Volume 112, Issue 7 p. 1538-1546
Original Article
Free Access

Liver disease induced by radioembolization of liver tumors

Description and possible risk factors

Bruno Sangro MD, PhD

Corresponding Author

Bruno Sangro MD, PhD

Liver Unit, Department of Internal Medicine, University Clinic and CIBERHD, Pamplona, Spain

Drs. Sangro, Martinez-Cuesta, and Bilbao have received lecture fees from Sirtex Medical Europe.

Fax: (011) 34 948 296 500

Liver Unit. Department of Internal Medicine, Clinica Universitaria de Navarra, Avda. Pio XII 36, 31008-Pamplona, Spain===Search for more papers by this author
Belen Gil-Alzugaray MD

Belen Gil-Alzugaray MD

Liver Unit, Department of Internal Medicine, University Clinic and CIBERHD, Pamplona, Spain

Search for more papers by this author
Javier Rodriguez MD, PhD

Javier Rodriguez MD, PhD

Department of Medical Oncology, University Clinic, Pamplona, Spain

Search for more papers by this author
Iosu Sola MD, PhD

Iosu Sola MD, PhD

Department of Pathology, University Clinic, Pamplona, Spain

Search for more papers by this author
Antonio Martinez-Cuesta MD, MSc, FRCR

Antonio Martinez-Cuesta MD, MSc, FRCR

Department of Interventional Radiology, University Clinic, Pamplona, Spain

Drs. Sangro, Martinez-Cuesta, and Bilbao have received lecture fees from Sirtex Medical Europe.

Search for more papers by this author
Antonio Viudez MD

Antonio Viudez MD

Department of Medical Oncology, University Clinic, Pamplona, Spain

Search for more papers by this author
Ana Chopitea MD

Ana Chopitea MD

Department of Medical Oncology, University Clinic, Pamplona, Spain

Search for more papers by this author
Mercedes Iñarrairaegui MD, PhD

Mercedes Iñarrairaegui MD, PhD

Liver Unit, Department of Internal Medicine, University Clinic and CIBERHD, Pamplona, Spain

Search for more papers by this author
Javier Arbizu MD, PhD

Javier Arbizu MD, PhD

Department of Nuclear Medicine, University Clinic, Pamplona, Spain

Search for more papers by this author
Jose I. Bilbao MD, PhD

Jose I. Bilbao MD, PhD

Department of Interventional Radiology, University Clinic, Pamplona, Spain

Drs. Sangro, Martinez-Cuesta, and Bilbao have received lecture fees from Sirtex Medical Europe.

Search for more papers by this author
First published: 19 March 2008
Citations: 296

Fax: (011) 34 948 296 500



To the authors' knowledge, liver damage after liver radioembolization with yttrium90-labeled microspheres has never been studied specifically.


Using a complete set of data recorded prospectively among all patients without previous chronic liver disease treated by radioembolization at the authors' institution from September 2003 to July 2006, patterns of liver damage were identified and possible risk factors were analyzed.


In all, 20% of patients developed a distinct clinical picture that appeared 4 to 8 weeks after treatment and was characterized by jaundice and ascites. Veno-occlusive disease was the histologic hallmark observed in the most severe cases. This form of sinusoidal obstruction syndrome was not observed among patients who never received chemotherapy or in those in whom a single hepatic lobe was treated. Relevant to treatment planning, a possible risk factor was a higher treatment dose in relation to the targeted liver volume. A transjugular intrahepatic stent shunt improved liver function in 2 patients with impending liver failure, although 1 of them eventually died from it.


Radioembolization of liver tumors, particularly after antineoplastic chemotherapy, may result in an uncommon but potentially life-threatening form of hepatic sinusoidal obstruction syndrome that presents clinically with jaundice and ascites. Cancer 2008. © 2008 American Cancer Society.

Radioembolization (RE) is a form of brachytherapy in which radiation is delivered by radioactive isotopes labeled in microspheres that are injected into the arteries that feed the tumors.1 Liver RE has been shown to produce significant rates of tumor growth control among patients with hepatocellular carcinoma (HCC)2-7 and liver metastases from other tumors, particularly colorectal cancer.8-17 After injection, microspheres are embolized in the tumor microvasculature in a larger proportion than in the nontumoral parenchyma. Although the clinical impact of damage caused by yttrium90 (90Y)-microspheres that do not reach the tumors is purportedly low, to our knowledge the information available in the literature regarding liver toxicity after RE is obscure and sometimes confusing. We analyzed liver damage occurring after RE among patients without previous chronic liver disease using a set of data recorded prospectively. Herein we describe the incidence, patterns, outcome, and possible risk factors of a clinical picture that is consistent with the definition of hepatic sinusoidal obstruction syndrome.18


From September 2003 to July 2006, 100 patients with liver tumors were evaluated for RE at our institution. Treatment was contraindicated in 20 patients and 35 patients were excluded from analysis, 28 because they had primary liver tumors with concurrent chronic liver disease (either confirmed histologically, or suspected because of clinical and imaging data), and 7 because of deficient follow-up. Therefore, the occurrence of liver damage was studied in 45 patients without chronic liver disease who underwent liver RE. Clinical and analytical data were recorded prospectively for all treated patients.

Treatment and Follow-up

Pretreatment evaluation and patient selection

RE was considered for those patients with liver tumors that could not be treated by surgical or percutaneous ablation, had already received standard therapy if available, had a good functional status (Eastern Cooperative Oncology Group [ECOG] score of 0–2), and had a histologic diagnosis of malignancy. Although most patients had isolated liver cancer, some had extrahepatic disease that was considered either under control or not likely to influence prognosis (eg, abdominal lymph node). The majority of patients had been previously treated with 2 or 3 lines of chemotherapy. RE was contraindicated in the presence of severe hypersplenism (neutrophil count <1.5/pL or platelet count <45/pL), altered liver function (bilirubin >3 mg/dL, or ascites), altered renal function (creatinine >2 mg/dL), or any contraindication to angiography.

Our protocol for RE of liver tumors has been explained in detail elsewhere7 and can be summarized as follows. After signing an informed consent, patients were studied within the 4-week period before treatment by means of: 1) an angiogram, to detect possible variants of arterial liver irrigation, identify the vessels that provide arterial blood supply to every liver tumor nodule, and assess portal vein blood flow; 2) a scintigraphy using macroaggregates of technetium-99m labeled human serum albumin (MAA) to measure the degree of shunt to the lung, predict any possible misplacement of microspheres in the gastrointestinal tract, and evaluate the relative distribution of radioactivity between the hepatic tumors and the nontumoral liver; 3) chest and abdominal dual-phase, spiral computed tomography (CT) or magnetic resonance imaging (MRI) to measure the volume of the tumor mass to be treated, and that of the nontumoral liver irrigated by the artery in which the tip of the catheter was to be placed; and 4) blood tests including complete blood count, liver function tests, creatinine, albumin, and prothrombin activity.

During angiography the original vascular anatomy was occasionally altered by occluding either those vessels with an origin in the hepatic arteries that supplied extrahepatic organs (to prevent their irradiation) or those aberrant hepatic vessels originating from nonhepatic arteries (so that flow redistribution might allow targeting all liver nodules from a single injection of microspheres). Theoretically, this last procedure may result in an unequal distribution of the microspheres and an increased risk of liver toxicity.

Since October 2004, a fusion of the single-photon emission CT (SPECT) images obtained after MAA injection and the images of contrast-enhanced MRI or CT studies was obtained for every patient. Patients were denied RE whenever the activity uptake observed in tumor lesions was very poor in comparison with nontumoral areas because we believed this could reflect an increased risk of liver damage because of radiation.

Treatment procedure and follow-up

All patients were treated with SIR-Spheres (Sirtex Medical Europe, Bonn, Germany), an implantable medical device consisting of resin microspheres labeled with 90Y. Two methods were used to calculate the activity of 90Y to be administered as explained elsewhere.7 In the BSA Method, activity depends on the body surface area (BSA) and the extent of tumor liver involvement, and in the Partition Model,19 the maximal activity that remains safe for the lung and the nontumoral liver is calculated. Although this model was originally designed for patients with single or discrete nodules, a modification easily applicable for those patients with multinodular tumors has been used as the main method for activity calculation.

On the treatment day the calculated activity of 90Y-microspheres was injected after confirming that there were no new collateral vessels connecting to the gastrointestinal tract. Patients then remained in the hospital overnight and supportive therapy consisted of a proton pump inhibitor that was prescribed for 4 weeks, and antiemetics and analgesics on demand.

Follow-up schedules after treatment varied depending on the treatment plan of each patient, but monthly assessments that included physical examination and blood tests were available. Abdominal imaging was performed for evaluation of response with a sequence that differed according to tumor type and individual treatment plan, although at least bimonthly assessments are available for all patients.

Selection of Variables and Statistical Analysis

More than 100 pretreatment and posttreatment variables that might have affected the occurrence of liver damage after RE were analyzed (Table 1). A commercial statistical software package (SPSS for Windows, version 13.0.1; SPSS Inc, Chicago, Ill) was used for data analysis. The results are presented as the mean (standard deviation) for continuous variables with a normal distribution, and as the median (interquartile range) for continuous variables with a nonnormal distribution. To investigate their correlation with the odds of having liver damage, a univariate analysis was first performed for all of the aforementioned variables using a Student t test or Mann-Whitney U test when testing associations of independent continuous variables, and the chi-square or Fisher exact test when testing associations of independent categoric variables. Because of the reduced number of cases, only the most clinically relevant factors among those statistically significant at the 0.1 level were then selected for multivariate analysis using multiple logistic regression. A backward selection procedure, with a P value < .1 used for retention in the model, was performed to identify important factors at the .05 level of statistical significance. The Hosmer-Lemeshow test was used to check the goodness-of-fit of the model.20 All the tests performed as part of the univariate and multivariate analysis were 2-sided. A P value < .5 was considered to indicate statistical significance.

Table 1. Variables Under Study
Factors linked to the patient but not to the neoplastic disease or its treatment
Age and sex
History of alcohol consumption (occasional or daily)
Radiologic signs of liver steatosis (yes/no)
Serum levels of transaminases at the time of diagnosis of the primary tumor (as possible indicators of pre-existing liver injury)
Factors linked to previous antineoplastic treatment
Major liver surgery (yes/no)
Systemic chemotherapy, hepatic artery embolization, and hepatic arterial infusion chemotherapy within the 2-y period previous to RE (yes/no)
Administration of antineoplastic agents at any time (yes/no)
Total dose in mg received prior to RE for the most common antineoplastic agents
Time in d from the last course of treatment containing each of these agents to RE
Having received each of these agents during the mo previous to RE (yes/no)
Administration of chemotherapy within the 2-mo period after RE (yes/no)
Factors linked to the clinical condition of the patient and the liver at the time of RE
Liver function tests
Blood cell count
BMI and body surface area
Findings observed or procedures done at the time of angiographic evaluation (yes/no), including signs of portal hypertension and flow redistribution
Total liver volume
Total tumor volume
Volume of the nontumoral liver potentially targeted by microspheres (eg, for a lobar treatment, this variable represents the volume of the lobe exclusive of the volume of the tumor)
Factors linked to the RE procedure
Extension of treatment (whole liver or lobar)
Total activity administered in GBq
Estimated dose of radiation delivered to the tumor and the nontumoral liver in Gy
Correlation between the activity actually administered and the tumor volume, the nontumoral liver potentially targeted by microspheres, the degree of liver involvement in %, the BMI, and the activity as calculated by body surface area
Post-treatment variables used to identify liver toxicity induced by RE
Liver function tests, coagulation tests, and detection of ascites or liver encephalopathy on physical examination, all of which were obtained at 1 mo, 2 mo, 3 mo, and 6 mo after treatment
The worst level of liver function tests observed within the first 2 mo after RE (because it most likely reflects better the intensity of liver damage)
  • RE indicates radioembolization; BMI, body mass index; GBq, gigabecquerel; Gy, grays.


Characteristics of the Study Population

The study population included 45 consecutive patients without chronic liver disease who were undergoing RE of liver tumors (Table 2). Twelve patients had no previous exposure to chemotherapy, including 10 patients with HCC and 2 patients with liver metastases from neuroendocrine tumors (the latter had progressed to somatostatin analogs and interferon therapies). Among the rest, therapeutic regimens most frequently included oxaliplatin, capecitabine, irinotecan, and gemcitabine (Table 3). Conditions that may alter the liver vascular network and thus potentially modify the distribution of microspheres within the liver were identified in 19 patients, including major liver surgery (8 patients), transarterial embolization (6 patients), and hepatic artery infusional chemotherapy (16 patients). The mean activity administered was 1.7 gigabecquerels (GBq) but the range varied widely from 0.31 GBq for a whole-liver treatment in a patient with minimal tumor burden to 3.05 GBq for a right lobe treatment in a patient with a bulky tumor.

Table 2. Characteristics of the 45 Patients Without Chronic Liver Disease Who Underwent Liver RE
Age, y 55 (10.6)
Sex, M/F 30/15
Primary tumor
 Colorectal 13
 Hepatocellular 12
 Neuroendocrine 5
 Breast 4
 Various* 11
Data at time of diagnosis of primary tumor
 Daily alcohol consumption 15.9%
 Liver steatosis 28.9%
 Serum ALT, IU/mL 24.0 (27.5)
 Serum GGTP, IU/mL 165 (235.7)
Previous treatments
 Liver surgery 17.8%
 Transarterial embolization 13.3%
 Systemic chemotherapy 73.3%
 Hepatic artery infusional chemotherapy 35.5%
Data at time of RE
 BMI, kg/m2 24.8 (3.64)
 Serum ALT, IU/mL 20 (19.6)
 Serum GGTP, IU/mL 112 (164)
 Serum total bilirubin, mg/dL 0.69 (0.39)
 Prothrombin activity, % 96 (9.0)
 Hemoglobin, g/dL 12.0 (1.9)
 WBC count, /pL 6.35 (2.96)
 Platelet count, /pL 222 (93.9)
 Tumor volume, mL 239 (734.2)
 Nontumor-targeted liver volume, mL 1471 (490.8)
 Flow redistribution 17.8%
 Extension (whole liver/right lobe/left lobe) 33/9/3
 Activity administered, GBq 1.62 (0.59)
Chemotherapy after RE 28.9%
  • RE indicates radioembolization; M/F, male/female; ALT, alanine aminotransferase; GGTP, γ-glutamyl transpeptidase; BMI, body mass index; WBC, white blood count; GBq, gigabecquerels.
  • Values are shown as the median (± standard deviation) unless otherwise explained.
  • * Various primary tumors include pancreatic (3 patients); renal (2 patients); gastric (2 patients); and ampuloma, jejunum, cholangiocarcinoma, and small cell lung cancer (1 patient each).
Table 3. Characteristics of the Chemotherapy Treatments Received Prior to Radioembolization*
Agent No. of patients REILD (n = 9) No REILD (n = 24)
Total dose, mg* Time since last course, days* Treatment within the last 30 days Total dose, mg* Time since last course, days* Treatment within the last 30 days
Oxaliplatin 19 1954 (1565) 74 (84) 3/5 1259 (658) 157 (156) 1/14
Capecitabine 19 375 × 103 (472 × 103) 38 (38) 4/6 409 × 103 (329 × 103) 156 (160) 3/13
Irinotecan 13 1410 (721) 199 (267) 1/2 2026 (1385) 49 (18) 2/11
5-fluorouracil 10 0 0
Gemcitabine 12 18733 (8869) 277 (253) 0/3 12334 (12282) 66 (63) 4/9
Bevacizumab 5 1 4
Cisplatin 3 0 0
Mitomycin C 6 0 0
Doxorubicin 5 0 0
Taxanes 5 0 0
  • REILD indicates radioembolization-induced liver disease.
  • * Values are shown as the mean (± standard deviation) unless otherwise indicated.

Liver Disease Induced by RE

In the first 60 days after RE, 9 patients developed a similar clinical picture characterized by jaundice and ascites in the absence of overt tumor progression or bile duct dilation. Liver function tests revealed an increase in total bilirubin >3 mg/dL in all cases, a more heterogeneous tendency toward rising alkaline phosphatase and γ-glutamyl transpeptidase (GGTP), and virtually no changes in transaminases. Among these patients, baseline and peak values within this 60-day period averaged 1.05 (0.56) mg/dL and 9.85 (9.57) mg/dL, respectively, for total bilirubin, 266.4 (133.4) UI/mL and 527.5 (232.0) UI/mL, respectively, for alkaline phosphatase, and 187.4 (142.2) UI/mL and 346.6 (188.6) UI/mL, respectively, for GGTP. Three patients had a rapid, progressive course, with bilirubin levels >20 mg/dL and declining liver function. As shown in Figure 1, the liver biopsy of 2 of these patients showed extensive sinusoidal congestion affecting perivenular areas with focal hepatic atrophy, areas of necrosis around central veins with fresh thrombosis, and some cholestasis in periportal areas. These findings were consistent with hepatic veno-occlusive disease (VOD) and because of the abating liver function a transjugular intrahepatic portosystemic stent shunt (TIPS) was performed 60 days and 67 days, respectively, after RE. In both cases, TIPS placement was followed by an improvement in liver function. One of the patients had liver metastases from breast cancer and was in radiologic complete response 26 months after RE. The other had liver metastases from colorectal cancer, and 11 days after TIPS placement died from sepsis-related multiorgan failure while his liver function was progressively improving. The third patient with intense, progressive liver dysfunction refused a liver biopsy and died 75 days after RE. The remaining 6 patients had a more benign course, as shown in Figure 2. Ascites was easily controlled with diuretics and bilirubin tended to plateau but did not return to normal values in any case. At the time of last follow-up, all the patients had died of tumor progression. Two of them developed hepatic encephalopathy with increased serum levels of ammonium that responded to oral lactulose.

Details are in the caption following the image

Hepatic biopsy from Patient 1 demonstrated (A) extensive sinusoidal congestion with focal hepatic atrophy and (B) an area of hepatocellular necrosis around a central vein with fresh thrombosis. Some degree of cholestasis in periportal areas was also noted. Biopsy findings from Patient 2 were similar but milder (C), for example, sinusoidal congestion affecting perivenular areas and (D) cholestasis and a venous lesion healed with eccentric intimal fibrosis. In both cases some microspheres were located in arterial and portal vein branches (arrowheads).

Details are in the caption following the image

Changes in serum bilirubin levels among patients with radioembolization-induced liver disease. Each solid line represents results from an individual patient.

Among patients without this distinct picture, a clinically irrelevant but statistically significant increase in bilirubin was observed after RE (mean values: 0.78 mg/dL at baseline and 1.03 mg/dL at acme; P < .05), but not in alkaline phosphatase, GGTP, or transaminases.

Risk Factors for Liver Disease Induced by RE

When compared with patients who did not, those who developed RE-induced liver disease (REILD) were younger (mean age of 45 years vs 56 years; P < .05), leaner (mean body mass index [BMI] of 23.7 vs 26.0; P < .1), and had received chemotherapy more frequently (100% vs 66%; P < .05). At the time of RE they had a lower tumor burden (mean tumor volume of 190 mL vs 590 mL; P < .1), a higher bilirubin (mean bilirubin of 1.05 mg/dL vs 0.79 mg/dL; P < .1), and were more likely to have a flow redistribution procedure performed at angiographic evaluation (44.4% vs 12.5%; P < .05). They did not receive a higher total activity (mean activity of 1.6 GBq vs 1.76 GBq; P value not significant), but this activity might have resulted (according to the rough estimation of the partition model) in a higher dose of radiation being delivered to the nontumoral liver (mean dose of 37 grays [Gy] vs 29 Gy; P < .1). Only patients exposed to chemotherapy developed REILD. As shown in Table 3, patients who developed REILD more likely had received oxaliplatin and capecitabine in the recent past, although none of these differences were found to be statistically significant. Not surprisingly, all 9 patients with REILD had been treated in a whole-liver fashion as compared with only 66% of the patients without liver disease (P < .05). In fact, on multivariate analysis of the entire cohort, age, bilirubin at baseline, treatment approach (whole-liver vs lobar), and the amount of activity administered relative to the total volume treated were found to be independent risk factors for the development of REILD. With regard to histology, patients with primary liver cancer did not develop REILD but it should be noted that virtually none of them had prior exposure to chemotherapy.

A univariate analysis of the group of patients undergoing whole-liver RE showed the same risk profile for developing REILD as in the entire cohort (Tables 4 and 5). Owing to the low number of cases, a multivariate analysis was performed after selecting those variables that were more clinically meaningful and statistically significant. In the final model, only age <55 years and having received an activity higher than 0.8 GBq per L of targeted liver volume were found to be independent risk factors for REILD.

Table 4. Risk Factors for REILD Among Patients Receiving Whole-Liver RE on Univariate Analysis
Variables Non-REILD (n = 24) REILD (n = 9) P
Age, y 55.1 (9.0) 44.8 (6.5) .004
BMI, kg/m2 25.6 (3.5) 23.7 (2.3) .15
ALT at the time of diagnosis, mg/dL 32.0 (27.0) 49.7 (27.5) .12
Hepatocellular carcinoma, % 33.3 0 .002
Previous chemotherapy
 Any chemotherapy, % 79.1 100 .13
 Bevacizumab, % 26.3 0 .13
 Time from last course of capecitabine, d 175 (174) 38 (38) .03
Blood tests at baseline
 Total bilirubin, mg/dL 0.70 (0.31) 1.05 (0.56) .10
 Leukocyte count, /pL 7.42 (2.98) 4.94 (1.64) .006
 Platelet count, /pL 247 (102) 171 (74) .05
Flow redistribution during evaluation, % 4.3 44.4 .004
Tumor volume, mL 546 (739) 190 (236) .04
Activity administered relative to:
 Total liver volume, GBq·1000/mL 0.08 (0.03) 0.10 (0.01) .003
 Tumor volume, GBq·100/mL 1.42 (2.01) 6.14 (11.2) .13
Estimated dose of radiation delivered to:
 Nontumoral liver, Gy 25.7 (12.1) 36.7 (12.3) .02
 Tumor, Gy 116.4 (64.2) 194.1 (88.8) .09
Chemotherapy after RE, % 41.1 55.5 .16
  • RE indicates radioembolization; REILD, RE-induced liver disease; BMI, body mass index; ALT, alanine aminotransferase; GBq, gigabecquerels; Gy, grays.
  • The following variables demonstrated a Pvalue > .1: sex; alcohol consumption, radiologic liver steatosis, and asparate aminotransferase and γ-glutamyl transpeptidase levels at the time of diagnosis of the primary tumor; liver surgery, hepatic arterial embolization, or hepatic artery infusional chemotherapy prior to RE; previous treatment with 5-fluorouracil, oxaliplatin, irinotecan, capecitabine, gemcitabine, cisplatin, mitomycin C, doxorubicin, or taxanes; total dose of the most relevant agents received prior to RE and the gap between the last course and RE for 5-fluorouracil, oxaliplatin, irinotecan, and gemcitabine; liver function tests (except for bilirubin) and hemoglobin at baseline; aberrant liver vascularization and embolization of collateral vessels directed toward extrahepatic organs; total liver volume, the volume of the nontumoral liver potentially targeted by microspheres, and the percentage of liver involvement; and the total activity administered.
  • Values are shown as the mean (± standard deviation).
Table 5. Risk Factors for REILD Among Patients Receiving Whole-Liver RE on Multivariate Analysis
Variable P OR 95% CI
Age <55 y .003 1.9 1.24–2.91
Activity relative to targeted liver volume >0.8 GBq/L .03 1.6 1.17–2.18
Capecitabine administered within the last month .07 5.5 0.74–40.8
Leukocyte count <4000/pL .16 3.5 0.55–22.0
  • REILD indicates radioembolization-induced liver disease; RE, radioembolization; OR, odds ratio; 95% CI, 95% confidence interval; GBq, gigabecquerels.

As shown in Figure 3, among patients treated in a whole-liver fashion, the severity of REILD was related, first, to the degree of tumor involvement. Hence, patients in whom the tumor represented <0.2% of the total liver volume developed a more severe disease. Second, it was related to the degree of hepatomegaly. Hence, patients with a total tumor volume >2 L developed a less severe disease. This may reflect a limitation of the methods currently used to calculate the therapeutic activity among the population of patients with very low tumor burden, particularly when affecting a liver that is near normal in size.

Details are in the caption following the image

The intensity of radioembolization-induced liver disease is correlated with the degree of tumor involvement and total liver volume.


The low tolerance of the liver to radiation has prevented external irradiation from being used in the treatment of liver tumors for decades.21 However, an antitumor effect of liver RE has been clearly shown in the past few years. Among patients with liver metastases from colorectal cancer, RE with resin 90Y-microspheres has been shown to induce tumor regression in a significant proportion of patients,16 and to improve the results of systemic11, 17 and regional9 chemotherapy in small randomized controlled trials. Among patients with HCC, RE with resin 90Y-microspheres results in significant tumor growth control,2, 7 particularly for highly vascularized tumors,22 and may be useful in downstaging patients for surgery.23 However, although RE is generally well tolerated, it may cause relevant toxic effects including liver injury.

To our knowledge, liver damage after RE has not been thoroughly studied previously. The majority of published series, including clinical trials, have used the National Cancer Institute Common Toxicity Criteria (available at: http://ctep.cancer.gov Accessed June 2007) to evaluate toxicity, but this individual evaluation of blood tests and symptoms may complicate the detection of a distinct syndrome. In all, 20% of the patients in the current series developed this picture of potentially life-threatening liver damage that we have called REILD. It is characterized by jaundice and ascites developing 4 to 8 weeks after treatment, with pathologic changes consistent with VOD in the most severe cases. REILD should be placed in the spectrum of the sinusoidal obstruction syndromes, together with the 2 well-known syndromes of subacute liver damage that appear after external irradiation24 and also share the pathologic hallmark of VOD.25 Radiation-induced liver disease appears after external irradiation of liver tumors and causes rapid weight gain, ascites, and an elevation in alkaline phosphatase that is in proportion with that of other liver enzymes (anicteric ascites). On the contrary, combined modality-induced liver disease appears after allogeneic bone marrow transplantation (when high-dose chemotherapy and total body irradiation are used as preparative procedures) and causes jaundice and weight gain, ascites, and a striking elevation of bilirubin in contrast to mildly elevated transaminases and alkaline phosphatase. REILD is much more similar to the latter, and the finding that REILD was only observed among patients that had received chemotherapy either before or after RE suggests that both pictures may share a common pathophysiologic pathway. It is interesting to note that the hepatic sinusoidal obstruction syndrome can also be caused by chemotherapeutic drugs such as oxaliplatin without concurrent irradiation.26

We cannot compare the incidence of liver disease in the current series with data from the literature because to our knowledge this clinical entity after RE has not been specifically described previously. In a retrospective series and in 2 clinical trials using resin 90Y-microspheres for the treatment of liver metastasis from colorectal cancer, 4% to 9%11, 13, 17 of patients had relevant liver toxicity. However, liver toxicity was reportedly absent in other series.10, 15, 27 Among patients with mostly HCC treated with glass 90Y-microspheres, up to 9% may develop progressive hepatic failure after treatment.3 All in all, the high incidence of REILD in the current series might be because of the finding that our patients with liver metastases had been very heavily pretreated, and in many cases RE was performed immediately after or even during their last course of chemotherapy.

The amount of activity administered relative to the liver volume targeted and age were independent predictors of REILD. However, any conclusion regarding the usefulness of these variables in defining the prognosis of patients after RE should be tempered because of the reduced sample size. We have no explanation for the apparently increased susceptibility of younger patients to REILD. VOD is more frequent among very young infants undergoing bone marrow transplantation but this is a completely different population. Conversely, our results suggest that activities calculated by current methods are suitable for the majority of patients but may be excessive for a subset of individuals treated in a whole-liver fashion. However, only prospective studies can determine whether reducing the activity administered to a maximum relative to the volume of targeted liver may result in a reduced incidence of moderate to severe REILD. When seeking lobar RE, the occurrence of REILD is not relevant, because by reducing the targeted liver volume one can even increase the dose without jeopardizing the outcome in what has been called “radiation segmentectomy.”23, 28

In summary, REILD is a complication of liver RE that can produce significant morbidity, with a potentially life-threatening course in a small proportion of patients. Clinical features and population analysis strongly suggest that REILD results from the combined effect of chemotherapy and radiation of the liver. It may be the result of a particular ‘seed and soil’ combination, so that only when a sufficient amount of radiation is delivered to liver tissue primed by chemotherapy does a hepatic sinusoidal obstruction syndrome appear. However, a merely additive or synergistic effect of radiation and chemotherapy could also explain this finding. All patients treated by liver RE, particularly if a whole-liver approach is indicated, should be assessed systematically for liver damage between 4 and 8 weeks after treatment. Medical treatment can rely on diuretics and general support in mild cases, but the use of steroids and defibrotide29, 30 warrants investigation for patients with proven VOD and preventive measures such as low-dose heparin, ursodeoxicolic acid, and pentoxyfylline should be considered for high-risk patients.18 Although the indication of TIPS in VOD is a matter of debate,31, 32 our results suggest it should be considered for patients with deteriorating liver function.