo time of 136 msec in a 39-year-old woman with a pleomorphic adenoma of the left parotid gland shows Cho (3.2 ppm), Cr (3.02 ppm), and broad lipid signals (0.90–2.02 ppm). The nominal voxel volume was 4.8 cm3. Lipids and residual water were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (25K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a. (a) Spectrum acquired at echo time of 136 msec in a 24-year-old man with an adenoid cystic carcinoma of the right parotid gland. Cho (3.2 ppm), Cr (3.02 ppm), and intense lipid (1.50–2.02 ppm) signals were detected. The nominal voxel volume was 2.5 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (184K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b. (a) Spectrum acquired at echo time of 136 msec in a 24-year-old man with an adenoid cystic carcinoma of the right parotid gland. Cho (3.2 ppm), Cr (3.02 ppm), and intense lipid (1.50–2.02 ppm) signals were detected. The nominal voxel volume was 2.5 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (117K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3c. (a) Spectrum acquired at echo time of 136 msec in a 24-year-old man with an adenoid cystic carcinoma of the right parotid gland. Cho (3.2 ppm), Cr (3.02 ppm), and intense lipid (1.50–2.02 ppm) signals were detected. The nominal voxel volume was 2.5 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 


View larger version (85K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3d. (a) Spectrum acquired at echo time of 136 msec in a 24-year-old man with an adenoid cystic carcinoma of the right parotid gland. Cho (3.2 ppm), Cr (3.02 ppm), and intense lipid (1.50–2.02 ppm) signals were detected. The nominal voxel volume was 2.5 cm3. Residual water resonance (4.65 ppm) and lipids were removed from the spectrum with time-domain Hankel-Lanczos singular value decomposition filtering. The bottom trace shows the result of a fitted spectrum with prior knowledge. (b) Transverse T1-weighted MR image (425/18) shows the positioning of the volume of interest (box) that was selected for 1H MR spectroscopy. (c) Coronal T2-weighted (2500/120) and (d) contrast-enhanced transverse T1-weighted (425/18) MR images show the tumor (arrows).

 

View this table:
[in this window]
[in a new window]

  		TABLE 2. P Values in the Comparison of Mean Cho/Cr and Cho/Water Ratios between Tumor Groups

 
In regard to the scatterplot (Fig 4a) of the Cho/Cr ratios obtained at an echo time of 136 msec for Warthin tumors, pleomorphic adenomas, the oncocytoma, and malignant tumors, the Cho/Cr ratios in all benign tumors were in the range of 2.44–8.20, and ratios in all benign tumors were above those in malignant tumors, for which the range was 1.19–2.28. Therefore, in this study by using a value greater than 2.4, the sensitivity, specificity, negative predictive value, and positive predictive value for predicting a benign SGT were all 100%. However, there were overlapping Cho/Cr ratios between Warthin tumors and pleomorphic adenomas. Receiver operating characteristic analysis was performed to examine whether using Cho/Cr ratios obtained at an echo time of 136 msec could separate these two groups of benign tumors. The area under the receiver operating characteristic curve for separating Warthin tumors from pleomorphic adenomas was 0.86 ± 0.09 (Fig 4b). When we used a Cho/Cr ratio of 4.5 or greater to differentiate between Warthin tumors and pleomorphic adenomas, the technique had a sensitivity of 71.4% (five of seven), a specificity of 84.6% (11 of 13), a positive predictive value of 71.4% (five of seven), and a negative predictive value of 84.6% (11 of 13).



View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. (a) Scatterplot of Cho/Cr values obtained at an echo time of 136 msec for Warthin tumors, pleomorphic adenomas, an oncocytoma, and malignant tumors. A Cho/Cr ratio greater than 2.4 (dotted line) completely separates benign SGTs from cancers. In discriminating between Warthin tumors and pleomorphic adenomas, a Cho/Cr ratio of 4.5 or greater (dashed line) would give a sensitivity of 71.4% and a specificity of 84.6%. (b) Receiver operating characteristic curve (dashed line) of Cho/Cr values obtained at an echo time of 136 msec for use in differentiating Warthin tumor from pleomorphic adenoma. The area under the curve is 0.86 ± 0.09.

 


View larger version (13K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. (a) Scatterplot of Cho/Cr values obtained at an echo time of 136 msec for Warthin tumors, pleomorphic adenomas, an oncocytoma, and malignant tumors. A Cho/Cr ratio greater than 2.4 (dotted line) completely separates benign SGTs from cancers. In discriminating between Warthin tumors and pleomorphic adenomas, a Cho/Cr ratio of 4.5 or greater (dashed line) would give a sensitivity of 71.4% and a specificity of 84.6%. (b) Receiver operating characteristic curve (dashed line) of Cho/Cr values obtained at an echo time of 136 msec for use in differentiating Warthin tumor from pleomorphic adenoma. The area under the curve is 0.86 ± 0.09.

 
Power Analysis
We observed that the standard deviation ranged from 0.47 to 1.86 for the Cho/Cr ratios at an echo time of 136 msec, and we postulated that the minimal difference needed to differentiate between malignant and benign tumors was 2.75 (midpoint of 1.73 to 3.76). We had more than 80% power to detect the postulated difference assuming that the standard deviation was 1.165 (ie, [0.47 + 1.86]/2).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
The majority of SGTs in this study, as well as those in the general population, arose in the parotid gland (10). Parotid gland tumors account for less than 3% of head and neck tumors (11), but they show a very diverse range of histologic types that influences both prognosis and treatment. Benign tumors constitute about 80% of these tumors, of which pleomorphic adenomas are the most common, followed by Warthin tumors (10). The distribution of benign tumors in this study was unusual in that Warthin tumors were the most common type of tumor. It is thought that the incidence of Warthin tumors is increasing (12) and that these tumors are generally more common in the Asian population and especially among the Chinese. In our local Chinese population, this tumor appears to be even more common than is quoted in the literature.

Malignant tumors are less frequent and are usually carcinomas of the mucoepidermoid, acinic cell, adenoid cystic, or adenocarcinoma histologic type. While malignant tumors form only a small percentage of parotid gland tumors, the surgical treatment can be potentially disfiguring. At cross-sectional imaging with computed tomography, MR imaging, or US, diagnosis of a malignant lesion is reliant on the identification of infiltration into the adjacent soft tissues and lymphadenopathy. The depiction of the extent of tumor infiltration along nerves, around vascular structures, and into bone is one of the reasons that MR imaging is preferred for presurgical tumor mapping. However, these conventional imaging techniques do not enable one to reliably distinguish between a benign and a malignant lesion (13,14), and, to our knowledge, positron emission tomography has not been able to improve the diagnostic accuracy (15).

Fine-needle aspiration cytologic analysis, or FNAC, has gained in popularity and has been shown to change the clinical approach in around 35% of patients with SGTs (16). FNAC is quick and safe and can be used to determine the correct diagnosis in up to 95% of patients (17,18). For these reasons, many centers at which US is widely available (including our own) combine US imaging with US-guided FNAC as the method of choice for preoperative diagnosis. However, the results of FNAC are dependent on the level of experience of the cytologist, and there are reports that FNAC may be less sensitive for malignant tumors of the parotid gland (19). In addition, expertise in head and neck US is not widely available in all regions of the world, and in many centers MR imaging may be the initial method of imaging investigation. In this situation, 1H MR spectroscopy offers the prospect of improving the characterization of SGTs at the same time that conventional MR imaging is performed.

Spectra were successfully obtained in most SGTs and all normal parotid glands because we were able to obtain satisfactory shimming. In less than 10% of patients in our study, spectroscopy was unsuccessful as a result of patient motion during data acquisition. Analysis of the spectra did not demonstrate any peaks that were specific for the different histologic groups of SGTs or for SGTs as a whole. We did not find evidence of lactate in the SGTs in our study, although lactate has been reported to be present in head and neck lesions (5). However, we cannot rule out its presence in SGTs for the following reasons: First, lactate resonates (1.32 ppm) in the same region as lipids (0.90–2.02 ppm), and, since these fatty acid signals were very broad (12.2 Hz ± 2.4) and intense, weak signal contribution from lactate might have been overshadowed by that from lipids. Second, we used a conventional spectroscopy pulse sequence in our study, which may not be optimal for the detection of lactate in SGTs because recent reports have shown that special lactate-editing techniques may be necessary in the identification of lactate in head and neck tumors in vivo (5,6) in the presence of strong lipid signals.

Acquisition of spectra for analysis of Cho/Cr ratios was most successful with an echo time of 136 msec, although even at this echo time Cr could be identified in only about half of SGTs thereby reducing the number of patients in whom the Cho/Cr ratios could be used to characterize tumors. However, despite the reduction in the number of cases for Cho/Cr analysis and the small total number of malignant tumors in this series, there was still a statistically significant difference in the ratios between benign tumors and malignant tumors, with higher ratios found in the benign tumors. Results of a previous study by Maheshwari et al (7) also showed a significantly elevated Cho/Cr ratio in benign tumors compared with that in malignant tumors of the head and neck.

An elevated Cho level, which is a marker of membrane turnover, is found not only in malignant tumors but also in benign tumors that are hypercellular and in inflammatory processes. Warthin tumors contain a large number of lymphocytes, and this may be why these benign tumors showed significantly higher ratios than did either malignant tumors or pleomorphic adenomas. Our initial results suggest that if the Cho/Cr ratio acquired at an echo time of 136 msec is greater than 2.4, the positive predictive value that the SGT is benign is 100%, and if the ratio is greater than 4.5, the positive predictive value that the SGT is a Warthin tumor is 71%. Cr peaks could not be identified as frequently by using an echo time of 272 msec; this is probably because of the effect of relaxation losses occurring at the longer echo time. Despite this, the differences in the mean Cho/Cr ratios for malignant versus benign tumors and for Warthin tumors versus pleomorphic adenomas remained statistically significant.

There were also statistically significant results when using the Cho/water ratios, and the use of water instead of Cr as the internal reference increased the number of patients in whom a metabolic ratio for a SGT was measurable. This ratio could be used as an alternative to the Cho/Cr ratio; however, there are a large number of factors that can influence the amount of MR-visible water, which may have accounted for the wider standard deviations of this ratio at both echo times.

The main limitation of this study was that about half of the diagnoses were made with fine-needle aspiration cytologic analysis and not histologic evaluation from the surgical specimen. Unfortunately this was unavoidable in our center, where many patients, especially those with Warthin tumors, elect not to undergo surgery. The other limitation was that the number of malignant SGTs was small, and larger studies are required to validate the results. In particular, a larger series is required to ensure that there is no overlap in the ratios between low-grade malignancies and pleomorphic adenomas, which is an area that currently causes the greatest difficulty at imaging and fine-needle aspiration cytologic analysis, and to evaluate a broader histologic range of malignant SGTs. In addition, we did not evaluate inflammatory and infectious salivary conditions to determine if there is an overlap with the ratios for SGT.

In conclusion, our results showed that Cho is detected in benign and malignant SGTs but not in the normal parotid gland. The major limitation of using 1H MR spectroscopy for the evaluation of SGTs is that it can be performed only in tumors that are larger than 1 cm3 in size, and even within this group, the Cho/Cr ratios can be obtained in less than half of these tumors. However, in those patients in whom Cho/Cr ratios are obtained, the preliminary results of our study suggest that the ratio may be used to characterize the SGTs. The ratios are highest in Warthin tumors, followed by pleomorphic adenomas and then malignant tumors. By using Cho/Cr ratios at an echo time of 136 msec, our results suggest that a ratio greater than 2.4 may be used to distinguish between benign and malignant tumors, while a ratio greater than 4.5 suggests that the lesion is probably a Warthin tumor. A larger study of malignant SGTs is required to validate these initial results.


   ACKNOWLEDGMENTS
 
We thank Eric M. C. Wong, MSc, and Abby Y. T. Tong, BSc, for their assistance with the statistical analysis in the preparation of the manuscript.


   FOOTNOTES
 

Abbreviations: Cho = choline • Cr = creatine • SGT = salivary gland tumor

Authors stated no financial relationship to disclose.

Author contributions: Guarantor of integrity of entire study, A.D.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, A.D.K., D.K.W.Y.; clinical studies, A.D.K., A.T.A., G.M.K.T., H.Y.Y., K.T.W., A.C.v.H.; statistical analysis, A.D.K., D.K.Y.; and manuscript editing, A.D.K., D.K.Y.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Eneroth CM. Salivary gland tumors in the parotid gland, submandibular gland and palate region. Cancer 1971;27:1415–1418.[CrossRef][Medline]
  2. Myers EN, Suen JY. Cancer of the head and neck. 3rd ed. Philadelphia, Pa: Saunders, 1996; 525–561.
  3. Mukherji SK, Schiro S, Castillo M, Kwock L, Muller KE, Blackstock W. Proton MR spectroscopy of squamous cell carcinoma of the extracranial head and neck: in vitro and in vivo studies. AJNR Am J Neuroradiol 1997;18:1057–1072.[Abstract]
  4. Adalsteinsson E, Spielman DM, Pauly JM, Terris DJ, Sommer G, Macovski A. Feasibility study of lactate imaging of head and neck tumors. NMR Biomed 1998;11:360–369.[CrossRef][Medline]
  5. Star-Lack J, Spielman D, Adalsteinsson E, Kurhanewicz J, Terris DJ, Vigneron DB. In vivo lactate editing with simultaneous detection of choline, creatine NAA, and lipid singlets at 1.5 T using PRESS excitation with applications to the study of brain and head and neck tumors. J Magn Reson 1998;133:243–254.[CrossRef][Medline]
  6. Star-Lack JM, Adalsteinsson E, Adam MF, et al. In vivo 1H MR spectroscopy of human head and neck lymph node metastasis and comparison with oxygen tension measurements. AJNR Am J Neuroradiol 2000;21:183–193.[Abstract/Free Full Text]
  7. Maheshwari SR, Mukherji SK, Neelon B, et al. The choline/creatine ratio in five benign neoplasms: comparison with squamous cell carcinoma by use of in vitro MR spectroscopy. AJNR Am J Neuroradiol 2000;21:1930–1935.[Abstract/Free Full Text]
  8. Vanhamme L, van den Boogaart A, van Huffel S. Improved method for accurate and efficient quantification of MRS data with use of prior knowledge. J Magn Reson 1997;129:35–43.[CrossRef][Medline]
  9. Van den Boogaart A, van Ormondt D, Pijnappel WWF, de Beer R, Ala-Korpela M. Removal of the water resonance from H-1 magnetic resonance spectra. In: McWhirter JG, ed. Mathematics in signal processing III. Oxford, England: Clarendon, 1994;175–195.
  10. Eveson JW, Cawson RA. Salivary gland tumours: a review of 2410 cases with particular reference to histological types, site, age and sex distribution. J Pathol 1985;146:51–58.[CrossRef][Medline]
  11. van der Wal JE, Leverstein H, Snow GB, Kraaijenhagen HA, van der Waal I. Parotid gland tumors: histologic reevaluation and reclassification of 478 cases. Head Neck 1998; 20:204–207.[CrossRef][Medline]
  12. Chung YF, Khoo ML, Heng MK, Hong GS, Soo KC. Epidemiology of Warthin's tumour of the parotid gland in an Asian population. Br J Surg 1999;86:661–664.[CrossRef][Medline]
  13. Yousem DM, Kraut MA, Chalian AA. Major salivary gland imaging. Radiology 2000;216:19–29.[Abstract/Free Full Text]
  14. Freling NJ, Molenaar WM, Vermey A, et al. Malignant parotid tumors: clinical use of MR imaging and histologic correlation. Radiology 1992;185:691–696.[Abstract/Free Full Text]
  15. McGuirt WF, Keyes JW Jr, Greven KM, Williams DW 3rd, Watson NE Jr, Cappellari JO. Preoperative identification of benign versus malignant parotid masses: a comparative study including positron emission tomography. Laryngoscope 1995;105:579–584.[Medline]
  16. Heller KS, Dubner S, Chess Q, Attie JN. Value of the fine needle aspiration biopsy of salivary gland masses in clinical decision-making. Am J Surg 1992;164:667–670.[CrossRef][Medline]
  17. Al-Khafaji BM, Nestok BR, Katz RL. Fine-needle aspiration of 154 parotid masses with histologic correlation: ten-year experience at the University of Texas M. D. Anderson Cancer Center. Cancer 1998; 84:153–159.[CrossRef][Medline]
  18. Raab SS, Sigman JD, Hoffman HT. The utility of parotid gland and level I and II neck fine-needle aspiration. Arch Pathol Lab Med 1998;122:823–827.[Medline]
  19. Stafford ND, Wilde A. Parotid cancer. Surg Oncol 1997;6:209–213.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 Am. J. Neuroradiol.Home page
J.A. Castelijns and C. R. Leemans
Closing the Uncertainty Gap in the Diagnosis of Parotid Tumors
AJNR Am. J. Neuroradiol., March 1, 2009; 30(3): 442 - 443.
[Full Text] [PDF]

Home page
 Am. J. Neuroradiol.Home page
Q. Yu, J. Yang, and P. Wang
Malignant Tumors and Chronic Infections in the Masticator Space: Preliminary Assessment with In Vivo Single-Voxel 1H-MR Spectroscopy
AJNR Am. J. Neuroradiol., April 1, 2008; 29(4): 716 - 719.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
J. F. Faria, S. M. Goldman, J. Szejnfeld, H. Melo, C. Kater, P. Kenney, M. P. Huayllas, G. Demarchi, V. V. Francisco, C. Andreoni, et al.
Adrenal Masses: Characterization with in Vivo Proton MR Spectroscopy Initial Experience
Radiology, December 1, 2007; 245(3): 788 - 797.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by King, A. D.
Right arrow Articles by van Hasselt, A. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by King, A. D.
Right arrow Articles by van Hasselt, A. C.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE