Potential conflict of interest: Dr. Mulcahy advises Genesis Technologies. She received grants from Sanofi-Aventis and MDS Nordion. Dr. Salem is a consultant for MDS Nordion.
We present the correlation between radiologic and pathologic findings in HCC patients who underwent radioembolization with yttrium-90 (90Y) microspheres prior to resection or transplantation. Thirty-five patients with a total of 38 lesions who underwent liver explantation after 90Y radioembolization were studied. Imaging surrogates following treatment were evaluated; the explants were examined for assessment of necrosis by pathology. The correlation betwen radiologic and histologic findings of the treated lesions was analyzed. Twenty-three of 38 (61%) target lesions showed complete pathologic necrosis. All target lesions demonstrated some degree of histologic necrosis at explant. Complete histologic necrosis was seen in 89% of lesions with pretreatment size <3 cm. Complete pathologic necrosis was seen in 100%, 78%, and 93% of the lesions that were shown to have complete response by European Association for the Study of the Liver (EASL) necrosis criteria, partial response by World Health Organizaton (WHO) criteria, or thin rim enhancement on posttreatment imaging, respectively. In contrast, complete pathologic necrosis was seen in only 52% and 38% of the lesions that showed partial response by EASL criteria and peripheral nodular enhancement, respectively. Conclusion: Post-radioembolization imaging findings of response by EASL and WHO criteria are predictive of the degree of pathologic necrosis. Rim enhancement was an imaging characteristic that correlated well with histologic necrosis. (HEPATOLOGY 2009.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Hepatocellular carcinoma (HCC) is the most common primary malignant neoplasia of the liver. It is the sixth most common malignancy worldwide and is the third most common cause of cancer-related mortality.1, 2 The last decade has seen significant advancement in the management of this disease. Numerous treatment modalities are now available.3 Liver transplantation remains the standard of care for patients within Milan criteria.4 The limited availability of donor organs prolongs the waiting time and thus increases the chance of dropout due to tumor progression.5, 6 Locoregional therapies such as percutaneous ethanol injection (PEI), transarterial chemoembolization (TACE) including drug eluting beads, radiofrequency ablation (RFA), and radioembolization with yttrium-90 (90Y) have the potential to bridge patients within Milan criteria and downstage patients to transplantation.6-8 All these treatments vary in their mode of action, response rates, and toxicity profiles.
Imaging plays an integral role in staging of HCC and assessment of treatment response.9, 10 The radiologic findings after radioembolization have been discussed in detail by Ibrahim et al.10 It is recognized that response to therapy in HCC does not necessarily correspond with an immediate decrease in size of the treated lesion. Therefore, the degree of necrosis on imaging is an important component in assessing response to treatment.9 The efficacy of TACE and RFA by estimating the degree of response at imaging with correlation to the degree of necrosis on pathological examination has been described.11 The purpose of this study was to correlate patterns of imaging findings following 90Y with the level of tumor necrosis achieved following treatment. Pathologic explants were used as the gold standard for assessing the degree of necrosis.
AFP, alpha-fetoprotein; BCLC, Barcelona Clinic Liver Cancer; CR, complete response; EASL, European Association for the Study of the Liver; HCC, hepatocellular carcinoma; NPV, negative predictive value; OLT, orthotopic liver transplantation; PEI, percutaneous ethanol injection; PPV, positive predictive value; PR, partial response; RFA, radiofrequency ablation; SD, stable disease; TACE, transarterial chemoembolization; WHO, World Health Organization; UNOS, United Network for Organ Sharing; 90Y, yttrium-90.
Patients and Methods
Between 2003 and 2008, patients were treated for HCC using 90Y according to a protocol approved by our Institutional Review Board. Thirty-five patients ultimately had liver explantation, thus permitting pathologic analysis of the explants after treatment with 90Y.
All patients underwent pretreatment assessment that consisted of a clinical history, appropriate laboratory workup (alpha-fetoprotein [AFP], liver function test) and imaging studies. In lesions >2 cm, HCC was confirmed using biopsy only if the lesion did not meet European Association for the Study of the Liver (EASL) or American Association for the Study of Liver Diseases criteria.3, 9, 12, 13 Lesions measuring 1 to 2 cm required biopsy for confirmation of HCC.13 All patients were classified according to Child-Pugh, United Network for Organ Sharing (UNOS), and Barcelona Clinic Liver Cancer (BCLC) classification systems.12, 14, 15 By our definition for this study, (1) UNOS T4a patients could never be downstaged, because locoregional therapies rarely result in complete disappearance of lesions, and (2) UNOS T4b patients could only be downstaged if there was complete resolution of portal vein thrombosis. Radiologists performing the baseline assessments were blinded to whether the patients had undergone orthotopic liver transplantation (OLT) or resection. This approach was used in order to minimize staging bias based on BCLC or UNOS. In other words, a radiologist with preexisting knowledge of eventual transplantation may be more inclined to understage the disease. The target lesions were identified and measured; untreated lesions were not included in this analysis.
The selection of 90Y over other treatments was based on the consensus of a multidisciplinary team in hepatoma conference at our institution. For this cohort of patients, smaller lesions were treated with 90Y rather than RFA on the basis of the size/shape of the lesion; location at the dome or caudate lobe; or proximity of the lesion to major ducts, vessels, or liver surface.11, 16
All patients underwent mapping angiography before the treatment to determine vascular anatomy and arterial variants.17 Prophylactic embolization of nontarget vessels was performed in order to avoid nontarget deposition of microspheres.18, 1999Tc-MAA (macro-aggregated albumin) was used to assess lung shunt fraction and splanchnic shunting.
90Y Glass Microspheres.
Radioembolization or transcatheter 90Y internal radiotherapy is a minimally invasive procedure in which radiation is delivered to the tumor by catheterization of the artery supplying the tumor. 90Y is a pure beta-emitter with a half-life of 64.2 hours.20 Therasphere (MDS Nordion, Ottawa, Ontario, Canada) consists of nonbiodegradable glass microspheres that range from 20 to 30 μm in diameter and have 90Y as an integral constituent.21 A dose of 50 Gy of radiation is given to 1 kg of tissue by administering 1 GBq of 90Y.
The dose given to the target site was calculated using the formula
where D is the dose administered to the target site, A is the actual activity delivered to the site, LSF is the lung shunt fraction, R is the residual activity in the vial, and M is the mass of the area infused. The technical details of the treatment have been described.22
Patients underwent cross-sectional imaging 1 month after treatment to assess response to therapy in the target lesion only, and were subsequently followed with imaging every 3 months until OLT or resection.9 Four radiologists who were well-versed in the radiologic changes following radioembolization reviewed the scans. All scans between the first radioembolization and explantation were evaluated. World Health Organization (WHO)/EASL partial responses were only categorized as such if they were maintained to the time of explantation.
Target lesions were classified into three groups according to size on pretreatment imaging (1-2.9 cm, 3-5 cm, >5 cm). This was done in order to perform response/pathological analyses stratified by baseline tumor size.
Response according to size was assessed using WHO criteria in which the minimum cross product of the lesion after treatment was compared with that of the baseline cross product for calculating percent change in size. Complete response (CR) was defined as a 100% decrease in size; partial response (PR) was defined as a ≥50% decrease in size of the target lesion; and progressive disease was defined as a >25% increase in the cross product. All other findings were defined as stable disease (SD).9, 23
There is evidence that response by measuring change in size alone is not an effective mode of assessing tumor necrosis.9, 24 Thus, enhancement characteristics were studied as well. The enhancement characteristics evident after treatment with locoregional therapy were classified into three groups. No enhancement was defined as the absence of any enhancing tissue. Thin rim enhancement was defined as the presence of a thin rim of enhancing tissue of uniform thickness around a central area of necrosis. Peripheral nodular enhancement was defined as the presence of irregular, enhancing tissue at the periphery of a central area of necrosis.
Response in the lesions was also classified according to EASL necrosis criteria, which quantifies the amount of enhancing tissue in the treated lesion. CR was defined as the absence of any enhancing tissue. PR was defined as the appearance of a >50% decrease in enhancing tissue from the last scan. SD was defined as a <50% decrease in enhancing tissue.9 Lesions with thin rim enhancement were classified as EASL CRs.
In order to assess the clinical impression of treatment effect by imaging, the overall radiologic response to treatment as described by the interpreting radiologist was classified as favorable or unfavorable on the basis of change in size and degree of necrosis. The presence of residual tumor based on posttreatment radiographic findings and the ability of the blinded radiology reviewer to predict viable tumor was also assessed.
Patients with available pretreatment and posttreatment AFP levels were studied for this analysis. Those with a baseline AFP >200 ng/mL were selected.25 The posttreatment AFP levels were compared with the baseline AFP levels. The response was correlated to the degree of histologic necrosis.
Necrosis was evaluated at explantation both for evidence of gross and histologic necrosis. Thirty-three of the patients underwent OLT. One patient underwent resection, and one patient underwent an autopsy. One-centimeter sections of the entire liver were prepared. Representative samples of the lesions were taken and stained with routine hematoxylin-eosin stains for histologic examination. The presence of coagulative necrosis was noted. The target lesion was thoroughly examined for the presence of viable neoplastic tissue. Percentage necrosis of the treated lesions was tabulated using the following schema as described by the pathologist: 100% histological necrosis (defined as absence of any viable tissue), >50% necrosis (defined as significant necrosis but with clusters of viable tissue), or <50% necrosis (defined as minimal necrosis).
Although other studies have defined >90% necrosis as complete necrosis, we classified complete histological necrosis only if 100% necrosis was in fact noted by the pathology report.26 The presence of any degree of differentiation of the tumor was interpreted as HCC.27, 28 In order to report most conservatively, lesions with significant necrosis but with clusters of viable cells were classified as >50% necrosis.
The Fisher's exact test was used to determine the statistical significance between the differences observed in the imaging parameters under study. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were also calculated for the favorable radiologic changes after treatment, and the gold standard was taken as 100% histologic necrosis on explant pathologic evaluation.
The demographics of the patient population are presented in Tables 1 and 2. The median age was 59 years (range, 43-75), most were male (83%), and the majority were Eastern Cooperative Oncology Group performance status of 0 (80%). Sixty percent of the patients were Child-Pugh class A. Sixty-four percent were diagnosed with HCC using imaging studies, whereas 36% were confirmed to have HCC by biopsy. The presence of portal vein thrombosis on imaging in these patients was considered to be tumor-related. There were five patients diagnosed with vascular invasion; only two of these had pathologic evidence of vascular invasion at explant. Forty percent of the patients were BCLC class A, 34% were BCLC class B, and 26% were BCLC class C. At the time of treatment, 43% were UNOS class T2, 29% were UNOS class T3, 14% were UNOS class T4a, and 14% were UNOS class T4b by blinded radiologic assessment. Following 90Y radioembolization, no change in UNOS stage was observed except for eight of the 10 T3 patients who were downstaged to T2.
Table 1. Patient Baseline Characteristics
Abbreviations: CT, computed tomography; ECOG, Eastern Cooperative Oncology Group; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MRI, magnetic resonance imaging.
Etiology of HCC
HCV + alcohol
Method of diagnosis
Imaging (CT or MRI)
ECOG performance status
Table 2. Patient Baseline Staging
Baseline Child-Pugh class
Treatment and Follow-up
The median number of treatments per patient was 1 (range, 1-4). The median number of treatments to the target lesion was 1 (range, 1-2). The second treatment to the same lesion was performed if the first treatment was thought to have incompletely targeted the lesion because of alternate blood supply from parasitized hepatic arteries. The median dose administered to the target site was 93.5 Gy (range, 27-203). The median lung dose per treatment was 5.13 Gy (range, 0.8-26). RFA was not performed in any case due to the lesion location in 58% of the cases (caudate, dome, or surface), tumor size (>3 cm) in 76% of the cases, or proximity to a major duct or vessel in 10% of the cases. The median time from time of first treatment to explantation of the liver was 6 months (range, 1.4-17.9). The median number of days from the time of the last scan before explantation to the time of explantation was 29 days (range, 7-116).
Following 90Y radioembolization, 35 patients underwent transplantation (n = 33), resection (n = 1), or autopsy (n = 1) yielding 38 treated lesions that were the focus of this analysis. Three patients had two target lesions. One patient underwent transplantation 12 days following treatment and had no follow-up imaging for evaluation of imaging response. Hence, 37 lesions formed the basis of the imaging-pathologic correlate, while 38 lesions formed the basis of the overall post 90Y pathologic analysis.
For the 37 lesions with imaging-pathologic correlates, 61% showed 100% histologic necrosis, 24% showed >50% necrosis, and 15% showed <50% histological necrosis. All lesions showed some degree of histologic necrosis on pathologic evaluation at explant. The patient who received transplantation 12 days after treatment demonstrated <50% histological necrosis.
Thirty-seven lesions had imaging prior to and following treatment before explantation. Contrast-enhanced magnetic resonance imaging was used to assess the imaging surrogates after treatment in all patients. The degree of necrosis by imaging, stratified by baseline tumor size, is summarized in Table 3. The 12 lesions with a maximum diameter >5 cm had a mean diameter of 7.6 cm (range, 5.2-13.5). Figure 1 represents a waterfall plot demonstrating the percent change in cross-product of the lesions from baseline following treatment. Thirty-four (92%) lesions showed at least some decrease in size after treatment. Eighteen (49%) reached partial response by WHO criteria. EASL CR and PR was achieved in 12 (32%) and 21 (57%) of the lesions, respectively.
Table 3. Posttreatment Imaging Findings Stratified by Pretreatment Size
Months from first treatment to transplantation, median (range)
Effect of Time Between Treatment and Explant.
The effect of the duration between the radioembolization and explantation of the liver on the degree of histologic necrosis seen in the treated lesion is shown in Table 4. The time between treatment and explantation was available for all 38 lesions. Organs that were explanted <3 months from treatment had <50% pathologic necrosis in 55% of the 11 target lesions. Organs explanted >6 months after treatment had complete histologic necrosis in 68% of the 19 target lesions. There seemed to be a trend for higher rate of necrosis as time elapsed from treatment.
Table 4. Histologic Necrosis Stratified According to Time Period between First Treatment and Explantation
Time from Treatment
Number of treatments to target lesion, median (range)
Histologic necrosis, n (%)
Effect of Time Between Treatment and Radiologic Evaluation.
The evolution of radiologic changes with time after first treatment is summarized in Table 5. The median time to WHO response was 126 days (95% confidence interval 80.2-313.2); the median time to EASL response (CR or PR) was 34 days (95% confidence interval 29-43). The percentage of lesions showing response (EASL and WHO) was seen to increase with time.
Table 5. Radiologic Response Evolving with Time
Time After Treatment
Abbreviations: EASL, European Association for the Study of the Liver; WHO, World Health Organization.
Explanted patients were excluded from response analysis.
Correlation of Radiologic and Pathologic Findings.
Table 6 presents the degree of histologic necrosis in tumors stratified according to the pretreatment maximum dimension. All 38 lesions were imaged prior to treatment. Lesions with a maximum diameter of 1 to 2.9 cm, 3 to 5 cm, and >5 cm had 89%, 65%, and 33% complete histologic necrosis, respectively.
Table 6. Histologic Necrosis Stratified According to Pretreatment Size
Histologic necrosis, n (%)
The radiologic findings of the 37 lesions were compared with the pathologic findings in order to determine the predictability of actual necrosis by imaging. Table 7 presents the degree of histologic necrosis in lesions according to the response observed by WHO criteria. PR, SD, and PD after radioembolization were found to have complete histologic necrosis in 78%, 53%, and 0% of the lesions, respectively.
Table 7. Histologic Necrosis According to WHO/Enhancement Characteristics/EASL
Abbreviations: EASL, European Association for the Study of the Liver; WHO, World Health Organization.
Histologic necrosis, n (%)
Thin Rim Enhancement
Peripheral Nodular Enhancement
Histologic necrosis, n (%)
Histologic necrosis, n (%)
Table 7 presents the degree of histologic necrosis seen in lesions according to the enhancement characteristics observed in the treated lesions. No enhancement, thin rim enhancement, and peripheral nodular enhancement exhibited 100%, 93%, and 38% complete histologic necrosis, respectively. Table 7 presents the degree of histologic necrosis observed in lesions according to response by the EASL necrosis criteria. CR, PR, and SD after 90Y had complete histologic necrosis in 100%, 52%, and 0% of the lesions, respectively. Figures 2 and 3 demonstrate an example of radiologic-pathologic correlation.
Table 8 summarizes the sensitivity, specificity, PPV, and NPV of favorable radiologic changes as predictors of necrosis. EASL CR had a 100% PPV and specificity. WHO PR showed a 78% PPV and 71% specificity, and these values increased upon application of stricter criteria (>60% decrease in size by WHO criteria). Rim enhancement was seen to have a 93% PPV and specificity.
Table 8. Sensitivity, Specificity, PPV, and NPV
Positive Predictive Value (%)
Negative Predictive Value (%)
Abbreviations: CR, complete response; EASL, European Association for the Study of the Liver; PR, partial response; WHO, World Health Organization.
*Gold standard taken as 100% histologic necrosis.
CR or PR
PR (>50% decrease)
Thirty-one lesions demonstrated an overall favorable radiologic response to treatment by independent radiology assessment; 21 (68%) showed complete histologic necrosis. Twenty-eight lesions showed possible evidence of residual disease by the same radiologic assessment on posttreatment imaging; 28 (46%) showed incomplete histologic necrosis.
Baseline AFP >200 ng/mL was observed in 13 patients. Complete histologic necrosis was seen in 67% of patients with >50% decline in baseline AFP (n = 11). One patient had an increase in AFP and was found to have <50% histologic necrosis.
The incidence of hepatocellular carcinoma is increasing worldwide.29 Progress has been made in the past decade in the management of HCC. The role of resection and transplantation in early HCC is well known,4, 30 and the availability of new systemic and locoregional therapies is promising.31, 32 The development of effective locoregional treatment modalities for HCC necessitates the ability to accurately assess response to treatment using imaging studies. The variability in the radiologic response seen after different locoregional therapies makes the adoption of a universal system for assessing response after treatment of HCC difficult. The response seen on imaging after radiofrequency ablation is represented by an area of necrosis that demonstrates lack of enhancement but often an increase in size of the treated area. Using WHO criteria, this may represent progressive disease. Response assessment following TACE has traditionally been difficult prior to the advent of magnetic resonance imaging, because (1) the measurement of necrosis is confounded by the accumulation of lipiodol and (2) lack of enhancement is artificially created from macroembolization by occlusive embolic particles.20 Internal radiation with 90Y represents a microembolization procedure that does not significantly alter blood flow.33 Arterial hypervascularity of tumor relative to normal tissue is the underlying principle of this therapy, resulting in minimal toxicity.2290Y has demonstrated a role in the management of HCC.34
This study included 35 patients treated at a single institution with 90Y glass microspheres. The patients then underwent radical therapies (with the exception of the autopsy). This provided the opportunity to study the treated lesion pathologically. Explant pathology was compared to imaging findings post radioembolization to better determine the accuracy of radiographic treatment response. This is a novel study which shows that favorable radiologic response has a reasonable predictability of assessing actual tumor necrosis in lesions treated with radioembolization. Utilizing a strict definition of complete pathologic necrosis, radioembolization led to complete necrosis in 61% of the treated lesions. This is in line with prior reports that employed other forms of liver-directed therapy for HCC. Complete histologic necrosis following TACE according to Maddala et al.6 was 15%; however, this was defined as absence of any lesion on explant, while 59% of the dominant lesions showed >90% necrosis. The complete histologic necrosis rate according to Lu et al.35 was 65.7% after RFA. A similar study by Veltri et al.26 showed >90% histologic necrosis after TACE in 24% of the lesions, in 80% of the lesions after PEI and in all the lesions after both TACE and PEI.
The length of time from initial radioembolization to examination of explant pathology predicted the degree of histologic necrosis. The patient who underwent transplantation within 2 weeks of treatment had minimal necrosis on examination of his explanted liver. As seen in Table 4, all of the lesions that had <50% necrosis were explanted <3 months from radioembolization. The data may suggest that longer time is necessary for the radiation to have its cytotoxic effect on tumor, although certainly tumor size is another relevant factor in this effect.7 Table 5 may suggest that the radiologic response (EASL and WHO) seen after radioembolization evolves with time. However, since the median time to EASL response is 34 days and the median time to WHO response is 126 days, the requirement of the 1-month scan and the confirmatory role of the 3-month scan appear to be supported by these data. The use of diffusion-weighted functional magnetic resonance imaging to calculate apparent diffusion coefficient seems to have a promising role in assessing response to treatment earlier than conventional anatomic imaging techniques and is anticipated to play a role in future treatment algorithms.36
The smaller lesion demonstrated superior response rates both histologically and radiologically. However, lesions >5 cm had complete histologic necrosis in 33% of the cases. By using WHO criteria, partial response rates have been between 20% and 40% in the published literature.7, 37 The cohort in this study demonstrated a WHO PR rate of 49%. A PR by WHO criteria was seen to have complete histologic necrosis in 78% of the lesions. SD by WHO criteria indicated complete histologic necrosis in 53% of the lesions. This further confirms that for locoregional therapies, change in lesion size alone cannot be used to assess treatment response.24
Enhancement characteristics should be concomitantly studied to improve the predictability of actual necrosis by radiologic evaluation. Fifteen of the 16 (94%) patients in this study who had partial or complete EASL response and absence of nodular enhancement or progression according to WHO criteria had complete histologic necrosis. The favorable enhancement characteristics of the lesion on posttreatment radiologic examination (such as no enhancement or rim enhancement) correlated well with the degree of actual necrosis. By using necrosis criteria as outlined by EASL,9 the response rates reported in the literature have been between 70% and 90% after radioembolization.34 The cohort in this analysis had an EASL response rate of 89%. Complete EASL response correlated with 100% histologic necrosis in all cases. SD by EASL necrosis criteria correlated with <50% histologic necrosis in 75% of the lesions.
In this study, the values representing the performance of EASL criteria in predicting complete histologic necrosis were found to be superior to those obtained using WHO criteria. EASL response (CR or PR) resulted in 100% sensitivity, while EASL CR resulted in 100% specificity. These criteria outperformed the use of WHO PR and the radiologic finding of rim enhancement.
There are limitations to this study. Although this was a comprehensive analysis of radiologic–pathologic correlation following internal radiation, we studied a relatively small number of patients. Explantation within a median time of 6 months of the first treatment restricted our ability to study long-term response or time to progression. The study was at the lesional level, and the response was measured as such. Therefore, lesions that were not treated or were not identified until the time of explant examination were not part of this analysis. Future studies will examine treatment response and risk of HCC recurrence after transplantation/resection. Well-defined criteria based on response to treatment according to change in size and enhancement following this treatment option need to be established. There is a need for prospective studies that ensure staging systems incorporate the effect of the systemic and locoregional therapies on the tumor, as well as a need to correlate these favorable responses seen radiographically to improved outcomes.38
In conclusion, the data presented in this study support the notion that a decrease in enhancement and size of the target lesion correspond to actual histologic necrosis. Peripheral nodular enhancement more commonly indicates the presence of viable neoplastic tissue. EASL CR and WHO PR correlated well with complete necrosis. The organs explanted quickly after radioembolization were found to have minimal pathologic necrosis in the target lesions, suggesting a temporal effect for necrosis to ensue. Lesions <3 cm were treated with internal radiation as they were not amenable to ablative techniques given their location. There was a trend for smaller lesions to exhibit better response to radioembolization, although one-third of lesions >5 cm had evidence of complete histologic necrosis.