Abstract
Aim: To retrospectively evaluate infectious hepatic complications of transarterial radioembolization (RE) of the liver in patients with pre-existing biliodigestive anastomosis. Patients and Methods: Patients who underwent RE were retrospectively analyzed. All patients had at least a contrast-enhanced magnetic resonance imaging or contrast-enhanced triple-phase computed tomographic scan before and 6-8 weeks after RE. Results: Overall, 143 patients (67 women, 76 men; median age=65±11.2 years) were analyzed. Nine patients had a biliodigestive anastomosis. The complications were as follows: one case of cholecystitis, three of gastroduodenal ulcer with one ulcer perforation, and six of radioembolization-induced liver disease. There were no infectious complications. There were no major or minor complications in the group with previous biliodigestive anastomosis. Conclusion: Pre-existing bilioenteric anastomoses are not a negative predictive factor for the development of infectious hepatic complications after RE. RE with 90Y microspheres can be safely performed following careful patient selection.
- Liver metastasis
- radioembolization
- biliodigestive anastomosis
- bilioenteric anastomosis
- hepaticojejunostomy
- complications
- liver abscess
The liver is one of the main organs to develop hematogenous metastases from various types of cancers, either via the arterial or portal circulation. Furthermore, a number of primary liver and biliary cancers exist. Surgery for liver malignancies represents the optimal treatment option, but a large number of patients with either primary or secondary liver cancer are not suitable for surgical treatment, either because of the extent of their disease or because of poor medical condition (1, 2). Chemotherapy is another therapeutic option, but some tumor entities lack an effective chemotherapeutic treatment or systemic toxicity limits this treatment option. Because of radiation-induced liver disease, external radiation therapy for the liver is rarely practiced (3). Interventional radiology provides promising new treatment options: transarterial locoregional therapies, such as transarterial (chemo-)embolization (TAE/TACE) and radioembolization (RE). They exploit the fact that liver malignancies receive their blood supply primarily from the liver artery while normal liver tissue receives its blood supply by the portal vein (4-6). Although prospective randomized trials are not yet available, there are some indicators for longer survival in patients undergoing RE compared to historic patient populations undergoing other minimally invasive procedures (7-11).
It is well-known that TACE can lead to hepatic and biliary damage, which mainly consists of hepatic insufficiency, liver abscess and intrahepatic bilioma formation (12-15). Biliodigestive anastomosis has been identified as a major determinant of liver abscess formation after TAE and TACE and antibiotic prophylaxis does not seem to prevent this complication effectively (16).
Since TACE and RE share some characteristics – both deliver an anti-tumoral agent via the hepatic artery and compromise arterial circulation, although to a different extent – the purpose of this study was to evaluate RE for infectious hepatic complications in patients with pre-existing biliodigestive anastomoses.
Patients and Methods
All consecutive patients treated by 90Y RE at our Department between January 2009 and October 2012 were included in this retrospective analysis. The indication for RE was discussed in an interdisciplinary tumor conference for all patients. RE was considered for patients with liver tumors that were not treatable by surgical resection or percutaneous ablation and who had already received standard therapy, if available. In patients with extrahepatic disease, the extrahepatic manifestations were not deemed to determine the survival by the referring physician or tumor conference. Histologically confirmed diagnosis of malignancy was another precondition for RE. Other criteria for performing the procedure were good functional status (Eastern Cooperative Oncology Group (ECOG) score of 0-2), liver function status of Child-Pugh A or B, bilirubin <2 mg/dl, sufficient renal function (serum creatinine <2 mg/dl), platelet count >50,000/nl, International normalized ratio <1.5, and partial thromboplastin time <50 s. If indicated, hemostatic function was improved (e.g. platelet transfusion). All patients gave written and oral informed consent.
Pre-treatment evaluation. All patients had an extensive pre-treatment evaluation consisting of the following: a discussion of technique, risks and chances of success with the patient; obtainment of the complete medical history; a chest and abdominal triple-phase computed tomographic (CT) scan no older than four weeks in order to evaluate celiac and mesenteric arteries as well as intra- and extrahepatic tumor manifestations; and blood tests including complete blood count, liver function parameters, creatinine, albumin and coagulation parameters. A pretreatment angiography with selective visceral catheterization was then conducted to study tumor and liver vascularization as well as portal vein status. In most cases, the vascular flow was changed during pretreatment angiography by occluding vessels with an origin from the hepatic arteries supplying extrahepatic organs (e.g. gastroduodenal artery) or aberrant hepatic vessels originating from celiac or mesenteric arteries. After the angiography, single-photon emission computed tomography (SPECT) with 150 to 200 MBq 99mtechnetium-albumin macro-aggregates (MAA) was performed to measure pulmonary shunt fraction, detect potential misplacement of MAA in the gastrointestinal tract and to evaluate distribution of MAA in the gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) was performed to determine hepatic tumor load and to serve as a baseline to measure treatment efficacy, if no contraindications were present.
Treatment. Patients were treated with SIR-Spheres (Sirtex SIR-Spheres Pty Ltd., Sydney, Australia). Depending on tumor distribution, patients received either bilobar or unilobar treatment. The activity of 90Y was calculated according to the body surface area (BSA) method (activity depends on BSA as well as liver and tumor volumes) as described elsewhere (17). Before applying the calculated activity of 90Y-microspheres, an angiography to confirm the state of the pre-treatment evaluation was performed. Treatment was conducted on an inpatient basis as patients remained in the hospital for a total of three nights. Before RE, patients received 15 mg piritramide, 250 mg prednisolone and 8 mg ondansetron intravenously. Supportive therapy for six weeks after RE consisted of proton pump inhibitors as well as systemic corticoid for patients without diabetes.
Follow-up. All patients had an outpatient follow-up organized by our clinic for minimally-invasive tumor therapy, which is a section of our Department. Gadoxetic acid-enhanced MRI (or triple phase CT if contraindications were present) was performed two times within 6-8 weeks and then every three months after treatment to evaluate treatment efficacy. All images were evaluated by a minimum of two clinical radiologists (4 and 16 years of experience) in consensus. Laboratory and clinical toxicity were classified by using the National Cancer Institute common toxicity criteria (version 3.0) (grade 0=normal, grade 1=more than the upper limit of normal (ULN) to 1.5-times ULN, grade 2=more than 1.5- to three-times ULN, grade 3=more than three- to 10-times ULN, grade 4=more than 10-times ULN). Procedural complications were classified by using the standards of the Society of Interventional Radiology (18). Gastrointestinal, biliary, infectious and pulmonary complications as well as hepatic dysfunction, and radio-embolization-induced liver disease (REILD) were assessed on clinical follow-up and follow-up imaging. REILD was defined according to Gil-Alzugaray et al. as jaundice and ascites appearing within two months after RE in the absence of tumor progression or bile duct occlusion (19).
Statistical analysis. Statistical analysis was performed using PASW Statistics 21 (IBM, Armonk, NY USA). To investigate risk factors for developing a complication or differences between two groups (e.g. biliodigestive anastomosis vs. normal bile drainage), a univariate analysis was performed for all of the variables using a Student t-test, Mann–Whitney U-test, chi-square or Fisher exact test according to the type of the variable. For selected variables, a multivariate analysis using multiple logistic regression was conducted. Power analysis was conducted by using GNU R, version 2.15.3 (The R Foundation for Statistical Computing, Vienna, Austria). Univariate survival analysis was performed using the Kaplan–Meier method. All tests were two-sided. A p-value of less than 0.05 was considered statistically significant. Quantitative data are expressed as the mean±standard error, unless otherwise indicated.
Results
Patients' characteristics. A total of 168 patients were evaluated for RE, eight patients were not eligible for RE treatment. Reasons were excessive lung shunting (n=4), poor medical condition at the time of treatment (n=2), extrahepatic manifestations (n=1) and misplacement of MAA in the gastrointestinal tract which was not correctable (n=1).
One hundred and sixty patients received a total of 178 administrations of 90Y microspheres. Seventeen patients were lost to first follow-up, consequently 143 patients with 156 administrations were analyzed. The median age of patients was 65±11.2 years; there were 67 women and 76 men.
Nine patients (6%) had a biliodigestive anastomosis with a hepaticojejunostomy (seven by pylorus-preserving pancreaticoduodenectomy, one by standard Whipple procedure, and one by extended right hemihepatectomy). For further characteristics see Tables I and II.
Treatment parameters. Mean pulmonary shunt, as measured by pretreatment scintigraphy, was 9.63±4.61%. Mean total serum bilirubin the day before RE was 0.61±0.43 (range=0.2-3.3) mg/dl. One patient was treated with a bilirubin levels greater than 2.0 mg/dl.
A total of 130 patients (91%) received a single treatment, 13 patients (9%) received two treatments (separate treatment for left and right side with an interval of about four weeks). Pretreatment vessel occlusion was necessary in 118 (83%) patients. Occlusion included gastroduodenal artery (n=114) (80%), right gastric artery (n=70) (49%), right phrenic artery (n=6) (4%), supraduodenal artery (n=6) (4%), cystic artery (n=2) (1%) and greater pancreatic artery (n=1) (1%).
90Y microspheres were administered into the proper hepatic artery (n=38) (27%), left hepatic artery (n=23) (16%) or right hepatic artery (n=45) (31%) or separated into the left and right hepatic artery (n=37) (26%).
The mean administered dose of 90Y was 1.49±0.44 (range=0.38-2.47) GBq, although only 19 patients (13%) received a dose of 2 GBq or more. Dose reduction because of pulmonary shunting was necessary in four patients.
Follow-up and adverse events. Seventeen patients (11%) were lost to follow-up. At least one follow-up MRI or CT within 4-6 weeks after RE was available for 143 patients. The median follow-up time was 123±201 days, and the mean was 90 days. Thirty-nine patients died within the follow-up period after a median of 169±183 days. Median survival was 294 days (95% confidence interval (CI)=142-446 days), mean survival was 453 days (95% CI=368-539 days). 30-Day survival was 97.9%, 60-day was 75.5%, 90-day was 65.7%, 180-day was 35% and 360-day survival was 15.4% (Figure 1).
Incidence of adverse events after RE was very low. There was one (0.7%) arteriovenous fistula after inguinal arterial puncture detected by an ultrasound examination the day after RE which resolved spontaneously (grade I). There were four complications related to misplacement of 90Y-microspheres: One (0.7%) case of gastric ulcer and one (0.7%) case of duodenal ulcer which were treated conservatively and classified as grade II; one (0.7%) case of cholecystitis which had to be treated with cholecystectomy 38 days after RE was classified as grade III; and one (0.7%) case of gastric ulcer perforation which was treated by surgical oversewing 55 days after RE (grade IV). Both patients who required surgery following RE died later because of tumor progression. All post-procedural complications were late complications appearing after more than 30 days. There were no infectious complications, especially no cases of liver abscess formation. No deaths within the follow-up period were attributable to the RE procedure.
According to the stated criteria, the incidence of REILD was 4.2% (six cases). Patients who developed REILD were significantly younger than the patients who did not (52.5±13.2 vs. 63±10.9 years, p<0.05). Serum bilirubin and administered activity did not differ between groups. Three patients with REILD had a whole-liver treatment and three did not. All six patients with REILD had had no previous liver surgery and no cirrhosis; five patients were female and one male, although this was not statistically significant (p=0.067). The tumor entity of patients with REILD was pancreatic neuroendocrine carcinoma in one case, cholangiocarcinoma in two and breast cancer in three. Tumor entity was not associated with the development of REILD. The multivariate analysis revealed no independent risk factors for the development of REILD (Table III).
Although laboratory follow-up was only available for 122 patients, we noted three (2.1%) increases in bilirubin toxicity, two from grade I to IV and one from grade 0 to IV.
In the patient cohort with a history of biliodigestive anastomosis (n=9), no adverse events were registered, especially no infectious complications.
Discussion
In our study, in 143 patients, we instigated the incidence of complications after RE with a special focus on patients with a biliodigestive anastomosis. Overall, we had a very low rate of complications, making RE a safe procedure: grade I complications had an incidence of 0.7%, grade II of 1.4%, grade III of 0.7% and grade IV of 0.7% respectively. This is in accordance with other published studies (20, 21).
In 2008 Sangro et al. first described a clinical syndrome occurring after RE treatment, which consists of jaundice and ascites appearing one to two months after treatment in the absence of tumor progression or bile duct occlusion and was called REILD (22). The incidence of REILD was 4.2% in our patients, which is similar to the results of Gil-Alzugaray et al. with their modified protocol (19). Interestingly, the patients who developed REILD did not have pre-existing cirrhosis, previous liver surgery or elevated bilirubin. Furthermore, whole-liver treatment was not associated with a higher risk of developing REILD. Reducing the prescribed activity is also not the key to eliminating the development of REILD, as there was no dose–response relationship, comparable to the study of Gil-Alzugaray et al. (19). Even a very conservative patient selection and the use of an optimized treatment protocol seems not to complete eliminate completely the risk of developing REILD.
Biliary complications may result either from direct radiation-induced injury (e.g. radiation cholecystitis) or from the microembolic effect of the delivery agents. It has been shown that the predominant cause of biliary injury and infectious complications after TACE is the microembolic effect (23, 24). The embolic effect of RE agents is minimal, which may be a possible explanation for the lower incidence of biliary complications (25). In our study, we only observed one case of radiation-induced cholecystitis, which is on a level with that of a study of Atassi et al. (26) who studied the presence of biliary sequelae after RE with 90Y in 327 patients with primary or secondary liver tumors. They found biliary sequelae diagnosed by imaging in 34 patients: two had cholecystitis and one had a liver abscess. Only six patients needed a surgical or interventional procedure. Radiation cholecystitis is effectively prevented by injecting microspheres distal to the origin of the cystic artery or – if not possible – temporary or permanent embolization of the cystic artery (27). Radiation-induced cholecystitis is specific to RE and is not related to the embolic effect of the therapeutic agent.
Several studies have investigated the incidence of biliary complications and their associated risk factors after TACE (14, 28). Biliodigestive anastomosis represents a major risk factor for the development of a liver abscess after TACE. In a study of Kim et al. (16), seven out of 157 patients developed a liver abscess approximately 2-4 weeks after TACE; six of them had a biliodigestive anastomosis, which translates into an odds ratio of 894 for the development of an abscess. Pancreaticoduodenectomy with biliodigestive anastomosis alone has also been reported to be a risk factor for liver abscess formation (29).
Previous biliodigestive anastomosis leads to a retrograde intestinal bacterial colonization of the biliary tree in more than 90% of cases (30). The biliary tree is supplied primarily by a capillary plexus, which is fed by hepatic arteries targeted during RE. TACE has been shown to cause bile duct necrosis in up to 13% of procedures (31). Biliary duct injuries during TACE and biliary tree contamination together provide an explanation for the increased risk of hepatic abscess in patients with a biliodigestive anastomosis. As biliary necrosis also occurs after RE, although only shown on imaging (26), RE should theoretically also pose a risk for liver abscess formation with a pre-existing biliodigestive anastomosis. This was, however, not shown in our study. One possible explanation could be that in RE, total embolization of the tumor feeding vessel is not the objective – in contrast to TACE – tumor treatment is not based on ischemia but on β-rays emitted by 90Yttrium. Therefore in contrast to TACE, usually, no partial liver necrosis occurs after RE. Bile ducts are dependent on an arterial blood supply (e.g. ischemic type biliary lesions after liver transplantation in patients with compromised hepatic arterial blood flow) (32, 33). Consequently, bile structures are compromised after TACE, resulting in biliary abscesses because of bile duct colonization after biliodigestive anastomosis. A similar increase of biliary abscesses in patients with biliodigestive anastomosis has been reported in thermal ablation procedures such an radiofrequency ablation (34). In this procedure, necrosis of the liver tumor and the surrounding liver tissue is induced by thermal energy. This thermal energy also disrupts bile ducts, leading to an increased rate of biliary abscesses in patients with a history of biliodigestive anastomosis.
The present study has some important limitations. Firstly, the number of patients in our collective with a pre-existing biliodigestive anastomosis is relatively low (6.3%), which can be explained by the overall low number of patients being suitable for RE after a Whipple procedure. Secondly, laboratory follow-ups were not available for all patients as patients had only imaging follow-up in our Department.
In the patient cohort with a history of biliodigestive anastomosis (n=9), no adverse events were registered, therefore a power analysis is not useful for our study. According to Kim et al., who investigated 157 patients after TACE, the incidence of developing a liver abscess in the group which had a biliodigestive anastomosis was 86% (16). Woo et al. reported an incidence of 48% (12 out of 25 patients with biliodigestive anastomosis) (35). This makes a pooled hypothetical incidence of 56% after TACE.
To conclude, RE with 90Y microspheres is an overall safe procedure and does not seem to have a higher complication rate in patients with a history of biliodigestive anastomosis, although study of a larger number of patients with biliodigestive anastomosis would be desirable.
- Received April 8, 2014.
- Revision received May 30, 2014.
- Accepted June 2, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved