In the adrenal glands, adenomas and metastases are the most common benign and malignant tumors, respectively. In some cases such as patients with primary malignant tumors, identification the incidentalomas’ biological characteristics is somewhat challenging. However, various spectral parameters generated by spectral imaging based on dual-energy CT systems may provide more helpful clues to tackle this clinical problem. Our results revealed that all spectral parameters were significantly different between adenomas and metastases in the venous phase (all p < 0.05). The combined spectral parameters (CTVNC value + s-SHC value + Z-eff value + ID value) showed a better diagnostic performance in the venous phase than in the arterial or delayed phase (all p < 0.05). The AUCs of iodine-to-CTVNC, CTVNC and s-SHC values were higher than those of other spectral parameters in the venous phase in differentiating adenomas, lipid-rich adenomas and lipid-poor adenomas from metastases, respectively.
In contrast enhanced CT scanning, adrenal adenomas typically display fast wash-in and rapid washout, whlie metastases reveal slow wash-in and relatively sustained enhancement [4]. Wash-out scan of adrenal commonly perform on 10 or 15 minutes after contrast injection requiring additional patient’s separate visit of CT scan. Several studies have shown the wash-out threshold of adenomas differring with the time of delayed phase scan[18–19, 21]. Our study started delayed phase scanning on a short-time of 180 second after contrast injection, which was similar to Ng’s and Liu’s study[18, 19]. They started delayed phase scanning on 120∼248 second and 200 second, respectively. The diagnostic performance of RPW were better than APW in difference of adrenal benign and malignant tumors, in accordance with previous studies. The 21% criterion of RPW for adenomas yielded 94% sensitivity and 90% specificity, while for lipid-rich adenomas yielded 100% sensitivity and 90% specificity.
There are multiple phases in enhanced adrenal CT examinations, but few studies have analyzed the diagnostic performances of spectral parameters in different scanning phases to differentiate adenomas from metastases. The above results indicated that all spectral parameters in the venous phase (CTVNC value, s-SHC value, Z-eff value, ID value, NID value and iodine-to-CTVNC value) were significantly different between adenomas and metastases. Furthermore, the diagnostic performance of the combined spectral parameters in the venous phase was superior to that in the arterial or delayed phase (P < 0.05).
VNC images are reconstructed by postprocessing spectral CT scans by removing iodine from tumors, tissues and vessels, thus generating images similar to true noncontrast (TNC) images[22] Connolly et al.[23] pointed out that the sensitivity and specificity of the CTVNC value in the diagnosis of adrenal adenoma are relatively low, i.e., 54% and 57%, respectively. To improve the diagnostic value, a study by Nagayama et al[20] employed the iodine-to-CTVNC ratio for the differential diagnosis of adrenal adenomas and metastases, and the AUC of the iodine-to-CTVNC value was 0.98. The sensitivity and specificity of iodine-to-CTVNC value were both 95%. Our results revealed that the AUC of the iodine-to-CTVNC ratio was 0.920. With threshold of 4.18, the sensitivity and specificity of the ratio were 92% and 74%, respectively. Differences in iodine density values for adrenal tumors may be responsible for the differences between Nagayama’s results (adenomas vs. metastases, 2.4 mg/mL vs. 1.7 mg/mL) and ours (adenomas vs. metastases, 1.75 mg/mL vs. 1.16 mg/mL). Differences in scanning protocols including the injection rate of the contrast agent and scanning phase in two studies may lead to differences in the iodine density of lesions.
The iodine density is one of the most commonly used spectral parameters and has shown promising results for the diagnoses of various research such as metastatic lymph node, pulmonary thromboembolism, pancreatitis, and adrenal tumors[24–27]. Martin et al.[26] showed that in the portal phase, the ID value of adenomas was significantly lower than that of metastases (1.3 ± 0.4 mg/ml vs. 3.2 ± 1.4 mg/ml, P < 0.001). Contrary to their study, our results revealed that the ID value of adenomas in the venous phase was higher than that of metastases (1.75 ± 0.93 mg/mL vs. 1.16 ± 0.55 mg/mL, P < 0.05). Reasons for the differences between the results of Martin’s study and ours may be as follows: (1) the majority of primary cancers in the metastasis group were renal carcinoma (12/26, 46.2%) in Martin’ s study. Previous findings[28] suggested that lipids are present in adrenal metastases from renal carcinoma, decreasing the ability of the metastases to uptake iodine, which may lead to differences between the two studies. (2) Martin’ s study started portal phase scanning 70 s after the contrast agent injection, while we used the operating mode of scan triggering. Individual variations in cardiac function and hemodynamic characteristics may be more obvious without scan triggering mode. (3) A dual-source CT (DSCT) system was employed in Martin’ s study, while a DLSCT system was used in this study. The measurement error of iodine density is the difference between the iodine density measured in a phantom and that measured on images generated by the CT system. Phantom studies[29] revealed that this error for the DSCT system is far greater than that of the DLSCT system.
Virtual monoenergetic image (VMI) derived from dual-energy CT provide great diagnostic value[30]. The s-SHC value can reflect the spectral curve shape of a specific tissue, and the slope of the 40∼100-keV curve is usually employed. Previous studies proposed that the spectral curves of adenomas generally have an ascending or descending shape, while most curves of metastases have a descending shape in plain images[31, 32]. Our study pointed out that s-SHC values of lipid-rich adenomas, lipid-poor adenomas and metastases were 0.71, 2.05 and 1.18, respectively. With threshold of 4.18, the AUC of the s-SHC value (0.920) was higher than that of other parameters in distinguishing lipid-poor adenomas and metastases, and the diagnostic sensitivity and specificity of the s-SHC value were 79.1% and 93.1%, respectively.
There were a few limitations in this research. First, this was a single-center clinical trial with a relatively small sample size. Second, most metastatic lesions were diagnosed based on imaging evidence. This was similar to many previous studies of adrenal imaging[12, 26, 33].Third, the spectral characteristics of adrenal metastases from different primary malignant tumors have not been analyzed owing to the small sample size, and further research is warranted for thorough investigation. Forth, only spectral parameters of adrenal adenomas and metastese were assessed in our research, further investigation of other incidentalomas including benign and malignant are needed.
In conclusion, on DLSCT, the combined spectral parameters in the venous phase are efficient in distinguishing adrenal adenomas from metastases. The iodine-to-CTVNC ratio, CTVNC value and s-SHC value (0.920, 0.966 and 0.920, respectively) had the highest AUCs in differentiating adenomas, lipid-rich adenomas and lipid-poor adenomas from metastases, respectively. Therefore, spectral imaging provides the morphological features, enhancement patterns, attenuations wash-out values, and spectral parameters of adrenal lesions, which is beneficial for the differential diagnoses of adrenal adenomas and metastases.