Hepatic Vein Tumor Thrombus as a Risk Factor for Excessive Pulmonary Deposition of Microspheres during TheraSphere Therapy for Unresectable Hepatocellular Carcinoma Chad J. Fleming, MD, James C. Andrews, MD, Gregory A. Wiseman, MD, Denise N. Gansen, CNMT, and Lewis R. Roberts, MBChB, PhD
PURPOSE: To evaluate the impact of identifiable hepatic vein tumor thrombus on the ability to safely deliver TheraSphere (yttrium 90 – containing glass microspheres) for the treatment of hepatocellular carcinoma (HCC). MATERIALS AND METHODS: A retrospective review was performed of 87 patients (71 men, 16 women; mean age, 64.5 years; age range, 25– 83 y) referred for TheraSphere therapy for HCC during a 2-year period between April 2005 and May 2007. Evaluation included contrast-enhanced computed tomography or magnetic resonance imaging, selective mesenteric angiography, and radionuclide perfusion scintigraphy to measure the arteriovenous shunting through the tumor. RESULTS: Of the 87 patients, 83 underwent angiography and perfusion scintigraphy; 53 were ultimately treated with 65 glass microsphere infusions. Twelve of 83 were identified as having tumor thrombus in a hepatic vein or extending into the inferior vena cava. The mean lung shunt for the patients with hepatic vein tumor thrombus was 30% (range, 11%– 60%), compared with 8.2% (range, 3%–23%) for patients without identifiable tumor thrombus. Two of the 12 patients were treated with reduced doses of glass microspheres, and the remaining 10 were offered alternative therapies. CONCLUSIONS: The presence of hepatic vein tumor thrombus is a risk factor for an increased lung shunt that may prohibit delivery of a therapeutic dose of TheraSphere to hepatic tumor. J Vasc Interv Radiol 2009; 20:1460 –1463 Abbreviations:
HCC ⫽ hepatocellular carcinoma, IVC ⫽ inferior vena cava
HEPATIC radioembolization, the hepatic arterial delivery of yttrium 90 – containing microspheres, is an emerging therapy for hepatic malignancies such as hepatocellular carcinoma (HCC) and colorectal metasta-
From the Department of Radiology, Division of Vascular and Interventional Radiology (C.J.F., J.C.A.); Division of Nuclear Medicine (G.A.W., D.N.G.); and Department of Internal Medicine, Division of Gastroenterology (L.R.R.), Mayo Clinic College of Medicine, 200 First Street SW, Rochester, MN 55905. Received April 15, 2008; final revision received July 13, 2009; accepted July 16, 2009. Address correspondence to J.C.A.; E-mail:
[email protected] None of the authors have identified a conflict of interest. © SIR, 2009 DOI: 10.1016/j.jvir.2009.07.033
1460
ses (1,2). These -emitting microspheres act as internal radiation delivered as micron-sized embolic particles. This therapy requires careful preadministration planning that includes clinical history, performance status, laboratory values, and cross-sectional imaging. Additionally, radiation dose calculation and planning and modification based on tumor and hepatic volume are key elements to patient safety and treatment efficacy (2). To permit safe delivery of the glass microspheres, intraabdominal perfusion must be limited to the liver, and the activity of microspheres delivered to the lungs by way of arteriovenous shunting through the tumor should result in less than 30 Gy of delivered pulmonary dose in a single treatment session (i.e., 50 Gy cumulative dose) to prevent the clinical aspects of radiation-induced lung dam-
age (3). Risk factors that may affect hepatopulmonary shunting include tumor histology, previous embolotherapy or radioembolotherapy, tumor volume, and hepatic arterial epinephrine (4). During our initial experience with TheraSphere (glass microsphere) radioembolization, we identified 12 patients with hepatic vein tumor thrombus and excessively high lung shunt fractions that impacted the ability to safely deliver this therapy. The purpose of this study is to evaluate the impact of identifiable hepatic vein tumor thrombus on the ability to safely deliver glass microspheres for the treatment of HCC.
MATERIALS AND METHODS Treatment of patients with HCC with TheraSphere (MDS Nordion; Ot-
Volume 20
Number 11
tawa, Ontario, Canada) was carried out under an institutional review board–approved protocol. Retrospective data analysis of patients who were considered candidates for this therapy between April 2005 and May 2007 was performed with institutional review board approval. Patient evaluation was carried out as previously described (2,3). Briefly, patients were considered for glass microsphere therapy only if their clinical history, laboratory data, and imaging findings supported the diagnosis of primary HCC. Exclusion criteria included (but were not limited to) poor performance status, inability to comply with the complex treatment regimen, and poor hepatic reserve. The diagnosis of HCC was made based on the American Association for the Study of Liver Diseases and European Association for the Study of the Liver guidelines and biopsies were not routinely performed. Patients were evaluated in a multidisciplinary hepatobiliary neoplasia clinic and were not candidates for surgical resection, ablation, or transplantation. All patients underwent contrast-enhanced abdominal computed tomography (CT) or magnetic resonance (MR) imaging for tumor staging and hepatic volume measurement. None of the evaluated tumors had previously undergone catheterdirected therapy. If, after the initial clinical and noninvasive imaging evaluation, patients were considered candidates for this therapy, they were referred for diagnostic angiography and perfusion scintigraphy. The angiographic evaluation consisted of selective celiac and superior mesenteric arteriograms to define the hepatic arterial anatomy and tumor vascularity and to assess the patency of the portal vein. Selective hepatic arteriograms with embolization of branches that might lead to intraabdominal, extrahepatic deposition of the microspheres (eg, right gastric and gastroduodenal arteries) were then performed when indicated. With the diagnostic catheter in the planned position for glass microsphere infusion, 5 mCi of technetium Tc 99m macroaggregated albumin was infused to define the regional perfusion and estimate the arteriovenous shunting through the liver and the tumor (2). Planar images were acquired at 5 minutes per acquisition within 60 minutes after infusion. The estimated treatment volume (ob-
Fleming et al
Lung Shunt and Delivered
•
1461
90
Y Doses Treated Patients
Measurement
No HV Tumor Thrombus (n ⫽ 51)
HV Tumor Thrombus (n ⫽ 2)
Excluded Patients with HV Tumor Thrombus (n ⫽ 10)
Mean lung shunt fraction (%) Mean delivered tumor dose (Gy) Mean delivered lung dose (Gy)
8.2 (3–23) 113 (86–137) 12.5 (1.9–29.3)
15 (11, 18) 76 (67, 85) 27 (25.4, 28.6)
33.2 (17–60) — —
Note.—HV ⫽ hepatic vein.
tained from the CT or MR scan) was used to calculate the required activity of 90 Y to achieve an absorbed dose of 100 – 120 Gy (3). If this activity, multiplied by the estimated lung shunt, would result in greater than 30 Gy delivered to the lungs, the patient was disqualified from this therapy. Lung shunt was calculated according to the following formula: Lung shunt ⫽ geometric mean of lungs ⁄ (geometric mean of liver ⫹ geometric mean of lungs) ⫻ 100 For this report, the pretreatment CT and MR images were assessed for the presence of identifiable tumor thrombus in the hepatic veins or inferior vena cava (IVC) and correlated with the angiographic findings and the estimated lung shunt. Tumor thrombus was identified as a filling defect in the venous system in continuity with the primary tumor mass, enlargement of the vein, early enhancement of thrombus, and/or neovascularization of the thrombus on CT or MR imaging (5,6). The angiographic finding was the classic “thread-andstreak” appearance of tumor vessels within the thrombus (7). The Wilcoxon rank-sum test was used to compare the groups with and without identifiable hepatic vein tumor thrombus.
RESULTS Between April 2005 and May 2007, 87 patients (71 men, 16 women; age, 25– 83 years; mean age, 64.5 y) were referred for consideration for glass microsphere therapy. Four patients were excluded as a result of progressive liver failure or
patient choice before the performance of any invasive procedures. Eighty-three patients underwent diagnostic angiography and perfusion scintigraphy. Of these, 53 received a total of 65 glass microsphere infusions. Twelve patients received two infusions each. Thirty-four patients were disqualified from this therapy (including the initial four who did not undergo the invasive portion of the evaluation) for a variety of reasons, including progressive liver failure during the evaluation process, vascular anatomy, excessive lung shunt, and patient choice. Twelve of 83 patients (14%) were identified as having tumor thrombus within a hepatic vein or extending into the IVC. The tumor thrombus was identified on preevaluation CT scans in 10 of 12 patients, on preevaluation MR images in one patient, and on diagnostic angiograms in nine of 12 patients. In two patients, the tumor thrombus was seen only on the angiogram and was not suspected based on noninvasive imaging. The tumor thrombus was limited to the hepatic vein in five of 12 patients and extended into the IVC in the other seven. The mean lung shunt for this group was an estimated 30% (range, 11%– 60%; median, 23%). For the treated patients without identifiable hepatic vein tumor thrombus (n ⫽ 51), the mean lung shunt measured 8.2% (range, 3%–23%; median, 5.8%; Table). This achieved significance (P ⬍ .0001) with a Wilcoxon rank-sum test comparing the two groups (Fig 1). Although the tumors in the patients with hepatic vein tumor thrombus were somewhat larger than those in patients without hepatic vein tumor thrombus (mean diameter, 9.5 cm [range, 2.4 –14 cm] vs 6.2 cm [range,
1462
•
Hepatic Vein Tumor Thrombus in Microsphere Radioembolization
Figure 2.
November 2009
JVIR
Plot of lung shunt fraction versus tumor size.
Figure 1. Scatter-plot demonstrates lung shunt percentage for patients with and without identifiable hepatic vein tumor thrombus.
1.5–13.3 cm]), no correlation between tumor diameter and lung shunt percentage was identified (Fig 2). The mean calculated delivered radiation dose for the treated patients without hepatic vein tumor thrombus was 113 Gy to the treated volume, and the estimated delivered lung dose was 12.5 Gy. Therapeutic target dose was 100 Gy. Two of the 12 patients with hepatic vein tumor thrombus were treated at reduced doses of glass microspheres, with estimated delivered hepatic doses of 67 and 85 Gy, respectively. Although this was less than the desired target dose, these two patients had no other good therapeutic options and desired this attempt at therapy. The estimated pulmonary doses for these two patients were 28.6 and 25.4 Gy, respectively (maximum allowed, 30 Gy). These data are summarized in the Table. The remaining 10 patients with hepatic vein tumor thrombus were referred for alternative therapies as the lung shunt precluded the delivery of a therapeutic dose of glass microspheres to the tumor. None of the patients in the study exhibited radiation pneumonitis clinically or based on imaging.
DISCUSSION Liver tumors have a greater arteriolar density than the surrounding normal liver at the microvascular level (8). The hepatic arterial infusion of 90Y-containing microspheres distribute proportional to blood flow and capitalizes on this difference, allowing the delivery of
high doses of radiation to hepatic tumors, with tolerable doses to the normal liver. A less favorable characteristic of many tumors is the greater degree of arteriovenous shunting than seen in the neighboring normal tissue. In general, this is most marked in vascular tumors, such as those most ideal for radioactive microsphere therapy. One study (9) reported a direct correlation between angiographic tumor vascularity and lung shunt fraction. These same authors (9) also noted the lung shunt decreased from 28.5% to 1% after tumor resection in one patient, confirming that the shunting occurs largely within the tumor. In a study of hepatic tumor blood flow manipulation with intraarterial epinephrine, as the tumor-to-liver perfusion ratio increased (ie, greater perfusion to tumor, less to normal liver), so did the lung shunt, again suggesting the shunts occur largely through the tumor (4). In this setting, the arteriovenous shunting results in pulmonary deposition of the therapeutic microspheres, potentially leading to radiation-induced pulmonary injury. The tolerable lung dose for radiation delivered by 90Y-containing microspheres has not yet been clearly determined in human subjects. For conventional external-beam radiation therapy, the likelihood of radiation-induced lung injury is related to the total dose, the fractionation schedule, and the amount of lung in the treatment field (10). The TheraSphere package insert (3) recommends maintaining the pulmonary dose at less than 30 Gy. It has been suggested in a recent review (2) to limit the
pulmonary dose to 30 Gy for a single course of therapy, or cumulatively 50 Gy for multiple infusions (2,11). In one report (12), radiation pneumonitis developed in five of 80 patients treated with 90Y-containing resin microspheres. These five patients received an estimated mean pulmonary dose of 25 Gy (range, 10.4 –36.4 Gy). The median lung shunt for these five patients was 23.7% (range, 13.1%– 45.6%). It is unclear as to the role that underlying pulmonary disease may have played in the development of radiation pneumonitis in these patients, although all were nonsmokers. The median lung shunt fraction for the 75 patients without radiation pneumonitis was 6% (range, 1%–15%). Based on these data, the authors recommend excluding patients with a lung shunt greater than 13%. This, unfortunately, is an overly simplistic approach. The lung shunt must be viewed in the context of the volume of liver to be treated. A large lung shunt may be tolerated if the volume of liver to be treated is relatively small, limiting the activity of 90Y needed to achieve the desired tumor dose. In the same way, if a large tissue volume must be treated, only a small lung shunt is tolerable. In another report (13), 58 of 403 patients treated with 90Y received greater than 30 Gy cumulative lung dose. None of those cases developed clinical or imaging radiation pneumonitis. The authors concluded that dose-escalation studies are needed to better define radiation tolerance of lung parenchyma with this mode of therapy (13). In our treatment group, none of the patients received a lung dose greater than 30 Gy
Volume 20
Number 11
at a single treatment session and none manifested treatment-related pulmonary radiation injury. We agree that continued investigation of an acceptable lung dose is needed to permit safe, therapeutic doses of 90Y delivery without creating pulmonary radiation toxicity. If the pulmonary dose limitations are increased as safety is demonstrated with a higher treatment dose, this will allow an increase in potential patients who would benefit from treatment. HCC is one of several tumor types that are well recognized to exhibit intravascular growth, most commonly into the portal system. Other tumors that exhibit intravascular growth include renal cell carcinoma, islet cell tumors of the pancreas, and adrenal and thyroid carcinoma. When the tumor grows into the portal system, patients may develop portal hypertension as a result of a combination of portal obstruction and increased flow from arterial-portal shunting (5). Although not as well described, it would follow that, when the tumor grows into the hepatic veins, there could be excessive hepatic arterial-to-IVC (and, ultimately, pulmonary artery) shunting as exhibited in 12 patients in the present study. Based on these preliminary data, it would appear that the presence of hepatic vein tumor thrombus in patients with HCC is an independent risk factor for increased lung shunts. Imaging findings suggestive of tumor thrombus include a filling defect in the venous system in continuity with the primary tumor mass, enlargement of the vein, early enhancement of the thrombus, neovascularization of the thrombus, and arterialization of the Doppler signal within the patent portion of the vein (5,6). Angiographically, tumor thrombus is diagnosed by the demonstration of tumor vessels within the thrombus. This has been termed the thread-andstreak sign (7). A recent report (14) identified tumor thrombus in the IVC in 184 of 4,785 patients (3.8%) undergoing chemoembolization for HCC. It is not clear why a greater fraction of our patients had tumor thrombus in the hepatic vein or IVC. It may be that, because TheraSphere administration represented a new therapy during the study period, patients were not referred until later in their disease course. Techniques have been described to allow treatment of patients with in-
Fleming et al creased pulmonary shunt fraction. One small series (15) described temporary balloon occlusion of the hepatic veins during 90Y infusion, with encouraging results. This would not be a viable option in our treatment group as the presence of tumor thrombus in the hepatic veins precludes placement of a balloon occlusion catheter. Other reports (11,16) have described modification and individualization of dosimetry in those with increased lung shunts by incorporating the concept of hypervascularity (ie, proportional tumor to normal tissue) to reduce the activity of the administered 90 Y, thereby achieving a tumoricidal dose to the tumor and a safe pulmonary dose. We concur with the authors’ conclusions that (i) an absolute cutoff in lung shunting fraction should not be used to decide whether a patient can undergo 90Y treatment and (ii) reducing lung shunting to allow delivery warrants continued investigation. A limitation of the present study is the lack of biopsy proof for confirmation of HCC or tumor thrombus. We used the classic features of tumor thrombus in cross-sectional and angiographic evaluation as described. The presence of hepatic vein bland thrombus in the setting of HCC is less common, and we would not expect this to confound our results. In conclusion, these preliminary data indicate that patients with hepatic vein tumor thrombus are at higher risk for excessive pulmonary shunting. We are using this information to counsel patients in whom we identify hepatic vein tumor thrombus on cross-sectional imaging that they may be at higher risk of exclusion from 90Y treatment based on a high lung shunt fraction. In this way, patients may be able to make a better-informed decision about proceeding with the invasive phase of the evaluation. References 1. Andrews J, Walker S, Ackermann R. Hepatic radioembolization with yttrium90 containing glass microspheres: preliminary results and clinical follow-up. J Nucl Med 1994; 35:1637–1644. 2. Salem R, Thurston K. Radioembolization with yttrium-90 microspheres: a stateof-the-art brachytherapy treatment for primary and secondary liver malignancies. Part 1: technical and methodologic considerations. J Vasc Interv Radiol 2006; 17: 1251–1278.
•
1463
3. TheraSphere yttrium-90 microspheres [package insert]. Kanata, ON, Canada: MDS Nordion. 4. Andrews J, Walker S, Juni J, Ensminger W. Modulation of liver tumor blood flow with hepatic arterial epinephrine: a SPECT study. Radiology 1989; 173: 645– 647. 5. Pozniak M, Baus K. Hepatofugal arterial signal in the main portal vein: an indicator of intravascular tumor spread. Radiology 1991; 180:663– 666. 6. Tublin M, Dodd G, Baron R. Benign and malignant portal vein thrombosis: differentiation by CT characteristics. AJR Am J Roentgenol 1997; 168:719 –723. 7. Okuda K, Jinnouchi S, Nagasaki Y. Angiographic demonstration of growth of hepatocellular carcinoma in the hepatic vein and inferior vena cava. Radiology 1977; 124:33–36. 8. Gyves J, Ziessman H, Ensminger W. Definition of hepatic tumor microcirculation by single photon emission computerized tomography (SPECT). J Nucl Med 1984; 25:972–977. 9. Leung W, Lau W, Ho S. Measuring lung shunting in hepatocellular carcinoma with intrahepatic-arterial technetium-99m macroaggregated albumin. J Nucl Med 1994; 35:70 –73. 10. Movasas B, Raffin T, Epstein A, Link C. Pulmonary radiation injury. Chest 1997; 111:1061–1076. 11. Ho S, Lau W, Leung T. Partition model for estimating radiation doses from yttrium-90 microspheres in treating hepatic tumours. Eur J Nucl Med 1996; 23:947–952. 12. Leung W, Lau W, Ho S. Radiation pneumonitis after selective internal radiation treatment with intra-arterial yttrium90 microspheres for inoperable hepatic tumors. Int J Radiat Oncol Biol Phys 1995; 33:919–924. 13. Salem R, Gates V, Lewandowski R. Radiation pneumonitis following yttrium-90 microspheres using current dosimetry models: fact or fiction? J Vasc Interv Radiol 2008; 19(suppl):S132. 14. Lee I, Chung J, Kim H. Extrahepatic collateral arterial supply to the tumor thrombi of hepatocellular carcinoma involving inferior vena cava: the prevalence and determinant factors. J Vasc Interv Radiol 2009; 20:22–29. 15. Bester L, Salem R. Reduction of arteriohepatovenous shunting temporary balloon occlusion in patients undergoing radioembolization. J Vasc Interv Radiol 2007; 18:1310 –1314. 16. Salem R, Thurston K. Radioembolization with yttrium-90 microspheres: a state-of-the-art brachytherapy treatment for primary and secondary liver malignancies. Part 2: special topics. J Vasc Interv Radiol 2006; 16:1425– 1439.