Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
Purpose:
To describe the prevalence of three relative enhancement patterns of parathyroid lesions on four-dimensional (4D) computed tomographic (CT) scans.
Materials and Methods:
The institutional review board approved this HIPAA-compliant study and waived the need for informed consent. The authors retrospectively reviewed preoperative 4D CT scans obtained from November 2012 to June 2014 in 94 patients with pathologically proven parathyroid adenomas or hyperplasia. Lesions were classified into one of three relative enhancement patterns. All patterns required lesions to be lower in attenuation than the thyroid on non–contrast materialenhanced images, but patterns differed in the two contrastenhanced phases. Type A lesions were higher in attenuation than the thyroid in the arterial phase, type B lesions were not higher in attenuation than the thyroid in the arterial phase but were lower in attenuation than the thyroid in the delayed phase, and type C lesions were neither higher in attenuation than the thyroid in the arterial phase nor lower in attenuation than the thyroid in the delayed phase. The prevalence of the relative enhancement patterns was compared. The t test was used to compare mean attenuation differences in Hounsfield units between the relative enhancement patterns.
Results:
Ninety-four patients had 110 parathyroid lesions, including 11 patients with multigland disease. The sensitivity for single-gland disease was 94% (78 of 83) and that for multigland disease was 59% (16 of 27). Type B enhancement was most common, with a prevalence of 57% (54 of 94), followed by type C (22% [21 of 94]) and type A (20% [19 of 94]). Five lesions were interpreted incorrectly as parathyroid adenoma (false-positive), and all lesions had the type C pattern. Relative to the thyroid, lesions categorized as type A by readers had mean attenuation difference (6 standard deviation) of 39 HU 6 13 in the arterial phase, and type B lesions had a difference of 258 HU 6 26 in the delayed phase. These values differed from the mean attenuation difference of lesions not in these categories (P , .001).
Conclusion:
Parathyroid adenomas and hyperplasia can be grouped into three relative enhancement patterns based on a protocol with a non–contrast-enhanced and two contrast-enhanced phases. The type B pattern is most common and could be diagnosed with two contrast-enhanced phases. However, almost one quarter of lesions have the type C pattern and thus could be missed without the non–contrast-enhanced phase.
1
From the Department of Radiology, Division of Neuroradiology, Duke University Medical Center, Box 3808, Erwin Rd, Durham, NC 27710. Received October 10, 2014; revision requested December 8; final revision received February 6, 2015; accepted March 2; final version accepted March 13. Address correspondence to J.K.H. (e-mail: jennykh@ gmail.com). RSNA, 2015
q
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
RSNA, 2015
q
1
and Neck Imaging
Manisha Bahl, MD, MPH Ali R. Sepahdari, MD Julie A. Sosa, MD, MA Jenny K. Hoang, MBBS
Original Research n Head
Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans: Three Patterns of Enhancement Relative to the Thyroid Gland Justify a Three-Phase Protocol1
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
F
our-dimensional (4D) computed tomography (CT) of the parathyroid is a technique for preoperative localization of parathyroid adenomas. It involves multi–detector row CT of the neck and upper chest during two or more contrast material–enhanced phases (1,2). Multiple phases are performed to determine the pattern of enhancement, which is one of the primary features used to identify a parathyroid adenoma and to distinguish
Advances in Knowledge nn The enhancement pattern of parathyroid adenomas and hyperplasia on four-dimensional (4D) CT scans can be grouped into three types on the basis of relative attenuation compared with that of the thyroid gland; each type is lower in attenuation than the thyroid on nonenhanced images, but patterns differ according to the arterial and delayed phase findings. nn Type A lesions have higher attenuation than the thyroid in the arterial phase; only 20% (19 of 94) of parathyroid lesions had this enhancement pattern. nn Type B lesions are not higher in attenuation than the thyroid in the arterial phase but are lower in attenuation than the thyroid in the delayed phase; more than half of lesions demonstrated the type B pattern (57% [54 of 94]). nn Type C lesions are neither higher in attenuation than the thyroid in the arterial phase nor lower in attenuation than the thyroid in the delayed phase; type C is the second most common pattern (22% [21 of 94]) and could be missed without the nonenhanced phase. nn All five false-positive lesions had the type C pattern. nn The sensitivity of 4D CT for single-gland disease was 94% and that for multigland disease was 59%; overall sensitivity for all lesions was 85%. 2
it from mimics, such as thyroid nodules and lymph nodes. The characteristic description of a parathyroid adenoma on 4D CT scan is a lesion with lower attenuation than that of the thyroid gland in the non–contrastenhanced phase, peak enhancement greater than that of the thyroid gland in the arterial phase, and washout of contrast material in the delayed phase (2–4). However, the characteristic enhancement pattern may not be evident in all cases, and failure to appreciate the variability in enhancement pattern may lead to missed lesions. In our clinical experience, we have observed three relative enhancement patterns of parathyroid lesions compared with the thyroid gland on 4D CT scans. To our knowledge, there is no literature about variable enhancement patterns of parathyroid lesions. Such studies would be valuable to radiologists
Implications for Patient Care nn Radiologists should be aware of the three relative enhancement patterns of parathyroid adenomas and hyperplasia when interpreting 4D CT studies because not all adenomas are best seen in the arterial phase, as previously described in the literature. nn In designing 4D CT protocols, it is important to recognize that categorization of relative enhancement patterns in adenomas uses all phases of CT in a threephase protocol. nn Lesions with a type C relative enhancement pattern may mimic thyroid tissue without a nonenhanced phase because they are not higher in attenuation than the thyroid in the arterial phase or lower in attenuation than the thyroid in the delayed phase. nn Omitting the nonenhanced phase from the 4D CT protocol could lead to missed lesions with type C enhancement and an increase in the rate of false-positive lesions.
Bahl et al
who are new to interpreting 4D CT images and for justifying the optimal number of phases for 4D CT (4–8). The aim of this study was to describe the prevalence of three types of enhancement patterns of parathyroid adenomas and hyperplasia on 4D CT scans.
Materials and Methods Patients Our institutional review board approved this study and waived the need for informed consent. The studied complied with the Health Insurance Portability and Accountability Act. We performed a retrospective study of patients with parathyroid adenomas and hyperplasia who underwent preoperative 4D CT of the parathyroid, with at least two postcontrast phases, between November 2012 and June 2014. Patients were identified by searching radiology text reports for CT scans with the term parathyroid. A total of 133 patients were identified. Patients were excluded if they did not have surgery (n = 34 [26%]), did not have a three-phase study (n = 3 [2%]), or had parathyroid carcinoma (n = 2 [2%]). Thus, the final cohort consisted of 94 patients. All patients had biochemical evidence of hyperparathyroidism (elevated calcium and parathyroid hormone levels). Fifty of 94 patients (53%) underwent ultrasonography before 4D CT, and 38 of Published online before print 10.1148/radiol.2015142393 Content codes: Radiology 2015; 000:1–9 Abbreviation: 4D = four-dimensional Author contributions: Guarantors of integrity of entire study, M.B., J.K.H.; 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; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, M.B., A.R.S., J.K.H.; clinical studies, M.B., A.R.S., J.A.S.; experimental studies, M.B., J.K.H.; statistical analysis, M.B., J.K.H.; and manuscript editing, all authors Conflicts of interest are listed at the end of this article.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
94 (40%) underwent nuclear medicine scintigraphy. Hospital electronic medical records were accessed to obtain the following information: patient age and sex, surgical pathology reports, and surgical reports. Surgical pathology reports were reviewed for diagnosis (adenoma vs hyperplasia), weight, and size. If more than one parathyroid lesion was identified at surgery, these lesions were reported as parathyroid hyperplasia.
Four-dimensional CT Imaging Technique Four-dimensional CT was performed with 64-row multi–row detector CT scanners (Discovery CT750 HD and LightSpeed VCT; GE Healthcare, Milwaukee, Wis). The first phase was nonenhanced to cover the thyroid gland, with the z-axis from the hyoid bone to the clavicular head. The next two phases were contrast material–enhanced phases from the angle of the mandible to the carina, performed after intravenous administration of 75 mL of iopamidol (Isovue-300; Bracco, Princeton, NJ) via a 20-gauge cannula in a right antecubital vein at a rate of 4 mL/sec, followed by a 25-mL saline chaser. Arterial phase images were acquired 25 seconds after the start of the injection, and delayed (venous) phase images were acquired 80 seconds from the start of the injection. The parameters for all three phases were as follows: detector configuration, 64 3 0.625 mm; tube rotation time, 0.4 second; pitch factor, 0.516:1; field of view, 20 cm; 120 kV (peak); and automatic tube current modulation (Smart mA; GE Healthcare) (noise index, 8; minimum of 100 mA and maximum of 700 mA for arterial phase, minimum of 100 mA and maximum of 500 mA for the other two phases to reduce radiation exposure). The 0.625-mmthick contiguous axial images in all three phases were sent to the picture archiving and communication system for interpretation and to allow manipulation of images on a three-dimensional workstation. Reformatted images in the arterial phases were sent to the picture archiving and communication system as 2.5-mm-thick contiguous images in the axial, coronal, and sagittal planes. On
the basis of a prior phantom study that measured organ doses with dosimeters, scanning a 73-kg patient with this protocol would result in dose length product of 1928 mGy · cm and calculated effective dose from organ doses of 28 mSv (9).
Image Interpretation Accuracy of 4D CT for localization of parathyroid adenomas.—The 4D CT images were reviewed on a picture archiving and communication system workstation (Centricity; GE Healthcare). One radiologist (with 12 years of experience) reviewed the 4D CT images before surgery to localize the parathyroid adenomas and generate the radiology report. Interpretation of 4D CT scans was considered correct if it enabled localization of the adenoma to the correct side and quadrant or to its ectopic location and if it approximated the size of the surgical specimen. If a lesion was reported in the impression section of the radiology report, this was regarded as a positive interpretation on 4D CT scans and this lesion was correlated with the surgical and pathology notes. Lesions that were not confirmed to be parathyroid adenomas or hyperplastic glands at surgery were considered to be false-positive cases. For false-negative cases (missed lesions), a radiologist (J.K.H., fellowshiptrained neuroradiologist with 12 years of experience) was given information about the location of the missed lesion at surgery, and images were reviewed again to determine whether a missed lesion could be seen in the exact quadrant. A second reader (M.B., a radiology resident with 4 years of experience) was required to agree with the findings and determine that the size at pathologic assessment matched the CT size. The retrospectively identified false-negative lesions were analyzed for relative enhancement patterns; however, they were not included in the final analysis with the true-positive and false-positive lesions because they were not seen at the original image interpretation. Classification of relative enhancement patterns.—Lesions were classified according to three relative enhancement
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Bahl et al
Figure 1
Figure 1: Classification of relative enhancement patterns of parathyroid adenomas and hyperplasia on 4D CT scans.
patterns by the two readers. The readers assigned the patterns subjectively and did not measure attenuation values. Images were independently reviewed on a picture archiving and communication system station. The readers could adjust the window settings if required. All three relative enhancement patterns required the adenoma to be lower in attenuation than the thyroid gland on nonenhanced images (Figs 1–4). In addition, with type A, the lesion was higher in attenuation than the thyroid gland in the arterial phase but could have any attenuation in the delayed phase (Fig 2). With type B, the lesion was not higher in attenuation than the thyroid gland in the arterial phase but was lower in attenuation than the thyroid gland in the delayed phase (Fig 3). With type C, the lesion was not higher in attenuation than the thyroid gland in the arterial phase and not lower in attenuation than the thyroid gland in the delayed phase (Fig 4). When the two readers disagreed about the enhancement classification, a consensus was reached by measurement of attenuation values in Hounsfield units. 3
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
at least 10 mm2 (when possible) with exclusion of cystic areas. To quantify the attenuation differences appreciated by the readers, the attenuation differences between the parathyroid lesions and the thyroid gland for each lesion were calculated by subtracting CT attenuation values of the thyroid from the CT attenuation values of the parathyroid lesion.
Figure 2
Figure 2: Axial 4D CT images (window level, 75 HU; width, 350 HU) in a 49-year-old woman with an inferior left parathyroid adenoma. (a) Non–contrast-enhanced image demonstrates a lesion (arrow) measuring up to 12 mm posterior to the inferior left thyroid lobe, adjacent to the trachea and medial to the common carotid artery. The lesion is lower in attenuation than the adjacent thyroid gland. (b) Arterial phase image demonstrates that the lesion (arrow) is higher in attenuation than the adjacent thyroid gland. This pattern is consistent with type A. (c) Delayed phase image demonstrates washout of the lesion (arrow) with similar attenuation to the thyroid gland.
After assigning categories of relative enhancement, a radiologist (M.B.) measured the attenuation values for the 4
Bahl et al
thyroid gland and parathyroid lesions for each imaging phase using the mean of three ellipsoid regions of interest of
Outcomes and Statistical Analysis The sensitivity of 4D CT for localization of parathyroid lesions was calculated on the basis of the original radiology report. The prevalence of enhancement patterns was calculated for pathologically proven parathyroid lesions that were correctly identified on the original radiology report. The characteristics of parathyroid lesions were compared for the three relative enhancement patterns. These characteristics included lesion size, ectopic location, single versus multigland disease, and prior failed surgery. Analysis of variance was used to compare continuous data. The x2 test or Fisher exact test was used for categorical data. The Fisher exact test was used when expected cell frequencies were fewer than five. For differences in enhancement in patients with multigland disease, a logistic regression model was fitted by using generalized estimating equations to account for multiple lesions in the same patient. Using the measured CT attenuation values for the thyroid gland and parathyroid lesions, the attenuation difference for each parathyroid lesion was calculated by subtracting the attenuation value of the thyroid gland from the attenuation value of the parathyroid lesion. The mean and standard deviation of the attenuation differences were then calculated for the three relative enhancement patterns. The t test was used to compare the mean of attenuation differences between the relative enhancement patterns. The mean of attenuation differences in the arterial phase for type A lesions (attenuation greater than that of the thyroid) was compared with the mean of attenuation differences for type B and C lesions
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
Figure 3
Bahl et al
(attenuation not greater than that of the thyroid). The mean of attenuation differences in the delayed phase for type B lesions (attenuation less than that of the thyroid) was compared with the mean of attenuation differences for type C lesions (attenuation not less than that of the thyroid). Type B lesions were not compared with type A lesions for the delayed phase because type A lesions could have any attenuation relative to the thyroid in the delayed phase. The Fleiss k value was calculated to determine interrater agreement for the enhancement pattern type. Agreement was regarded as poor with Fleiss k 0.20, as slight with k of 0.21–0.40, as moderate with k of 0.41–0.60, as substantial with k of 0.61–0.80, and as almost perfect with k of 0.81–1.00 (10). The data were entered into an Excel spreadsheet (2007 version; Microsoft, Redmond, Wash). Statistical analyses were performed by using SAS Enterprise (version 4.2; SAS Institute, Cary, NC) and R computing platform (www.r-project.org). P values of , .05 were considered to indicate statistically significant results.
Results Patients Of 94 patients (67 women; mean age, 63 years; range, 34–88 years) with pathologically proven parathyroid lesions, 83 (88%) had single-gland disease and 11 (12%) had multigland disease, resulting in a total of 110 lesions. The 110 lesions had a mean (6 standard deviation) maximal diameter at pathologic evaluation of 15.9 mm 6 6.8 and mean weight of 0.58 g 6 0.57.
Figure 3: Axial 4D CT images (window level, 75 HU; width, 350 HU) in a 50-year-old man with a superior left parathyroid adenoma. (a) Non–contrast-enhanced image demonstrates a lesion (arrow) measuring up to 7 mm posterior to the superior left thyroid lobe. The lesion is lower in attenuation than the adjacent thyroid gland. (b) Arterial phase image demonstrates that the lesion (arrow) is similar in attenuation to the adjacent thyroid gland. This pattern is not consistent with type A. (c) Delayed phase image demonstrates washout of the lesion (arrow) with attenuation lower than that of the adjacent thyroid gland. This pattern is consistent with type B. Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Accuracy of 4D CT for Localization of Parathyroid Adenoma In 78 of 83 patients with single-gland disease, lesions were correctly localized with 4D CT (sensitivity, 94%; 95% confidence interval: 89%, 99%). Eleven patients with multigland disease had 27 hyperplastic glands, 16 of which were correctly localized with 4D CT (sensitivity, 59%; 95% confidence interval: 41%, 78%). Of 110 total lesions, 94 5
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
lesions (85%) were identified (78 single-gland and 16 multigland) and 16 lesions (15%) were missed (five singlegland and 11 multigland). Among the patients with multigland disease, two or more glands were correctly identified in five patients. The overall sensitivity for single-gland and multigland disease was thus 85% (94 of 110; 95% confidence interval: 79%, 92%). There were five false-positive lesions in four patients; these represent lesions reported by the radiologist that were not confirmed to be adenomas at surgery in patients with a biochemical cure. The mean maximal diameter at pathologic evaluation of the 16 missed lesions was 11.9 mm 6 4.8.
Figure 4
Figure 4: Axial 4D CT images (window level, 75 HU; width, 350 HU) in a 63-year-old man with a right parathyroid adenoma. (a) Non–contrast-enhanced image demonstrates a lesion (arrow) measuring up to 12 mm posterior to the midportion of the right thyroid lobe. The lesion is lower in attenuation than the adjacent thyroid gland. (b) Arterial phase image demonstrates that the lesion (arrow) is similar in attenuation to the adjacent thyroid gland. (c) Delayed phase image demonstrates that the lesion (arrow) is similar in attenuation to the adjacent thyroid gland. This pattern is consistent with type C. 6
Bahl et al
Classification of Relative Enhancement Patterns Ninety-four parathyroid lesions detected at initial review and five false-positive lesions were categorized according to the relative enhancement pattern. The characteristic pattern of enhancement (type A) was present in 20% (19 of 94) of parathyroid lesions. Type B was seen in 57% (54 of 94) and type C lesions in 22% (21 of 94). All five falsepositive lesions were type C. There was no statistically significant difference in lesion size, ectopic location, multigland disease status, and prior failed surgery for the three relative enhancement patterns (Table 1). Table 2 summarizes the measured attenuation and mean of attenuation differences between the thyroid gland and parathyroid lesions. Type A lesions were categorized by readers as higher in attenuation than the thyroid gland in the arterial phase. The mean of attenuation differences between the thyroid gland and type A lesions in the arterial phase was 39 HU 6 13 compared with 232 HU 6 36 for lesions not classified as type A (P , .001). Type B lesions were categorized by readers as lower in attenuation than the thyroid gland in the delayed phase. The mean of attenuation differences between the thyroid gland and type B lesions in the delayed phase was 258 HU 6 26 compared with 25 HU 6 19 for type C lesions (not lower in attenuation than the thyroid in the delayed phase) (P , .001).
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
The two readers agreed on the enhancement categorization in 89 of 99 lesions (90%). The Fleiss k (an index of interrater agreement) was 0.85 (95% confidence interval: 0.76, 0.94), which indicates almost perfect agreement. In the 10 cases of disagreement between the two readers, the attenuation differences between the thyroid gland and
the candidate lesion ranged from 1 to 34 HU. Nine of 16 surgically confirmed lesions that were missed on 4D CT scans (false-negative results) were identified retrospectively. The mean maximal diameter of false-negative lesions was 9.2 mm 6 3.5 on CT scans and 11.2 mm 6 3.8 at pathologic evaluation. One lesion
Table 1 Characteristics of Parathyroid Adenoma or Hyperplasia according to Relative Enhancement Patterns Characteristic Adenoma or hyperplasia Single-gland disease Multigland disease False-positive lesions Mean maximal diameter at CT (mm)# Mean maximal diameter at pathologic evaluation (mm)# Mean weight (g)# Ectopic location‡‡ History of unsuccessful surgery
All Lesions
Type A
Type B
Type C
P Value
94* 78† 16§ 5 13.2 6 7.8
19 (20) 16 (21) 3 (19) 0 12.0 6 5.6
54 (57) 45 (58) 9 (56) 0 14.0 6 8.6
21 (22) 17 (22) 4 (25) 5 (100) 10.1 6 4.4
.96‡ .62|| ,.001‡ .11**
16.8 6 7.1
16.0 6 5.8
17.3 6 7.6
15.2 6 5.7
.48**
0.64 6 0.61 8 12
0.55 6 0.66 1 (13) 4 (33)
0.67 6 0.64 5 (63) 7 (58)
0.59 6 0.46 2 (26) 1 (8)
.74** .85‡ .31‡
Note.—Unless otherwise indicated, values are expressed as number of lesions, with percentages in parentheses. Percentages may not add up to 100% owing to rounding. Mean values are accompanied by standard deviation. * Sixteen of 110 surgically confirmed parathyroid lesions were not identified on 4D CT scans before surgery. †
Five of 83 surgically confirmed parathyroid lesions were not identified on 4D CT scans before surgery.
‡
Calculated with the Fisher exact test to test difference in proportions for the enhancement patterns.
§
Eleven of 27 surgically confirmed parathyroid lesions were not identified on 4D CT scans before surgery.
||
Calculated by fitting a logistic regression model and using generalized estimating equations to account for correlations due to multiple lesions measured in five of 94 patients. #
Data are means 6 standard deviation.
** Calculated with analysis of variance test to test difference in lesion size for the enhancement patterns. ††
Five were mediastinal, two retropharyngeal, and one parapharyngeal in location.
Bahl et al
had the type A enhancement pattern, five had the type B pattern, and three had the type C pattern.
Discussion In this study, we describe an enhancement categorization method for parathyroid adenomas based on three enhancement patterns relative to the thyroid tissue. The most common pattern, type B, accounted for more than half of all lesions and could be diagnosed confidently with two contrastenhanced CT phases alone. However, almost one quarter of lesions had the type C pattern, which could be challenging to differentiate from thyroid tissue without the nonenhanced phase. Previous studies have quantified the enhancement of adenomas and its mimics with attenuation values (3,4,6,11,12). Beland et al (3) compared enhancement of lymph nodes with that of adenomas and found adenomas to be significantly higher in attenuation at 30 and 60 seconds relative to other phases. Vu et al (11) found that the percentage change in attenuation from baseline to the arterial phase was the most powerful discriminatory feature for parathyroid adenomas and that contrast material washout from arterial peak to the venous phase was a less powerful discriminator. Hunter et al (12) proposed a multinomial logistic regression model based on inherent tissue attenuation and vascular
Table 2 Attenuation Values and Mean of Differences between Thyroid Gland and Parathyroid Lesions in Nonenhanced, Arterial, and Delayed Phases Nonenhanced Phase Type Type A Type B Type C
Arterial Phase
Delayed Phase
No. of Lesions
Thyroid*
Lesion*
Mean of Differences†
Thyroid*
Lesion*
Mean of Differences†
Thyroid*
Lesion*
Mean of Differences†
19 54 21
99 6 20 101 6 25 97 6 31
54 6 21 40 6 17 40 6 30
245 6 27 261 6 31 257 6 22
179 6 62 184 6 39 175 6 47
217 6 60 150 6 43 148 6 33
39 6 13 234 6 35 227 6 39
142 6 29 139 6 22 113 6 27
110 6 21 81 6 27 108 6 29
232 6 20 258 6 26 25 6 19
Note.—Categorization of relative enhancement patterns is based on subjective visual comparison of parathyroid lesions with the thyroid without performing attenuation measurements. * Data are mean attenuation values expressed in Hounsfield units 6 standard deviation. †
Attenuation difference for each lesion equals attenuation of parathyroid lesion minus attenuation of thyroid gland. The mean and standard deviation of the attenuation differences were calculated for the three relative enhancement patterns.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
7
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
information that could be used to estimate the probability that a lesion is a parathyroid adenoma. All of these methods involve placing regions of interest to measure attenuation and performing additional calculations to quantify change between phases. In practice, these measurements can be tedious for radiologists. The advantage of our relative enhancement classification system is that it consists of three simple categories that compare attenuation of the suspected parathyroid lesion with that of the adjacent thyroid without the need to quantify attenuation with measurements. Deviation of enhancement in parathyroid lesions from the typical description is valuable information to radiologists who are new to interpreting 4D CT images. Prior studies did not capture the variability in relative enhancement in individual cases because they quantified enhancement as average values for tissue types across multiple lesions in multiple patients. In our study, reasons for the variation are not related to lesion size, location, or multigland disease status. However, other possible factors that we did not measure are differences in a patient’s cardiac output (affecting the delivery of contrast material to the adenoma), local blood supply to the parathyroid lesion, and image noise related to patient body habitus. Given concerns about radiation exposure from multiphase imaging, some authors have advocated omission of the non–contrast-enhanced phase to reduce radiation exposure, while others have suggested using only the arterial phase (4–7). Although it may be true that lesions can still be detected with fewer phases, this study demonstrates that the use of three phases captures all possible relative enhancement patterns. Reducing the 4D CT protocol to fewer than three phases could reduce confidence and even lead to missed lesions because the identification of type A pattern lesions depends on an arterial phase, type B pattern lesions require a delayed phase, and type C pattern lesions mimic thyroid tissue without non–contrast-enhanced imaging. With fewer than three phases, the 8
radiologist may report a lesion as “a possible parathyroid adenoma” rather than “consistent with a parathyroid adenoma,” which makes a considerable difference to the surgeon if minimally invasive parathyroidectomy is planned and if 4D CT is one of the first-line modalities. Two recent studies advising that the 4D CT protocol be reduced to two phases did not evaluate diagnostic confidence in interpretation of the scans (7) or consisted of patients in whom 4D CT was performed as a third-line modality (8). This study had several limitations. First, the study had selection bias because we included only patients who had surgically proven parathyroid adenoma or hyperplasia. This may overestimate the sensitivity of 4D CT because patients who did not undergo surgery may be more likely to have a negative or indeterminate result on 4D CT scans. Exclusion of patients without surgery may also affect the prevalence of relative enhancement because lesions in nonsurgical patient may be smaller and difficult to categorize by enhancement. Second, this was a retrospective study at a single institution with a single neuroradiologist reading all 4D CT studies. It would be ideal to determine sensitivity of 4D CT by a blinded review by more than one reader without knowledge of prior surgery or imaging results. However, given that only one radiologist interprets all 4D CT images at our institution, there would be recall bias if the images were reinterpreted. The performance of 4D CT could differ at other institutions with more readers and less experienced readers, but our sensitivity for diagnosis was similar to that reported in other studies (13–15). The experience of the reader did not affect the classification of relative enhancement patterns; a second reader with only 4 years of CT experience also categorized enhancement, and the interrater agreement was high and verified by the attenuation measurements. Third, the attenuation measurements were made by one radiologist and may differ from measurements
Bahl et al
made by another reader, as the thyroid gland can be heterogeneous in attenuation and the parathyroid lesion may have cystic components that should not be included in the region of interest. Despite this, the objective measures of attenuation correlated well with the subjective categorization of relative enhancement patterns. Finally, our study is limited by having very few false-positive lesions, but it does suggest that different types of enhancement may have different degrees of diagnostic confidence. Because all five false-positive lesions had the type C pattern, it appears that type A and B patterns may be more specific for parathyroid adenomas. In a future study, these relative enhancement patterns, combined with morphologic features of parathyroid lesions (such as size, polar vessel, or cystic component), could be valuable in formulating a confidence score (2,16). The implication would be that if a lesion is confidently consistent with a parathyroid adenoma on 4D CT scans, additional imaging with other modalities may not be necessary, thereby reducing the cost of diagnostic evaluation and minimizing radiation exposure (2,16,17). In conclusion, parathyroid lesions can be grouped into three distinct relative enhancement patterns based on attenuation compared with the thyroid gland, which highlights the importance of a protocol with a non–contrast-enhanced and at least two contrast-enhanced phases. The most common pattern, type B, accounted for more than half of all lesions and could be diagnosed with two contrast-enhanced phases alone. However, almost one quarter of lesions had the type C pattern and thus could be challenging to detect with confidence if not for the non–contrast-enhanced phase. Radiologists should recognize these relative enhancement patterns in interpreting 4D CT images and designing 4D CT protocols at their institutions. Disclosures of Conflicts of Interest: M.B. disclosed no relevant relationships. A.R.S. disclosed no relevant relationships. J.A.S. disclosed no relevant relationships. J.K.H. disclosed no relevant relationships.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
References 1. Rodgers SE, Hunter GJ, Hamberg LM, et al. Improved preoperative planning for directed parathyroidectomy with 4-dimensional computed tomography. Surgery 2006;140(6): 932–940; discussion 940–941. 2. Hoang JK, Sung WK, Bahl M, Phillips CD. How to perform parathyroid 4D CT: tips and traps for technique and interpretation. Radiology 2014;270(1):15–24. 3. Beland MD, Mayo-Smith WW, Grand DJ, Machan JT, Monchik JM. Dynamic MDCT for localization of occult parathyroid adenomas in 26 patients with primary hyperparathyroidism. AJR Am J Roentgenol 2011;196(1):61–65. 4. Gafton AR, Glastonbury CM, Eastwood JD, Hoang JK. Parathyroid lesions: characterization with dual-phase arterial and venous enhanced CT of the neck. AJNR Am J Neuroradiol 2012;33(5):949–952. 5. van Dalen A, Smit CP, van Vroonhoven TJ, Burger H, de Lange EE. Minimally invasive surgery for solitary parathyroid adenomas in patients with primary hyperparathyroidism: role of US with supplemental CT. Radiology 2001;220(3):631–639. 6. Randall GJ, Zald PB, Cohen JI, Hamilton BE. Contrast-enhanced MDCT characteristics of parathyroid adenomas. AJR Am J Roentgenol 2009;193(2):W139–W143.
7. Raghavan P, Durst CR, Ornan DA, et al. Dynamic CT for parathyroid disease: are multiple phases necessary? AJNR Am J Neuroradiol 2014;35(10):1959–1964. 8. Noureldine SI, Aygun N, Walden MJ, Hassoon A, Gujar SK, Tufano RP. Multiphase computed tomography for localization of parathyroid disease in patients with primary hyperparathyroidism: how many phases do we really need? Surgery 2014;156(6):1300– 1306; discussion 13006–13007. 9. Hoang JK, Reiman RE, Nguyen GB, et al. Lifetime attributable risk of cancer from radiation exposure during parathyroid imaging: comparison of 4DCT and parathyroid scintigraphy. AJR Am J Roentgenol 2015; 204(5):W579–W585. 10. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33(1):159–174. 11. Vu TH, Guha-Thakurta N, Harrell RK, et al. Imaging characteristics of hyperfunctioning parathyroid adenomas using multiphase multidectector computed tomography: a quantitative and qualitative approach. J Comput Assist Tomogr 2011;35(5):560–567. 12. Hunter GJ, Ginat DT, Kelly HR, Halpern EF, Hamberg LM. Discriminating parathyroid adenoma from local mimics by using inherent tissue attenuation and vascular information obtained with four-dimensional CT: for-
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Bahl et al
mulation of a multinomial logistic regression model. Radiology 2014;270(1):168–175. 13. Hunter GJ, Schellingerhout D, Vu TH, Perrier ND, Hamberg LM. Accuracy of fourdimensional CT for the localization of abnormal parathyroid glands in patients with primary hyperparathyroidism. Radiology 2012;264(3):789–795. 14. Brown SJ, Lee JC, Christie J, et al. Four-dimensional computed tomography for parathyroid localization: a new imaging modality. ANZ J Surg 2014 Mar 27. [Epub ahead of print] 15. Cheung K, Wang TS, Farrokhyar F, Roman SA, Sosa JA. A meta-analysis of preoperative localization techniques for patients with primary hyperparathyroidism. Ann Surg Oncol 2012;19(2):577–583. 16. Bahl M, Muzaffar M, Vij G, Sosa JA, Choudhury KR, Hoang JK. Prevalence of the polar vessel sign in parathyroid adenomas on the arterial phase of 4D CT. AJNR Am J Neuroradiol 2014;35(3):578–581. 17. Harris R, Ryu H, Vu T, et al. Modern approach to surgical intervention of the thyroid and parathyroid glands. Semin Ultrasound CT MR 2012;33(2):115–122.
9
Purpose:
To describe the prevalence of three relative enhancement patterns of parathyroid lesions on four-dimensional (4D) computed tomographic (CT) scans.
Materials and Methods:
The institutional review board approved this HIPAA-compliant study and waived the need for informed consent. The authors retrospectively reviewed preoperative 4D CT scans obtained from November 2012 to June 2014 in 94 patients with pathologically proven parathyroid adenomas or hyperplasia. Lesions were classified into one of three relative enhancement patterns. All patterns required lesions to be lower in attenuation than the thyroid on non–contrast materialenhanced images, but patterns differed in the two contrastenhanced phases. Type A lesions were higher in attenuation than the thyroid in the arterial phase, type B lesions were not higher in attenuation than the thyroid in the arterial phase but were lower in attenuation than the thyroid in the delayed phase, and type C lesions were neither higher in attenuation than the thyroid in the arterial phase nor lower in attenuation than the thyroid in the delayed phase. The prevalence of the relative enhancement patterns was compared. The t test was used to compare mean attenuation differences in Hounsfield units between the relative enhancement patterns.
Results:
Ninety-four patients had 110 parathyroid lesions, including 11 patients with multigland disease. The sensitivity for single-gland disease was 94% (78 of 83) and that for multigland disease was 59% (16 of 27). Type B enhancement was most common, with a prevalence of 57% (54 of 94), followed by type C (22% [21 of 94]) and type A (20% [19 of 94]). Five lesions were interpreted incorrectly as parathyroid adenoma (false-positive), and all lesions had the type C pattern. Relative to the thyroid, lesions categorized as type A by readers had mean attenuation difference (6 standard deviation) of 39 HU 6 13 in the arterial phase, and type B lesions had a difference of 258 HU 6 26 in the delayed phase. These values differed from the mean attenuation difference of lesions not in these categories (P , .001).
Conclusion:
Parathyroid adenomas and hyperplasia can be grouped into three relative enhancement patterns based on a protocol with a non–contrast-enhanced and two contrast-enhanced phases. The type B pattern is most common and could be diagnosed with two contrast-enhanced phases. However, almost one quarter of lesions have the type C pattern and thus could be missed without the non–contrast-enhanced phase.
1
From the Department of Radiology, Division of Neuroradiology, Duke University Medical Center, Box 3808, Erwin Rd, Durham, NC 27710. Received October 10, 2014; revision requested December 8; final revision received February 6, 2015; accepted March 2; final version accepted March 13. Address correspondence to J.K.H. (e-mail: jennykh@ gmail.com). RSNA, 2015
q
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
RSNA, 2015
q
1
and Neck Imaging
Manisha Bahl, MD, MPH Ali R. Sepahdari, MD Julie A. Sosa, MD, MA Jenny K. Hoang, MBBS
Original Research n Head
Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans: Three Patterns of Enhancement Relative to the Thyroid Gland Justify a Three-Phase Protocol1
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
F
our-dimensional (4D) computed tomography (CT) of the parathyroid is a technique for preoperative localization of parathyroid adenomas. It involves multi–detector row CT of the neck and upper chest during two or more contrast material–enhanced phases (1,2). Multiple phases are performed to determine the pattern of enhancement, which is one of the primary features used to identify a parathyroid adenoma and to distinguish
Advances in Knowledge nn The enhancement pattern of parathyroid adenomas and hyperplasia on four-dimensional (4D) CT scans can be grouped into three types on the basis of relative attenuation compared with that of the thyroid gland; each type is lower in attenuation than the thyroid on nonenhanced images, but patterns differ according to the arterial and delayed phase findings. nn Type A lesions have higher attenuation than the thyroid in the arterial phase; only 20% (19 of 94) of parathyroid lesions had this enhancement pattern. nn Type B lesions are not higher in attenuation than the thyroid in the arterial phase but are lower in attenuation than the thyroid in the delayed phase; more than half of lesions demonstrated the type B pattern (57% [54 of 94]). nn Type C lesions are neither higher in attenuation than the thyroid in the arterial phase nor lower in attenuation than the thyroid in the delayed phase; type C is the second most common pattern (22% [21 of 94]) and could be missed without the nonenhanced phase. nn All five false-positive lesions had the type C pattern. nn The sensitivity of 4D CT for single-gland disease was 94% and that for multigland disease was 59%; overall sensitivity for all lesions was 85%. 2
it from mimics, such as thyroid nodules and lymph nodes. The characteristic description of a parathyroid adenoma on 4D CT scan is a lesion with lower attenuation than that of the thyroid gland in the non–contrastenhanced phase, peak enhancement greater than that of the thyroid gland in the arterial phase, and washout of contrast material in the delayed phase (2–4). However, the characteristic enhancement pattern may not be evident in all cases, and failure to appreciate the variability in enhancement pattern may lead to missed lesions. In our clinical experience, we have observed three relative enhancement patterns of parathyroid lesions compared with the thyroid gland on 4D CT scans. To our knowledge, there is no literature about variable enhancement patterns of parathyroid lesions. Such studies would be valuable to radiologists
Implications for Patient Care nn Radiologists should be aware of the three relative enhancement patterns of parathyroid adenomas and hyperplasia when interpreting 4D CT studies because not all adenomas are best seen in the arterial phase, as previously described in the literature. nn In designing 4D CT protocols, it is important to recognize that categorization of relative enhancement patterns in adenomas uses all phases of CT in a threephase protocol. nn Lesions with a type C relative enhancement pattern may mimic thyroid tissue without a nonenhanced phase because they are not higher in attenuation than the thyroid in the arterial phase or lower in attenuation than the thyroid in the delayed phase. nn Omitting the nonenhanced phase from the 4D CT protocol could lead to missed lesions with type C enhancement and an increase in the rate of false-positive lesions.
Bahl et al
who are new to interpreting 4D CT images and for justifying the optimal number of phases for 4D CT (4–8). The aim of this study was to describe the prevalence of three types of enhancement patterns of parathyroid adenomas and hyperplasia on 4D CT scans.
Materials and Methods Patients Our institutional review board approved this study and waived the need for informed consent. The studied complied with the Health Insurance Portability and Accountability Act. We performed a retrospective study of patients with parathyroid adenomas and hyperplasia who underwent preoperative 4D CT of the parathyroid, with at least two postcontrast phases, between November 2012 and June 2014. Patients were identified by searching radiology text reports for CT scans with the term parathyroid. A total of 133 patients were identified. Patients were excluded if they did not have surgery (n = 34 [26%]), did not have a three-phase study (n = 3 [2%]), or had parathyroid carcinoma (n = 2 [2%]). Thus, the final cohort consisted of 94 patients. All patients had biochemical evidence of hyperparathyroidism (elevated calcium and parathyroid hormone levels). Fifty of 94 patients (53%) underwent ultrasonography before 4D CT, and 38 of Published online before print 10.1148/radiol.2015142393 Content codes: Radiology 2015; 000:1–9 Abbreviation: 4D = four-dimensional Author contributions: Guarantors of integrity of entire study, M.B., J.K.H.; 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; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, M.B., A.R.S., J.K.H.; clinical studies, M.B., A.R.S., J.A.S.; experimental studies, M.B., J.K.H.; statistical analysis, M.B., J.K.H.; and manuscript editing, all authors Conflicts of interest are listed at the end of this article.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
94 (40%) underwent nuclear medicine scintigraphy. Hospital electronic medical records were accessed to obtain the following information: patient age and sex, surgical pathology reports, and surgical reports. Surgical pathology reports were reviewed for diagnosis (adenoma vs hyperplasia), weight, and size. If more than one parathyroid lesion was identified at surgery, these lesions were reported as parathyroid hyperplasia.
Four-dimensional CT Imaging Technique Four-dimensional CT was performed with 64-row multi–row detector CT scanners (Discovery CT750 HD and LightSpeed VCT; GE Healthcare, Milwaukee, Wis). The first phase was nonenhanced to cover the thyroid gland, with the z-axis from the hyoid bone to the clavicular head. The next two phases were contrast material–enhanced phases from the angle of the mandible to the carina, performed after intravenous administration of 75 mL of iopamidol (Isovue-300; Bracco, Princeton, NJ) via a 20-gauge cannula in a right antecubital vein at a rate of 4 mL/sec, followed by a 25-mL saline chaser. Arterial phase images were acquired 25 seconds after the start of the injection, and delayed (venous) phase images were acquired 80 seconds from the start of the injection. The parameters for all three phases were as follows: detector configuration, 64 3 0.625 mm; tube rotation time, 0.4 second; pitch factor, 0.516:1; field of view, 20 cm; 120 kV (peak); and automatic tube current modulation (Smart mA; GE Healthcare) (noise index, 8; minimum of 100 mA and maximum of 700 mA for arterial phase, minimum of 100 mA and maximum of 500 mA for the other two phases to reduce radiation exposure). The 0.625-mmthick contiguous axial images in all three phases were sent to the picture archiving and communication system for interpretation and to allow manipulation of images on a three-dimensional workstation. Reformatted images in the arterial phases were sent to the picture archiving and communication system as 2.5-mm-thick contiguous images in the axial, coronal, and sagittal planes. On
the basis of a prior phantom study that measured organ doses with dosimeters, scanning a 73-kg patient with this protocol would result in dose length product of 1928 mGy · cm and calculated effective dose from organ doses of 28 mSv (9).
Image Interpretation Accuracy of 4D CT for localization of parathyroid adenomas.—The 4D CT images were reviewed on a picture archiving and communication system workstation (Centricity; GE Healthcare). One radiologist (with 12 years of experience) reviewed the 4D CT images before surgery to localize the parathyroid adenomas and generate the radiology report. Interpretation of 4D CT scans was considered correct if it enabled localization of the adenoma to the correct side and quadrant or to its ectopic location and if it approximated the size of the surgical specimen. If a lesion was reported in the impression section of the radiology report, this was regarded as a positive interpretation on 4D CT scans and this lesion was correlated with the surgical and pathology notes. Lesions that were not confirmed to be parathyroid adenomas or hyperplastic glands at surgery were considered to be false-positive cases. For false-negative cases (missed lesions), a radiologist (J.K.H., fellowshiptrained neuroradiologist with 12 years of experience) was given information about the location of the missed lesion at surgery, and images were reviewed again to determine whether a missed lesion could be seen in the exact quadrant. A second reader (M.B., a radiology resident with 4 years of experience) was required to agree with the findings and determine that the size at pathologic assessment matched the CT size. The retrospectively identified false-negative lesions were analyzed for relative enhancement patterns; however, they were not included in the final analysis with the true-positive and false-positive lesions because they were not seen at the original image interpretation. Classification of relative enhancement patterns.—Lesions were classified according to three relative enhancement
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Bahl et al
Figure 1
Figure 1: Classification of relative enhancement patterns of parathyroid adenomas and hyperplasia on 4D CT scans.
patterns by the two readers. The readers assigned the patterns subjectively and did not measure attenuation values. Images were independently reviewed on a picture archiving and communication system station. The readers could adjust the window settings if required. All three relative enhancement patterns required the adenoma to be lower in attenuation than the thyroid gland on nonenhanced images (Figs 1–4). In addition, with type A, the lesion was higher in attenuation than the thyroid gland in the arterial phase but could have any attenuation in the delayed phase (Fig 2). With type B, the lesion was not higher in attenuation than the thyroid gland in the arterial phase but was lower in attenuation than the thyroid gland in the delayed phase (Fig 3). With type C, the lesion was not higher in attenuation than the thyroid gland in the arterial phase and not lower in attenuation than the thyroid gland in the delayed phase (Fig 4). When the two readers disagreed about the enhancement classification, a consensus was reached by measurement of attenuation values in Hounsfield units. 3
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
at least 10 mm2 (when possible) with exclusion of cystic areas. To quantify the attenuation differences appreciated by the readers, the attenuation differences between the parathyroid lesions and the thyroid gland for each lesion were calculated by subtracting CT attenuation values of the thyroid from the CT attenuation values of the parathyroid lesion.
Figure 2
Figure 2: Axial 4D CT images (window level, 75 HU; width, 350 HU) in a 49-year-old woman with an inferior left parathyroid adenoma. (a) Non–contrast-enhanced image demonstrates a lesion (arrow) measuring up to 12 mm posterior to the inferior left thyroid lobe, adjacent to the trachea and medial to the common carotid artery. The lesion is lower in attenuation than the adjacent thyroid gland. (b) Arterial phase image demonstrates that the lesion (arrow) is higher in attenuation than the adjacent thyroid gland. This pattern is consistent with type A. (c) Delayed phase image demonstrates washout of the lesion (arrow) with similar attenuation to the thyroid gland.
After assigning categories of relative enhancement, a radiologist (M.B.) measured the attenuation values for the 4
Bahl et al
thyroid gland and parathyroid lesions for each imaging phase using the mean of three ellipsoid regions of interest of
Outcomes and Statistical Analysis The sensitivity of 4D CT for localization of parathyroid lesions was calculated on the basis of the original radiology report. The prevalence of enhancement patterns was calculated for pathologically proven parathyroid lesions that were correctly identified on the original radiology report. The characteristics of parathyroid lesions were compared for the three relative enhancement patterns. These characteristics included lesion size, ectopic location, single versus multigland disease, and prior failed surgery. Analysis of variance was used to compare continuous data. The x2 test or Fisher exact test was used for categorical data. The Fisher exact test was used when expected cell frequencies were fewer than five. For differences in enhancement in patients with multigland disease, a logistic regression model was fitted by using generalized estimating equations to account for multiple lesions in the same patient. Using the measured CT attenuation values for the thyroid gland and parathyroid lesions, the attenuation difference for each parathyroid lesion was calculated by subtracting the attenuation value of the thyroid gland from the attenuation value of the parathyroid lesion. The mean and standard deviation of the attenuation differences were then calculated for the three relative enhancement patterns. The t test was used to compare the mean of attenuation differences between the relative enhancement patterns. The mean of attenuation differences in the arterial phase for type A lesions (attenuation greater than that of the thyroid) was compared with the mean of attenuation differences for type B and C lesions
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
Figure 3
Bahl et al
(attenuation not greater than that of the thyroid). The mean of attenuation differences in the delayed phase for type B lesions (attenuation less than that of the thyroid) was compared with the mean of attenuation differences for type C lesions (attenuation not less than that of the thyroid). Type B lesions were not compared with type A lesions for the delayed phase because type A lesions could have any attenuation relative to the thyroid in the delayed phase. The Fleiss k value was calculated to determine interrater agreement for the enhancement pattern type. Agreement was regarded as poor with Fleiss k 0.20, as slight with k of 0.21–0.40, as moderate with k of 0.41–0.60, as substantial with k of 0.61–0.80, and as almost perfect with k of 0.81–1.00 (10). The data were entered into an Excel spreadsheet (2007 version; Microsoft, Redmond, Wash). Statistical analyses were performed by using SAS Enterprise (version 4.2; SAS Institute, Cary, NC) and R computing platform (www.r-project.org). P values of , .05 were considered to indicate statistically significant results.
Results Patients Of 94 patients (67 women; mean age, 63 years; range, 34–88 years) with pathologically proven parathyroid lesions, 83 (88%) had single-gland disease and 11 (12%) had multigland disease, resulting in a total of 110 lesions. The 110 lesions had a mean (6 standard deviation) maximal diameter at pathologic evaluation of 15.9 mm 6 6.8 and mean weight of 0.58 g 6 0.57.
Figure 3: Axial 4D CT images (window level, 75 HU; width, 350 HU) in a 50-year-old man with a superior left parathyroid adenoma. (a) Non–contrast-enhanced image demonstrates a lesion (arrow) measuring up to 7 mm posterior to the superior left thyroid lobe. The lesion is lower in attenuation than the adjacent thyroid gland. (b) Arterial phase image demonstrates that the lesion (arrow) is similar in attenuation to the adjacent thyroid gland. This pattern is not consistent with type A. (c) Delayed phase image demonstrates washout of the lesion (arrow) with attenuation lower than that of the adjacent thyroid gland. This pattern is consistent with type B. Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Accuracy of 4D CT for Localization of Parathyroid Adenoma In 78 of 83 patients with single-gland disease, lesions were correctly localized with 4D CT (sensitivity, 94%; 95% confidence interval: 89%, 99%). Eleven patients with multigland disease had 27 hyperplastic glands, 16 of which were correctly localized with 4D CT (sensitivity, 59%; 95% confidence interval: 41%, 78%). Of 110 total lesions, 94 5
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
lesions (85%) were identified (78 single-gland and 16 multigland) and 16 lesions (15%) were missed (five singlegland and 11 multigland). Among the patients with multigland disease, two or more glands were correctly identified in five patients. The overall sensitivity for single-gland and multigland disease was thus 85% (94 of 110; 95% confidence interval: 79%, 92%). There were five false-positive lesions in four patients; these represent lesions reported by the radiologist that were not confirmed to be adenomas at surgery in patients with a biochemical cure. The mean maximal diameter at pathologic evaluation of the 16 missed lesions was 11.9 mm 6 4.8.
Figure 4
Figure 4: Axial 4D CT images (window level, 75 HU; width, 350 HU) in a 63-year-old man with a right parathyroid adenoma. (a) Non–contrast-enhanced image demonstrates a lesion (arrow) measuring up to 12 mm posterior to the midportion of the right thyroid lobe. The lesion is lower in attenuation than the adjacent thyroid gland. (b) Arterial phase image demonstrates that the lesion (arrow) is similar in attenuation to the adjacent thyroid gland. (c) Delayed phase image demonstrates that the lesion (arrow) is similar in attenuation to the adjacent thyroid gland. This pattern is consistent with type C. 6
Bahl et al
Classification of Relative Enhancement Patterns Ninety-four parathyroid lesions detected at initial review and five false-positive lesions were categorized according to the relative enhancement pattern. The characteristic pattern of enhancement (type A) was present in 20% (19 of 94) of parathyroid lesions. Type B was seen in 57% (54 of 94) and type C lesions in 22% (21 of 94). All five falsepositive lesions were type C. There was no statistically significant difference in lesion size, ectopic location, multigland disease status, and prior failed surgery for the three relative enhancement patterns (Table 1). Table 2 summarizes the measured attenuation and mean of attenuation differences between the thyroid gland and parathyroid lesions. Type A lesions were categorized by readers as higher in attenuation than the thyroid gland in the arterial phase. The mean of attenuation differences between the thyroid gland and type A lesions in the arterial phase was 39 HU 6 13 compared with 232 HU 6 36 for lesions not classified as type A (P , .001). Type B lesions were categorized by readers as lower in attenuation than the thyroid gland in the delayed phase. The mean of attenuation differences between the thyroid gland and type B lesions in the delayed phase was 258 HU 6 26 compared with 25 HU 6 19 for type C lesions (not lower in attenuation than the thyroid in the delayed phase) (P , .001).
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
The two readers agreed on the enhancement categorization in 89 of 99 lesions (90%). The Fleiss k (an index of interrater agreement) was 0.85 (95% confidence interval: 0.76, 0.94), which indicates almost perfect agreement. In the 10 cases of disagreement between the two readers, the attenuation differences between the thyroid gland and
the candidate lesion ranged from 1 to 34 HU. Nine of 16 surgically confirmed lesions that were missed on 4D CT scans (false-negative results) were identified retrospectively. The mean maximal diameter of false-negative lesions was 9.2 mm 6 3.5 on CT scans and 11.2 mm 6 3.8 at pathologic evaluation. One lesion
Table 1 Characteristics of Parathyroid Adenoma or Hyperplasia according to Relative Enhancement Patterns Characteristic Adenoma or hyperplasia Single-gland disease Multigland disease False-positive lesions Mean maximal diameter at CT (mm)# Mean maximal diameter at pathologic evaluation (mm)# Mean weight (g)# Ectopic location‡‡ History of unsuccessful surgery
All Lesions
Type A
Type B
Type C
P Value
94* 78† 16§ 5 13.2 6 7.8
19 (20) 16 (21) 3 (19) 0 12.0 6 5.6
54 (57) 45 (58) 9 (56) 0 14.0 6 8.6
21 (22) 17 (22) 4 (25) 5 (100) 10.1 6 4.4
.96‡ .62|| ,.001‡ .11**
16.8 6 7.1
16.0 6 5.8
17.3 6 7.6
15.2 6 5.7
.48**
0.64 6 0.61 8 12
0.55 6 0.66 1 (13) 4 (33)
0.67 6 0.64 5 (63) 7 (58)
0.59 6 0.46 2 (26) 1 (8)
.74** .85‡ .31‡
Note.—Unless otherwise indicated, values are expressed as number of lesions, with percentages in parentheses. Percentages may not add up to 100% owing to rounding. Mean values are accompanied by standard deviation. * Sixteen of 110 surgically confirmed parathyroid lesions were not identified on 4D CT scans before surgery. †
Five of 83 surgically confirmed parathyroid lesions were not identified on 4D CT scans before surgery.
‡
Calculated with the Fisher exact test to test difference in proportions for the enhancement patterns.
§
Eleven of 27 surgically confirmed parathyroid lesions were not identified on 4D CT scans before surgery.
||
Calculated by fitting a logistic regression model and using generalized estimating equations to account for correlations due to multiple lesions measured in five of 94 patients. #
Data are means 6 standard deviation.
** Calculated with analysis of variance test to test difference in lesion size for the enhancement patterns. ††
Five were mediastinal, two retropharyngeal, and one parapharyngeal in location.
Bahl et al
had the type A enhancement pattern, five had the type B pattern, and three had the type C pattern.
Discussion In this study, we describe an enhancement categorization method for parathyroid adenomas based on three enhancement patterns relative to the thyroid tissue. The most common pattern, type B, accounted for more than half of all lesions and could be diagnosed confidently with two contrastenhanced CT phases alone. However, almost one quarter of lesions had the type C pattern, which could be challenging to differentiate from thyroid tissue without the nonenhanced phase. Previous studies have quantified the enhancement of adenomas and its mimics with attenuation values (3,4,6,11,12). Beland et al (3) compared enhancement of lymph nodes with that of adenomas and found adenomas to be significantly higher in attenuation at 30 and 60 seconds relative to other phases. Vu et al (11) found that the percentage change in attenuation from baseline to the arterial phase was the most powerful discriminatory feature for parathyroid adenomas and that contrast material washout from arterial peak to the venous phase was a less powerful discriminator. Hunter et al (12) proposed a multinomial logistic regression model based on inherent tissue attenuation and vascular
Table 2 Attenuation Values and Mean of Differences between Thyroid Gland and Parathyroid Lesions in Nonenhanced, Arterial, and Delayed Phases Nonenhanced Phase Type Type A Type B Type C
Arterial Phase
Delayed Phase
No. of Lesions
Thyroid*
Lesion*
Mean of Differences†
Thyroid*
Lesion*
Mean of Differences†
Thyroid*
Lesion*
Mean of Differences†
19 54 21
99 6 20 101 6 25 97 6 31
54 6 21 40 6 17 40 6 30
245 6 27 261 6 31 257 6 22
179 6 62 184 6 39 175 6 47
217 6 60 150 6 43 148 6 33
39 6 13 234 6 35 227 6 39
142 6 29 139 6 22 113 6 27
110 6 21 81 6 27 108 6 29
232 6 20 258 6 26 25 6 19
Note.—Categorization of relative enhancement patterns is based on subjective visual comparison of parathyroid lesions with the thyroid without performing attenuation measurements. * Data are mean attenuation values expressed in Hounsfield units 6 standard deviation. †
Attenuation difference for each lesion equals attenuation of parathyroid lesion minus attenuation of thyroid gland. The mean and standard deviation of the attenuation differences were calculated for the three relative enhancement patterns.
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
7
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
information that could be used to estimate the probability that a lesion is a parathyroid adenoma. All of these methods involve placing regions of interest to measure attenuation and performing additional calculations to quantify change between phases. In practice, these measurements can be tedious for radiologists. The advantage of our relative enhancement classification system is that it consists of three simple categories that compare attenuation of the suspected parathyroid lesion with that of the adjacent thyroid without the need to quantify attenuation with measurements. Deviation of enhancement in parathyroid lesions from the typical description is valuable information to radiologists who are new to interpreting 4D CT images. Prior studies did not capture the variability in relative enhancement in individual cases because they quantified enhancement as average values for tissue types across multiple lesions in multiple patients. In our study, reasons for the variation are not related to lesion size, location, or multigland disease status. However, other possible factors that we did not measure are differences in a patient’s cardiac output (affecting the delivery of contrast material to the adenoma), local blood supply to the parathyroid lesion, and image noise related to patient body habitus. Given concerns about radiation exposure from multiphase imaging, some authors have advocated omission of the non–contrast-enhanced phase to reduce radiation exposure, while others have suggested using only the arterial phase (4–7). Although it may be true that lesions can still be detected with fewer phases, this study demonstrates that the use of three phases captures all possible relative enhancement patterns. Reducing the 4D CT protocol to fewer than three phases could reduce confidence and even lead to missed lesions because the identification of type A pattern lesions depends on an arterial phase, type B pattern lesions require a delayed phase, and type C pattern lesions mimic thyroid tissue without non–contrast-enhanced imaging. With fewer than three phases, the 8
radiologist may report a lesion as “a possible parathyroid adenoma” rather than “consistent with a parathyroid adenoma,” which makes a considerable difference to the surgeon if minimally invasive parathyroidectomy is planned and if 4D CT is one of the first-line modalities. Two recent studies advising that the 4D CT protocol be reduced to two phases did not evaluate diagnostic confidence in interpretation of the scans (7) or consisted of patients in whom 4D CT was performed as a third-line modality (8). This study had several limitations. First, the study had selection bias because we included only patients who had surgically proven parathyroid adenoma or hyperplasia. This may overestimate the sensitivity of 4D CT because patients who did not undergo surgery may be more likely to have a negative or indeterminate result on 4D CT scans. Exclusion of patients without surgery may also affect the prevalence of relative enhancement because lesions in nonsurgical patient may be smaller and difficult to categorize by enhancement. Second, this was a retrospective study at a single institution with a single neuroradiologist reading all 4D CT studies. It would be ideal to determine sensitivity of 4D CT by a blinded review by more than one reader without knowledge of prior surgery or imaging results. However, given that only one radiologist interprets all 4D CT images at our institution, there would be recall bias if the images were reinterpreted. The performance of 4D CT could differ at other institutions with more readers and less experienced readers, but our sensitivity for diagnosis was similar to that reported in other studies (13–15). The experience of the reader did not affect the classification of relative enhancement patterns; a second reader with only 4 years of CT experience also categorized enhancement, and the interrater agreement was high and verified by the attenuation measurements. Third, the attenuation measurements were made by one radiologist and may differ from measurements
Bahl et al
made by another reader, as the thyroid gland can be heterogeneous in attenuation and the parathyroid lesion may have cystic components that should not be included in the region of interest. Despite this, the objective measures of attenuation correlated well with the subjective categorization of relative enhancement patterns. Finally, our study is limited by having very few false-positive lesions, but it does suggest that different types of enhancement may have different degrees of diagnostic confidence. Because all five false-positive lesions had the type C pattern, it appears that type A and B patterns may be more specific for parathyroid adenomas. In a future study, these relative enhancement patterns, combined with morphologic features of parathyroid lesions (such as size, polar vessel, or cystic component), could be valuable in formulating a confidence score (2,16). The implication would be that if a lesion is confidently consistent with a parathyroid adenoma on 4D CT scans, additional imaging with other modalities may not be necessary, thereby reducing the cost of diagnostic evaluation and minimizing radiation exposure (2,16,17). In conclusion, parathyroid lesions can be grouped into three distinct relative enhancement patterns based on attenuation compared with the thyroid gland, which highlights the importance of a protocol with a non–contrast-enhanced and at least two contrast-enhanced phases. The most common pattern, type B, accounted for more than half of all lesions and could be diagnosed with two contrast-enhanced phases alone. However, almost one quarter of lesions had the type C pattern and thus could be challenging to detect with confidence if not for the non–contrast-enhanced phase. Radiologists should recognize these relative enhancement patterns in interpreting 4D CT images and designing 4D CT protocols at their institutions. Disclosures of Conflicts of Interest: M.B. disclosed no relevant relationships. A.R.S. disclosed no relevant relationships. J.A.S. disclosed no relevant relationships. J.K.H. disclosed no relevant relationships.
radiology.rsna.org n Radiology: Volume 000: Number 0— 2015
HEAD AND NECK IMAGING: Parathyroid Adenomas and Hyperplasia on Four-dimensional CT Scans
References 1. Rodgers SE, Hunter GJ, Hamberg LM, et al. Improved preoperative planning for directed parathyroidectomy with 4-dimensional computed tomography. Surgery 2006;140(6): 932–940; discussion 940–941. 2. Hoang JK, Sung WK, Bahl M, Phillips CD. How to perform parathyroid 4D CT: tips and traps for technique and interpretation. Radiology 2014;270(1):15–24. 3. Beland MD, Mayo-Smith WW, Grand DJ, Machan JT, Monchik JM. Dynamic MDCT for localization of occult parathyroid adenomas in 26 patients with primary hyperparathyroidism. AJR Am J Roentgenol 2011;196(1):61–65. 4. Gafton AR, Glastonbury CM, Eastwood JD, Hoang JK. Parathyroid lesions: characterization with dual-phase arterial and venous enhanced CT of the neck. AJNR Am J Neuroradiol 2012;33(5):949–952. 5. van Dalen A, Smit CP, van Vroonhoven TJ, Burger H, de Lange EE. Minimally invasive surgery for solitary parathyroid adenomas in patients with primary hyperparathyroidism: role of US with supplemental CT. Radiology 2001;220(3):631–639. 6. Randall GJ, Zald PB, Cohen JI, Hamilton BE. Contrast-enhanced MDCT characteristics of parathyroid adenomas. AJR Am J Roentgenol 2009;193(2):W139–W143.
7. Raghavan P, Durst CR, Ornan DA, et al. Dynamic CT for parathyroid disease: are multiple phases necessary? AJNR Am J Neuroradiol 2014;35(10):1959–1964. 8. Noureldine SI, Aygun N, Walden MJ, Hassoon A, Gujar SK, Tufano RP. Multiphase computed tomography for localization of parathyroid disease in patients with primary hyperparathyroidism: how many phases do we really need? Surgery 2014;156(6):1300– 1306; discussion 13006–13007. 9. Hoang JK, Reiman RE, Nguyen GB, et al. Lifetime attributable risk of cancer from radiation exposure during parathyroid imaging: comparison of 4DCT and parathyroid scintigraphy. AJR Am J Roentgenol 2015; 204(5):W579–W585. 10. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33(1):159–174. 11. Vu TH, Guha-Thakurta N, Harrell RK, et al. Imaging characteristics of hyperfunctioning parathyroid adenomas using multiphase multidectector computed tomography: a quantitative and qualitative approach. J Comput Assist Tomogr 2011;35(5):560–567. 12. Hunter GJ, Ginat DT, Kelly HR, Halpern EF, Hamberg LM. Discriminating parathyroid adenoma from local mimics by using inherent tissue attenuation and vascular information obtained with four-dimensional CT: for-
Radiology: Volume 000: Number 0— 2015 n radiology.rsna.org
Bahl et al
mulation of a multinomial logistic regression model. Radiology 2014;270(1):168–175. 13. Hunter GJ, Schellingerhout D, Vu TH, Perrier ND, Hamberg LM. Accuracy of fourdimensional CT for the localization of abnormal parathyroid glands in patients with primary hyperparathyroidism. Radiology 2012;264(3):789–795. 14. Brown SJ, Lee JC, Christie J, et al. Four-dimensional computed tomography for parathyroid localization: a new imaging modality. ANZ J Surg 2014 Mar 27. [Epub ahead of print] 15. Cheung K, Wang TS, Farrokhyar F, Roman SA, Sosa JA. A meta-analysis of preoperative localization techniques for patients with primary hyperparathyroidism. Ann Surg Oncol 2012;19(2):577–583. 16. Bahl M, Muzaffar M, Vij G, Sosa JA, Choudhury KR, Hoang JK. Prevalence of the polar vessel sign in parathyroid adenomas on the arterial phase of 4D CT. AJNR Am J Neuroradiol 2014;35(3):578–581. 17. Harris R, Ryu H, Vu T, et al. Modern approach to surgical intervention of the thyroid and parathyroid glands. Semin Ultrasound CT MR 2012;33(2):115–122.
9
