Fax: (412) 624-6666
Evaluation of responses to chemoembolization in patients with unresectable hepatocellular carcinoma
Version of Record online: 3 FEB 2003
Copyright © 2003 American Cancer Society
Volume 97, Issue 4, pages 1042–1050, 15 February 2003
How to Cite
Ebied, O. M., Federle, M. P., Carr, B. I., Pealer, K. M., Li, W., Amesur, N. and Zajko, A. (2003), Evaluation of responses to chemoembolization in patients with unresectable hepatocellular carcinoma. Cancer, 97: 1042–1050. doi: 10.1002/cncr.11111
- Issue online: 3 FEB 2003
- Version of Record online: 3 FEB 2003
- Manuscript Accepted: 19 SEP 2002
- Manuscript Revised: 17 JUL 2002
- Manuscript Received: 19 APR 2002
- efficacy studies;
- computed tomography;
The authors used computed tomography (CT) scans to correlate the changes in tumor vascularity, necrosis, and size with response and survival after transcatheter arterial chemoembolization (TACE) in patients with advanced, unresectable, hepatocellular carcinoma (HCC).
The authors studied 72 patients with biopsy-proven, unresectable HCC and focused on 186 individual tumor masses. A baseline, multiphase, helical CT was performed and at least three follow-up CT scans were performed after treatment by TACE. Tumors were classified as hypervascular or hypovascular and patients were classified as responders or nonresponders based on CT evidence of altered tumor size, tumor necrosis, and the appearance of new tumors. A new scoring system was used to monitor patient response to TACE.
Thirty-eight patients were responders and 34 were nonresponders. Patient survival was significantly increased (P = 0.009) in patients who were hypervascular responders. Survival also was increased in hypervascular nonresponders compared with hypovascular nonresponders (P = 0.008) and in hypovascular responders compared with hypovascular nonresponders (P = 0.002). Response to chemoembolization was found to be significantly (P = 0.02) and inversely proportional to tumor size, but the number of tumor foci in an individual patient was not predictive.
TACE appears to result in improved survival among HCC patients with hypervascular tumors who responded to therapy. However, even patients classified by CT as hypervascular nonresponders and hypovascular responders have improved survival. Cancer 2003;97:1042–50. © 2003 American Cancer Society.
Malignant hepatic tumors are a common cause of cancer mortality and few patients with primary malignancies are candidates for curative resection.1–4 Without treatment, nonresectable patients with hepatocellular carcinoma (HCC) have a median survival period of less than 6 months. Intravenous chemotherapy and radiation therapy are usually ineffective in these patients.2, 5, 6 Selective transcatheter arterial chemoembolization (TACE) has been used to treat HCC patients for almost 20 years and has shown some promise.7, 8 Chemoembolization delivers high concentrations of chemotherapeutic agents to the tumor. Delivery of temporary arterial occlusive material before and after the anticancer agent produces a high concentration of the medication within the tumor for a prolonged time with relatively minor systemic toxicity.2, 3, 5–7, 9 Transcatheter arterial chemoembolization is usually given as a series of multiple embolization procedures performed over a period of months.10, 11
Relatively little is known about criteria that might predict which patients with HCC would most benefit from TACE. A recent study by Katyal et al.11 reported on patients with HCC who received TACE. They found that patients with hypervascular HCC (as defined by an enhancement pattern on an arterial-phase computed tomographic [CT] scan) were more likely to respond to TACE with a reduction in tumor size and longer survival than patients with less vascular tumors. We studied a different group of patients using the same diagnostic and therapeutic protocols to verify and extend these findings. We propose a new scoring system to assess HCC patient response to treatment.
We treated a large sample of HCC patients with TACE. It has been our experience that the traditional oncologic measure of tumor response to therapy, namely, a reduction in the cross-sectional area or dimension of certain index lesions, did not always appear to correlate with other outcome measures, including the degree of tumor necrosis and survival. In addition, we tried to correlate other tumor characteristics found on baseline CT scans (i.e., the size, number, vascularity, and necrosis of the tumor) with objective measures of treatment response and survival. We performed baseline and serial helical multiphase CT scans on 72 patients with HCC who were treated with TACE. Tumor response was quantified using measures of tumor size, tumor necrosis, tumor vascularity, and patient survival.
MATERIALS AND METHODS
KMP, OME, and BIC reviewed the electronic medical records (radiology and pathology reports and discharge summaries) of 102 patients with HCC who had been treated at our institution from October 1997 to July 2000. These patients were treated with TACE using cisplatin and Gelfoam (Ethican CJ&J, Somerville, NJ). Cisplatin was selected because it is an active drug in HCC patients with relatively marrow sparing. This is helpful in patients with splenomegaly. Nine patients were lost to follow-up and did not complete three TACE treatments. We excluded four patients who had been treated with iodized oil (Lipiodol; Savage Laboratories, Melville, NY) because this precluded evaluation of tumor necrosis. In addition, iodized oil did not provide any added clinical benefits.24 We included patients who had histologic proof of HCC, which was achieved in all cases by percutaneous ultrasound or by CT-guided 18-gauge core needle biopsy. All patients had either Child Class A cirrhosis or no cirrhosis to be eligible for TACE therapy. We also included only patients who had helical multiphase CT scans at baseline (before treatment) and at three follow-up visits. Seventeen patients were excluded because they did not undergo helical CT before therapy. Each follow-up CT scan was obtained on the same day and just before the next TACE session. Follow-up CT scans were performed before each TACE session, either on the day before or occasionally on the same day. Patients had an average of seven treatment cycles.
Seventy-two patients were examined in this retrospective analysis. They had 186 measurable liver tumors at baseline. There were 52 male and 20 female patients, with an average age of 60 years (range, 19–82 years).
The tumors varied widely in size from less than 1 cm to 23 cm in maximum dimension: 37 lesions were smaller than 3 cm in diameter, 53 lesions were 3–5 cm in greatest dimension, 54 lesions were larger than 5–10 cm in greatest dimension, and 42 lesions were larger than 10 cm in greatest dimension. Of the 186 lesions, 128 were in the right hepatic lobe alone, 23 were only in the left lobe, and 15 were large tumors that involved both the left and right lobes.
All patient records were reviewed by the liver tumor board and all patients were judged to have unresectable HCC either because of tumor location and stage or because of poor hepatic reserve. All patients subsequently were treated by the same oncologist (BIC).
This retrospective analysis was approved by the institutional review board. Patient consent was obtained for performance of the TACE but was not required for the retrospective review of records.
All CT studies were performed with helical scanners (Hispeed Advantage, GE Medical Systems, Milwaukee, WI). Patients had oral contrast and both unenhanced and intravenous contrast-enhanced imaging through the abdomen. They received 125–150 mL of 60% iodinated contrast medium (iothalamate meglumine [Conray 60] or ioversol [Optiray 350]; Mallinckrodt Medical, St. Louis, MO) administered intraveously at a rate of 3–5 mL/sec with a power injector (Model OP 100; Medrad, Pittsburgh, PA). Scan delay lasted 30–35 seconds for the hepatic arterial phase and 60–70 seconds for portal venous phase imaging. Collimation was 5 mm and pitch was 1.5, which were sufficient to obtain each CT phase with suspended breathing.
Patients abstained from solid food but continued fluid intake the night before the procedure. On the morning of the TACE, patients received intravenous antiemetics, antibiotics, sedatives, and fluids. Informed consent for the procedure was obtained in all cases. A femoral artery (usually the right) was catheterized using the Seldinger technique and a 5-F angiography catheter was introduced into the hepatic arteries to map the vessels and to assess portal venous flow. Angiographic findings and CT scans determine the hepatic lobe to be treated and the vessel to be embolized. Usually, only one hepatic lobe was treated with chemoembolization in each TACE procedure, although the chemotherapeutic agent was administered to each lobe in patients with bilobar tumors, but at different treatment sessions. A slurry of surgical gelatin particles (Gelfoam) and contrast medium was injected under fluoroscopic control to decrease the inflow and washout of the tumor. The chemotherapeutic agent in 150 mL of saline was then infused at a rate to allow the entire volume to be administered in 30–45 minutes. All patients received cisplatin at starting doses of 125 mg/m2. If no change was found in the weekly complete blood counts or platelets at follow-up, the next dose was increased to 150 mg/m2. The final dose was increased again to 175 mg/m2. Any decrease in the level of leukocytes or platelets or an increase in the bilirubin level resulted in subsequent cisplatin doses at one dose down. Additional embolization with gelatin particles was performed to further decrease or temporarily occlude arterial flow to the tumor. Contrast was used to localize the catheter and Gelfoam was injected in three to four aliquots, typically in 6 mL saline plus 6 mL contrast. A typical total amount of Optiray (< 30 mL) contrast was used. By contrast, the initial baseline angiogram used about 100 mL of contrast. Gelfoam aliquots typically were given before, during, and at the end of the chemotherapy infusion. The aim was to achieve vascular slowing, but not complete occlusion.
The TACE treatments were repeated every 6–12 weeks until the tumor resolved completely, became totally necrotic as judged by CT criteria, or was judged to be ineffective as shown by tumor growth or the appearance of new tumors. Arterial embolization was used to diminish, but never to completely occlude, arterial flow to the tumor. Patients were followed for a minimum of 24 months from the time of diagnosis or until death, whichever occurred first. The average hospital stay was 23 hours. Complications included transient abdominal pain (25 patients), inguinal bruising (3 patients), peptic ulcer (2 patients), and transient hypotension (4 patients).
Patient treatment decisions were based on clinical and CT assessment at the time of therapy. For this study, two of the authors (O.M.E. and M.P.F.) retrospectively reviewed the CT scans on a workstation. We evaluated the baseline (before treatment) and all follow-up CT scans for the following features: number of tumor lesions, change in size, distribution, appearance of new lesions, change in necrosis, and the degree of vascularity of each lesion as determined on the arterial and portal venous phase CT. The data for this review were obtained at the CT scan done after the third treatment, i.e., at the time of the fourth treatment.
All lesions were classified into three categories according to the degree of contrast enhancement on the arterial phase baseline CT scan images: 1) homogeneous hypervascular (tumors uniformly hyperattenuated to the liver parenchyma); 2) heterogeneous hypervascular (some tumor components hyperattenuated and some hypoattenuated to the liver); and 3) hypovascular (all parts of the tumor were less enhanced than the liver on arterial phase CT scans).
A patient with only one hypervascular or hypovascular lesion is classified as either hypervascular or hypovascular. A patient with many hypovascular and hypervascular lesions is classified as hypervascular if all or most of the lesions are hypervascular. Similarly, a patient with lesions that are all or mostly hypovascular is classified as hypovascular. A patient with an equal number of hypovascular and hypervascular lesions is classified as mixed vascularity. None of our patients had lesions of mixed vascularity.
Tumor size was measured by electronic calipers as the maximum perpendicular diameters of each tumor (a × b). The following formula was used to calculate the change in total tumor size between the baseline (pretreatment) and the third posttreatment CT scan.
where a × b represents the product of the longest perpendicular diameters of each tumor before treatment and a′ × b′ is the product of the diameters of the tumors following treatment. For the posttreatment CT scans, the change in tumor size compared with the original size as detected at baseline CT scan was calculated.
Necrosis was defined radiologically as areas of very low attenuation. These areas show no contrast enhancement at arterial or portal venous phases of biphasic CT scans in comparison to the normal adjacent liver parenchyma.
The Hounsfield unit (HU) density was measured for all lesions on both preembolization and postembolization CT scans. The region of interest (ROI) marker was placed in the center of each lesion to cover 50% of the area of the tumor. On the same slice, using the same size ROI marker, control density from the normal liver parenchyma was obtained. Measurements were taken using the same anatomic plane for each lesion and for the corresponding normal liver as a control in each phase (noncontrast, arterial, and portal venous). The increase in HU density of the lesion and normal liver was calculated by subtracting the noncontrast HU density from the arterial or the portal venous HU density. Failure of the lesion to increase in density in the arterial and portal venous phases compared with the noncontrast phase indicates a lack of perfusion and thus coagulative necrosis or successful treatment.
The necrotic areas were measured by the longest perpendicular diameter of all tumors, summed, and the percent of necrosis was calculated using the following formula:
where a′ and b′ are the perpendicular diameters of the necrotic areas (summed) and a and b are the perpendicular diameters of the whole tumor, summed, measured on the baseline CT scan.
New lesions constituted additional tumor masses that developed within the hepatic lobe being treated by TACE. New tumor growth was usually not histologically confirmed but the diagnosis was based on the detection of new focal hepatic lesions that had similar CT scan characteristics as the tumor in the same patient on baseline CT scan.
Definition of Responders
To account for the variables of change in tumor necrosis, size reduction, and appearance of new lesions, we devised a scoring system (Table 1). For patients whose total tumor size showed more than a 25% reduction, we assigned 1 point. For total tumor necrosis of more than 50%, we assigned 1 point. If no new hepatic lesions developed by the time of the third follow-up CT scan, we assigned 1 point. If the point total was 2 or 3, the patient was a responder. If the total was 0 or 1, the patient was a nonresponder.
|Criteria||1 point||0 points|
|Size reduction increase||≥ 25%||< 25% reduction|
|Necrosis||≥ 50%||≤ 50% necrosis|
|New lesions||No new lesions||New lesions|
The chi-square test was used to test independence in contingency tables where the row and column levels are not ordered. The Kruskal–Wallis test was used for contingency tables in which the column (response) levels are ordered. The Jonckheere–Terpstra test was used for contingency tables where the row and column levels are both ordered. Survival curves were produced using the Kaplan–Meier method. The log rank statistic was used to test the equality of the survival curves.
Among the 72 patients with 186 individually evaluable tumor masses, 59 tumors were classified as homogeneously hypervascular, 94 as heterogeneously hypervascular, and 33 as hypovascular. Table 1 illustrates the method by which we derived our modified criteria to judge response to therapy based on a combination of tumor size reduction, tumor necrosis, and the development of new malignant hepatic lesions during therapy. Using the traditional criterion of reduction in tumor size only as the measure of response to TACE therapy, there was no significant difference (P = 0.93) in response among the three vascular categories of tumor (Table 2).
|Size decrease (%)||Homogeneous/hypervascular (n = 59) (%)||Heterogeneous/hypervascular (n = 94) (%)||Hypovascular (n = 33) (%)|
|51–100||26 (44)||35 (37)||14 (42)|
|25–50||10 (17)||24 (26)||7 (21)|
|0–24||8 (14)||19 (20)||7 (21)|
|Progression||15 (25)||16 (17)||5 (15)|
|Total no. of lesions||59||94||33|
If the two subcategories of hypervascular lesions were combined, there were 153 hypervascular and 33 hypovascular tumors. Tables 3, 4, and 5 show the respective relationships between tumor vascularity, tumor size reduction, percent necrosis and development of new tumors during treatment. None of the trends toward a greater response among the hypervascular tumors reached statistical significance.
|Decrease in tumor size (%)||Hypervascular (n = 153) (%)||Hypovascular (n = 33) (%)|
|50–100||61 (40)||14 (42)|
|25–< 50||38 (25)||7 (21)|
|0–< 25||23 (15)||7 (21)|
|Progression||31 (20)||5 (15)|
|Increase in tumor necrosis (%)||Hypervascular (n = 153) (%)||Hypovascular (n = 33) (%)|
|50–100||42 (28)||10 (30)|
|25–< 50||12 (7)||Zero|
|> 1–24||32 (21)||1 (3)|
|0||67 (44)||22 (67)|
|Hypervascular (n = 62 patients) (%)||Hypovascular (n = 10 patients) (%)|
|16 (26)||3 (30)|
Using the modified criteria, 38 patients were classified as responders and 34 patients were nonresponders (Table 6). As shown in Figure 1 and Table 7, patient survival was significantly increased (P = 0.009) in patients who were hypervascular responders versus those who were hypervascular nonresponders. Patient survival was also significantly increased in hypovascular responders versus hypovascular nonresponders (P = 0.002) and in hypervascular nonresponders versus hypovascular nonresponders (P = 0.008).
|12 mos||18 mos||24 mos|
|38 Responders (53%)||71||40||16|
|34 Nonresponders (47%)||41||9||3|
|Survival (mos)||Hypervascular responders (n = 34) (%)||Hypervascular nonresponders (n = 28) (%)||Hypovascular responders (n = 4) (%)||Hypovascular nonresponders (n = 6) (%)|
|6||34 (100)||25 (89)||4 (100)||5 (83)|
|12||28 (82)||18 (44)||2 (50)||0 (0)|
|18||16 (47)||4 (14)||2 (50)||0 (0)|
|24||9 (27)||3 (11)||1 (25)||0 (0)|
Figure 2 shows a significant (P < 0.0001) survival benefit among HCC responders to TACE. Survival among HCC nonresponders was significantly less, but still improved in comparison to historical controls of patients with advanced HCC who received no treatment.
Among the 42 patients with single lesions, 54% responded to TACE. Among the 30 patients with multiple lesions, 49% (n = 19) responded, but the difference was not significant (P = 0.31). Figure 3 shows that the survival of patients with single and multiple lesions was the same. Figures 4, 5, and 6 show CT scan images of three HCC patients before and after TACE treatment.
Among the 186 total individual tumor masses, therapeutic response to TACE was significantly (P = 0.02) and inversely proportional to initial tumor size (Table 8).
|Tumor size (cm)||Responders (%)||Nonresponders (%)|
|< 3 (n = 33)||23 (70)||10 (30)|
|3–5 (n = 55)||31 (56)||24 (44)|
|5–10 (n = 56)||28 (50)||28 (50)|
|> 10 (n = 42)||18 (43)||24 (57)|
|Total (n = 186)|
Patients with unresectable HCC have a poor prognosis. Few HCC patients in the United States are candidates for surgical resection, which is considered the only potentially curative therapy.13, 14 Although new therapies (e.g., ablative techniques) are being developed, the most extensive experience has been reported for TACE.2, 4 The goals of TACE are to deliver concentrated chemotherapy directly to the tumor(s) and to reduce the blood supply to the tumor, causing ischemic necrosis.15 Many investigators have reported their results of tumors treated with TACE, usually using cross-sectional imaging (ultrasonography, CT scan, or magnetic resonance imaging [MRI] scan) to measure tumor size before and after therapy and using the World Health Organization (WHO) criteria of tumor size to determine the response rate. By these criteria, complete response indicates complete disappearance of the lesion; partial response means reduction of size (the product of two perpendicular diameters) of greater than 50%; minor response means a reduction of 25–50%; and no change means less than 25% increase or decrease in tumor size. Progressive disease indicates a greater than 25% increase in tumor size.12, 16–18
Some investigators have noted certain discrepancies or apparent illogic in applying the WHO criteria in judging the response of HCC patients to TACE. Santis et al.18 and Takayasu et al.16 noted decreased HCC tumor size in response to TACE, but persistent evidence of tumor vascularity on CT scan, angiography, and/or MRI scan, indicating viable tumors.
Although many necrotic tumors will decrease in size, the rate of size decrease varies substantially, probably due to the varying rates of resorption of the necrotic tissue.17, 20 Others20, 21 have proposed using tumor necrosis, rather than size, as the criterion for judging response to TACE. Still others7 have proposed judging therapeutic response by angiographic demonstration of a decrease in tumor vessels (neovasculature).
Computed tomographic scan is widely available and helical multiphase CT scan allows us to evaluate the number, size, necrosis, and vascularity of hepatic tumors. We believe that our proposed criteria provide a logical and simple means to predict and measure the response of various hepatic tumors to TACE.
Many potential factors determine a patient's response to TACE.12, 18, 20, 22 One of the more important variables is the degree of cirrhosis and not the tumor itself. In our institution, TACE is not used to treat patients with Child C cirrhosis and is rarely used to treat patients with Child B cirrhosis. The patients in this study had either Child A cirrhosis or no cirrhosis. Therefore, the results are more likely to reflect the tumor characteristics and its treatment. Another concern is worsened survival due to hepatotoxicity in the compromised, cirrhotic liver. To limit this eventuality, we used Gelfoam to slow the hepatic arterial flow to the tumor, but not to complete occlusion. We reasoned that the tumor occludes the portal vein and complete occlusion of the hepatic artery in that setting can result in major liver necrosis. Larger tumors at baseline were less likely to respond to TACE.18, 23 In the current study, 70% of tumors less than 3 cm in diameter responded to TACE compared with only 43% of tumors more than 10 cm in diameter.
The maximum benefit of TACE may not be achievable in patients with severely compromised liver function. The time interval between TACE procedures was prolonged in patients who developed hyperbilirubinemia, which is an important but unavoidable variable in trying to assess the therapeutic effect of TACE on various tumors. Patients with HCC often have decreased hepatic reserve. Therefore, they often receive less aggressive TACE than patients with minimal or no impairment of hepatic function, depending on the center, which may have an important influence on tumor response and patient survival. In the current study, all patients had Child A cirrhosis or no cirrhosis and were treated by the same protocol.
Katyal et al.11 reported that HCCs that were predominantly hypervascular at baseline helical arterial phase CT scan were more likely to respond to TACE and that these patients had prolonged survival compared with patients who had less vascular HCCs. We confirmed the survival benefit among patients with hypervascular HCCs in a larger group of patients. Of our 72 patients with HCC, 38 fit our revised criteria of responders and 34 were nonresponders. The survival rates for the responders at 12, 18, and 24 months were 71%, 40%, and 16%, respectively. For nonresponders, the patient survival rates at 12, 18, and 24 months were 41%, 9%, and 3%, respectively (Table 6).
One important difference between the Katyal et al. study and ours was our evaluation of tumor response and patient survival in relation to tumor size change, necrosis, and tumor vascularity. Hypervascular tumorsmay be more likely to respond to TACE, but we were unable to confirm this as a significant relationship in our study. However, we did demonstrate significantly improved survival among patients with predominantly hypervascular tumors, using our combined criteria for responders.
It is noteworthy that patients with hypervascular HCCs have a survival benefit from TACE, even if they are classified as nonresponders by size criteria or our combined criteria. The 12-month, 18-month, and 24-month survival rates for patients classified as hypervascular nonresponders (44%, 17%, and 10%, respectively) compare favorably with historical controls of patients with untreated HCC.5, 9, 10 All patients in our study had advanced Stage III or IV tumors (TNM classification) and a life expectancy of a few months without treatment. Our results indicate that TACE results in a survival benefit to responding patients with unresectable HCC. We suggest a modified scoring system that takes into account not only changes in tumor size, but also changes in tumor necrosis and the appearance of new lesions.
- 21Transcatheter oily chemoembolization for the treatment of hepatocellular carcinoma: a multivariate analysis of prognostic factors. Hepatology. 1994; 19: 1117–1123., , , et al.