1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

The aim of this study was to compare radiological and pathological changes and test the adjunct efficacy of Sorafenib to Y90 as a bridge to transplantation in hepatocellular carcinoma (HCC). 15 patients with 16 HCC lesions were randomized to Y90 without (Group A, n = 9) or with Sorafenib (Group B, n = 7). Size (WHO, RECIST), enhancement (EASL, mRECIST) and diffusion-weighted imaging criteria (apparent diffusion coefficient, ADC) measurements were obtained at baseline, then at 1 and every 3 months after treatment until transplantation. Percentage necrosis in explanted tumors was correlated with imaging findings. 100%, 50%-99% and <50% pathological necrosis was observed in 6 (67%), 1 (11%), and 2 (22%) tumors in Group A and 3 (42%), 2 (28%), and 2 (28%) in Group B, respectively (P = 0.81). While ADC (P = 0.46) did not change after treatment, WHO (P = 0.06) and RECIST (P = 0.08) response at 1 month failed to reach significance, but significant responses by EASL (P < 0.01/0.03) and mRECIST (P < 0.01/0.03) at 1 and 3 months were observed. Response was equivalent by EASL or mRECIST. No difference in response rates was observed between groups A and B at 1 and 3 months by WHO, RECIST, EASL, mRECIST or ADC measurements. Despite failing to reach significance, smaller baseline size was associated with complete pathological necrosis (CPN) (RECIST: P = 0.07; WHO: P = 0.05). However, a cut-off size of 35 mm was predictive of CPN (P = 0.005). CPN could not be predicted by WHO (P = 0.25 and 0.62), RECIST (P = 0.35 and 0.54), EASL (P = 0.49 and 0.46), mRECIST (P = 0.49 and 0.60) or ADC (P = 0.86 and 0.93). Conclusion: The adjunct of Sorafenib did not augment radiological or pathological response to Y90 therapy for HCC. Equivalent significant reduction in enhancement at 1 and 3 months by EASL/mRECIST was noted. Neither EASL nor mRECIST could reliably predict CPN. (HEPATOLOGY 2013;58:1655–1666)


American Association for the Study of Liver Diseases


apparent diffusion coefficient




Barcelona Clinic Liver Cancer


complete pathological necrosis


complete response


computed tomography


diffusion-weighted imaging


European Association for the Study of the Liver




hepatocellular carcinoma


locoregional therapy


modified Response Evaluation Criteria in Solid Tumors


magnetic resonance imaging




orthotopic liver transplant


progressive disease


partial response


portal venous thrombosis




stable disease


Response Evaluation Criteria in Solid Tumors


region of interest


transarterial chemoembolization


time to progression


technetium-99 macroaggregated albumin


gradient echo T1-weighted


turbo spin echo T2-weighted


University of California San Francisco


United Network for Organ Sharing


World Health Organization


yttrium-90 radioembolization

The development of surrogate markers for locoregional therapies (LRTs) in hepatocellular carcinoma (HCC) is desirable to improve treatment planning and accelerate design and endpoints in clinical trials. Before validation, early imaging surrogate markers face different challenges, including methodological considerations, reproducibility, accuracy to detect real treatment response, and, potentially most important, detection of a survival benefit. In comparison with survival, surrogate endpoints (time to progression [TTP] and progression-free survival) offer the advantage of potentially less-confounding effect by concomitant liver (i.e., cirrhosis, fibrosis) or systemic diseases as well as previous or subsequent locoregional or systemic treatment.[1]

The European Association for the Study of the Liver (EASL) guidelines (2011) advocate the use of enhancing tissue to assess imaging response of HCC.[2] Modified Response Evaluation Criteria in Solid Tumors (mRECIST) were devised with keeping this concept in mind and are currently being proposed as the standard methodology of radiological response in HCC.[3] However, few radiological-pathological studies support these criteria; our research group has previously highlighted the relevance of these important correlative concepts for both chemo- and radioembolization.[4-6] Uni- and bidimensional measurements of the entire treated tumor (Response Evaluation Criteria in Solid Tumors [RECIST] and World Health Organization [WHO] criteria) are often criticized, given their lack of correlation with viable tumor. Emerging functional imaging parameters also exist; diffusion-weighted imaging (DWI), representing the motion restriction of free water molecules, is one of the most commonly discussed techniques.[7]

The aim of this study was to compare radio- and pathological changes and test the adjunct efficacy of sorafenib to yttrium-90 radioembolization (Y90) as a bridge to transplantation in HCC. We tested WHO, EASL, RECIST, mRECIST and apparent diffusion coefficient (ADC) values (DWI parameter) as surrogate markers of complete pathological response after randomization to Yttrium-90 radioembolization (Y90) with or without Sorafenib.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
Patient Sample

This is a detailed imaging analysis from a prospective, randomized study of Y90 radioembolization ± sorafenib in HCC patients being bridged to orthotopic liver transplant (OLT). Patients were randomized 1:1 to Y90 alone (group A) or in combination with sorafenib (group B). The trial was approved by the Northwestern University Institutional Review Board (Chicago, IL), compliant with the Health Insurance Portability and Accountability Act, and has been registered (NCT00846131). Clinical effects (adverse events, tolerability, and dose reductions) of combining Y90 with sorafenib are beyond the scope of this imaging analysis and are being reported in a separate article focused on clinical outcomes.

Inclusion criteria for the study included HCC confirmed by American Association for the Study of Liver Diseases (AASLD) guidelines, Child-Pugh score ≤B8, and candidates for OLT (up to University of California San Francisco [UCSF] criteria).[2] Patients with performance status >2, metastatic disease, tumor-related portal vein thrombosis (PVT), and/or biological or clinical abnormality contraindicating sorafenib or radioembolization were not study candidates. By protocol, patients receiving >2 Y90 treatments were withdrawn from the analysis. Despite being classified as advanced HCC by Barcelona staging (Barcelona Clinic Liver Cancer; BCLC), patients with performance status >0, but with imaging findings of BCLC A, were still considered for transplantation.

Between February 2009 and October 2012, 23 patients (group A: N = 12; group B: N = 11) were enrolled in the study (study flow chart; Fig. 1). Two did not receive therapy: One patient from group A did not have confirmed angiographic hypervascularity at angiography (despite meeting diagnostic criteria), with a subsequent biopsy being negative for malignancy, and 1 from group B died before treatment (ruptured HCC). One patient from group A withdrew consent; that patient was treated off-study with Y90, followed by OLT. The 20 remaining patients comprise the intention-to-treat patient sample (group A: N = 10; group B: N = 10). The study was officially closed on February 7, 2013, when the last remaining patient in group A died of cardiac causes while awaiting transplantation.


Figure 1. Study flow chart.

Download figure to PowerPoint

Y90 Procedure

Radioembolization treatment was preceded by a simulation procedure, during which technetium-99 macroaggregated albumin (99Tc-MAA) was injected into the hepatic arterial vasculature, simulating Y90 microspheres distribution to estimate the degree of extrahepatic deposition. Coiling of extrahepatic arteries was performed, when required, to avoid inadvertent deposition. Glass microspheres loaded with 90Yttrium (TheraSphere; Nordion, Ottowa, Ontario, Canada) were used in this study per standard methodology. Patients were observed for 2 hours (arterial closure device) and subsequently discharged.[8-11]

Sorafenib Treatment

For group B, sorafenib 400 mg (2 × 200 mg tablets) was administered orally, initially twice-daily (total, 800 mg daily/4 tablets) before Y90 (median, 20 days; range, 13-35). Dose was adjusted per guidelines, and sorafenib treatment never exceeded 12 months. Sorafenib was stopped when imminent transplantation (#1 on transplant list) was expected according to patient's model for end-stage liver disease score. Detailed reporting of adverse events combining Y90 and sorafenib will be reported on elsewhere; in brief, no unexpected toxicities combining Y90 and sorafenib were noted.

Imaging Response Assessment

All radiological assessment was performed using magnetic resonance imaging (MRI). One patient in group B who had 2 Y90 sessions, and 2 patients who had 3 Y90 procedures (1 patient in group A and 1 in group B) were not transplanted. One patient who had 2 Y90 procedures and chemoembolization in group B was excluded. Eight and seven patients in groups A and B, respectively, were transplanted. Consequently, we performed our radiological/pathological study on 15 patients (group A: N = 8; group B: N = 7) for a tumor-by-tumor analysis on 16 HCC lesions (study flow chart; Fig. 1).

MRI protocol included gradient echo T1-weighted (T1 GRE) fat suppressed sequences before and after intravenous injection of gadolinium (Gd) agent, turbo spin-echo T2-weighted (T2 TSE) sequences and multishot PROPELLER diffusion-weighted sequences, as described extensively in Supporting Table 1. Measurements were repeated at 1-month and 3-month follow-up MRI scans post-Y90 and on all subsequent MRI scans until OLT. To evaluate the possible adjunct efficacy of sorafenib over Y90, tumor response after Y90 was compared to pre-Y90 MRI scans for both groups.

We measured all treated lesions on the arterial phase of post-Gd T1 GRE dynamic sequences according to WHO and RECIST criteria, respectively, measuring the percentage of change in the sum of the maximal bidimensional perpendicular diameters and the maximal unidimensional diameter, including viable and nonenhancing areas within the tumor, and EASL and mRECIST criteria, respectively, measuring the percentage of change in the sum of the maximal bidimensional diameters and the maximal unidimensional diameter, including only the enhancing portion of the tumor. For these response criteria, radiologic interpretation was classified as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD) according to cutoffs defined in Supporting Table 2.

As a functional imaging parameter, ADC values were calculated for all treated tumors using the same methodology. On the corresponding post-Gd T1-GRE sequence for the selection of the image level, a circular region of interest (ROI) was positioned in the enhancing portion of the tumors (presumably viable) or in the center of the lesion, if no viable tumor was identified. A similar ROI was then transferred at the same position on the low b (s/mm) (b 0 or b 50) and high b (b 400 or b 500) sequences, and mean ADC values (mm2/s) were calculated using the following formula:

  • display math

Arbitrarily, we imposed a minimal size of 1.00 cm2 to minimize the error resulting from the low number of voxels included in the ROI sample and hence minimize the random distribution of our measurements. According to the ADC change of tumors reported in different studies after sorafenib or Y90, a response (R) was considered as an increase in ADC values of 5% or more and a nonresponse (NR) as a less than 5% increase in ADC values, compared to the appropriate baseline measurement.[12, 13]

Finally, we decided to perform a subjective response assessment. Three investigators (M.V., F.H.M., and R.S.) independently analyzed the pre- and post-Gd T1 GRE dynamic MRI sequences, estimated the percentage nonenhancing tumor, considering these radiological patterns as necrotic tissue, and classified subjectively tumor response as CR (no enhancement), PR (>50%, but not 100%), SD (between PR and PD), or PD (worsening enhancement) at 1 and 3 months after Y90 treatment for every treated tumor, compared to baseline imaging, without knowledge of the final pathology report. One of them (F.H.M.) also used DWI sequences in borderline cases.

Pathologic Findings on Explant

Explanted livers were analyzed by surgical pathology in our institution, with sectioning of liver tissue at 0.5-1.0 cm. Pathological response was classified as 100% complete pathological necrosis (CPN) and 50%-99% or <50% necrosis per our previous description.[4-6]

Statistical Analysis

All data were summarized using appropriate descriptive statistics (count and frequency for categorical variables and median and range for continuous variables). Uni- or multivariate analysis using Mann Whitney's U test, the Student t test, chi-square test, or Fischer's exact test were used where appropriate to compare radiological parameters between groups (group A versus group B and CPN versus non-CPN) at baseline to identify any potential cofounders as well as after Y90. Scatter graphics representing the percentage of change in WHO, RECIST, EASL, mRECIST, and ADC measurements for groups A and B were built, considering 1 and 3 months post-Y90 and all subsequent imaging follow-up until OLT. Whisker box plots showing median, range, and interquartile values, as well as analysis of variance by Friedman's two-tailed test and Wilcoxon's test, were used to demonstrate 1- and 3-month post-Y90 changes, controlling for baseline values, in WHO, RECIST, EASL, mRECIST, and ADC values. Bonferroni's correction was applied if significant P values were observed when multiple hypotheses were tested for the same populations. Tumor-by-tumor radio- and pathological response classification was represented by summary table and graphical methods. For all tests, a P value <0.05 was considered statistically significant. All analyses were conducted using MedCalc software (MedCalc Software, Mariakerke, Belgium).


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Patient Sample

Baseline characteristics are described in Table 1. Median age was 57 years. One patient in group B presented with bilobar disease (2 lesions in segment 6 and 1 in segment 4a) and was classified as United Network for Organ Sharing (UNOS) stage T3. The groups were otherwise well balanced.

Table 1. Baseline Characteristics
Variable Y90 Alone (n = 8)Y90 + Sorafenib (n = 7)Total (n = 15)P Value
  1. Abbreviations: ECOG, Eastern Cooperative Oncology Group; NASH, nonalcoholic steatohepatitis; PBC, primary biliary cirrhosis.

Age, yearsMedian5956570.30
African American101
Alcohol + NASH101
Method of diagnosisImaging56111.0
Baseline AFP, ng/mLMedian13.211.911.90.60
Portal hypertensionAbsent0110.47
Tumor distributionUnilobar86141.0
UNOS stageT10110.20

Imaging Follow-up

Patients from group B (n = 7) underwent a pre-sorafenib MRI scan, with a maximum of 32 days before initiation of sorafenib (median, 18; range, 8-32) and a maximum of 53 days before Y90 (median, 42; range, 21-53).

Median time from baseline MRI to Y90 procedure for group A was 18 days (range, 14-42). Seven of the eight patients in group B had a baseline MRI scan on the day of Y90 treatment immediately preceding the procedure, translating into a median time from imaging to Y90 of 0 days. For both groups, the pre-Y90 MRI scan served as the baseline.

Median time from last MRI scan to transplant was 25 days (range, 5-93). Findings on the last pre-OLT scan were consistent with the 3-month scan for all 16 lesions.

Pathology Findings

CPN as well as 50%-99% and <50% necrosis was observed in 6 (67%), 1 (11%), and 2 (22%) tumors in group A and 3 (42%), 2 (28%), and 2 (28%) in group B, respectively (P = 0.41; Table 2).

Table 2. Summary Radiological Classification at 1 and 3 Months With Pathological Results
 Anatomic CriteriaFunctional Criteria
1 month (n = 16)
*1 missing data resulting from artifact on MRI
3 months (n = 14)
*2 missing data resulting from artifact on MRI
Pathology Necrosis Rate by Group
 Group A Y90 (%)Group B Y90 + Sorafenib (%)Total
100% necrosis6 (67)3 (42)9
50%-99% necrosis1 (11)2 (28)3
<50% necrosis2 (22)2 (28)4

Imaging Findings


Grouping all tumors, response by size criteria was observed by RECIST (P = 0.08) and WHO (P = 0.06), despite failing to reach significance (Fig. 2). Corrected P value of Wilcoxon's test, comparing 1 month post-Y90 to baseline, showed a significant reduction of WHO (P = 0.047), but failed to reach significance for RECIST (P = 0.077).


Compared to baseline, a significant decrease in enhancing tumor diameter (P < 0.01 and 0.03) and the sum of the longest and largest viable tumor diameter (P < 0.01 and 0.03) was observed at 1 and 3 months, suggesting that EASL and mRECIST were equivalent (Fig. 2).

At 1 month, CRs by EASL and mRECIST were noted in 4 of 16 lesions; these corresponded to CPN in 2 of 4 of cases. At 3 months, CRs by EASL and mRECIST were noted in 7 of 14; this corresponded to CPN in 3 of 7 of cases (Table 2; Fig. 3).


Figure 2. Imaging parameter changes.

Download figure to PowerPoint


Figure 3. Tumor-by-tumor radiological-pathological classification.

Download figure to PowerPoint

At 1 month, PRs by EASL and mRECIST were noted in 8 of 16 lesions; these corresponded to CPN in 5 of 8 of cases. At 3 months, PRs by EASL and mRECIST were noted in 3 of 14 and 4 of 14 lesions; this corresponded to CPN in 1 of 3 and 2 of 4 cases (Table 2; Fig. 3).


Compared to baseline, ADC (P = 0.46) values did not differ at 1 or 3 months (Fig. 2). With response defined as an ADC increase ≥5% from baseline, 9 of 15 and 8 of 12 lesions were classified as responders at 1 and 3 months, respectively (Table 2), but without being able to predict pathological results. CPN as well as 50%-99% and <50% necrosis were observed in 5, 3, and 1 ADC responding lesions and 4, 1, and 1 ADC nonresponding lesions at 1 month (P = 0.47); at 3 months, it was 4, 2, and 2 ADC responding lesions and 2, 1, and 1 ADC nonresponding lesions (P = 0.73; Fig. 3).

Subjective Assessment

The subjective response assessment showed good results in predicting pathological results, particularly for one of the investigators (F.M.), who used DWI sequences as a complementary tool: Uncertain tumor response in portions of the tumors exhibiting irregular enhancing patterns could sometimes be better classified by this investigator, given diffusion restriction estimation in suspicious areas with somehow better sensitivity, specificity, and positive predictive and negative predictive values at 1 and 3 months than other investigators (respectively, 56%, 86%, 83%, and 60% and 71%, 43%, 56%, and 60%; Fig. 3).


Comparing group A to B at baseline and 1 and 3 months posttreatment, no difference in terms of WHO, RECIST, EASL, mRECIST, or ADC measurements was observed (Table 3). The percentage of change in WHO, RECIST, EASL, mRECIST, and ADC measurements after Y90 until OLT in both groups is illustrated in Supporting Fig. 1.

Table 3. Tumor Measurements Pretreatment and Posttreatment
  Group A (n = 9) (Median, Range)Group B (n = 7) (Median, Range)Total (n = 16) (Median, Range)P Value (Mann-Whitney)
RECIST (mm)Baseline28.7, 13.5–55.933.3, 14.9–42.529.5, 13.5–55.90.68
1 month26.4, 9.9–45.928.4, 13.3–36.928.4, 9.9–45.90.91
3 months30.5, 5.6–45.627.7, 9.0–63.929.1, 5.6–63.90.71
WHO (mm)Baseline54.6, 24.4–105.857.6, 25.9–76.356.8, 24.4–105.80.68
1 month50.5, 19.6–90.255.0, 22.2–70.753.5, 19.6–90.21
3 months61.0, 10.1–85.048.9, 16.5–119.555.0, 10.1–119.50.71
mRECIST (mm)Baseline28.7, 13.5–55.928.8, 14.9–36.928.8, 13.5–55.91
1 month9.5, 0.0–25.513.0, 0.0–34.310.0, 0.0–34.30.46
3 months9.5, 0.0–26.30.0, 0.0–63.94.8, 0.0–63.90.84
EASL (mm)Baseline54.6, 24.4–105.856.2, 22.3–72.255.4, 22.3–105.80.76
1 month15.1, 0.0–50.518.6, 0.0–58.016.3, 0.0–58.00.40
3 months16.9, 0.0–47.70.0, 0.0–119.57.8, 0.0–119.50.74
ADC (mm2/s × 10−3)Baseline1.5, 1.2–2.21.5, 1.0–1.61.5, 1.0–2.20.76
1 month1.5, 0.7–2.91.2, 1.1–2.21.5, 0.7–2.90.69
3 months1.8, 1.1–2.71.3, 1.2–1.51.5, 1.1–2.70.06

Complete Pathological Response

Although not reaching significance, a trend of smaller lesions at baseline (RECIST, P = 0.07; WHO, P = 0.05) was observed in CPN lesions. However, 1 and 3 months after Y90, CPN could not be predicted by WHO (P = 0.25 and 0.62), RECIST (P = 0.35 and 0.54), EASL (P = 0.49 and 0.46), mRECIST (P = 0.49 and 0.60), or ADC (P = 0.86 and 0.93) (Table 4). A cut-off size at baseline of 35 mm was found to be highly significant (P = 0.005) in the prediction of CPN (Table 5); this cutoff was not affected by the addition of sorafenib. Summary pathological results and radiological classification at 1 and 3 months are summarized in Table 2 and Fig. 3.

Table 4. CPN Versus Non-CPN
  CPN (n = 9) (Median, Range)Non-CPN (n = 7) (Median, Range)Total (n = 16) (Median, Range)P Value (Mann-Whitney)
  1. Abbreviations: CPN, complete pathological necrosis; n, number of tumors.

RECIST (mm)Baseline28.7, 13.5–33.336.9, 19.7–55.929.5, 13.5–55.90.07
1 month26.4, 9.9–34.328.4, 17.3–45.928.4, 9.9–45.90.35
3 months30.5, 9.0–63.927.7, 5.6–44.529.1, 5.6–63.90.54
WHO (mm)Baseline54.6, 24.4–57.872.2, 31.9–105.856.8, 24.4–105.80.05
1 month50.5, 19.6–58.056.5, 31.4–90.253.5, 19.6–90.20.25
3 months61.0,16.5–119.548.9, 10.1–85.055.0, 10.1–119.50.62
mRECIST (mm)Baseline28.7, 13.5–33.336.5, 16.5–55.928.8, 13.5–55.90.25
1 month8.0, 0.0–34.313.7, 0.0–28.410.0, 0.0–34.30.49
3 months9.5, 0.0–63.90.0, 0.0–26.04.8, 0.0–63.90.60
EASL (mm)baseline54.6, 24.4–57.865.6, 22.3–105.855.4, 22.3–105.80.41
1 month14.4, 0.0–58.025.0, 0.0–55.016.3, 0.0–58.00.49
3 months16.9, 0.0–119.50.0, 0.0–40.67.8, 0.0–119.50.46
ADC (mm2/s × 10−3)baseline1.5, 1.2–2.21.4, 1.0–1.61.5, 1.0–2.20.35
1 month1.5, 0.7–2.91.4, 1.1–2.21.5, 0.7–2.90.86
3 months1.5, 1.1–2.71.4, 1.2–2.01.5, 1.1–2.70.93
Table 5. Pathological Necrosis by Size Range
 ≤25 mm>25 mmP Value
  1. Abbreviation: CPN, complete pathological necrosis.

CPN (%)3 (60)6 (54)1.0
Non-CPN (%)2 (40)5 (46)
 ≤30 mm>30 mm 
CPN (%)6 (75)3 (38)0.31
Non-CPN (%)2 (25)5 (62)
 ≤35 mm>35 mm 
CPN (%)9 (82)0 (0)0.005
Non-CPN (%)2 (18)5 (100)


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

To our knowledge, this study constitutes one of the only prospective radiological/pathological studies for HCC.[4-6] This is of clinical relevance, because imaging guidelines in HCC lack these gold-standard correlative studies. As a subset of the Y90 ± sorafenib study, we tested the hypothesis of sorafenib treatment adjunct efficacy on Y90 as a neoadjuvant treatment or bridge to transplantation in HCC candidates for liver transplantation. No change in lesional aspect on imaging at 1 and 3 months nor difference in pathological results could be observed between patients treated with sorafenib and Y90 and those treated by Y90 alone. Hence, we concluded that on a tumor-by-tumor analysis, sorafenib did not improve imaging or pathological outcome in transplanted patients. However, sorafenib, as a cytostatic and antiangiogenic agent, has a potential role in controlling the background liver disease or lesions not treated by LRT; a survival gain of nearly 3 months was noted in the SHARP trial.[14] Although sorafenib could be considered as a treatment option after OLT, this discussion is beyond the scope of this study.[15] In relation to its antiangiogenic effect, other imaging parameters, such as perfusion computed tomography (CT) or MRI, as well as serum biomarkers (vascular endothelial growth factor, epithelial growth factor receptor, platelet-dewrived growth factor, and hypoxia-inducible factor 1 alpha) could also be more appropriate for the response assessment to sorafenib.[16-18] Similarly, alpha-fetoprotein (AFP) serum level was demonstrated to be a strong predictive marker of response in AFP producer HCC patients.[19, 20] Unfortunately, as shown in Table 1, baseline AFP levels were insignificant and could not be tested.

Interestingly, a strong trend of smaller lesions at baseline was observed in the group of complete pathological response. This can be explained by the smaller tumor burden, resulting in a higher ratio of radiation/tumor volume and improved treatment efficacy. A cut-off size at baseline of 35 mm was found to be highly significant in the prediction of CPN. This tumor size is somewhat comparable to other relevant studies and recommendations with radiofrequency ablation.[21-23]

In accord with HCC guidelines and other studies, our results support that measurement of the viable portions of tumor at 1 and 3 months is likely the best way to establish tumor response of HCC treated by targeted or locoregional therapies.[4, 24-26] However, this study suggests that older WHO and RECIST criteria are not to be considered obsolete for response assessment after Y90, a reduction in uni- and bidimensional measurements being observed at 1 month post-Y90. This observation can be explained by the lack of detection of intratumoral changes, such as necrosis or decrease of cellularity. EASL and mRECIST show clear methodology limitations as well: when, how, and what enhancing area to measure? Anatomical changes (size, density, and nodularity) after LRT are a dynamic phenomenon. These evolve within several months after Y90, and it is common that tumor borders exhibit a pseudonodular area of enhancement, or that intratumoral enhancing septa are being observed. These borderline appearances may be persistent at 3-month follow-up. We experienced these difficulties when performing EASL and mRECIST assessment; it was routine for the three readers to use different areas of enhancing tissue to perform the measurements, with consensus adjudication being common when performing EASL/mRECIST measurements.

In comparison with transarterial chemoembolization (TACE), the absence of lipiodol infusion in the treated area facilitated the depiction of the enhancing tissue, even though Shim et al. demonstrated, in a retrospective study, that lipiodol could be considered necrotic tissue on CT, and that mRECIST and EASL were found to be good predictors of pathological response.[27] However, with respect to differences in baseline size range of treated tumors (10-137 mm) and treatment technique, we advocate that it is often impossible to differentiate persistent tumor from an inflammatory or regenerative process in enhancing tissue. We observed this phenomenon in our study, where enhancing tissue on one scan disappeared on subsequent imaging, likely suggesting an inflammatory and remodeling nature to the enhancement. For all aforementioned reasons, even if EASL and mRECIST criteria showed a significant change at 1 month, our clinical practice policy is to assess imaging response after 3 months of follow-up. We believe two imaging follow-up scans permit a better understanding of the response timeline and more confident decision-making approach for potential ulterior treatments.

Riaz et al. showed that combining measurements of the entire tumor and of its enhancing portion (especially WHO and EASL) could increase complete pathological response detection.[4] Considering imaging as a potential surrogate marker of pathological response to liver-directed therapies, we advocate that combining anatomical and functional criteria are currently to be considered the next steps of research. Among these, DWI could play an important role as a potential adjunct tool in response assessment after Y90. However, when used as unique response criteria, ADC calculation was disappointing for detection of pathological response. However, our results necessitate some comments. ADC calculation methodology was heterogeneous in the literature and highly debated. Also, DWI sequences parameters are still to be defined (use of b-values, echo-planar versus spin-echo, and single versus multishot sequences). Some researchers propose calculating ADC values in the entire lesion (necrotic or viable), wheras others advocate studying only the borders. Even if automated segmentation software is available, some prefer a manual drawing of the ROI. Finally, a choice must be made between measurements directly performed on ADC maps or calculated after measurements on both low and high b-value sequences series, that is, bypassing automated postprocessing ADC calculation (we chose this latest methodology to optimize the accuracy of series coregistration). Whatever the chosen methodology, we have to accept advantages and disadvantages. As a potential optional tool in response assessment for borderline cases, we opted for a more restrictive and discriminant technique; when possible, we placed our ROI on the suspected viable portions of the tumors. However, baseline and posttreatment ADC values in our study (baseline: median, 1.5; range, 1.0-2.2; 1 month: median, 1.5; range, 0.7-2.9; 3 months: median, 1.5; range, 1.1-2.7) were consistent with other studies evaluating ADC changes after TACE and sorafenib. Despite equivocal results in our study, we recognize that ADC could constitute a useful optional tool in clinical practice for borderline cases. For instance, one of the investigators (F.M.) showed better results in subjectively estimating CPN, partially because of DWI as ancillary data.

Further improvements in ADC methodology and software (i.e., volumetric ADC mapping) would be beneficial. The use of ADC after sorafenib may be problematic because patients may develop hemorrhagic necrosis as a favorable treatment response, which can decrease ADC values and hence mimic residual tumor.[12, 13]

There are strengths to this study. This is the first radiological/pathological correlative study generated from a prospective, randomized trial; these are rare. Second, the analysis was comprehensive and investigated relevant parameters, including size (WHO and RECIST), enhancement (EASL and mRECIST), and functional imaging criteria (DWI). Third, we used the gold-standard pathology analysis for quantification of necrosis. Fourth, the time from last scan to explant was <30 days, an acceptable time when imaging is reflective of pathology. There are weaknesses. Although the study is limited by sample size, finding clinical trial candidates being bridged to transplant, with solitary lesions (most in our study had solitary lesions strengthening the analysis), and eligible for both Y90 and sorafenib (randomized trial) is extremely challenging. Second, we did not identify any effect of sorafenib on imaging or pathology; this may have been because of the relatively small size of the tumors. Reports of sorafenib decreasing enhancement and vascularity are usually illustrated in advanced disease (infiltrative or large tumors, ± vascular invasion). Third, it was clear that, given the irregularity of tumoral enhancement posttreatment, there was a subjective element to measuring the longest enhancing tissue; these may be improved with (semi-) automated volume software. Fourth, whereas we observed that CPN could not be predicted by WHO and RECIST response classifications, we observed that smaller lesions were nevertheless more likely to exhibit CPN at explant. This is explained by the fact that measurement of treatment response by size criteria (ignoring enhancement) almost never reaches zero; there is always a measurable defect after treatment. Finally, none of the imaging parameters evaluated in our study, including EASL and mRECIST, could reliably detect CPN at a microscopic level, highlighting the limitations of imaging methodologies that, despite being advocated by HCC guidelines, remain imperfect.

On a tumor-by-tumor analysis, the adjunct of sorafenib to Y90 for HCC does not augment radio- or pathological response to therapy in HCC patients being bridged to transplantation. A reduction in standard size criteria (WHO and RECIST) at 1 month and a significant reduction in enhancing tumor at 1 and 3 months was observed, but failed to reach significance, likely a result of the cytotoxic effect of Y90. Response to treatment was equivalent when measuring by EASL or mRECIST, neither of which could reliably detect CPN. A trend of smaller lesions at baseline (35-mm cutoff) was predictive of CPN. Diffusion-weighted imaging (ADC) did not change after treatment. Standardization of ADC measurements, automated volumetric software (measurement enhancing portions of tumors), and the combination of response criteria (anatomic plus functional) should be considered as future areas of research to improve the detection of CPN.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
  • 1
    Llovet JM, Di Bisceglie AM, Bruix J, Kramer BS, Lencioni R, Zhu AX, et al. Design and endpoints of clinical trials in hepatocellular carcinoma. J Natl Cancer Inst 2008;100:698-711.
  • 2
    Bruix J, Sherman M. Management of hepatocellular carcinoma: an update. Hepatology 2011;53:1020-1022.
  • 3
    Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment for hepatocellular carcinoma. Semin Liver Dis 2010;30:52-60.
  • 4
    Riaz A, Memon K, Miller FH, Nikolaidis P, Kulik LM, Lewandowski RJ, et al. Role of the EASL, RECIST, and WHO response guidelines alone or in combination for hepatocellular carcinoma: radiologic-pathologic correlation. J Hepatol 2011;54:695-704.
  • 5
    Riaz A, Kulik L, Lewandowski RJ, Ryu RK, Giakoumis Spear G, Mulcahy MF, et al. Radiologic-pathologic correlation of hepatocellular carcinoma treated with internal radiation using yttrium-90 microspheres. Hepatology 2009;49:1185-1193.
  • 6
    Riaz A, Lewandowski RJ, Kulik L, Ryu RK, Mulcahy MF, Baker T, et al. Radiologic-pathologic correlation of hepatocellular carcinoma treated with chemoembolization. Cardiovasc Intervent Radiol 2010;33:1143-1152.
  • 7
    Luypaert R BS, Sourbron S, Osteaux M. Diffusion and perfusion MRI: basic physics. Eur J Radiol. 2001;38:19-27.
  • 8
    Salem R, Lewandowski RJ, Gates VL, Nutting CW, Murthy R, Rose SC, et al. Research reporting standards for radioembolization of hepatic malignancies. J Vasc Interv Radiol 2011;22:265-278.
  • 9
    Salem R, Thurston KG. Radioembolization with 90Yttrium microspheres: a state-of-the-art brachytherapy treatment for primary and secondary liver malignancies: part 1: technical and methodologic considerations. J Vasc Interv Radiol 2006;17:1251-1278.
  • 10
    Salem R, Lewandowski RJ, Mulcahy MF, Riaz A, Ryu RK, Ibrahim S, et al. Radioembolization for hepatocellular carcinoma using Yttrium-90 microspheres: a comprehensive report of long-term outcomes. Gastroenterology 2010;138:52-64.
  • 11
    Salem R, Lewandowski RJ, Kulik L, Wang E, Riaz A, Ryu RK, et al. Radioembolization results in longer time-to-progression and reduced toxicity compared with chemoembolization in patients with hepatocellular carcinoma. Gastroenterology 2011;140:497-507.e2.
  • 12
    Schraml C, Schwenzer NF, Martirosian P, Bitzer M, Lauer U, Claussen CD, Horger M. Diffusion-weighted MRI of advanced hepatocellular carcinoma during sorafenib treatment: initial results. AJR Am J Roentgenol 2009;193:W301-W307.
  • 13
    Vossen JA BM, Kamel IR. Assessment of tumor response on MR imaging after locoregional therapy. Tech Vasc Interv Radiol 2006;9:125-132.
  • 14
    Llovet JM, Ricci S, Mazzaferro V, Hilgard P, Gane E, Blanc JF, et al. Sorafenib in advanced hepatocellular carcinoma. N Engl J Med 2008;359:378-390.
  • 15
    Weinmann A, Niederle IM, Koch S, Hoppe-Lotichius M, Heise M, Düber C, et al. Sorafenib for recurrence of hepatocellular carcinoma after liver transplantation. Dig Liver Dis 2012;44:432-437.
  • 16
    Miyahara K, Nouso K, Tomoda T, Kobayashi S, Hagihara H, Kuwaki K, et al. Predicting the treatment effect of sorafenib using serum angiogenesis markers in patients with hepatocellular carcinoma. J Gastroenterol Hepatol 2011;26:1604-1611.
  • 17
    Jiang T, Kambadakone A, Kulkarni NM, Zhu AX, Sahani DV. Monitoring response to antiangiogenic treatment and predicting outcomes in advanced hepatocellular carcinoma using image biomarkers, CT perfusion, tumor density, and tumor size (RECIST). Invest Radiol 2012;47:11-17.
  • 18
    Salem R, Miller FH, Yaghmai V, Lewandowski RJ. Response assessment methodologies in hepatocellular carcinoma: complexities in the era of local and systemic treatments. J Hepatol 2013 Jan 24. doi: 10.1016/j.jhep.2013.01.021.
  • 19
    Riaz A, Ryu RK, Kulik LM, Mulcahy MF, Lewandowski RJ, Minocha J, et al. Alpha-fetoprotein response after locoregional therapy for hepatocellular carcinoma: oncologic marker of radiologic response, progression, and survival. J Clin Oncol 2009;27:5734-5742.
  • 20
    Memon K, Kulik L, Lewandowski RJ, Wang E, Ryu RK, Riaz A, et al. Alpha-fetoprotein response correlates with EASL response and survival in solitary hepatocellular carcinoma treated with transarterial therapies: a subgroup analysis. J Hepatol 2012;56:1112-1120.
  • 21
    Shiina S, Tateishi R, Arano T, Uchino K, Enooku K, Nakagawa H, et al. Radiofrequency ablation for hepatocellular carcinoma: 10-year outcome and prognostic factors. Am J Gastroenterol 2012;107:569-577.
  • 22
    Waki K, Aikata H, Katamura Y, Kawaoka T, Takaki S, Hiramatsu A, et al. Percutaneous radiofrequency ablation as first-line treatment for small hepatocellular carcinoma: results and prognostic factors on long-term follow up. J Gastroenterol Hepatol 2010;25:597-604.
  • 23
    Lam VW, Ng KK, Chok KS, Cheung TT, Yuen J, Tung H, et al. Risk factors and prognostic factors of local recurrence after radiofrequency ablation of hepatocellular carcinoma. J Am Coll Surg. 2008;207:20-29.
  • 24
    Spira D, Fenchel M, Lauer UM, Claussen CD, Gregor M, Bitzer M, Horger M. Comparison of different tumor response criteria in patients with hepatocellular carcinoma after systemic therapy with the multikinase inhibitor sorafenib. Acad Radiol 2011;18:89-96.
  • 25
    Gillmore R, Stuart S, Kirkwood A, Hameeduddin A, Woodward N, Burroughs AK, Meyer T. EASL and mRECIST responses are independent prognostic factors for survival in hepatocellular cancer patients treated with transarterial embolization. J Hepatol 2011;55:1309-1316.
  • 26
    Kim BK, Kim KA, Park JY, Ahn SH, Chon CY, Han KH, et al. Prospective comparison of prognostic values of modified Response Evaluation Criteria in Solid Tumours with European Association for the Study of the Liver criteria in hepatocellular carcinoma following chemoembolisation. Eur J Cancer 2013;49:826-834.
  • 27
    Shim JH, Han S, Shin YM, Yu E, Park W, Kim KM, et al. Optimal measurement modality and method for evaluation of responses to transarterial chemoembolization of hepatocellular carcinoma based on enhancement criteria. J Vasc Interv Radiol 2013;24:316-325.

Supporting Information

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Additional Supporting Information may be found in the online version of this article.

hep26487-sup-0001-supptables.doc252KSupplementary Table 1: MRI protocol

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.