Multi-Detector-Row CT Diagnosis of Adrenal Incidentaloma in Patients with Hepatocellular Carcinoma

We investigate the diagnostic reliability of differentiating between lipid-poor adrenal adenomas and metastatic adrenal tumors originating from hepatocellular carcinoma (HCC) using a routine dynamic CT protocol for liver imaging. Eighteen metastatic adrenal tumors originating from HCC and 13 lipid-poor adrenal adenomas were identified. Dynamic CT data were analyzed for CT attenuation of adrenal lesions before and after contrast administration. When a cutoff of 36 HU was set for adrenal lesions at pre-contrast attenuation, the sensitivity and specificity for the diagnosis of metastatic lesions were 94.4% and 92.3%, respectively. Attenuation criteria on pre-contrast CT may help optimize the differentiation between these lesions.


Introduction
Hepatocellular carcinoma (HCC) results in between 250,000 and one million deaths globally per year [1][2][3][4].Hepatocellular carcinoma is a devastating tumor, with a mean survival time of much less than 1 year, if left untreated [5].The treatment of HCC depends entirely on tumor stage.Potentially curative partial hepatic resection or orthotopic liver transplantation should be reserved for patients with stage I or II tumors [6].Patients with an advanced tumor staging or with poor hepatocellular reserves are not candidates for these curative surgical treatments, and therefore have palliative care as their only option.Patients with advanced-stage disease have lower survival rates than patients with early-stage HCC tumors.Thus, the detection of extrahepatic metastatic disease becomes a crucial factor in planning potential therapy for patients with HCC, as it is desirable to avoid unnecessary surgical interventions.The adrenal glands are a common site for extrahepatic metastases, which accounted in one study for 19 (8.4%) of 232 consecutive cases of hepatocellular carcinoma autopsied [7].
Recent studies have predicted that an adrenal incidentaloma will be identified in 4% -5% of abdominal CTs [8].However approximately 75% of adrenal masses in patients with no history of cancer are cortical adenomas, whereas adrenal metastases account for 45% -73% of adrenal masses in cancer patients [8].Computed tomography has become the imaging modality of choice to differentiate between benign adrenal adenomas and metastases in oncology patients [9].As approximately 30% of adenomas are lipid-poor and difficult to characterize on unenhanced CT scans [10], it has been proposed that if the attenuation of the adrenal gland exceeds 10 HU upon non-enhanced CT, contrast material-enhanced CT should be performed, and the washout calculated.With a combination of unenhanced and delayed enhanced CT, nearly all adrenal masses can be correctly categorized as adenomas or non-adenomas [11,12].
In patients with HCC, adrenal incidentalomas may be depicted if the patient is undergoing dynamic CT for an HCC protocol that would include the hepatic arterial, portal, and equilibrium phases after contrast administration, in addition to non-enhanced CT.Lipid-rich adrenal adenomas may be diagnosed by analyzing unenhanced CT images.On the other hand, it remains difficult to differentiate metastatic adrenal tumors from lipid-poor adrenal adenomas via CT images using an HCC protocol without a 10-minute delayed CT scan.As regards the CT-imaging results of adrenal tumors metastasized from HCC, Katyal and colleagues reported finding that contrast material-enhancement characteristics ranged from the typical hypoattenuation of soft tissue (seen in adrenal metastases from other primary tumors) to a marked hyperattenuating enhancement [13].However, to the best of our knowledge, there has been no detailed analysis of the attenuation value of adrenal tumors metastasized from HCC as observed on dynamic CT.We hypothesized that it would be possible to differentiate between metastatic adrenal HCC lesions and lipid-poor adenomas using a CT protocol that is routinely performed for patients with HCC.Therefore, the purpose of this study was to investigate the diagnostic value of a routine HCC imaging protocol to differentiate between lipid-poor adenomas and metastatic adrenal tumors originating from HCC.

Ethical Issues
We adapted the requirements of our institutional review board for a retrospective observation study, and the requirement for informed consent was waived.

Patient Selection
Diagnostic reports from dynamic CT examinations performed between January 2002 and December 2009 that contained the words "hepatocellular carcinoma", "adrenal adenoma", or "adrenal metastasis" were searched retrospectively.In 2888 reports identified, the diagnosis of adrenal metastasis was based on rapid growth (more than 50% growth in the largest diameter) of a mass in 6 months.Metastases exceeding 30 mm in the largest diameter were excluded, because adrenal adenomas larger than 30 mm are uncommon.Cases involving malignant tumors in organs other than the liver were excluded to prevent inclusion in the analysis of any confounding data from metastatic tumors that had originated in other organs.When there were multiple follow-up studies of metastatic HCC patients, the CT study (usually the latest) involving the largest tumor with a diameter of less than 30 mm was selected for analysis.A diagnosis of adenoma was established on the basis of stable size of the mass for more than half a year.Adrenal adenomas of less than 5 mm in the largest diameter were excluded to avoid partial volume averaging with densitometry.Lipid-rich adenomas with an average CT number of less than 20 HU were excluded, because the study was designed with a focus on lipid-poor adenomas.In our selection of lipidpoor adenomas, adrenal nodules with cysts, macroscopic fat, or calcifications were also excluded.We identified 12 patients (8 men, 4 women; age range: 57 -81 years; mean age: 68.4 years) with 13 lipid-poor adenomas, and 15 patients (12 men, 3 women; age range: 41 -80 years; mean age: 61.1 years) with 18 adrenal metastases from HCC.In 10 cases of metastases, histological proof was available, and 4 cases were bilateral.

CT scan
Dynamic CT scans were performed with patients in the supine position.Breath-hold CT examinations of the abdomen were performed with a 4-section or a 64-section multi-detector Aquilion Scanner (Toshiba Medical Systems Corporation, Otawara, Tochigi, Japan).The following scan parameters were used for the 4-section multi-detector scanner: tube current, 140 mAs; voltage, 120 kV; collimation, 4 × 2 mm; reconstructed slice thickness, 5 mm; and reconstruction increment, 5 mm.The following scan parameters were used for the 64-section multidetector scanner: tube current-via automatic exposure control; voltage, 120 kV; collimation, 64 × 0.5 mm; reconstructed slice thickness, 5 mm; and reconstruction increment, 5 mm.Scans of the abdomen were performed in the craniocaudal direction with a protocol that included the hepatic arterial phase (approximately 40 seconds after the onset of contrast injection) and equilibrium phase (150 seconds after the onset of contrast injection), with the scan timing adjusted by a bolus tracking system.The dosage of the contrast material was 450 mgI/kg (body weight) of iodine contrast agent.It is important to emphasize that the fixed injection duration of 30 seconds, used for the present injection protocol, was selected over a fixed injection rate in order to establish an optimal contrast-enhanced dynamic CT protocol of the liver; in general, injection duration is considered the most important of the two measures, and is the only technical factor capable of predicting the scan timing for each phase [14].

Image Analysis
Computed tomographic images were retrieved from the institutional picture archiving and communication system (PACS) to a standard viewer (Vox-Base; J-Mac Systems, Sapporo, Japan).Average attenuation values on the precontrast, hepatic arterial contrast-enhanced, and equilibrium contrast-enhanced images were recorded.A circular region of interest (ROI) was placed in the center of the adrenal mass, in a section where the mass appeared largest; the region of interest covered approximately one-half to two-thirds of the mass.Care was taken to avoid the partial-volume effect and beam-hardening artifacts.All images were retrospectively reviewed by a radiologist (HT) with 2 years of training in abdominal imaging.When more than one adrenal mass was present, all measurements were obtained for each mass.We determined both the relative percentage washout (RPW) and absolute percentage washout (APW) rates.The RPW and APW were calculated as follows: RPW = 100•(HA − EqA)/HA and APW = 100•([HA − EqA]/[HA − PA]), where HA is attenuation on hepatic arterial contrast-enhanced scans, EqA is attenuation on equilibrium contrast-enhanced scans, PA is precontrast attenuation, and all attenuation meas-urements are in Hounsfield units.

Statistical Analysis
Statistical analysis was performed using MedCalc statistical software, version 7.2.0.2 (MedCalc Software, Mariakerke, Belgium).Quantitative variables are given as the average and range.The independent t-test was performed to assess differences in values.To determine optimal cutoffs and calculate the area under the curve (AUC), sensitivity and specificity, and receiver operating characteristics curve (ROC) analyses were performed.Any p value less than 0.05 was considered statistically significant.

Results
Detailed data on the size, laterality, CT attenuation (precontrast CT, hepatic arterial phase, and equilibrium phase), APW, and RPW for the lipid-poor adenoma and metastasis groups are shown in Table 1.Size, CT at-tenuation, and washout percentage were all significantly larger in the HCC group than in the adenoma group, with the exception of the CT attenuation on the arterial phase.Figure 1 shows the scatterplot distribution of the precontrast attenuation values of lipid-poor adenomas and metastases.When we set the cutoff value at 36 HU, sensitivity and specificity were 94.4% and 92.3%, respectively.The results and a comparison ROC analysis for differentiating between the two groups are shown in Table 2 and Figure 2. Precontrast CT attenuation showed the largest AUC value of 0.97.Out of 18 nodules classified as metastatic adrenal tumors, histopathological analysis was possible in 10 cases, 8 tumors were diagnosed as predominantly moderately differentiated HCC with 2 of these 8 cases having a poorly differentiated component, and with 1 case of well-differentiated component.Two tumors were diagnosed as poorly differentiated HCC.Both typical and atypical examples of images of lipidpoor adenomas and metastases are shown in Figures 3-6.

Discussion
In this study, we attempted to characterize metastatic adrenal tumors originating from HCC in terms of a comparison with lipid-poor adrenal adenomas.Although we did not find any clear definition in the literature of "lipidrich" or "lipid-poor" adenomas in terms of pre-contrast CT attenuation, previous researchers have investigated attenuation threshold values of 10 -25 HU when distinguishing adenomas from other masses on pre-contrast CT; in the past decade, a sensitivity of 36% -89% and a specificity of 95% -100% was reported [15][16][17][18][19][20].Here, we adopted a pre-contrast CT attenuation value of 20 HU to define lipid-poor adrenal adenomas; lesions with less than 20 HU were excluded as these were considered typical lipid-rich adenomas (or some with negative HU, myelolipomas), and lesion with 20 HU or greater attenuation on noncontrast CT that were stable over 6 months or greater were judged to be lipid-poor adenomas.We demonstrated that lipid-poor adrenal adenomas tend to have lower attenuation characteristics than that of metastatic adrenal tumors.When the cutoff value was set at 36 HU, sensitivity and specificity were 94.4% and 92.3%, respectively.The attenuation characteristics of adrenal tumors metastasized from HCC tended to be higher than that of lipid-poor adenomas, possibly because metastatic tumors originating from HCC are composed primarily of moderately differentiated components that barely contain lipids.
Interestingly, we identified one case with a poorly differentiated component in a metastatic tumor under the cutoff of 36 HU (Figure 4).As regards this particular case, differentiation from a lipid-poor adenoma was difficult due to the use of pre-contrast attenuation alone.
As expected, the RPW and APW threshold with the HCC protocol were less useful than with the 10-or 15minute protocol, due to the shorter amount of time for de-enhancement.
The majority of benign adrenal lesions are adenomas.In choosing reliable threshold values for distinguishing between benign and malignant lesions, the recognition of malignancy is paramount, even at the cost of subjecting some patients with benign lesions to biopsy.It is widely considered to be better to biopsy a few benign lesions than to miss any malignant lesion.
Several limitations to our study bear mention.This was a retrospective review of data from a relatively limited number of patients.However, it is the largest series of which we are aware, in which 18 small (less than 25 mm), clinically or histopathologically diagnosed metastatic adrenal tumors originating from HCC were evaluated.In our study, histological evidence was obtained in 10 of 18 metastatic tumors, whereas the majority of the adrenal masses were not pathologically investigated and thus required imaging follow-up for characterization, according to the accepted method of classifying benign and malignant lesions in previous studies [6,8,9,18].We cannot eliminate the possibility of metastases from other sites, although we found no description of the existence of any malignancies other than HCC in the 8 cases of malignancy examined.No pheochromocytomas or adrenal carcinomas were evaluated in our study, and therefore it remains uncertain whether or not use of this protocol will enable the differentiation of these rare tumors.When we consider that patients with chronic hepatitis and/or HCC commonly undergo regular follow-up examination using dynamic CT or MR imaging, interval increase in adrenal size may be helpful in arriving at a diagnosis of a metastatic adrenal tumor originating from HCC.It is likely that using pre-contrast CT, the diagnostic criteria presented here may increase the confidence level for a diagnosis of a metastatic tumor originating from HCC, which may be of critical importance when encountering an incidentaloma upon an initial workup for HCC, i.e., when previous images would not typically be available for comparison.
In conclusion, our results obtained with a multi-detector row CT protocol for HCC establish attenuation criteria for use in known HCC patients with adrenal incidentalomas; the present protocol and attenuation criteria will help optimize the differentiation between lipid-poor adenomas and metastatic adrenal masses originating from HCC in this group of patients.We therefore propose the following principles for imaging studies of HCC patients with adrenal nodules exhibiting a largest diameter of less than 30 mm.First, all non-calcified, nonhemorrhagic adrenal lesions with a pre-contrast attenuation of greater than 35 HU should be considered as suspicious for metastases originating from HCC, namely, as candidates for invasive interventions such as biopsy or resection, even in the absence of a washout study with more than a 10-minute delay.Second, in adrenal lesions with a precontrast attenuation of less than 35 HU, the chance of a metastatic adrenal tumor originating from HCC is unlikely; however, a washout study with a delay of more than 10-minutes may be needed in such cases, because there remains the possibility of metastasis from another primary site, or that of a primary adrenal tumor such as an adrenal carcinoma, pheochromocytoma, or adrenal adenoma.

Table 2 . Receiver operating characteristics (ROC) analysis for differentiating between lipid-poor adenomas and adrenal me- tastases originating from HCC.
AUC, area under the curve.For APW and RPW, see Table1.