KIDNEYS, URETERS, BLADDER, RETROPERITONEUM
Check for updates
Diagnostic performance of hepatic CT and chemical-shift MRI to discriminate lipid-poor adrenal adenomas from hepatocellular carcinoma metastases
Yasunori Nagayama1(D . Hidetaka Hayashi1 . Narumi Taguchi1 . Ryuya Yoshida1 . Ryota Harai1 . Masafumi Kidoh1 . Seitaro Oda1 . Takeshi Nakaura1 . Toshinori Hirai1
Received: 28 November 2023 / Revised: 27 January 2024 / Accepted: 29 January 2024 / Published online: 8 March 2024 @ The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024
Abstract
Purpose To evaluate the diagnostic performance of multiphase hepatic CT parameters (non-contrast attenuation, absolute and relative washout ratios [APW and RPW, respectively], and relative enhancement ratio [RER]) and chemical-shift MRI (CS-MRI) for discriminating lipid-poor adrenal adenomas (with non-contrast CT attenuation > 10 HU) from metastases in patients with hepatocellular carcinoma (HCC).
Methods This retrospective study included HCC patients with lipid-poor adrenal lesions who underwent multiphase hepatic CT between January 2010 and December 2021. For each adrenal lesion, non-contrast attenuation, APW, RPW, RER, and signal-intensity index (SI-index) were measured. Each parameter was compared between adenomas and metastases. The area under the receiver operating characteristic curves (AUCs) and sensitivities to achieve 100% specificity for adenoma diagnoses were determined.
Results 104 HCC patients (78 men; mean age, 71.8±9.6 years) with 63 adenomas and 48 metastases were identified; CS- MRI was performed in 66 patients with 49 adenomas and 21 metastases within one year of CT. Lipid-poor adenomas showed lower non-contrast attenuation (22.9±7.1 vs. 37.9±9.4 HU) and higher APW (40.5% ± 12.8% vs. 23.7% ±17.4%), RPW (30.0% ± 10.2% vs. 12.4%±9.6%), RER (329%±152% vs. 111%±43.0%), and SI-index (43.3±20.7 vs. 10.8±13.4) than HCC metastases (all p <. 001). AUC for non-contrast attenuation, APW, RPW, RER, and SI-index were 0.894, 0.786, 0.904, 0.969, and 0.902, respectively. The sensitivities to achieve 100% specificity were 7.9%, 25.4%, 30.2%, 63.5%, and 24.5%, respectively. Combined RER and APW achieved the highest sensitivity of 73.0%.
Conclusion Multiphase hepatic CT allows for better discrimination between lipid-poor adrenal adenomas and metastases relative to CS-MRI, especially when combined with RER and washout parameters.
☒ Yasunori Nagayama y.nagayama1980@gmail.com
1 Department of Diagnostic Radiology, Graduate School of Medical Sciences, Kumamoto University, 1-1-1, Honjo, Chuo-ku, Kumamoto 860-8556, Japan
Keywords Adrenocortical adenoma . Adrenal neoplasm . Hepatocellular carcinoma . Multidetector computed tomography . Chemical shift imaging
Diagnostic Performance of CT vs CS-MRI to Discriminate Lipid-poor Adrenal Adenomas from Metastases in HCC Patients
100
· Multiphase hepatic CT provides better discrimination between lipid-poor adrenal adenomas and metastases than CS-MRI in HCC patients.
80
Sensitivity (%)
60
40
Non-contrast
APW
RPW
· Combining RER with APW yielded the highest sensitivity of 73% to achieve 100% specificity.
20
RER
SI-index
0
0
20
40
60
80
100
100-Specificity (%)
Abdominal Radiology The Official Journal of the Society of Abdominal Radiology www.abdominalradiology.org
Nagayama Y et al; 2024
Introduction
Hepatocellular carcinoma (HCC) is the most prevalent pri- mary malignant hepatic tumor and the fourth leading cause of cancer-related deaths worldwide [1]. Multiphase CT is a first line imaging modality for HCC diagnosis [2-4]. Alongside assessing intrahepatic tumor burden, the identi- fication of extrahepatic metastases is crucial for optimizing treatment strategies [1, 2]. The adrenal gland is the fourth most common site of extrahepatic metastasis, following the lung, lymph nodes, and bone [5, 6]. While the major- ity of adrenal lesions are nonfunctional adenomas that do not require treatment, distinguishing them from metastases is imperative, especially when neither other extrahepatic spread nor macrovascular invasion is identified.
Adrenal adenomas typically contain intracytoplasmic fat, which decreases non-contrast CT attenuation. The adrenal lesions with non-contrast attenuation ≤ 10 HU are diagnosed as lipid-rich adenomas [7]. However, there remains a challenge in distinguishing lipid-poor adenomas (> 10 HU) from metastases due to overlapping attenuation values. In such cases, the calculation of absolute or rela- tive percentage washout (APW or RPW, respectively) on dedicated adrenal CT scans may be selected as subsequent diagnostic procedure, given that adenomas demonstrate
more rapid contrast washout compared to most metastases [7, 8]. However, the diagnostic efficacy of this approach may be compromised in patients with HCC due to over- lapped washout characteristics between adenomas and metastases [9]. Furthermore, adrenal washout CT requires a longer delay time (15-min) than delayed scans on routine hepatic CT (3-5 min), restricting its practical applicability. Chemical-shift MRI (CS-MRI) is an alternative diagnostic approach that enables more sensitive detection of micro- scopic fat compared to non-contrast CT [10-12]. However, adrenal metastases from HCC can contain microscopic fat, potentially mimicking adenomas [13, 14]. Although this diagnostic pitfall has been repeatedly described in the lit- eratures [15-17], no prior investigations have addressed the actual diagnostic performance of CS-MRI for discrimi- nating lipid-poor adenomas from HCC metastases.
Recently, the utility of combined non-contrast and por- tal venous phase (PVP) CT has been proposed and dem- onstrated for the diagnosis of lipid-poor adrenal adenomas [18-22]. This approach relies on two CT imaging features of adenomas; low non-contrast attenuation and relatively high contrast wash-in during the PVP. Although several investi- gations have shown its diagnostic efficacy [18-22], no prior investigations have focused on patients with HCC, whose adrenal metastases can show low non-contrast attenuation due to intratumoral fat [13, 14] and contrast kinetics similar
to adenomas [9]. Given that the performance of adrenal imaging can be affected by patient characteristics, such as the presence of extra-adrenal primary malignancy [12, 23, 24], the type of underlying malignancy [9], or the preva- lence of hypervascular adrenal lesions [25, 26], a dedicated performance assessment for this specific population would be of clinical value.
The purpose of this study was to evaluate the diagnos- tic performance of multiphase hepatic CT parameters in discriminating between lipid-poor adrenal adenomas and metastases in patients with HCC and to compare it with that of CS-MRI.
Materials and methods
This retrospective study was approved by the institutional review board; the requirement for written informed consent was waived.
Patients
The radiological database at the study institution was searched to identify consecutive HCC patients with adre- nal lesions who underwent multiphase hepatic CT, includ- ing the non-contrast and contrast-enhanced scanning dur- ing portal-venous and 3-min delayed-phases (PVP and DP, respectively) between January 2010 and December 2021. All intrahepatic HCC was diagnosed with histopathology and/or typical imaging findings [2, 4]. Exclusion criteria included (a) adrenal lesions showing non-contrast attenua- tion of ≤ 10 HU (i.e. lipid-rich adenomas) or macroscopic fat (i.e. myelolipomas); (b) lesions without a solid component, defined as attenuation change between pre-and post-contrast imaging of> 10 HU (i.e., cysts or hematomas); (c) lack of an adequate reference standard (see reference standard section); (d) acquisition protocols deviating from our standard mul- tiphase hepatic CT (see image acquisition); (e) lesions with prior systemic or focal therapy; and (f) lesions < 10 mm in maximum diameter. The size threshold was chosen to follow the relevant guidelines [7, 27] and to prevent measurement errors caused by partial averaging at the section thickness used in this study. The diagnostic performance of CT param- eters was compared in the entire study sample. In addition, a subgroup of patients who underwent CS-MRI within one year of a CT scan was identified and used to compare the diagnostic performance of CT with CS-MRI.
Reference standard
The final diagnosis of adrenal lesions was based on his- topathologic findings or imaging criteria [7]. Adrenal lesions that fulfilled the following criteria were classified as
lipid-poor adenomas: histopathologic diagnosis or exhibit- ing size stability, which was defined as a maximum diameter growth of 3 mm/year or less for a minimum of 12-months during imaging follow-up [28]. Adrenal lesions were cat- egorized as metastases if they fulfilled at least one of the following criteria: histopathological diagnosis, demonstrated new development, exhibited an increase in size (with a mini- mum 8 mm growth in maximum diameter) within 12-months [24] and/or showed a decrease in size after systemic chemo- therapy. Adrenal lesions that exhibited a maximum diameter increase of 4-7 mm within a given 12-month were excluded from the analysis, unless there were additional follow-up images available to confirm the final diagnosis (exclusion criteria: lack of a reference standard).
Image acquisition
CT images were acquired using either a 64- or 320-row scanner (Brilliance 64, Philips; IQon, Philips; or Aquilion ONE, Canon Medical Systems). Scanning parameters varied according to the scanner used, but a tube voltage of 120-kVp and automated tube current modulation were utilized for all acquisitions. After non-contrast image acquisition, contrast medium with a concentration of 300-370 mgI/mL (iodine dose: 600 mgI/kg) was intravenously administered over 30 s via antecubital veins. Late arterial phase and PVP images were acquired 18 and 55 s after reaching a threshold of 150 HU at the abdominal aorta (the times from the start of con- trast medium injection of PVP was approximately 70-80 s). DP images were acquired 3-min after the start of contrast medium injection. Late arterial phase images were acquired to assess arterial phase hyperenhancement of intrahepatic HCC but were not used for adrenal lesion assessment in this study. Axial images were reconstructed with a section thick- ness of 5 mm. For each scan, a size-specific dose estimate was calculated to estimate radiation dose exposure.
CS-MRI was performed on 3.0-T systems (Achieva, Philips, Ingenia, Philips, Manetom PrismaFit, Siemens, or Magnetom Trio Tim, Siemens). For all examinations, axial in-phase (IP) and opposed-phase (OP) images were acquired using either a 2D dual gradient echo or a 3D-Dixon tech- nique in a single breath-hold. The OP echo was performed before the IP echo, utilizing the first echo pair to minimize the T2* decay effect. Slice thickness ranged from 2 to 8 mm.
Image analysis
A board-certified radiologist (N.T) with 10 years of clinical experience in abdominal imaging, performed image analyses while blinded to the final diagnoses of the adrenal lesions. To measure the lesion attenuation on non-contrast, PVP, and DP images, circular or ovoid regions of interest (ROIs) were manually placed on each adrenal lesion at the section level
showing the greatest lesion diameter. The retroperitoneal fat, normal adrenal parenchyma, calcification, artifacts, vessels, and necrotic or cystic areas lacking contrast enhancement (defined as attenuation differences less than 10 HU between non-contrast and contrast-enhanced images) were excluded from the ROIs. The ROIs were initially placed on the PVP images and then copy and pasted onto the non-contrast and DP images, with appropriate corrections made if necessary. The absolute enhancement during PVP and DP was calcu- lated by subtracting non-contrast attenuation from contrast- enhanced attenuation at each enhancement phase. APW was calculated as ([attenuation difference between PVP and DP]/ absolute enhancement during PVP)× 100. RPW was cal- culated as ([attenuation difference between PVP and DP]/ contrast-enhanced attenuation during PVP)× 100. The rela- tive enhancement ratio (RER) was calculated as (absolute enhancement during PVP/non-contrast attenuation × 100). In CS-MRI, the signal intensity (SI) was measured on IP and OP images by placing the ROI on adrenal lesions at the sec- tion level showing the greatest lesion diameter, with avoid- ance of the peripheral low signal lines on the OP images. The SI-index was calculated as ([SI on IP-SI on OP]/SI on IP) × 100 [29]. To evaluate inter-observer reliability, a radiologist (H.H), with four years of clinical experience in abdominal imaging and no knowledge of final diagnoses and quantitative data from another radiologist, performed all quantifications.
Statistical analysis
Statistical analyses were performed using MedCalc soft- ware (MedCalc Software, Mariakerke, Belgium). Continu- ous variables determined to be normally or non-normally distributed through the Kolmogorov-Smirnov test were expressed as mean ± standard deviation or median (inter- quartile range), respectively, and categorical variables were expressed as numbers (percentages). Mann-Whitney U-test or Fisher’s exact test was used to compare patient charac- teristics, radiation doses, and imaging parameters between adenomas and metastases. The correlations between non- contrast attenuation and APW, RPW, RER, or SI-index were calculated for adenomas using Pearson’s-correlation-coeffi- cients to clarify the impact of non-contrast attenuation on the quantitative values and discriminative performance of each parameter. This correlation analysis is expected to provide important insights regarding which parameters are more suitable for diagnosing adenomas with hypo- and hyper- attenuation on non-contrast CT. Interobserver reliability for the quantification of all imaging parameters was evaluated using intraclass-correlation-coefficients (ICC). ICC values were interpreted as follows [30]: < 0.400, poor; 0.400-0.599, fair; 0.600-0.749, good; 0.750-1.000, excellent. Receiver operating characteristics (ROC) analysis was performed to
determine the area under the ROC curve (AUC) for discrim- inating lipid-poor adenomas from HCC metastases. AUC and the sensitivity to achieve 100% specificity of CT param- eters (non-contrast attenuation, APW, RPW, and RER) were compared with the DeLong method and McNemar’s test, respectively. Holm method was used to correct p-value. In the subgroup of patients who underwent CS-MRI, the AUC and sensitivity to achieve 100% specificity were calculated for the SI-index and compared with a CT parameter that achieved the highest performance in entire sample using the DeLong method and McNemar’s test, respectively. The 100% specificity level was selected because perfect identi- fication of metastases in patients with known malignancy is crucial, even though some adenomas may require further examinations. The sensitivity of APW, RPW, RER, and SI- index was analyzed for hypo- and hyper-attenuating adeno- mas showing non-contrast attenuation of ≤25 HU and> 25 HU, respectively. P-values <0.05 were considered statisti- cally significant.
Results
Patient characteristics
A total of 171 HCC patients with adrenal lesions were iden- tified. Of these, the following were excluded: 24 with lipid- rich adenomas (all lesions demonstrated size stability during mean follow-up period of 70±48 [range, 14-207] months), three with myelolipomas, two with adrenal hematomas, one with an adrenal cyst, six imaged with different protocols, 15 with previously treated lesions, eight with insufficient refer- ence standards, and eight with lesions smaller than 1 cm. The final sample included 104 patients with 111 adrenal lesions (78 men and 26 women; mean age, 71.8±9.6 years; mean body mass index: 24.3 +4.2 kg/m2, Fig. 1). The diag- nosis of HCC in liver was based on imaging findings in 52 patients, while histopathologic confirmation was obtained in remaining 52 patients. Sixty-three adrenal lesions were diag- nosed as lipid-poor adenomas based on size stability (mean follow-up: 56±34 [range, 12-149] months). The remaining 48 adrenal lesions were classified as metastases based on histopathology (n=7) or interval imaging (n=41). Three and four patients in the adenoma and metastasis groups, respectively, had bilateral adrenal lesions. The extra-adrenal metastases were more frequently observed in the metastasis group than in the adenoma group (31.8% [14/44] vs. 6.7% [4/60], p <0.001). Patient characteristics of both groups are summarized in Table 1. Among 104 patients, 66 individu- als (48 male, 18 women; mean age, 72.3 ±9.6 year) with 49 adenomas and 21 metastases underwent both CT and CS-MRI within a 1-year interval. Median interval between CT and CS-MRI was 73 (interquartile range: 25-170) days.
171 HCC patients with adrenal lesions underwent multiphase hepatic CT between January 2010 to December 2021
· Lipid-rich adenoma (n=24)
· Myelolipoma (n=3)
· Cyst or hematoma (n=3)
· Previously treated (n=15)
104 HCC patients with 111 lipid-poor adrenal lesions (>10 HU on non-contrast CT)
· Different protocols (n=6)
· Lesions < 10 mm (n=8)
· No reference standard (n=8)
. Lack of CS-MRI within 1 year of CT (n=38)
70 lipid-poor adrenal lesions in 66 patients who underwent both CT and CS-MRI within 1 year
63 lipid-poor adenomas in 60 patients
48 HCC metastases in 44 patients
49 lipid-poor adenomas in 46 patients
21 HCC metastases in 20 patients
Analysis for CT parameters
Analysis for CT parameters vs. CS-MRI
| Patient characteristics | Lipid-poor adenoma (n=60) | Metastases* (n=44) | p value |
|---|---|---|---|
| Age (years) | 72.1±7.9 | 71.4±11.6 | .87 |
| Male: Female | 42 (70.0): 18 (30.0) | 36 (81.8): 8 (18.2) | .25 |
| Body mass index (kg/m2) | 24.6±4.9 | 23.9±3.1 | .72 |
| Adrenal lesion | |||
| Unilateral: bilateral | 57 (95.0): 3 (5.0) | 40 (90.9): 4 (9.1) | .45 |
| Tumor diameter (mm) | 15.8±4.0 | 31.0±20.0 | <. 001 |
| BCLC stage, 0: A: B: C: D | 9 (15.9): 26 (43.3): 16 (26.7): 8 (13.3): 1 (1.7) | 0 (0): 0 (0): 0 (0): 39 (88.6): 5 (11.4) | <. 001 |
| Child-Pugh, A: B: C | 55 (91.7): 5 (8.3): 0 (0) | 23 (52.3): 17 (38.6): 4 (9.1) | <. 001 |
| T1: T2: T3: T4 | 12 (20.0): 27 (45.0): 11 (18.3): 10 (16.7) | 2 (4.5): 8 (18.2): 12 (27.3): 21 (47.7) | <. 001 |
| N0: N1 | 57 (95.0)/3 (5.0) | 36 (81.8)/8 (18.2) | .0499 |
| M0: M1 | 58 (96.7)/2 (3.3) | 0 (0)/44 (100) | <. 001 |
| Extra-adrenal metastasis | |||
| Presence: absence | 4 (6.7): 56 (93.3) | 14 (31.8): 30 (68.2) | <. 001 |
| AFP (ng/ml) | 8.3 (3.6-36.9) | 21.4 (6.5-426) | .049 |
| PIVKA-II (mAU/mL) | 31.5 (19.8-180) | 623 (118-5370) | <. 001 |
| Underlying liver etiology | |||
| Chronic hepatitis C | 28 (46.7) | 21 (47.7) | .30 |
| Chronic hepatitis B | 12 (20.0) | 6 (13.6) | |
| Nonalcohol steatohepatitis | 3 (5.0) | 3 (6.8) | |
| Alcohol | 7 (11.7) | 4 (9.1) | |
| Autoimmune | 1 (1.7) | 0 (0) | |
| Indeterminate | 9 (15.0) | 10 (22.7) |
Data presented as mean ± standard deviation, median (interquartile range), or number (%). Two adrenal metastases in a patient were developed after transplantation
BCLC Barcelona Clinic Liver Cancer, AFP alpha-fetoprotein, PIVKA-II protein induced by vitamin K absence or antagonist
| Parameters | Lipid-poor adenoma (n=63) | HCC metastasis (n=48) | p values |
|---|---|---|---|
| CT attenuation (HU) | |||
| Non-contrast phase | 22.9±7.1 | 37.9±9.4 | <. 001 |
| Portal venous phase | 90.0±17.9 | 77.5±16.6 | <. 001 |
| 3-min delayed phase | 62.6±12.8 | 67.1±12.3 | .045 |
| Absolute enhancement (HU) | |||
| Portal venous phase | 67.1±17.5 | 39.6±12.1 | <. 001 |
| 3-min delayed | 39.7±11.8 | 29.2±7.4 | <. 001 |
| phase | |||
| APW (%) | 40.5±12.8 | 23.7±17.4 | <. 001 |
| RPW (%) | 30.0±10.2 | 12.4±9.6 | <. 001 |
| RER (%) | 328.7±151.6 | 110.6±43.2 | <. 001 |
| SI-index* | 43.3±20.7 | 10.8±13.4 | <. 001 |
Data presented as mean ± standard deviation
APW absolute percentage washout, RPW relative percentage washout, RER relative enhancement ratio, SI-index: signal intensity index
*Data for patients with 49 adenomas and 21 metastases who under- went chemical-shift MRI within 1 year of CT scans
Image analysis
Tables 2 and Fig. 2 summarize the results of image analy- sis. Non-contrast attenuation was lower in lipid-poor ade- nomas than in HCC metastases (22.9±7.1 vs. 37.9±9.4 HU, p<0.001). Of the 63 lipid-poor adenomas, 39 and 27 lesions exhibited non-contrast attenuation of ≤25 HU and> 25 HU, respectively. Of the 48 metastases, four and 44 lesions demonstrated non-contrast attenuation of ≤ 25 and> 25 HU, respectively. Compared with metastases, contrast-enhanced attenuation and absolute enhancement were higher in adenomas during both PVP (90.0±17.9 vs. 77.5 ± 16.6 HU and 67.1 ±17.5 vs. 39.6±12.1 HU, respectively, both p<0.001) and DP (62.6±12.8 vs. 67.1 ±12.3 HU and 39.7±11.8 vs. 29.2±7.4 HU, respec- tively, both p <0.001). Adenomas exhibited higher APW (40.5% ± 12.8% vs. 23.7%±17.4%), RPW (30.0%±10.2% vs. 12.4% ±9.6%), RER (329% ±152% vs. 111%±43.0%), and SI-index (43.2+20.9 vs. 10.8±14.4) than metastases (all, p<0.001). The ICC for all imaging parameters ranged from 0.849 to 0.964, indicating excellent interobserver reli- ability (Supplemental Table 1).
Correlation of non-contrast attenuation with APW, RPW, and RER in adenomas
In entire sample, APW, RPW, and RER of adenomas showed no correlation (r= - 0.06 [95% CI: - 0.190 to 0.303]),
60
70
50
700
80
A
B
C
D
E
60
70
50
40
600
50
60
Non-contrast attenuation (HU)
500
40
40
30
50
APW (%)
30
RPW (%)
RER (%)
400
SI-index
40
30
20
20
300
30
20
10
10
20
200
0
10
10
0
100
-10
0
0
-20
-10
0
-10
Adenoma
Metastasis
Adenoma
Metastasis
Adenoma
Metastasis
Adenoma
Metastasis
Adenoma
Metastasis
weak negative correlation (r= - 0.222 [95%-CI: - 0.445 to 0.027]), and strong negative correlation (r= - 0.772 [95%- CI: - 0.856 to - 0.648]) with non-contrast attenuation, respectively. In subgroup patients who underwent CS-MRI, SI-index showed strong negative correlation with non-con- trast attenuation (r= - 0.664 [95%-CI: - 0.471 to - 0.800]). These results indicate that hypo-attenuating adenomas tended to exhibit higher RPW, RER, and SI-index values
100
80
Sensitivity (%)
60
40
Non-contrast
APW
20
RPW
RER
0
0
20
40
60
80
100
100-Specificity (%)
compared to hyper-attenuating adenomas, whereas APW was not influenced by baseline non-contrast attenuation.
Diagnostic performance of CT parameters
Figure 3 displays the ROC curves of CT parameters in entire sample. Table 3 presents the AUC and sensitivity of each CT parameter to achieve 100% specificity. RER exhib- ited the AUC of 0.969 (95%-CI: 0.917-0.993), which was higher than the other CT parameters (all, p ≤0.035). The second highest AUC was obtained with RPW (0.904 [95% CI: 0.833-0.952]), followed by non-contrast attenuation (0.894 [95%-CI: 0.821-0.944]) and APW (0.786 [95%-CI: 0.698-0.858]). The sensitivity of non-contrast attenuation was 7.9% (5/63, 95%-CI: 2.6-17.6%, threshold ≤13 HU), while that of APW and RPW was 25.4% (16/63, 95%-CI: 15.3-37.9%, threshold>48.5%) and 30.2% (19/63, 95%-CI: 19.2-43.0%, threshold> 33.3%), respectively. Using con- ventional thresholds for APW (≥60%) and RPW (≥40%), their sensitivities were 4.8% (3/63, 95% CI: 1.0-13.3%) and 17.5% (11/63, 95%-CI: 9.1-29.1%), respectively, with 100% specificity for both. The sensitivity of RER to achieve 100% specificity was 63.5% (40/63, 95%-CI: 50.4-75.3%, thresh- old> 239%), which was higher than of the other parameters (all, p<0.001). Among the 27 adenomas undiagnosable with RER, six and three fulfilled the adenoma criteria of APW and RPW, respectively. Combined RER with APW or RPW achieved overall sensitivities of 73.0% (46/63, 95% CI: 60.4-83.4%) or 68.3% (43/63, 95%-CI: 55.3-79.4%), respectively, with 100% specificity.
Diagnostic performance of CS-MRI
Table 4 and Fig. 4 show the diagnostic performance for 66 patients with 70 lesions who underwent CS-MRI. SI-index yielded AUC 0.902 (95%-CI: 0.807-0.960) and sensitivity 24.5% at threshold of>61.1 (12/49, 95%-CI: 13.3-38.9%), which were inferior to those of RER (AUC 0.977
| Parameters | Thresholds* | Sensitivity at 100% specificity | AUC |
|---|---|---|---|
| Non-contrast attenuation | ≤ 13 HU | 7.9% (5/63) [2.6%-17.6%] | 0.894 [0.821-0.944] |
| APW | >48.5% | 25.4% (16/63) [15.3%-37.9%] | 0.786 [0.698-0.858] |
| RPW | >33.3% | 30.2% (19/63) [19.2%-43.0%] | 0.904 [0.833-0.952] |
| RER | >238% | 63.5% (40/63) [50.4%-75.3%] | 0.969 [0.917-0.993] |
| Combined RER and APWt | 73.0% (46/63) [60.4%-83.4%] | 0.974 [0.924-0.995] | |
| Combined RER and RPW# | 68.3% (43/63) [55.3%-79.4%] | 0.974 [0.924-0.995] |
Data in parentheses and brackets are numerators/denominators and 95% confidence intervals, respectively *Thresholds to attain the highest sensitivity at 100% specificity to discriminate adenomas from metastases ¡ adenomas with RER>238% and/or APW>48.5%, # adenomas with RER> 238% and/or RPW> 33.3% APW absolute percentage washout, AUC area under the receiver operating characteristic curve, RPW rela- tive percentage washout, RER relative enhancement ratio
| Parameters | Thresholds | Sensitivity | AUC |
|---|---|---|---|
| Non-contrast attenu- ation | ≤ 15 HU* | 18.4% (9/49) [8.9%-32.0%] | 0.890 [0.792-0.952] |
| ≤ 13 HUt | 4.1% (2/49) [0.5%-14.0%] | ||
| APW | >48.5%* | 22.5% (11/49) [11.8%-36.6%] | 0.815 [0.704-0.898] |
| >48.5%+ | 22.5% (11/49) [11.8%-36.6%] | ||
| RPW | >31.9%* | 44.9% (22/49) [30.7%-59.8%] | 0.917 [0.827-0.970] |
| >33.3%+ | 32.7% (19/49) [19.9%-47.5%] | ||
| RER | >206%* | 81.6% (40/49) [68.0%-91.2%] | 0.977 [0.909-0.998] |
| >238%+ | 63.3% (31/49) [48.3%-76.6%] | ||
| SI-index | >61%* | 24.5% (22/49) [13.3%-38.9%] | 0.902 [0.807-0.960] |
Data in parentheses and brackets are numerators/denominators and 95% confidence intervals, respectively *Thresholds determined at patients (49 adenomas and 21 metastases) who underwent CS-MRI to attain 100% specificity with the highest sensitivity
+Thresholds determined at entire study sample (63 adenomas and 49 metastases) to attain 100% specificity with the highest sensitivity
APW absolute percentage washout, AUC area under the receiver operating characteristic curve, RPW rela- tive percentage washout, RER relative enhancement ratio, CS-MRI chemical-shift MRI
100
80
Sensitivity (%)
60
Non-contrast
40
APW
RPW
20
RER
SI-index
0
0
20
40
60
80
100
100-Specificity (%)
[95%-CI: 0.909-0.998]; sensitivity 63.3% [31/49, 95%-CI: 48.3-76.6%] at thresholds of> 237% determined on entire sample [n = 111] and 81.6% [40/49, 95%-CI: 68.0-91.2%] at threshold of > 206% determined on subgroup patients [n =70], all p<0.001). When using previously reported
threshold (> 16.5%), the sensitivity and specificity of SI- index were 87.8% (42/49, 95%-CI: 74.8-95.3%) and 71.4% (15/21, 95%-CI: 47.8-88.7%), respectively.
Diagnostic performance of CT and CS-MRI according to non-contrast attenuation
Supplemental Table 2 shows the subgroup sensitivity for hypo- and hyper-attenuating adenoma. In entire sample, APW and RPW had a sensitivity of 25.6% (10/39, 95%- CT: 13.0-42.1%) and 41.0% (16/39, 95%-CT: 25.6-57.9%), respectively, for hypoattenuating adenomas, while the sen- sitivity of both was 25.0% (6/24, 95%-CT: 9.7-46.7%) for hyperattenuating adenomas. RER yielded the sensitivity of 87.2% (34/39, 95%-CT: 72.6-95.7%) and 25.0% (6/24, 95%-CT: 9.7-46.7%) for hypo- and hyper-attenuating adenomas, respectively. Adding either APW or RPW to RER improved the sensitivity for hyper-attenuating adeno- mas (50.0% [12/24, 95%-CI: 29.1-70.9%] or 37.5% [9/24, 95%-CI: 18.8-59.4%], respectively) but did not improve sensitivity for hypo-attenuating adenomas (87.2% [34/39, 95%-CI: 72.6-95.7%]). In the subgroup of patients who underwent CS-MRI, the SI-index provided a sensitivity of 36.7% (11/30, 95%-CI: 19.9-56.1%) for hypo-attenuating adenomas and 5% (1/19, 95%-CI: 0.1-26.0%) for hyper- attenuating adenomas. Representative cases are shown in Fig. 5 and 6.
Discussion
Characterizing lipid-poor adrenal lesions in patients with HCC may pose a diagnostic challenge, because both adenomas and HCC metastases can include fat contents
Non-contrast
PVP
DP
IP
OP
Adenoma
17 HU
70 HU
69 HU
928
487
Metastasis
16 HU
49 HU
43 HU
810
315
low non-contrast attenuation (17 and 16 HU) and signal drop at OP (signal intensity index: 47.5 and 61), indicative of microscopic fat content. The APW, RPW, and RER for the adenoma were 2, 1, and 312%, respectively, while for the metastasis they were 18, 12, and 206%, respectively. In the case of adenoma, neither APW nor RPW met the adenoma criteria (>48.5% and> 33.3%, respectively), while RER could accurately discriminate from metastasis with a threshold of 238%
Non-contrast
PVP
DP
IP
OP
Adenoma
28 HU
96 HU
76 HU
587
500
Metastasis
32 HU
76 HU
63 HU
499
400
and exhibit overlapping contrast washout characteristics. Our results showed that lipid-poor adenomas had lower non-contrast attenuation (22.9±7.1 vs. 37.9±9.4 HU),
contrast attenuation (28 and 32 HU), while signal intensity index on chemical-shift MRI was higher in metastasis than in adenoma (19.8 vs. 14.8), making the diagnosis impossible. The APW, RPW, and RER for the adenoma were 29, 21, and 243%, respectively, while for the metastasis they were 30, 17, and 137%, respectively. In the case of adenoma, neither APW nor RPW met the adenoma criteria (>48.5% and> 33.3%, respectively), while RER could accurately discriminate from metastasis with a threshold of 238%
higher APW (40.5% ± 12.8% vs. 23.7% ±17.4%), higher RPW (30.0% ±10.2% vs. 12.4%±9.6%) on multiphase hepatic CT, and higher SI-index on CS-MRI (43.3±20.7
vs.10.8 ± 14.4) than HCC metastases (all, p<0.001). Nev- ertheless, their sensitivities to achieve 100% specificity for adenoma diagnoses were limited (7.9% [11/63], 25.4% [16/63], 30.2% [19/63], and 24.5% [22/49], respectively). RER incorporating non-contrast attenuation and contrast wash-in kinetic was higher in adenomas than in metastases (329% ±152% vs. 111%±43.0%), with the highest sensitiv- ity of 63.5% (40/63). Combined washout parameters with RER allowed for adenoma diagnosis with a sensitivity of 73.0% (46/63) while maintaining 100% specificity, demon- strating the potential of multiphase hepatic CT as a conveni- ent and reliable diagnostic alternative.
The observed lower non-contrast attenuation in adeno- mas compared to HCC metastases is consistent with a few previous reports. Yasaka et al. demonstrated that adenomas exhibited lower non-contrast attenuation than metastases (16.8±12.9 vs. 43.1±9.4 HU), with a sensitivity of 46.2% and specificity of 100% at a threshold < 17 HU [31]. The inferior sensitivity in our study (7.9%) can be explained by the fact that we included only lipid-poor adenomas (> 10 HU on non-contrast CT). Harada et al. also demonstrated a lower non-contrast attenuation in lipid-poor adenomas compared to HCC metastases (29.0±6.4 vs. 48.1 ±8.5 HU), achiev- ing a sensitivity of 94.4% and specificity of 92.3% with a threshold of <36 HU [32]. The superior performance in comparison to our study may be attributed to the smaller sample size of their study, which consisted of 13 adenomas and 18 metastases.
Adrenal metastases from HCC can contain microscopic fat, which resembles adenomas [13, 14]. The intratumoral fat is less frequently observed in biologically aggressive or progressed HCC, which is prone to metastasis, compared to early or well-differentiated HCC [33-36]. This explains why most HCC metastases showed higher non-contrast attenua- tion and higher SI-index than adenomas. Nevertheless, a few HCC metastases in this study did exhibit low non-contrast attenuation and high SI-index, likely due to the intratumoral fat. As a result, both parameters demonstrated insufficient sensitivity to achieve 100% specificity. In particular, CS- MRI can detect microscopic fat in metastases more sensi- tively than non-contrast CT, suggesting that its interpretation requires increased caution in patients with fat-containing HCC.
Previous literatures have described adrenal HCC metas- tases as “hypervascular” [9, 16, 37, 38], whereas the paucity of data has confirmed this description. Choi et al. demon- strated that hypervascular metastases showed contrast kinet- ics comparable to adenomas [9]. However, their findings were mainly obtained from renal cell carcinoma metastases (n=13), with only three HCC metastases included, and no subgroup analysis was performed. The other study suggested that early absolute enhancement was lower in HCC metas- tases than in lipid-poor adenomas (37.1 vs. 51.7 HU) [32],
which is consistent with our findings. HCC with a high met- astatic potential (e.g. poorly differentiated and aggressive subtypes) tends to exhibit low vascularity and intra-tumoral necrosis [39], which may account for the observed low abso- lute enhancement in adrenal HCC metastases.
In the context of the contrast kinetic parameters, it’s important to note their distinct relationships with non- contrast attenuation and how they influence the assessment of hypo- and hyper-attenuating adenomas. APW is deter- mined by the ratio of early contrast wash-in and delayed wash-out, yet it is not influenced by baseline non-contrast attenuation, despite the latter being part of the formula. This independence with non-contrast attenuation (r=0.06) explains why APW showed nearly identical performance for adenomas with hyper- and hypo-attenuation on non-contrast CT (sensitivity: 25.6% vs. 25.0%). On the contrary, RPW indirectly incorporates non-contrast attenuation, because the denominator of the formula equals non-contrast attenu- ation plus absolute enhancement during the PVP. Although the formula doesn’t explicitly include non-contrast attenu- ation, RPW was negatively correlated with non-contrast attenuation (r =- 0.222) and tended to exhibit lower values for hypo-attenuating adenomas than for hyper-attenuating adenomas. This suggests that the washout characteristic of hypo-attenuating adenomas is more emphasized in RPW, making them more distinguishable compared to hyper- attenuating adenomas, as evidenced in this study (sensitiv- ity: 41.0% vs. 25.0%). Nevertheless, the overall sensitivity of washout parameters was insufficient, especially when using currently accepted thresholds. This may be attributed to the short delay time [40, 41] and the possible overlapping wash- out kinetics between the adenomas and HCC metastases.
RER can reflect both low non-contrast attenuation and early high contrast wash-in features of adenomas, contrib- uting to a relatively high overall sensitivity (63.5%). How- ever, RER was strongly affected by baseline non-contrast attenuation, decreasing as non-contrast attenuation levels rise (r =- 0.772). Consequently, there was a notable decline in performance for hyper-attenuating adenomas compared to hypo-attenuating ones (25% vs. 87.2%). Despite this limita- tion, several hyper-attenuating adenomas missed by RER can be correctly diagnosed by APW, and the combined use of both parameters achieves an overall sensitivity of 73%, indicating complementary diagnostic roles. By employing this combined approach, additional diagnostic procedures, such as a dedicated 15-min delay scan, PET/CT, or biopsy, can be eliminated for most lipid-poor adrenal lesions identi- fied in HCC patients.
Our study introduces several advancements compared to previous research. Firstly, this investigation assessed the diagnostic performance of the RER specifically in HCC patients. This specificity is crucial, as the diagnostic accu- racy of adrenal imaging can vary significantly based on
the targeted population. Secondly, our study evaluated the diagnostic performance of CS-MRI in discriminating lipid- poor adrenal adenomas from HCC metastases. Although the diagnostic pitfall of adrenal CS-MRI in HCC patients has been reported, no prior investigations have explored its actual performance. This fills a significant gap in the existing literature and enhances the applicability of our findings to clinical practice. Thirdly, this is the largest study evaluat- ing the diagnostic performance of the washout parameters in HCC patients, underscoring the unique contributions of our research. These assessments allowed the clarification of the complementary diagnostic role of RER and washout parameters on routine hepatic CT and their possible superi- ority to CS-MRI.
This study has limitations. First, this was a retrospective single-center study with a relatively small sample size. Sec- ond, most diagnoses were based on imaging criteria rather than histopathology, as in most adrenal imaging studies. Third, optimal cut-off and diagnostic performance of CT parameters may not be directly extrapolatable to images obtained using different iodine doses and scan timings. Fourth, seven patients had bilateral adrenal lesions, which could introduce clustering effects into the statistical analy- sis. Fifth, we were unable to correlate anatomopathological results with image findings regarding the fat component, primarily due to the absence of corresponding descriptions in the pathology reports. Finally, adrenal tumors other than adenomas and metastases, such as pheochromocytomas and adrenocortical carcinomas, could not be identified in the study sample due to their rarity [42].
In conclusion, RER discriminated lipid-poor adrenal adenomas from HCC metastases with higher sensitivity at 100% specificity than non-contrast attenuation and contrast- washout parameters on multiphase hepatic CT, as well as SI- index on CS-MRI. The combination of RER with washout parameters yielded the highest sensitivity, which may elimi- nate the need for additional diagnostic procedures for most lipid-poor adrenal lesions identified in patients with HCC.
Supplementary Information The online version contains supplemen- tary material available at https://doi.org/10.1007/s00261-024-04228-5.
Declarations
Conflict of interest No conflict of interest to declare.
References
1. Villanueva A (2019) Hepatocellular Carcinoma. N Engl J Med 380:1450-1462
2. Marrero JA, Kulik LM, Sirlin CB et al (2018) Diagnosis, Stag- ing, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology 68:723-750
3. Chernyak V, Fowler KJ, Kamaya A et al (2018) Liver Imaging Reporting and Data System (LI-RADS) Version 2018: Imag- ing of Hepatocellular Carcinoma in At-Risk Patients. Radiology 289:816-830
4. Lee Y-T, Wang JJ, Zhu Y, Agopian VG, Tseng H-R, Yang JD (2021) Diagnostic Criteria and LI-RADS for Hepatocellular Car- cinoma. Clinical Liver Disease 17:409-413
5. Katyal S, Oliver JH, 3rd, Peterson MS, Ferris JV, Carr BS, Baron RL (2000) Extrahepatic metastases of hepatocellular carcinoma. Radiology 216:698-703
6. Uchino K, Tateishi R, Shiina S et al (2011) Hepatocellular carci- noma with extrahepatic metastasis. Cancer 117:4475-4483
7. Mayo-Smith WW, Song JH, Boland GL et al (2017) Manage- ment of Incidental Adrenal Masses: A White Paper of the ACR Incidental Findings Committee. J Am Coll Radiol 14:1038-1044
8. Caoili EM, Korobkin M, Francis IR et al (2002) Adrenal masses: characterization with combined unenhanced and delayed enhanced CT. Radiology 222:629-633
9. Choi YA, Kim CK, Park BK, Kim B (2013) Evaluation of adre- nal metastases from renal cell carcinoma and hepatocellular carcinoma: use of delayed contrast-enhanced CT. Radiology 266:514-520
10. Haider MA, Ghai S, Jhaveri K, Lockwood G (2004) Chemical shift MR imaging of hyperattenuating (>10 HU) adrenal masses: does it still have a role? Radiology 231:711-716
11. Aldhufian M, Sheinis Pickovsky J, Alfaleh H, Melkus G, Schieda N (2023) Prevalence of ‘Fat-Poor’ Adrenal Adenomas at Chemi- cal-Shift MRI. Can Assoc Radiol J. https://doi.org/10.1177/08465 371231179881:8465371231179881
12. Taffel MT, Petrocelli RD, Rigau D et al (2023) Prevalence of Malignancy in Adrenal Nodules With Heterogeneous Microscopic Fat on Chemical-Shift MRI: A Multiinstitutional Study. AJR Am J Roentgenol 220:86-94
13. Sydow BD, Rosen MA, Siegelman ES (2006) Intracellular lipid within metastatic hepatocellular carcinoma of the adrenal gland: a potential diagnostic pitfall of chemical shift imaging of the adre- nal gland. AJR Am J Roentgenol 187:W550-551
14. Tariq U, Poder L, Carlson D, Courtier J, Joe BN, Coakley FV (2012) Multimodality Imaging of Fat-Containing Adrenal Metas- tasis From Hepatocellular Carcinoma. Clinical Nuclear Medicine 37:e157-e159
15. Adam SZ, Nikolaidis P, Horowitz JM et al (2016) Chemical Shift MR Imaging of the Adrenal Gland: Principles, Pitfalls, and Appli- cations. Radiographics 36:414-432
16. Schieda N, Siegelman ES (2017) Update on CT and MRI of Adre- nal Nodules. AJR Am J Roentgenol 208:1206-1217
17. Elsayes KM, Elmohr MM, Javadi S et al (2020) Mimics, pitfalls, and misdiagnoses of adrenal masses on CT and MRI. Abdom Radiol (NY) 45:982-1000
18. Nagayama Y, Inoue T, Kato Y et al (2021) Relative Enhancement Ratio of Portal Venous Phase to Unenhanced CT in the Diagnosis of Lipid-poor Adrenal Adenomas. Radiology 301:360-368
19. Lee HY, Oh YL, Park SY (2021) Hyperattenuating adrenal lesions in lung cancer: biphasic CT with unenhanced and 1-min enhanced images reliably predicts benign lesions. European Radiology 31:5948-5958
20. Jin S, Zhang H, Pan W et al (2022) Diagnostic value of the rela- tive enhancement ratio of the portal venous phase to unenhanced CT in the identification of lipid-poor adrenal tumors. Abdominal Radiology 47:3308-3317
21. Kim MJ, Kang KA, Kim CK, Park SY (2023) Inter-method agree- ment between wash-in and wash-out computed tomography for characterizing hyperattenuating adrenal lesions as adenomas or non-adenomas. Eur Radiol 33:2218-2226
22. Loonis A-ST, Yu H, Glazer DI, Bay CP, Sodickson AD (2023) Dual Energy-Derived Metrics for Differentiating
Adrenal Adenomas From Nonadenomas on Single-Phase Contrast-Enhanced CT. American Journal of Roentgenology 220:693-704
23. Akbulut S, Erten O, Kahramangil B et al (2021) A Critical Analy- sis of Computed Tomography Washout in Lipid-Poor Adrenal Incidentalomas. Annals of Surgical Oncology 28:2756-2762
24. Corwin MT, Badawy M, Caoili EM et al (2022) Incidental Adre- nal Nodules in Patients Without Known Malignancy: Prevalence of Malignancy and Utility of Washout CT for Characterization- A Multiinstitutional Study. American Journal of Roentgenology 219:804-812
25. Jin S, Zhang H, Pan W et al (2022) Diagnostic value of the rela- tive enhancement ratio of the portal venous phase to unenhanced CT in the identification of lipid-poor adrenal tumors. Abdominal Radiology. https://doi.org/10.1007/s00261-022-03593-3
26. Nagayama Y, Hirai T (2023) Letter to the editor on “diagnos- tic value of the relative enhancement ratio of the portal venous phase to unenhanced CT in the identification of lipid-poor adrenal tumors”. Abdom Radiol (NY) 48:1841-1842
27. Fassnacht M, Arlt W, Bancos I et al (2016) Management of adre- nal incidentalomas: European Society of Endocrinology Clinical Practice Guideline in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol 175:G1-g34
28. Corwin MT, Navarro SM, Malik DG et al (2019) Differences in Growth Rate on CT of Adrenal Adenomas and Malignant Adrenal Nodules. American Journal of Roentgenology 213:632-636
29. Fujiyoshi F, Nakajo M, Fukukura Y, Tsuchimochi S (2003) Char- acterization of adrenal tumors by chemical shift fast low-angle shot MR imaging: comparison of four methods of quantitative evaluation. AJR Am J Roentgenol 180:1649-1657
30. Cicchetti DV (1994) Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment 6:284-290
31. Yasaka K, Gonoi W, Akai H et al (2013) Differentiation of adre- nal tumors in patients with hepatocellular carcinoma: adrenal adenoma versus metastasis. Eur J Radiol 82:1213-1218
32. Harada T, Kamishima T, Terae S, Onodera Y, and Shirato H (2013) Multi-Detector-Row CT Diagnosis of Adrenal Inciden- taloma in Patients with Hepatocellular Carcinoma. Advances in Computed Tomography 2:34-40
33. Sano K, Ichikawa T, Motosugi U et al (2011) Imaging study of early hepatocellular carcinoma: usefulness of gadoxetic acid- enhanced MR imaging. Radiology 261:834-844
34. Siripongsakun S, Lee JK, Raman SS, Tong MJ, Sayre J, Lu DS (2012) MRI detection of intratumoral fat in hepatocellular carci- noma: potential biomarker for a more favorable prognosis. AJR Am J Roentgenol 199:1018-1025
35. Calderaro J, Couchy G, Imbeaud S et al (2017) Histological sub- types of hepatocellular carcinoma are related to gene mutations and molecular tumour classification. J Hepatol 67:727-738
36. Min JH, Kim YK, Lim S, Jeong WK, Choi D, Lee WJ (2015) Prediction of microvascular invasion of hepatocellular carci- nomas with gadoxetic acid-enhanced MR imaging: Impact of intra-tumoral fat detected on chemical-shift images. Eur J Radiol 84:1036-1043
37. Nandra G, Duxbury O, Patel P, Patel JH, Patel N, Vlahos I (2020) Technical and Interpretive Pitfalls in Adrenal Imaging. Radio- graphics 40:1041-1060
38. Chung R, Garratt J, Remer EM et al (2023) Adrenal Neoplasms: Lessons from Adrenal Multidisciplinary Tumor Boards. Radio- graphics 43:e220191
39. Asayama Y, Yoshimitsu K, Nishihara Y et al (2008) Arterial blood supply of hepatocellular carcinoma and histologic grading: radio- logic-pathologic correlation. AJR Am J Roentgenol 190:W28-34
40. Sangwaiya MJ, Boland GW, Cronin CG, Blake MA, Halpern EF, Hahn PF (2010) Incidental adrenal lesions: accuracy of charac- terization with contrast-enhanced washout multidetector CT- -10-minute delayed imaging protocol revisited in a large patient cohort. Radiology 256:504-510
41. Park S-W, Kim TN, Yoon JH et al (2012) The washout rate on the delayed CT image as a diagnostic tool for adrenal adenoma verified by pathology: a multicenter study. International Urology and Nephrology 44:1397-1402
42. Ebbehoj A, Li D, Kaur RJ et al (2020) Epidemiology of adrenal tumours in Olmsted County, Minnesota, USA: a population-based cohort study. The Lancet Diabetes & Endocrinology 8:894-902
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.