Downloaded from www.ajronline.org by Tufts University on 07/26/17 from IP address 130.64.11.153. Copyright ARRS. For personal use only; all rights reserved
Utility of MRI to Differentiate Clear Cell Renal Cell Carcinoma Adrenal Metastases From Adrenal Adenomas
Nicola Schieda1 Satheesh Krishna1 Matthew D. F. McInnes1 Bardia Moosavi1 Abdulmohsen Alrashed2 Robert Moreland1 Evan S. Siegelman3
Keywords: adenoma, adrenal, clear cell renal cell carcinoma, MRI, texture analysis
DOI:10.2214/AJR.16.17649
Received November 13, 2016; accepted after revision January 27, 2017.
1Department of Medical Imaging, The Ottawa Hospital, The University of Ottawa, 1053 Carling Ave, Ottawa, ON K1Y 4E9, Canada. Address correspondence to N. Schieda (nschieda@toh.on.ca).
2Department of Radiology and Medical Imaging, King Saud University Medical City, King Khaled University Hospital, Riyadh, Kingdom of Saudi Arabia.
3Department of Radiology, The Hospital of The University of Pennsylvania, The University of Pennsylvania, Philadelphia, PA.
WEB
This is a web exclusive article.
AJR2017; 209:W1-W8 0361-803X/17/2093-W1
@ American Roentgen Ray Society
OBJECTIVE. The purpose of this study is to compare MRI features of clear cell renal cell carcinoma (RCC) adrenal metastases and adenomas.
MATERIALS AND METHODS. Fifteen clear cell RCC adrenal metastases imaged with MRI were compared to 29 consecutive adenomas between 2006 and 2015. Two blinded radiologists assessed homogeneity (homogeneous vs heterogeneous), signal intensity (SI) de- crease on chemical-shift MRI, and T2-weighted SI (isointense, mildly hyperintense, or mark- edly hyperintense) relative to muscle. A third blinded radiologist measured the chemical-shift SI index, adrenal-to-spleen SI ratio, T2-weighted SI ratio, AUC for contrast-enhanced MRI, and histogram analysis. Analyses were performed using chi-square, linear regression, ROC, and logistic regression modeling.
RESULTS. Clear cell RCC metastases were larger than adenomas (mean [+ SD], 5.0 ± 4.2 cm [range, 1.1-15 cm] vs 2.0 ± 0.7 cm [range, 1.2-3.7 cm]; p < 0.0001). Subjectively, 33.3% (5/15) of metastases and 68.9% (20/29) of adenomas showed an SI decrease on chemical-shift MRI (p = 0.0421; K = 0.76). Chemical-shift SI index (mean, 9.2% + 20.6%; range, -30.0% to 57.9%) and adrenal-to-spleen SI ratio (0.94 + 0.23 [range, 0.44-1.33]) for metastases differed significantly from those for adenomas (47.3% + 27.8% [range, -9.4% to 86%] and 0.52 ± 0.28 [range, 0.13-1.11], respectively) (p <0.0001). Twenty percent (3/15) of metastases had chemi- cal-shift SI index in the adenoma range (> 16.5%). Metastases had higher T2-weighted SI than did adenomas, both quantitatively (5.1 + 3.0 [range, 1.5-10.6] vs 1.8 ± 0.8 [range, 0.5-3.8]; p < 0.0001) and subjectively (p < 0.0001; K = 0.89). Metastases had higher entropy than did ad- enomas (6.76 ± 0.61 vs 6.1 ± 0.74; p = 0.0051) and were subjectively more heterogeneous (p < 0.0001; K = 0.86). The contrast-enhanced MRI AUC, skewness, and kurtosis did not differ between groups (p>0.05). The ROC AUCs were 0.91 (95% CI, 0.79-1.0) for T2-weighted SI ratio and 0.85 (95% CI, 0.68-1.0) for entropy. The logistic regression model of T2-weighted SI ratio plus entropy improved accuracy (ROC AUC, 0.97; 95% CI, 0.93-1.0]) compared with either feature alone (p = 0.0215).
CONCLUSION. Increased T2-weighted SI and heterogeneity are features that can dif- ferentiate clear cell RCC adrenal metastases from adenomas using quantitative and subjec- tive analysis.
C lear cell renal cell carcinoma (RCC) is the most common ma- lignant renal tumor and is over- whelmingly the most common RCC subtype that can metastasize [1]. Clear cell RCC commonly metastasizes systemi- cally to the lung, bone, lymph nodes, liver, adrenal gland, and brain. Adrenal gland me- tastases from clear cell RCC are reported to occur in 5-10% of patients [2, 3] and may be ipsilateral or contralateral to the site of the primary renal tumor [4]. Metastases may be detected at first presentation or during imag- ing follow-up of disease. At first presentation
and in instances where no prior imaging is available to document growth, clear cell RCC metastases are difficult to differentiate from the more ubiquitous adrenal adenoma. At unenhanced CT, an adrenal nodule mea- suring less than 10 HU in attenuation is high- ly specific for adenoma; however, 30% of ad- enomas are lipid poor and will show attenuation values above 10 HU [5, 6]. Al- though most lipid-poor adenomas can be characterized using CT washout analysis [5, 6], clear cell RCC metastases are hypervas- cular and can also show washout of contrast agent in the adenoma range [7]. With MRI,
A
B
C
although the detection of decreases in signal intensity (SI) on opposed-phase compared with in-phase chemical-shift MRI is charac- teristic for the diagnosis of lipid-rich adrenal adenoma (because of the presence of intra- cellular lipid) [8, 9], clear cell RCC metasta- ses can also show an SI decrease on opposed- phase chemical-shift MRI due to intracellular lipid content [10-14]. It is estimated that 20% of clear cell RCC metastases may show an SI decrease on chemical-shift MRI, and this finding may not necessarily be present in the primary tumor [15], which could further confound interpretation.
The differentiation of adrenal adenomas from clear cell RCC adrenal metastases is critical. Surgical management with intent to cure for clear cell RCC attempts to remove all sites of tumor; therefore, mistaking a be- nign adenoma for an adrenal metastasis could
result in unnecessary adrenalectomy. Con- versely, a failure to diagnose metastatic dis- ease preoperatively would leave the patient with residual tumor postoperatively and ne- cessitate future intervention. The ability to distinguish adrenal metastases from clear cell RCC is, therefore, desirable because the ab- sence of reliable imaging features to differ- entiate them from adrenal adenomas typical- ly requires short-term follow-up imaging [8] (which may not be practical when planning surgery) or biopsy for histologic diagnosis.
Clear cell RCCs characteristically show increased SI at T2-weighted MRI [16, 17]; conversely, adrenal adenomas typically show low-to-intermediate SI at T2-weighted MRI [18]. Clear cell RCC have also been shown to be more heterogeneous when compared with benign lesions [19, 20]. Previously, Woo et al. [13] found that clear cell RCC
metastases were subjectively of higher SI at T2-weighted MRI (relative to liver parenchy- ma) compared with adrenal adenomas. Their method was limited by using only subjec- tive analysis (which introduces interobserv- er variability) and because liver was used as the comparison organ, which can be prob- lematic in the presence of hepatic steatosis (because fat will appear bright on fast spin- echo techniques due to disruption of J-cou- pling effects). Histogram and texture anal- ysis are previously described methods that can be used to quantitatively evaluate tu- mor characteristics beyond subjective visu- al analysis. Previous authors have used these methods when evaluating renal masses [19, 21, 22]; however, to our knowledge, quanti- tative analysis of T2-weighted MRI and MR histogram or texture features have not been previously compared between clear cell RCC
A
B
C
Use of MRI to Distinguish RCC Metastases From Adenoma
adrenal metastases and benign adrenal ade- nomas. The purpose of this study is to evalu- ate both quantitative and subjective MRI for the diagnosis of clear cell RCC adrenal me- tastases compared with adrenal adenomas.
Materials and Methods
Patients
This retrospective study was approved by the research ethics board of The Ottawa Hospi- tal Research Institute, which waived the need for informed consent. Between January 2006 and December 2015, we performed a consecutive ret- rospective search of our PACS database from a sin- gle institution for the terms “renal cell carcinoma or RCC” and “adrenal nodule or metastasis or me- tastases” under the search filter “MRI.” We identi- fied 17 patients who met the search criteria. All pa- tients underwent MRI for primary staging of RCC with at least one indeterminate adrenal nodule de- tected at previous CT examination. No patient had undergone surgery, chemotherapy, or radiotherapy before MRI. A diagnosis of metastatic disease was established using the following two criteria: histo- logic confirmation from adrenalectomy specimen
or 18-gauge core needle biopsy, or short-term in- terval (<6 month) growth of the nodule in the con- text of progression of metastatic disease (new me- tastases or > 20% increase in the sum of disease [23]) elsewhere in the same patient. Seven patients were excluded from the analysis because they did not have histologic or imaging follow-up to con- firm a diagnosis of metastatic disease. Ten patients met the inclusion criteria with 15 adrenal metas- tases (three patients with two metastases, one pa- tient with three metastases, and the other six pa- tients with a solitary metastasis) confirmed by histologic examination (n = 9) or growth (n = 6). The mean (+ SD) percentage increase in the size of adrenal nodules (measured by dividing the differ- ence in long-axis measurements between scans di- vided by the initial measurement) on follow-up was 49.3% ± 18.3% (range, 29.4-81.3%) with a mean interval between scans of 147 ± 31.6 days (range, 120-180 days).
For the control group, we used an existing da- tabase of 29 patients with diagnosis of adrenal ad- enoma who underwent MRI during the same study period who were previously evaluated for MRI fea- tures that could distinguish adenomas from pheo-
chromocytomas [24]. The reference standard for adenoma was established by the absence of a histo- ry of malignancy and negative biochemical testing for pheochromocytoma, a minimum 1-year inter- val with stability in size, and appropriate imaging findings including for lipid-rich adenomas (i.e., at- tenuation < 10 HU or chemical-shift SI index > 16.5%) and for lipid-poor adenomas (absolute and relative CT washout of > 60% and > 40%, respec- tively) [14].
MRI Technique
Thirty-eight of 39 patients underwent MRI at a single tertiary-care referral center. MRI was per- formed using one of three clinical 1.5- or 3-T sys- tems (n = 20, Symphony, Siemens Healthcare; n = 13, TRIO, Siemens Healthcare; n = 5, Discovery 750 W, GE Healthcare) using a torso phased-ar- ray coil (6-element anterior array for the TRIO, 4-element anterior array for the Symphony, and 16-element array for the Discovery 750 W). Pulse sequence parameters of T2-weighted imaging, T1-weighted chemical-shift in- and opposed- phase gradient-recalled echo (GRE) imaging, and unenhanced and gadolinium-enhanced fat-sup-
| Pulse Sequence | Dual-Echo T1-Weighted GRE | T2-Weighted TSE or FSE | Volume Interpolated T1-Weighted 3D GREª | ||||
|---|---|---|---|---|---|---|---|
| 2D GRE | 3D GRE | TSE or FSE | Single-Shot TSE or FSE | 3T | 1.5 T | ||
| 3T | 1.5 T | ||||||
| Physiology | Breath-hold | Breath-hold | Breath-hold | Respiratory triggered | Breath-hold | Breath-hold | |
| Duration | 21 s | 16 s | 20 s | 3-4 min | 22 s | 20 s | |
| Fat suppression | NA | NA | NA | NA | NA | Chemical or spectral inversion recovery | |
| TR (ms) | 160-180 | 5.5 and 4.0 | 7.6 | 5000 | 1030 | 4.0-4.5 | 4.3 |
| TE (ms) | 4.6 (in-phase), 2.3 (opposed-phase) | 2.5 (in-phase), 1.3 (opposed- phase); or 2.2 (in-phase), 1.1 (opposed-phase) | 4.6 (in-phase), 2.3 (opposed-phase) | 83-92 | 83-88 | 1.7-2.5 | 1.4 |
| Flip angle (°) | 70 | 10-12 | 10 | 150 | 180 | 10-12 | 10-12 |
| Bandwidth (Hz) | 260 | 700 | 313 | 260 | 450 | 325-460 | 488 |
| No. of excitations | 1 | 0.7-1 | 1 | 2 | Half-Fourier | 1 | 1 |
| Acceleration factor | 2 | 2 | 1 | 1 | 2 | 2 | 2 |
| Matrix size | 256/320 × 134/152 | 294 × 224 | 192×320 | 256/320 × 100/132 | 170× 256 | 256×320 | 132×320 |
| FOV (cm) | 25×35 | 25×35 | 25× 35 | 25×35 | 25×35 | 25×35 | 25× 35 |
| Slice thickness (mm) | 5-6 | 3-4 | 3-5 | 5 | 5 | 2.5-4 | 2.5-4 |
Note-Imaging was performed on clinical 1.5-T (Symphony or Avanto, Siemens Healthcare) or 3-T (TRIO, Siemens Healthcare; Discovery 750 W, GE Healthcare Healthcare) systems. NA = not applicable.
aVIBE (Siemens Healthcare) or LAVA (GE Healthcare).
pressed GRE imaging are provided in Table 1. Gadolinium-enhanced GRE sequences were per- formed dynamically after the administration of 1.0/0.1 mmol/kg of gadopentetate dimeglumine or gadobutrol (Magnevist or Gadovist, Bayer HealthCare) injected at a rate of 2 mL/s followed by a 20-mL saline flush. Axial contrast-enhanced images were obtained after an empirical 30- to 40-second delay followed by successive acqui- sitions every minute for 3-5 minutes. The final patient underwent MRI at an outside institution using a 1.5-T clinical scanner (Avanto, Siemens Healthcare) with imaging parameters similar to those already described.
MRI ROI Analysis
A blinded abdominal radiology fellow mea- sured SI values. On T2-weighted images, an ROI was placed in the largest portion of the lesion en- compassing at least two-thirds of the lesion. A fixed diameter (10 mm) ROI was placed in the ip- silateral erector muscle. T2-weighted lesion-to- muscle SI ratio was calculated by dividing the le- sion SI by the muscle SI [25]. For chemical-shift MR images, an ROI was placed in the lesion as was performed for T2-weighted images, avoiding the edges of the lesion so as not to include areas of chemical-shift artifact of the second kind in the measurement [26]. A fixed diameter (10 mm) ROI was also placed in the spleen at the same level. Chemical-shift SI index and chemical-shift adre- nal-to-spleen SI ratio were calculated as described elsewhere [25, 27, 28]. For gadolinium-enhanced MRI, evaluation was performed by measuring the initial contrast-enhanced MRI AUC, as has been reported elsewhere [24, 29]. The contrast-en- hanced MRI AUC was calculated by measuring
the area under the corrected SI (y-axis) over time (x-axis) curve [30, 31].
MRI Histogram Analysis
For each tumor, the axial image used for ROI analysis of T2-weighted MRI was selected by the blinded radiology fellow. Patient-identifying in- formation was removed from each image obtained from the 44 adrenal lesions, and these images were exported in DICOM format from the PACS to an independent workstation for lesion analysis using Image J (version 1.48, National Institutes of Health). Each image was manually contoured by the fellow to define the outer margin of each adre- nal nodule, avoiding extralesional structures (Fig. 1). Histogram analysis was performed, and three specific parameters were studied. Kurtosis (a mea- sure of histogram flatness) and skewness (a mea- sure of histogram asymmetry) were extracted di- rectly from Image J software. Entropy (a measure of histogram irregularity) was obtained by ana- lyzing the histogram values using the following equation: - sum[p × log2(p)], where p denotes the histogram pixel SI values, by using an in-house software plugin created for Excel (version 14.0, Microsoft). T2-weighted MRI was used to mea- sure histogram features because it is a relatively robust imaging pulse sequence that provides reli- able image quality between patients, does not rely on the injection of gadolinium, and has been pre- viously investigated in RCC [30].
MRI Subjective Analysis
Two blinded genitourinary radiologists with 12 and 15 years of experience evaluated MRI for the following features: homogeneity using T2-weighted MRI (binary assessment: homoge-
neous [uniform SI] or heterogeneous [mixed ar- eas of low and high SI]), SI decrease on opposed- phase compared with in-phase chemical-shift MR images (binary outcome: present or absent), and SI on T2-weighted MRI compared with ipsilater- al erector muscle (isointense, mildly hyperintense [SI approaching spleen], or markedly hyperintense [SI approaching retroperitoneal fat or CSF]). Dis- crepancies were resolved by consensus after the initial independent blinded review.
Statistical Analysis
All data are presented as mean (± SD) with range also provided for independent variables. De- mographic variables and subjective outcomes were compared using the chi-square test of proportions, and parametric data were compared using linear regression analysis. ROC analysis was also per- formed to determine the diagnostic accuracy of sta- tistically significant MRI features, and multivariate logistic regression modeling was performed to eval- uate and compare diagnostic accuracy. A threshold of p < 0.05 indicated a statistically significant dif- ference. Interobserver agreement was assessed us- ing Cohen kappa statistics. Statistical analysis was performed with STATA data analysis and statistical software (version 13, StataCorp).
Results
There were 14 women in the adenoma group and three women in the metastases group (p = 0.3214). Patients in the adenoma group were older (mean age, 60.8 ± 9.9 years) than patients in the metastases group (mean age, 52.8 ± 10.1 years; p = 0.0332). Metastases were larger than adenomas (mean, 5.0 ± 4.2 cm [range, 1.1-15 cm] vs 2.0 ± 0.67 cm [range,
100
Chemical-Shift Signal Intensity Index
50
0
-50
Adenoma
Metastases
10
T2 Signal Intensity Ratio
8
6
4
2
0
Adenoma
Metastases
Use of MRI to Distinguish RCC Metastases From Adenoma
| Parameter | Adrenal Metastases (n= 15) | Adrenal Adenomas (n= 29) | pa |
|---|---|---|---|
| Size (cm) | 5.0 ± 4.4 (1.1-15) | 2.0 ± 0.7 (1.2-3.7) | <0.0001 |
| Chemical-shift SI indexb | 9.2 ± 20.6 (-30.0 to 57.9) | 47.3 ± 27.8 (-9.4 to 86.0) | <0.0001 |
| Chemical-shift adrenal-to-spleen SI ratioc | 0.94 ± 0.23 (0.44-1.33) | 0.52 ± 0.28 (0.13-1.11) | <0.0001 |
| T2-weighted SI ratiod | 5.1 ± 3.0 (1.5-10.6) | 1.8 ±0.8 (0.5-3.8) | <0.0001 |
| Contrast-enhanced MRI AUC | 369.8 ± 455.1 (-0.99 to 1333.8) | 276.2 ± 127.9 (145.3-609.1) | 0.4256 |
| Skewness | 0.34 ± 0.53 (-0.56 to 1.32) | 0.47 ± 0.51 (-0.39 to 1.77) | 0.4113 |
| Kurtosis | 0.19 ± 0.95 (-0.74 to 2.31) | 0.74 ± 0.95 (-0.35 to 3.57) | 0.0734 |
| Entropy | 6.76 ± 0.61 (5.19-7.41) | 6.10 ± 0.74 (3.79-6.87) | 0.0051 |
Note-Data are mean + SD (range).
aComparisons were performed using multivariate regression.
bChemical shift signal intensity (SI) index = [(SI,
SI
tumorIP tumorOp) / SI tumorIP
× 100, where SI tumorIP is tumor SI on in-phase imaging, and SI tumorOP is tumor SI on opposed-
phase imaging.
“Chemical shift adrenal-to-spleen SI index = [(SI, tumorOP bleenOp//3tun tumorIP spleenIP
SI ) /(SI
SI )] × 100, where SI, tumorOP
SI on opposed-phase imaging, SItumorip is tumor SI on in-phase imaging, and SIspleenlp is spleen SI on in-phase imaging.
is tumor SI on opposed-phase imaging, SI spleenOp is spleen dT2-weighted SI ratio = SItumor/ SI muscle where SItumoris SI in the tumor and SImuscle is SI in the muscle.
1.2-3.7]; p <0.0001); however, there was over- lap between groups, and 60.0% (9/15) of me- tastases were smaller than 4 cm (Fig. 2).
Results for quantitative MRI analysis are provided in Table 2. The SI index was higher and the adrenal-to-spleen SI ratio was lower in adenomas compared with me- tastases (p < 0.0001); 79.3% (23/29) of ad- enomas and 20.0% (3/15) of metastases had an SI index greater than 16.5% (Fig. 3). The T2-weighted SI ratio was higher in metasta- ses than in adenomas (p <0.0001; Figs. 1, 2, 4, and 5). The AUC for diagnosis of metasta- ses using the T2-weighted SI ratio was 0.91 (95% CI, 0.79-1.0) with an optimal sensitiv- ity of 80.0% and specificity of 93.1% using a value greater than 2.8 (Fig. 6). There was no difference in contrast-enhanced MRI AUC for metastases to adenomas (p>0.05). There was no difference in T2-weighted skewness or kurtosis between adenomas and metasta- ses (p>0.05); however, T2-weighted entro- py was higher in metastases compared with
adenomas (p = 0.0051). The AUC for the diagnosis of metastases using T2-weighted entropy was 0.85 (95% CI, 0.68-1.0) with an optimal sensitivity of 86.7% and spec- ificity of 86.2% using a value greater than 6.76 (Fig. 6). A logistic regression mod- el combining T2-weighted SI ratio and en- tropy was studied and resulted in an AUC for diagnosis of metastases of 0.97 (95% CI, 0.93-1.0) with optimal sensitivity of 93.3% and specificity of 86.21% (Fig. 6), which was significantly more accurate than either vari- able alone (p = 0.0215).
Results for subjective analysis of MRI fea- tures are provided in Table 3. Subjectively, 33.3% (5/15) of metastases showed an SI de- crease on chemical-shift MRI compared with 68.9% (20/29) of adenomas (p=0.0421). Me- tastases were more heterogeneous compared with adenomas (p < 0.0001; K = 0.86) and showed higher T2-weighted SI (p < 0.0001). Interobserver agreement was substantial to almost perfect (K = 0.76-0.89).
Discussion
This study evaluated the ability of MRI to differentiate between clear cell RCC adrenal metastases and adrenal adenomas. Our results indicate that a diagnosis of clear cell RCC me- tastases can be achieved using both quantita- tive and subjective analysis of MRI findings with high degrees of accuracy. In this study, a significant proportion of metastases showed an SI decrease on opposed-phase chemical- shift MRI but it occurred less commonly and to a lesser degree compared with adenomas. Clear cell RCC metastases were significant- ly larger, had increased T2-weighted SI, and were more heterogeneous compared with ad- enomas. A model combining T2-weighted SI ratio and entropy achieved excellent diagnos- tic accuracy.
In our study, approximately one-third of metastases showed an SI decrease at chemi- cal-shift MRI using subjective analysis, and one-quarter showed an SI decrease in the ad- enoma range (> 16.5%) using the chemical-
| Type of Lesion | T2-Weighted Signal Intensityª | Signal Intensity Decrease on Chemical-Shift MRIb | Homogeneity on T2-Weighted MRIc | ||||
|---|---|---|---|---|---|---|---|
| Isointense | Hyperintense | Very Hyperintense | Present | Absent | Homogeneous | Heterogeneous | |
| Adrenal metastases (n= 15) | 0 | 3 (20.0) | 12 (80.0) | 5 (33.3) | 10 (66.7) | 2 (13.3) | 13 (86.7) |
| Adrenal adenomas (n= 29) | 13 (44.8) | 16 (55.2) | 0 | 20 (68.9) | 9 (31.0) | 24 (82.8) | 5 (17.2) |
Note-Data are number (%) of lesions. Subjective MRI features are based on consensus review of two blinded radiologists’ interpretation of findings.
aSignal intensity of lesion compared with that of ipsilateral erector muscle. p< 0.0001 (chi-square statistic) for adenomas versus metastases. For interobserver agreement, K = 0.89.
bBinary outcome was signal decrease present or absent on oppose-phase compared with in-phase chemical-shift MRI. p= 0.0421 (chi-square statistic) for adenomas versus metastases. For interobserver agreement, K = 0.76.
cp < 0.0001 (chi-square statistic) for adenomas versus metastases. For interobserver agreement, K = 0.86.
A
B
C
shift SI index. It has previously been widely reported that adrenal metastases from clear cell RCC can mimic lipid-rich adenomas on chemical-shift MRI because clear cell RCCs also contain intracellular lipid [10-14]. Woo et al. [13] found that adenomas showed more profound SI decreases on opposed-phase MRI compared with clear cell RCC metasta- ses, and our quantitative results also support these findings. Our control group included both lipid-rich and lipid-poor adenomas (be- cause they were consecutively identified), and 20-30% of adenomas in our study did not show an SI decrease at chemical-shift MRI or meet the SI index threshold of 16.5%, which is concordant with the reported percentage of lipid-poor adenomas in the literature [5,6].
A limitation of our study was that it includ- ed imaging at both 1.5 and 3 T, and it could be argued that, by applying a 16.5% SI index threshold at 3 T, some of the lipid-poor ade- nomas may have actually been lipid-rich, be- cause it has been previously shown that SI in- dex thresholds at 3 T are lower than those at 1.5 T [32]. In our study, two of the six lipid- poor adenomas were imaged at 3 T, and, in both instances, the SI index was below the previously described threshold of 7.4% at 3 T, as described in the study by Ream et al. [32]. In one metastasis, the SI index was extremely low because of the presence of internal hem- orrhage, which resulted in T2* effects on in- phase GRE images, a finding that has been pre- viously described in RCCs [33-35]. Whether this observation could be used to differentiate hemorrhagic metastases from adenomas will require further analysis in larger sample sizes. In this study, the contrast-enhanced MRI AUC did not differ between lesions, which also sup- ports previous results that used washout CT and found that adenomas and RCC metasta- ses overlap in their enhancement features [7].
Adrenal metastases from clear cell RCC were larger than adenomas in this study, which supports the results of Sasaguri et al. [4], who showed that adrenal nodule size greater than 4 cm was highly predictive of RCC metastasis. Distinguishing between ad- enomas and metastases in large adrenal mass- es is impractical, and, although it could be ar- gued that a 4-cm threshold is useful because this would generally require surgical inter- vention regardless of other imaging findings, the use of size alone as a discriminating fea- ture between metastases and adenomas can be problematic. For example, in our study, nearly two-thirds of metastases were small- er than 4 cm. All metastases that were diag- nosed on the basis of progression of disease increased by greater than 20% between serial scans obtained less than 6 months apart. We could not compare the rate of growth to ad- enomas in our study because of our inclusion criteria for the control group (that adenomas
show stability in size). Nevertheless, these results support the common practice of per- forming short-term (< 6 month) imaging fol- low-up to differentiate metastases from adre- nal adenomas in patients with clear cell RCC.
We evaluated T2-weighted SI as a po- tential discriminating feature between me- tastases and adenomas because increased T2-weighted signal is a characteristic imag- ing finding of clear cell tumors [16, 17] and adenomas are typically of low-to-interme- diate T2-weighted signal. In our study, me- tastases showed higher T2-weighted SI ra- tios than did adenomas and also differed significantly using subjective analysis. These results support the conclusions by Woo et al. [13], who, using subjective analysis of T2-weighted MRI, found that clear cell RCC adrenal metastases were either slightly or markedly hyperintense relative to liver com- pared with adrenal adenomas, which were ei- ther isointense or slightly hyperintense.
1
1
Sensitivity
- - - - T2 SI Ratio
Entropy
T2 SI Ratio + Entropy
0
0
1
1 - Specificity
Use of MRI to Distinguish RCC Metastases From Adenoma
It has been previously shown that RCCs are more heterogeneous than benign re- nal masses [19, 20]; therefore, heterogeneity could be a valuable feature to distinguish be- tween RCC adrenal metastases and adeno- mas. Using subjective assessment, Sasaguri et al. [4] showed that RCC adrenal metasta- ses were more heterogeneous than adenomas at CT. Previously, Woo et al. [13] also showed that clear cell RCC metastases showed significantly higher rates of cystic change, necrosis, and hemorrhage compared with adenomas. We compared lesion heterogene- ity using histogram analysis and subjective analysis. In our study, metastases showed significantly higher entropy than adenomas, and entropy showed a high degree of accu- racy for the diagnosis of metastatic disease. Entropy has been previously found to be a valuable texture feature, which may be valu- able to diagnose RCC [19, 20] and predict higher-grade disease in RCC [21, 22]. To our knowledge, the use of quantitative histogram or texture analysis for adrenal lesions has yet to be studied, and our results suggest this may be an important area of future research for adrenal gland imaging. Subjectively, me- tastases were also significantly more hetero- geneous than adenomas. A model combin- ing T2-weighted SI and entropy achieved the best diagnostic accuracy and outperformed analysis using individual imaging features.
There are limitations to our study. Our study was retrospective, and this introduces bias into the study population. Our sample size is relatively small and, despite showing strong levels of significance when comparing variables between groups and using a similar sample size to what has been reported pre- viously in this patient population [13], vali- dation in a larger patient cohort is required. The inclusion of patients with multiple clear cell RCC metastases was used to increase our sample size, although this also introduces cluster bias, which is another limitation. The reference standard for adrenal adenomas used in this study can be considered a limitation; however, histopathologic analysis is not a practical reference standard for adenoma. We consider the control group used in our study to be a strength because we included consecu- tive adenomas in patients with no history of malignancy and with stability in size on at least 1-year follow-up, in addition to a mix of both lipid-rich and lipid-poor adenomas.
A final limitation is the variation in MRI performed in the study population. Although this variation is expected, given the long
study period required to capture a sufficient patient population for analysis, it limits the quantitative analysis in our study. The com- bined use of both 1.5- and 3-T scanners and the retrospective use of images with inher- ent noise reduction and surface coil correc- tion algorithms further limits the use of ROI and texture analysis. The applicability of ab- solute quantitative thresholds from our study into clinical practice will require further val- idation. Nevertheless, we think that this lim- itation was partially mitigated through the use of subjective analysis in our study, which confirmed the results from quantitative anal- ysis and may be more readily applicable.
In conclusion, MRI can differentiate clear cell RCC adrenal metastases from adenomas with high degrees of accuracy. Metastases are larger, show increased T2-weighted SI, and are more heterogeneous (with higher entropy) compared with adenomas. A model combin- ing T2-weighted SI and entropy provided ex- cellent diagnostic accuracy for clear cell RCC adrenal metastases. MRI analysis may replace the need for histologic sampling to establish diagnosis in clinical practice; however, this will require further analysis in a larger pa- tient cohort. Because of differences in techni- cal parameters, the subjective results from our study may be more readily applicable for clin- ical practice compared with our quantitative results, which require further study.
References
1. Reuter VE, Tickoo SK. Differential diagnosis of renal tumours with clear cell histology. Pathology 2010; 42:374-383
2. Kuczyk M, Wegener G, Jonas U. The therapeutic value of adrenalectomy in case of solitary meta- static spread originating from primary renal cell cancer. Eur Urol 2005; 48:252-257
3. O’Malley RL, Godoy G, Kanofsky JA, Taneja SS. The necessity of adrenalectomy at the time of radical nephrectomy: a systematic review. J Urol 2009; 181:2009-2017
4. Sasaguri K, Takahashi N, Takeuchi M, Carter RE, Leibovich BC, Kawashima A. Differentiation of benign from metastatic adrenal masses in patients with renal cell carcinoma on contrast-enhanced CT. AJR 2016; 207:1031-1038
5. Berland LL, Silverman SG, Gore RM, et al. Man- aging incidental findings on abdominal CT: white paper of the ACR incidental findings committee. J Am Coll Radiol 2010; 7:754-773
6. Boland GW, Blake MA, Hahn PF, Mayo-Smith WW. Incidental adrenal lesions: principles, tech- niques, and algorithms for imaging characteriza- tion. Radiology 2008; 249:756-775
7. Choi YA, Kim CK, Park BK, Kim B. Evaluation of adrenal metastases from renal cell carcinoma and hepatocellular carcinoma: use of delayed contrast- enhanced CT. Radiology 2013; 266:514-520
8. Siegelman ES. Adrenal MRI: techniques and clinical applications. J Magn Reson Imaging 2012; 36:272-285
9. Davarpanah AH, Israel GM. MR imaging of the kidneys and adrenal glands. Radiol Clin North Am 2014; 52:779-798
10. Outwater EK, Bhatia M, Siegelman ES, Burke MA, Mitchell DG. Lipid in renal clear cell carci- noma: detection on opposed-phase gradient-echo MR images. Radiology 1997; 205:103-107
11. Shinozaki K, Yoshimitsu K, Honda H, et al. Meta- static adrenal tumor from clear-cell renal cell car- cinoma: a pitfall of chemical shift MR imaging. Abdom Imaging 2001; 26:439-442
12. Yoshimitsu K, Honda H, Kuroiwa T, et al. MR de- tection of cytoplasmic fat in clear cell renal cell carcinoma utilizing chemical shift gradient-echo imaging. J Magn Reson Imaging 1999; 9:579-585
13. Woo S, Cho JY, Kim SY, Kim SH. Adrenal adeno- ma and metastasis from clear cell renal cell carci- noma: can they be differentiated using standard MR techniques? Acta Radiol 2014; 55:1120-1128
14. Schieda N, Al Dandan O, Kielar AZ, Flood TA, McInnes MD, Siegelman ES. Pitfalls of adrenal imaging with chemical shift MRI. Clin Radiol 2014; 69:1186-1197
15. Moosavi B, Shabana WM, El-Khodary M, et al. Intracellular lipid in clear cell renal cell carcino- ma tumor thrombus and metastases detected by chemical shift (in and opposed phase) MRI: radio- logic-pathologic correlation. Acta Radiol 2016; 57:241-248
16. Oliva MR, Glickman JN, Zou KH, et al. Renal cell carcinoma: T1 and T2 signal intensity characteris- tics of papillary and clear cell types correlated with pathology. AJR 2009; 192:1524-1530
17. Chung MS, Choi HJ, Kim MH, Cho KS. Com- parison of T2-weighted MRI with and without fat suppression for differentiating renal angiomyoli- pomas without visible fat from other renal tumors. AJR 2014; 202:765-771
18. Varghese JC, Hahn PF, Papanicolaou N, Mayo- Smith WW, Gaa JA, Lee MJ. MR differentiation of phaeochromocytoma from other adrenal le- sions based on qualitative analysis of T2 relax- ation times. Clin Radiol 1997; 52:603-606
19. Hodgdon T, McInnes MD, Schieda N, Flood TA, Lamb L, Thornhill RE. Can quantitative CT texture analysis be used to differentiate fat-poor renal an- giomyolipoma from renal cell carcinoma on unen- hanced CT images? Radiology 2015; 276:787-796
20. Sasaguri K, Takahashi N, Gomez-Cardona D, et al. Small (< 4 cm) renal mass: differentiation of onco- cytoma from renal cell carcinoma on biphasic
Downloaded from www.ajronline.org by Tufts University on 07/26/17 from IP address 130.64.11.153. Copyright ARRS. For personal use only; all rights reserved
Downloaded from www.ajronline.org by Tufts University on 07/26/17 from IP address 130.64.11.153. Copyright ARRS. For personal use only; all rights reserved
contrast-enhanced CT. AJR 2015; 205:999-1007
21. Schieda N, Thornhill RE, Al-Subhi M, et al. Diag- nosis of sarcomatoid renal cell carcinoma with CT: evaluation by qualitative imaging features and texture analysis. AJR 2015; 204:1013-1023
22. Doshi AM, ReAm JM, Kierans AS, et al. Use of MRI in differentiation of papillary renal cell car- cinoma subtypes: qualitative and quantitative analysis. AJR 2016; 206:566-572
23. Nishino M, Jagannathan JP, Ramaiya NH, Van den Abbeele AD. Revised RECIST guideline ver- sion 1.1: what oncologists want to know and what radiologists need to know. AJR 2010; 195:281-289
24. Schieda N, Alrashed A, Flood TA, Samji K, Shabana W, McInnes MD. Comparison of quanti- tative MRI and CT washout analysis for differen- tiation of adrenal pheochromocytoma from adre- nal adenoma. AJR 2016; 206:1141-1148
25. Sasiwimonphan K, Takahashi N, Leibovich BC, Carter RE, Atwell TD, Kawashima A. Small (<4 cm) renal mass: differentiation of angiomyolipoma without visible fat from renal cell carcinoma utiliz- ing MR imaging. Radiology 2012; 263:160-168
26. Ramamurthy NK, Moosavi B, McInnes MD,
Schieda et al.
Flood TA, Schieda N. Multiparametric MRI of solid renal masses: pearls and pitfalls. Clin Radiol 2015; 70:304-316
27. Hindman N, Ngo L, Genega EM, et al. Angiomyoli- poma with minimal fat: can it be differentiated from clear cell renal cell carcinoma by using standard MR techniques? Radiology 2012; 265:468-477
28. Karlo CA, Donati OF, Burger IA, et al. MR imag- ing of renal cortical tumours: qualitative and quantitative chemical shift imaging parameters. Eur Radiol 2013; 23:1738-1744
29. Hakim SW, Schieda N, Hodgdon T, McInnes MD, Dilauro M, Flood TA. Angiomyolipoma (AML) without visible fat: ultrasound, CT and MR imag- ing features with pathological correlation. Eur Radiol 2016; 26:592-600
30. Ferl GZ, Xu L, Friesenhahn M, Bernstein LJ, Barboriak DP, Port RE. An automated method for nonparametric kinetic analysis of clinical DCE-MRI data: application to glioblastoma treat- ed with bevacizumab. Magn Reson Med 2010; 63:1366-1375
31. Walker-Samuel S, Leach MO, Collins DJ. Evalua- tion of response to treatment using DCE-MRI: the
relationship between initial area under the gadolini- um curve (IAUGC) and quantitative pharmacoki- netic analysis. Phys Med Biol 2006; 51:3593-3602
32. Ream JM, Gaing B, Mussi TC, Rosenkrantz AB. Characterization of adrenal lesions at chemical- shift MRI: a direct intraindividual comparison of in- and opposed-phase imaging at 1.5 T and 3 T. AJR 2015; 204:536-541
33. Childs DD, Clingan MJ, Zagoria RJ, et al. In- phase signal intensity loss in solid renal masses on dual-echo gradient-echo MRI: association with malignancy and pathologic classification. AJR 2014; 203:[web]W421-W428
34. Yoshimitsu K, Kakihara D, Irie H, et al. Papillary renal carcinoma: diagnostic approach by chemical shift gradient-echo and echo-planar MR imaging. J Magn Reson Imaging 2006; 23:339-344
35. Murray CA, Quon M, McInnes MD, et al. Evalua- tion of T1-weighted MRI to detect intratumoral hemorrhage within papillary renal cell carcinoma as a feature differentiating from angiomyolipoma without visible fat. AJR 2016; 207:585-591
36. Taner AT, Schieda N, Siegelman ES. Pitfalls in ad- renal imaging. Semin Roentgenol 2015; 50:260-272