Consecutive HCC patients who underwent TACE at our hospital from February 2013 to October 2014 were included. After excluding patients with cardiovascular or respiratory disease, poor renal function, allergy to contrast media, or contraindications for artery puncture, a total of 30 patients were finally included. There were 19 men and 11 women, and their age ranged from 36 to 72 years, with a mean value of 57.2 years. TACE was performed according to the standards for the diagnosis and treatment of primary liver cancer (updated in 2011). Follow-up and treatment interval were based on the same standards. The study protocol was approved by the local ethics committee, and all patients provided written informed consent prior to TACE and GSI scans. The 30 patients received 1-3 TACE procedures, and GSI scans were performed 2-4 days before TACE treatment.

Dual-phase GSI CT scans: A GE discovery CT750 HD CT scanner was used for plain and GSI CT scans. All patients were fasted for at least 8 h before the scans and took 800-1000 mL water before examination to fill the upper gastrointestinal tract to help identify the relationship between organs and avoid possible missed diagnosis of abdominal masses. The scanning range was from the dome of the diaphragm to the inferior poles of the two kidneys. After acquiring the scanogram, plain abdominal scanning was performed. The nonionic contrast medium iohexol (350 mg/mL, 80 mL) was then injected via the elbow vein at a flow rate of 2.9 mL/s. The amount of iohexol used for weak patients was calculated based on the body weight (kg), and the flow rate and the total amount of contrast medium were reduced correspondingly. The arterial and portal venous phase scans were started 35 s and 70 s after the start of injection of the contrast medium, respectively. Subsequently, the GSI mode was used to conduct dual-phase contrast enhanced scanning to acquire arterial and portal venous phase QC images and spectral images. Scan parameters included: voltage, 80/140 kVp; switching within 0.5 ms; default mA; helical pitch, 1.375:1; collimation width, 40 mm; slice thickness, 5 mm; reconstruction slice thickness, 1.25 mm and 0.625 mm.

DSA of the liver: Percutaneous femoral artery puncture was performed using the Seldinger technique. DSA of the celiac trunk, common hepatic artery, and superior mesenteric artery was performed. DSA images were acquired in the arterial, parenchymal, and venous phases. After determining tumor location, size, number and tumor-feeding arteries, the left hepatic artery, right hepatic artery, and tumor-feeding arteries were super-selectively catheterized with a microcatheter for infusion chemotherapy and embolization with iodized oil emulsion. The presence of active residual lesions and recurrent or metastatic lesions was judged based on the presence of enlarged tumor-feeding arteries, hypervascularity, tumor stain, and iodized oil deposition [standards for the diagnosis and treatment of primary liver cancer (updated in 2011)].

QC images and spectral images were acquired at one time of scan. Images were transferred to the AW4.6 workstation and processed using GSI-viewer software. One hundred and one monochromatic images at kV levels from 40 to 140 keV, iodine (water)-based images, and spectral characterization diagrams (spectral curve, scatter plot, etc.) were reconstructed. Based on the carrier-to-noise ratio (CNR), the images in which the lesion and adjacent liver tissue had the best CNR were selected from the 101 monochromatic images. The areas that had dense iodized oil deposition or radial artifacts should be avoided. On the QC images, optimal monochromatic images and iodine-based images, regions of interest (ROIs) of the same size were selected in the central slice of normal liver tissue and lesions in largest diameter to measure the CT value and effective iodine content and plot spectral characterization charts (spectral curve, scatter plot, etc.) to observe the characteristics of the spectral characterization charts (including slope of curve and distribution characteristics).

Two experienced radiologists who were blinded to the clinical details independently evaluated the status of tumor blood supply and the presence of active residual lesions and recurrent or metastatic lesions in two reading sessions (A and B), and the results were compared with DSA findings. For session A, QC images acquired in plain scans and in the arterial and venous phases (equivalent to conventional CT) were observed. For session B, optimal monochromatic images, iodine (water)-based images and spectral characteristics were observed. The consistency of the results of sessions A and B with DSA findings was evaluated. Discordant findings were resolved by consensus review by a third radiologist.

All statistical analyses were performed using SPSS17.0 software. Using DSA results as the reference golden standard, the sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic coincidence rate for sessions A and B were calculated. The diagnostic performance between the two sessions was compared using the χ² test, and Kappa test was used to evaluate the consistency of the two sessions with DSA. P values < 0.05 were considered statistically significant.

DSA can accurately detect tumor vessels, tumor stain, arteriovenous fistula and collateral circulation in active residual lesions after TACE treatment, and is therefore the most sensitive and specific method for evaluating the efficacy of TACE[11]. However, DSA is invasive, and postoperative monitoring relying on repeated DSA examinations is not practical. Currently, there are many imaging modalities available for efficacy evaluation in HCC after TACE, with CT and MRI being the main ones. MRI is not affected by iodized oil deposition, and it has been reported[12,13] that the sensitivity, specificity and accuracy of MRI are superior to those of CT and ultrasound; however, many contraindications and long scanning time limit its use. In addition, due to some overlap between DWI signals for viable tumor area and tumor necrotic area, it may be difficult to identify some residual tumors and lesions lacking rich blood supply. MSCT can better show the status of enhanced blood vessels and tumor blood supply in HCC, and it is the first choice of imaging modality for postoperative monitoring in HCC after TACE treatment[14,15]. However, conventional CT is based on mixed energy, can produce beam hardening artifacts, and cause the “drift” of CT value. Moreover, due to the impact of many other factors such as postoperative hemodynamic changes, uneven iodized oil deposition, radial artifacts around iodized oil deposition, degree of cell differentiation in lesions, non-rich arterial blood supply, arteriovenous double blood supply, and cirrhotic background, there may be poor contrast between lesions and the surrounding tissue. As a result, in some cases, conventional plain and contrast-enhanced CT scans may show equidensity or atypical enhancement pattern, thus weakening the ability of conventional CT to observe lesions around embolized lesions and evaluate residual tumor tissue, and decreasing the performance of CT in efficacy evaluation. In addition, due to the lack of blood supply in necrotic tumor tissue or the presence of coagulation necrosis after arterial embolization, defects of iodized oil deposition may be present. The defects of iodized oil deposition do not necessarily indicate the presence of residual tumors or recurrence[16]. Therefore, conventional CT has a low detection rate for small lesions and lesions lacking blood supply, and missed diagnosis is not uncommon[17]. During the process of evaluation of polychromatic images (equivalent to conventional CT) in this study, although ROIs were placed to avoid areas of dense iodized oil deposition and radial artifacts as far as possible, there were still 12 cases that were misdiagnosed as having enhancement due to reasons including iodized oil deposition. Of the 70 missed lesions, 28 were due to lack of blood supply without enhancement, 22 were identified to have equidensity due to iodized oil interference and cirrhotic background, and 18 were misdiagnosed as cysts.

GSI adopts new high-purity and high-permeability detector materials, realizing fast kVp switching and single energy CT imaging. Therefore, GSI has obvious advantages and breakthroughs compared with conventional CT[18,19]. By optimizing the image quality and choosing the optimal monochromatic images, GSI is conducive to eliminating beam hardening artifacts produced in conventional CT and avoiding average attenuation effect, thus increasing the contrast-noise ratio of images. Based on the difference in absorption coefficients of different materials under different single photon energy, GSI can also produce different characteristic absorption curves, thereby realizing the separation and identification of materials[20], i.e., the spectral curves of different materials at different levels of KeV have different resolutions. Iodine based images can be used to directly determine iodine content and normalized iodine concentration, thereby indirectly reflecting blood supply status in different pathological conditions and in different stages of disease progression. Studies[21,22] have shown a significant correlation between the measured value and true value of iodine content in different concentrations of iodine-containing contrast medium, and that there was no significant difference between the measured value and true value. In the spectral curve for iodine, the difference between the CT values of iodine-containing hepatic artery and hepatic parenchyma increases with the decrease of the single energy KeV, thus improving the contrast between the lesions and normal liver tissue and the contrast between tissues, and exerting amplifying and highlighting effects. Therefore, GSI can, under the premise of without increasing radiation dose, improve the detection rate of lesions lacking blood supply, small lesions and multiple lesions, increase the qualitative accuracy, and avoid lesion missing[23-25], thus providing more robust information for accurate characterization, clear positioning, rapid diagnosis and timely treatment of diseases. In the present study, GSI showed that the optimal monochromatic images for lesions were at 60-70 KeV, and this finding is consistent with that reported in the literature[26]. The number of lesions detected by GSI was higher than that detected in polychromatic images (97 vs 72), and GSI findings had a good consistency with those of DSA. The sensitivity and specificity for the GSI group were significantly higher than those of the QC group. Three lesions showing enhancement in GSI were located around tumor lesions and iodized oil deposition areas, but DSA showed negative results. As non-nodular or non-mass lesions, their effective iodine contents and curve slopes were located between those for tumor lesions showing enhancement and normal parenchyma. We speculate that this may be due to reactive granulation tissue formation and/or iodized oil deposition after TACE for HCC. In this study, three lesions showed no enhancement in optimal monochromatic images by iodine quantitative measurement, but showed hypodensity in water (iodine)-based images. They were therefore misdiagnosed as cysts. Postoperatively, iodized oil deposition was visible, suggesting that they were metastases with no/little blood supply. In addition, conventional CT mainly observes the morphology of tumors and the change of CT values, but a significant change of tumor size often occurs 3 and 6 mo after chemotherapy[27]. GSI allows for analysis of the similarity of spectral characteristics between the pre-treated lesion and the post-treated lesion, thus providing objective information about the nature and origin of the lesion and aiding in the differential diagnosis[28-30]. In this study, it was found that the features of spectral curves and scatter plots for most of the detected lesions were substantially consistent with those for the pre-treated lesions, which suggests a homogeneity between them. However, there were few cases of inconsistency, and this was not caused by tumor cells themselves, but may be due to tumor liquefaction necrosis, few residual tumor cells, and many liquefaction necrosis foci in the ROIs.

Taken together, tumor morphological changes usually occur later than blood flow changes in HCC after TACE treatment, and therefore, morphological evaluation does not meet the requirements of efficacy evaluation. Compared to traditional CT value, effective iodine content in GSI has a higher correlation with iodine concentration, has a smaller error, can more accurately analyze material composition, can reflect real iodine concentration, and is more convincing. By measuring the amount of iodine accumulating in local tissue and comparing the change in the contents of iodine between pre-treated and post-treated lesions, GSI can indirectly reflect blood perfusion in tissues/organs and hemodynamic changes in lesions, and is therefore more sensitive and can detect more small lesions. By comparing the features of spectral characterization diagrams between suspicious lesions and primary lesions, GSI allows for more objective judgment and identification of residual lesions and recurrent or metastatic lesions, which is more accurate than conventional CT. However, this study has several limitations. Although a metal artifacts reduction system (MARS) was used in GSI, and areas of dense iodized oil deposition and radial artifacts were avoided as much as possible in the selection of ROIs in this study, some lesions were still misdiagnosed or missed due to iodized oil deposition; few lesions with no/little blood supply showing hypodensity in iodine (water) based images were misdiagnosed as cysts; the selection of size and placement of the ROIs were arbitrary, which may also lead to the change in the slopes of spectral curves. Therefore, future clinical studies are needed to address these issues carefully and optimize this technique.

GSI allows for multi-parametric imaging and overcomes the limitations of conventional CT, which relies on morphology and CT value and is a nonparametric imaging modality. Thus, GSI realizes the transition from morphological to functional imaging, and can therefore more accurately evaluate efficacy and detect active residual tumors and recurrent or metastatic lesions after TACE treatment for HCC, representing a simple, minimally invasive, efficient imaging modality for efficacy evaluation and follow-up after TACE treatment for HCC. GSI also provides reliable imaging evidence for early treatment and is beneficial to the improvement of curative effect and survival.

The incidence and mortality of primary hepatocellular carcinoma (HCC) in China are both among the highest in the world. Because of its low resectable and high recurrent rates, transarterial chemoembolization (TACE) has been an effective non-surgical modality and widely used in clinical practice. However, many HCC patients still have residual tumors after TACE, which results in the easy recurrence and requires repeated treatments. Therefore, early detection of active residual lesions and recurrent or metastatic lesions after TACE are of great clinical significance. Although many imaging modalities are currently available for evaluation in HCC patients after TACE and among which DSA is the golden standard, it is invasive and not suitable for routine monitoring. In this study, we evaluate the efficacy of gemstone spectral imaging (GSI) in hepatocellular cancer after TACE.

GSI has become a hot research topic in the fields of imaging and clinical oncology, for its ability to comprehensively analyze the spectral characteristics of tumor lesions without increasing the radiation dose at the same time. Few previous studies contain evaluations in HCC patients after TACE with GSI. The results of this study contribute to clarifying the potential of GSI for efficacy evaluation and follow-up in HCC patients after TACE.

In this study, GSI was a useful tool to detect the residual and recurrent or metastasis lesions of HCC after TACE. The sensitivity and specificity were 97% (97/100) and 94.4% (51/54), respectively. The results had a good consistency with those of DSA with a diagnostic coincidence rate of 74% (114/154) and were significantly higher than those of the QC group. This emphasizes the accuracy and efficacy of GSI for identifying residual tumors and recurrent or metastatic lesions.

This study suggests that GSI is useful for detecting the residual and recurrent or metastasis lesions of HCC after TACE. HCC patients after TACE can get timely and early further treatment if active residual lesions and recurrent or metastasis lesions are early and accurately detected based on the efficacy evaluation of the GSI.

As the authors indicated, the timely and accurate evaluation of TACE efficacy and early detection of active residual lesions and recurrent or metastatic lesions are of great clinical significance. In this manuscript, the authors assessed the value of gemstone spectral imaging in HCC after TACE treatment. The study is well designed and the results are interesting