Hepatocellular carcinoma (HCC) is the most frequent malignant liver tumor accounting for about 90% of all liver cancers (1). It is a heterogeneous tumor entity classified according to the Barcelona-Clinic Liver Cancer (BCLC) classification into early HCC (stage 0 and A), intermediate (stage B) or advanced stage (stage C) and very poor life expectancy (stage D) (2,3).

Transarterial radioembolization (TARE) is one transarterial treatment option for HCC patients (2-12). TARE can be performed using either glass (TheraSphere^®, Boston Scientific Corporation, Marlborough, MA, USA) or resin (SIR-Spheres^®, Sirtex Medical Inc., Sydney, Australia) microspheres including the β-emitter Y-90 (5,6). The potential clinical benefits of TARE have primarily been proven in advanced stage HCC (4). Prediction of the individual benefit of TARE is an important medical need, as alternative locoregional or systemic treatments become increasingly available (2,4). Current approaches are mainly based on clinical and laboratory parameter scoring systems, such as MELD, ALBI or Child-Pugh score that are strongly influenced by liver function (2-4).

In contrast, imaging-based methods for individual outcome prediction for TARE have not yet been established. Diffusion-weighted imaging (DWI) is an imaging sequence of magnetic resonance imaging (MRI), which quantifies the random water movement in tissues (13-15). Thereby, it is able to reflect the microstructure of tissues, predominantly cellularity (16). DWI based techniques can also be used for quantification of muscle fiber architecture (17,18). Indeed, those techniques have been shown to allow for estimation of muscle strength, training effect on muscles and muscle damage (19-22).

Several studies have reported that sarcopenia is associated with poor prognosis in cancer patients, including those with HCC (23-28). In previous studies, it was demonstrated that sarcopenia, defined by CT images as low-skeletal muscle area (LSMM) or low muscle density, is of prognostic relevance in patients with HCC undergoing transarterial chemoembolization (29,30). However, to date the potential association of muscle status as measured by ADC with survival in HCC patients has not yet been investigated. We therefore aimed to evaluate the association of ADC values of the paraspinal and psoas muscles with survival outcomes in patients with HCC undergoing TARE.

Patient sample acquisition. For our retrospective single center study, we included all patients with HCC treated with TARE between 2016 and 2020. Inclusion criteria of the present analysis was TARE treatment in intermediate HCC stage, regardless of previous treatment, follow-up documentation for at least 12 months after TARE and availability of MRI within one month before TARE. In our center, TARE is considered for patients with large tumor burden, macrovascular invasion, and liver dominant disease.

Clinical and serological parameters. For every patient several clinical and serological parameters were retrieved from the clinical database. Regarding clinical parameters, the presence of cirrhosis, portal vein thrombosis and BCLC stage (2) were retrieved from the clinical records. The following routine laboratory parameters were retrieved: platelets, bilirubin, quick, INR, AST, ALT, albumin, alpha-fetoprotein, and MELD score (31).

TARE. TARE was performed with a standardized technique using 90Y resin microspheres (SIR-Spheres; Sirtex Medical Limited) or 90Y glass microspheres (TheraSphere, Boston Scientific Corp.) (5-7). Forty-two patients were treated with resin spheres (79.2%) and 10 patients (20.8%) with glass spheres. In all cases, the indication for TARE was given by a multidisciplinary tumor board.

In the first angiography setting, 99mTc-MAA was given from a microcatheter located in a hepatic artery to assess lung shunt fraction (32). For the glass microspheres, a lung dose fraction >30 Gy per treatment setting was set as a limit for the treatment. In cases of more TARE settings, a total of 50 Gy was defined as a dose limit. Regarding resin microspheres, 20% of lung shunt was set as a limit, and reduced activity was recommended for patients with 10-20% with lung shunt. The single photon emission/CT was not essential in dosimetry but could demonstrate the effect the of the tumor response.

DWI and ADC measurement. In all patients, MRI of the liver was performed using one clinical 1.5 T scanner (Aera, Siemens Healthcare, Erlangen, Germany). The imaging protocol was comprised of a T2 weighted fat-suppressed short tau inversion recovery sequence, a half-Fourier acquisition single-shot turbo spin-echo T2 weighted sequence, and T1 weighted spin-echo (SE) images prior and after intravenous application of contrast medium with a dynamic acquisition.

In all cases, DWI was acquired with a multi-shot SE-echo planar sequence (repetition time: 7200 ms, echo time: 50 ms, slice thickness: 5 mm, matrix: 88×134, field of view: 450 mm). The sequences were obtained with the b-values of 50, 400 and 800 s/mm. The resulting apparent diffusion coefficient (ADC) maps were provided by the software. For ADC measurements four polygonal regions of interest (ROI) were drawn along the boundaries of the iliopsoas and paravertebral muscles on both sides. The ROIs were carefully placed to avoid large macroscopic fat areas and vessels. The reader was blinded to the clinical outcomes of the patients. The ADC values for iliopsoas and paravertebral muscles were calculated separately as a mean value of both sides. Moreover, a cumulative mean ADC value of the skeletal musculature was estimated out of the 4 ROIs, as described previously (33). Figure 1 displays a case patient for illustration purposes.

Clinical outcomes and assessments. The primary outcome of the present study was overall survival (OS) after 12 months, which was measured from the initial date of HCC diagnosis to the date of death from all causes. All survival data were obtained from the medical reports. The clinical efficacy of TARE was determined by follow-up contrast enhanced dynamic CT or MRI in every patient. Patients without residual viable tumor after TARE were monitored by contrast enhanced CT or MRI every 3 months.

Statistical analysis. The SPSS statistical software package (SPSS 29, SPSS Inc., Chicago IL, USA) was used for statistical analysis. Continuous variables are described by mean values, median and standard deviation. Categorical variables are provided with relative frequencies. ADC values between the tumor groups were compared using the Mann–Whitney U-test (two-group comparison). Multivariate survival analysis was conducted with the Cox regression model. Kaplan-Meier survival analysis and Log-rank test was further used for the 12-month survival analysis. Correlation analysis was performed with Spearman’s correlation coefficient after testing for normal distribution. In all instances, p-values below 0.05 were considered statistically significant.

Patients and treatment outcomes. A total of 52 patients were enrolled into the study (patient characteristics are shown in Table I). Most patients were staged as BCLC B and C [28 (52.8%) and 22 (41.5%) cases, respectively]. Thirty-nine patients (75%) had liver cirrhosis. Overall, 40 patients (76.9%) of the patient cohort died within the 12-month observation period after TARE. The mean OS after TARE was 10.9±10.5 (range=1-53) months. Table II provides data regarding previous patient treatments. Almost half of the patients were treatment naïve (n=25, 48.1%), followed by patients with previous transarterial chemoembolization (n=21, 40.4%).

ADC values before TARE. The ADC values before TARE were 1.27±0.17×10^–3 mm²/s (range=0.69-1.56×10^–3 mm²/s) for psoas muscles, 1.37±0.13×10^–3 mm²/s (range=1.07-1.63×10-3 mm²/s) for the paraspinal muscles and 1.31±0.13×10^–3 mm²/s (range=0.93-1.60×10^–3 mm²/s) for all muscles. The ADC values of the paraspinal muscles were statistically significantly higher (p=0.0031). There were no differences in regard to the mean ADC values of all muscles (p=0.76), psoas muscles (p=0.63), or paraspinal muscles (p=0.19) between survivors and non-survivors within 12 weeks after TARE.

Correlation of ADC values with outcomes after TARE. In the comparison analysis between survivors and non-survivors, there were no differences in regard to ADCmean of all muscles (p=0.76), psoas muscles (p=0.63) or paraspinal muscles (p=0.19). In the multivariate Cox regression analysis adjusted for age, BCLC stage, grading, and presence of liver cirrhosis, only the ADCmean of all muscles was associated with survival after 12 months (HR=0.98, 95% CI=0.97-0.99, p=0.04).

In the Kaplan-Meier curve, a statistical trend for better survival after 12 months can be seen for patients with an ADC value above the median of 1.3×10^–3 mm²/s (Figure 2).

Correlation of ADC values with other outcome markers after TARE. There were only weak correlations between ADC values and currently used markers for outcomes of TARE in HCC patients. There was an intermediate positive correlation between ADCmean and platelet count before (r=0.37, p=0.005) and after TARE treatment (r=0.35, p=0.01), while there was a negative correlation with serum bilirubin before (r=–0.30, p=0.03) and after treatment (r=–0.44, p=0.03) (Figure 3). There was no correlation of ADC values with MELD score (r=0.08, p=0.55), ALBI score (r=–0.09, p=0.50) or with Child-Pugh stage (r=0.05, p=0.71) before TARE.

Many studies providing evidence of the efficacy of TARE have been published over the last years (5-9). The reported overall survival (OS) ranges widely for unresectable HCC patients undergoing TARE from 7 to 30 months (9-11). This is exceptionally wide, due to the heterogeneity of unresectable HCC, which includes disparate disease burdens, ranging from solitary lesions to multifocal or metastatic disease (9-12). As a direct consequence, there is a definite need for appropriate patient selection, which can also be considered to be key in improving the clinical outcomes of this combined treatment.

Established prognosis factors for patients undergoing TARE are portal venous tumor thrombus extension, tumor bulk and bilirubin levels. They have an independent influence on long-term outcomes (12). A recent score system incorporated four parameters comprising maximal tumor size, tumor number, albumin, and portal vein tumor thrombosis to develop the SNAP-HCC score (8). In our study, we have for the first time used ADC values of the paraspinal and psoas muscles to test for association with 12 months outcomes in patients with intermediate stage HCC undergoing TARE. We found that in our patient population pre-treatment ADC values were associated with 12-month survival. Of note, only the ADC values of all muscles combined reached statistical significance, whereas the ADC value of the paraspinal and psoas muscles independently were not significantly associated with survival.

Elucidating novel prognostic biomarkers is crucial for modern treatment planning. This is especially important for HCC patients, who are treated in a multimodal manner and for whom decision making across an increasing number of treatment options is warranted in clinical practice (1-4). In our study, we could show that ADC values have the potential to be such a tool.

Sarcopenia is a key factor for survival in patients with liver cirrhosis or HCC (23,29,30,34). Another study investigated sarcopenia assessed by CT images and could demonstrate that sarcopenia was more commonly observed in HCC patients with disease progression after TARE (38.6% vs. 11.9%; p=0.006) (29). In another study using the skeletal muscle index, sarcopenia was considered an emerging prognostic factor in patients with HCC undergoing bland TAE therapy as it was associated with increased mortality (30). In comparison to CT based approaches to quantify sarcopenia, ADC values have the advantage that they can be derived from MRI which is the standard imaging method for many primary liver cancers (1-3). The development of ADC as a clinical tool could fill a gap for the large number of HCC patients for whom only MRI and no CT images are available. Moreover, it could provide more insight into the microstructure of the muscles compared to conventional CT images. It is still unknown how the ADC values of the muscles are directly associated with the conventional sarcopenia-defining skeletal muscle index derived from CT images. In another recent study, a German group elucidated the prognostic value of T2-weighted images to reflect the muscle infiltration as a muscle quality surrogate parameter derived from MRI with promising results (35). The authors derived an imaging marker, called fat-free muscle area, which was associated with tumor burden and even overall survival (35). It would be very interesting to combine the morphological information of the T2-weighted images with the functional characteristics of the DWI sequence to better characterize the muscle quality by MRI.

Other approaches are currently investigated for their potential in predicting outcomes of transarterial treatments in HCC patients. Thus, in a recent study by Nam et al., four independent risk factors were associated with OS, including maximal tumor size, tumor number, serum albumin, and portal vein tumor thrombosis, which were used to develop the SNAP-HCC score (8). This score and other scoring systems that were included into our analysis as MELD, ALBI, and Child-Pugh scores, largely reflect liver function and they therefore may be subject to wide fluctuations that may limit their accuracy. Besides our study, ADC values were to date only validated in another study investigating 114 patients with liver cirrhosis listed for transplantation (33). It has been shown that ADC values are intermediately correlated with MELD-score (r=0.47, p=0.0001) and that ADC values of patients with liver cirrhosis are significantly higher compared to healthy controls (33). Established prognosis factors for patients undergoing TARE are portal venous tumor thrombus extension, tumor bulk and bilirubin level, all of which are independently associated with long-term outcomes (12). As validated by imaging findings, quantitative tumor assessment by RECIST, mRECIST, EASL and qEASL are independent response biomarkers that predict OS after TARE (36). Our findings suggest that ADC may represent another promising and easy to apply tool for estimating individual survival chances and for tailoring personalized treatment concepts. Indeed, ADC value quantification of the paraspinal muscle could be easily translated into clinical routine, as DWI is a standard sequence within the liver MRI protocol (37). As such, the ADC value of the paraspinal muscle would be a by-product with no additional time or cost. Beyond that, in recent years, the systemic treatment options have been expanded and have been incorporated into earlier BCLC stages (38).

The present analysis has some limitations to address. Firstly, it is a retrospective study with a known inherent bias. However, the calculation of the ADC values was performed blinded to the clinical results and was standardized to reduce possible bias. Secondly, the results are based on a single center, which limits possibly external validity. There is a definite need to confirm the present results in different patients. Thirdly, resin as well as glass spheres were used in the present analysis, which could be a possible confounder. However, this reflects the clinical routine better than only one kind of sphere. Fourthly, the patient group was a heterogeneous HCC group with a low 12-month survival rate, which highlights the advanced stage of the included patients. These results might not be transferable to patients with less extensive HCC burden. Fifthly, different TARE indications, according to local preferences, could lead to earlier or later start of systemic treatment.

ADC values of the abdominal wall muscles could be used as a prognostic biomarker in patients with HCC undergoing TARE. ADC values of the muscles could be used a as a novel surrogate parameter for sarcopenia assessment. Our results need to be validated in further studies using external validation cohorts and other treatments for HCC.

The Authors declare that they have no conflicts of interest related to the contents of this article.

A.S., C.E., E.B., K.S., O.S., D.S. performed imaging and data analyses; T.D., T.B., F.B supplied resources and supervised the study; C.E. and HJ.M. conceived the study; HJ.M. drafted the manuscript; all the Authors contributed to the discussion and review of the final manuscript and approved the final manuscript.