Fax: (011) 61-2-9350 3997
Echogenicity of liver metastases from colorectal carcinoma is an independent prognostic factor in patients treated with regional chemotherapy
Version of Record online: 15 MAR 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 6, pages 1753–1759, 15 March 2002
How to Cite
Gruenberger, T., Zhao, J., King, J., Chung, T., Clingan, P. R. and Morris, D. L. (2002), Echogenicity of liver metastases from colorectal carcinoma is an independent prognostic factor in patients treated with regional chemotherapy. Cancer, 94: 1753–1759. doi: 10.1002/cncr.10386
- Issue online: 15 MAR 2002
- Version of Record online: 15 MAR 2002
- Manuscript Accepted: 26 NOV 2001
- Manuscript Revised: 22 OCT 2001
- Manuscript Received: 13 JUL 2001
- colorectal carcinoma;
- unresectable liver metastases;
- hepatic artery infusion;
Echogenicity of liver metastases was found to be a predictive biologic factor influencing long-term outcome after curative liver resection. The current analysis focuses on the influence of echogenicity on survival in patients treated with intraarterial chemotherapy for unresectable colorectal carcinoma liver metastases.
A retrospective analysis of prospectively collected data at the Department of Surgery at the University of New South Wales-affiliated St. George Hospital was performed. Two hundred twelve consecutive patients with unresectable hepatic metastases from colorectal carcinoma treated between May 1992 and September 2000 were analyzed. Echogenicity of metastases was measured intraoperatively using a 5 MHz probe. Overall survival difference was compared between hyper- and hypoechoic metastases on an intention-to-treat basis.
At a median followup of 15.1 months, 47 patients (22%) were alive and 165 (78%) had died. A significant survival benefit was observed in patients having hyperechoic lesions (median survival 16.2 months, 95% confidence interval [CI] 13.9–18.5) compared to hypoechoic lesions (median survival 11.6 months, 95% CI 8–15.2), P < 0.01. Other prognostic factors were differentiation of the primary tumor (P < 0.02), percentage hepatic replacement (P < 0.05) and carcinoembryonic antigen decrease (P < 0.03). Echogenicity was identified as an independent prognostic factor in multivariate analysis (P < 0.009).
Echogenicity is an important prognostic survival parameter. Cancer 2002;94:1753–9. © 2002 American Cancer Society.
Overall survival after resection of colorectal carcinoma is mostly dependent on the stage of the disease at presentation. The current accepted staging system is the Dukes classification, which is the most important independent risk factor when analyses of colorectal carcinoma surveillance have been undertaken. Around 25% of patients with colorectal carcinoma present with liver metastases, and an additional 40% of the remaining patients will develop liver metastases during followup. The survival of untreated patients with colorectal carcinoma liver metastases (CRCLM) is limited to a median of a little over six months.1, 2 The best treatment option for patients with CRCLM remains potentially curative resection or destruction of tumors, which leads to median survival figures of up to 40 months, with a 5 year survival rate of 25–45%.3, 4 Unfortunately, only approximately 20% of patients with CRCLM are candidates for a curative procedure at presentation.5 Several attempts have been made to increase the numbers of patients suitable for potentially curative treatment, including neoadjuvant chemotherapy,6 portal vein embolization,7 contralateral lobe tumor destruction,8 and increasing the upper limit of the number of metastases treated at one stage with curative intent.9
The high percentage of CRCLM patients unsuitable for curative treatment are left with three basic treatment options: systemic chemotherapy, intrahepatic chemotherapy, or best supportive care.
We recently identified echogenicity of CRCLM as an additional prognostic factor in patients undergoing either liver resection or cryotherapy with an independent influence on the long-term prognosis of these patients.10 It is unknown whether echogenicity of CRCLM is prognostic for other groups of patients.
The current report summarizes the results of intraarterial hepatic chemotherapy for patients with unresectable CRCLM with specific reference to the echogenic appearance of the metastases during intraoperative ultrasound and the influence of that echogenic appearance on prognosis. Comparison to other prognostic factors was performed.
MATERIALS AND METHODS
This prospective single center study consisted of two hundred twelve patients with unresectable CRCLM treated at the liver unit of the Department of General Surgery, St. George Hospital, affiliated with the University of New South Wales. All patients were classified on pre and/or intraoperative assessment as unsuitable for curative liver resection or cryotherapy, and a hepatic artery catheter was placed into the gastro-duodenal artery connected to either a subcutaneous port or a pump for hepatic arterial infusion chemotherapy (HAI). All patients had a histologic diagnosis of colorectal carcinoma and a preoperative workup of abdominal computed tomography (CT), including intravenous contrast, chest CT, a bone scan, and carcinoembryonic antigen (CEA) measurement. Recruitment of study subjects took place between July 1992 and October 2000. There was no restriction on the amount of liver replacement by metastases, but, in patients with high volume liver disease (> 75%) and poor performance status, a preoperative response assessment was performed. This consisted of percutaneous intraarterial catheter placement through a femoral approach, followed by one cycle of fluorouracil (5-FU) treatment (4 g over four days). Response was assessed by CEA measurement before and after treatment (Day −1 and +7). Extrahepatic disease was not a contraindication for the placement of a hepatic artery catheter, as long as the liver was the organ with the main tumor load. Intracolonic or local recurrence, positive hilar lymph nodes, or low volume peritoneal disease was resected along with the catheter placement.
At laparotomy the extent of liver involvement was assessed and an intraoperative ultrasound performed using an ALOKA SSD-650 CL machine with a 5 MHz T-shaped linear array probe (Aloka Co. Ltd., Tokyo, Japan). Echogenicity assessment was performed in all patients by the same investigator (D.L.M.). The echogenic appearances of the metastases were recorded in the operation notes as being hyper– or hypoechoic compared to the surrounding liver parenchyma (Figs. 1 and 2). The catheter was inserted into the gastroduodenal artery after distal ligation, and the right gastric artery was divided. The gastroduodenal artery was used for catheter placement whenever available. Accessory right or left hepatic arteries were ligated to achieve cross perfusion. If the gastroduodenal artery was absent or too small, the catheter was placed into the right gastric artery or the divided left hepatic artery retrograde. If all attempts to use original arteries were unsuccessful, the catheter was placed into a sidearm goretex-graft, which was sutured onto the main hepatic artery. Homogenous hepatic perfusion was confirmed with the injection of 5 mL of methylene blue through the subcutaneous placed port or pump. Careful ligation of additional branches supplying the stomach, duodenum, or pancreas was carried out. The port or pump was locked with heparin, and intraarterial chemotherapy was started within a few days of surgery.
The regional chemotherapy regime consisted either of 1000 mg/m2/day of 5-FU or 0.2 mg/kg/day fluorodeoxyuridine (FUDR) plus 8 mg dexamethasone and folonic acid 15 mg three times daily (tds) for seven days every second week if an external pump was used. In the case of a subcutaneous implanted pump, 0.2 mg/kg/day FUDR were given over a two week period, followed by 14 days of normal saline/heparin infusion; the dose was reduced to 0.1 mg/kg/day if toxicity occurred, and a dose escalation to 0.3 mg/kg/day was tested in patients primarily not responding. Dose variation was necessary in less than 10% of the group. Treatment was continued until either complete response was attained or a generalized progression of disease (intrahepatic and/or extrahepatic) was noted. Technical catheter problems were dealt with depending on response to HAI. If adequate response was achieved prior to blockage, a second attempt at catheter placement was undertaken; the new catheter was usually placed into a side arm graft, which was anastomosed end-to-side to the common hepatic artery. Early catheter blockage occurred in only 5 patients, late blockage (after a minimum of 2 treatment cycles) occurred in an additional 16 patients (9.9% total). Port or pump side infections were managed in 10 patients (4.7%) and were usually dealt with locally, and an additional decontamination infusion was given into the system. Late malperfusion occurred in 30 patients (14.1%), and the additional artery was embolized super-selectively whenever possible. Late local artery problems occurred in 13 patients (6.1%) and consisted of a hepatic artery aneurysm in 4 patients and sclerosing cholangitis in 9 patients.
Patients had followups every three months, which included clinical examinations and serum CEA measurements. In CEA-nonsecretors and patients with CEA response, this was accompanied by a contrast CT of the liver. In CEA non-responders a complete workup was performed, starting with a CT of the liver to distinguish between local (liver) and wide spread disease progression. Response was categorized as the sum of the products of the diameters of measurable lesions and recorded using World Health Organization criteria for complete or partial response and stable disease or disease progression. Complete response was defined as disappearance of intrahepatic tumor on CT and/or normalization of a previously elevated CEA level. Partial response was defined as a reduction of tumor volume on CT and/or CEA by > 50%. Patients were followed until February 2001 or death.
The statistical analysis was performed on an intention-to-treat basis. Survival estimates were calculated by the Kaplan-Meier method. Possible prognostic factors were assessed for their impact on survival by the log rank test (categorical variables) or Cox regression model (continuous variables) for unifactorial analysis (a significant difference was assumed for P < 0.05). The Cox regression model was used for multifactorial analysis. Statistical analysis was performed using the program SPSS version 10.0 (SPSS Inc, Chicago, IL).
From 1992 to 2000 adequate information on the echogenic appearance of CRCLM treated by HAI for survival analyses was available for 212 patients. There were 148 male and 64 female patients, with a mean age of 61 years (range, 34-82). In half of the patients, CRCLM were diagnosed through routine followup CT scans. In the other half, CRCLM were either diagnosed through staging of the primary tumor (30%), rising CEA levels during followup(12%), or the presentation of symptoms (8%). The total number of CRCLM at laparotomy was under 5 in 21% of patients, 5 to 10 in 44%, and > 10 in 35%. Hepatic replacement by tumor during the initial operation was < 50% in 133 patients (62%); 4% of the patients had a tumor burden of more than 75%. A percutaneous treatment cycle to evaluate tumor response was initiated in 16 patients (8.5%). Minor and resectable extrahepatic disease was found at laparotomy in 32% of patients. The primary colorectal carcinoma was well differentiated in only 3% of the study group, moderate in 73% and poor in 24%.
Partial response to intraarterial hepatic chemotherapy was achieved in 106 patients; in 44 patients (21%) postoperative CEA levels returned to normal. Carcinoembryonic antigen response was not measurable in 39 patients (18%) due to a lack of pretreatment CEA levels and in 34 patients (16%) defined as non-secretors. A measurable CEA response was therefore achieved in 50% of the patient group. Patients amenable to CEA and CT response evaluation were usually corresponding, but CEA values indicate response earlier.
Overall survival was 61% at one year, 24% at two years, and 11% at three years, with a median survival of 15.2 months. Echogenic appearance of the CRCLM was found to be hyperechoic in 138 patients (65%) and hypoechoic in 74 patients (35%). Echogenic appearance of CRCLM on ultrasound was able to predict a survival advantage for patients with hyperechoic lesions: 16.2 months versus 11.6 months for patients with hypoechoic lesions (P < 0.01) (Fig. 3). Distribution of prognostic factors between the two echogenic groups is shown in Table 1. A significantly higher number of patients in the hyperechoic group had primary tumors that were well or moderately differentiated; in contrast, twice the percentage of patients had hepatic replacement by tumor > 50% in the hyperechoic group. The number of CRCLM, the presence of extrahepatic disease, gender and CEA response was found equally distributed between the two groups. A significantly different percentage of patients returned to normal CEA levels in the two echogenic groups (hyperechoic, 25.4% versus hypoechoic, 12.2%, P < 0.02).
|Good||5 (3.8%)||1 (1.4%)|
|Moderate||104 (78.2)||45 (63.4)|
|Poor||24 (18)||25 (35.2)|
|< 25%||37 (27.6)||30 (41.1)|
|25–50%||37 (27.6)||29 (39.7)|
|> 50–75%||52 (38.8)||13 (17.8)|
|> 75%||8 (6)||1 (1.4)|
|≤ 50%||74 (55.2)||59 (80.8)|
|> 50%||60 (44.8)||14 (19.2)|
|< 5||31 (22.6)||13 (17.6)|
|5–10||60 (43.8)||34 (45.9)|
|> 10||46 (33.6)||27 (36.5)|
|Yes||41 (29.7)||26 (35.1)|
|No||97 (70.3)||48 (64.9)|
|No||21 (15.2)||12 (16.2)|
|Nonsecretor||16 (11.6)||18 (24.3)|
|Yes||40 (29)||22 (29.7)|
|To normal||35 (25.4)||9 (12.2)|
|Not available||26 (18.8)||13 (17.6)|
|Female||45 (32.6)||19 (25.7)|
|Male||93 (67.4)||55 (74.3)|
With regard to survival benefits, several prognostic factors were investigated (univariate analyses, Table 2). Survival figures were significantly influenced by the differentiation of the primary tumor: patients with well differentiated tumors had a median survival of 20 months, compared with 16 months for moderately differentiated and 8.8 months for poorly differentiated tumors (P = 0.0001). The percentage of hepatic replacement also significantly influenced survival: patients with < 25% metastases had a median survival of 17.7 months, compared to 13.8 months in the group with 25–50% replacement, 10.6 months in the group with 51–75% replacement, and 9.1 months in patients with a tumor burden over 75% (P = 0.01). Categorizing the numbers of metastases in the three groups showed a survival advantage for patients with less than 5 tumors: for those with under 5 CRCLM, median survival was 18 months, versus 15.7 months for those with 5-10 CRCLM and 9.7 months for patients with over 10 CRCLM (P = 0.03). The presence of extrahepatic disease had no significant influence on survival (only patients with macroscopically resectable disease were considered): median survival for patients without extrahepatic disease was 15.7 months, versus 12.1 months for patients with extrahepatic involvement (P = 0.09). Carcinoembryonic antigen response to HAI significantly influenced survival: the median survival was 8.1 months for non-responders, 17.7 months for responders, 19.9 months for patients whose CEA returned to normal, and 12.9 months for non-secretors (P < 0.0001).
|Characteristics||No. of patients||Median survival time||P|
|< 25%||67||17.7||(15.24, 20.16)|
|> 50–75%||65||10.57||(4.87, 16.27)|
|> 75%||9||9.1||(4.91, 13.29)|
|< 5||44||17.87||(14.86, 20.87)|
|> 10||73||9.77||(7.79, 11.74)|
|To normal||44||19.9||(13.46, 26.34)|
All prognostic factors used in univariate analysis were subsequently entered into a multivariate Cox regression analysis. Differentiation of the primary tumor (P = 0.001), percentage of hepatic replacement by CRCLM (P = 0.003), echogenicity (P = 0.009), and the presence of extrahepatic disease at laparotomy (P = 0.02) all independently influenced survival.
The natural history of unresectable metastatic colorectal carcinoma provides a median survival time of approximately six months.11 Systemic chemotherapy achieves a response rate of up to 30%, depending on the drug regimen used, and provides a median survival of approximately 12 months.12–14 The leading organ of metastatic involvement is the liver, especially in primary colon and upper rectum carcinoma. Liver metastases are almost exclusively supplied by the arterial blood stream,15 and this phenomenon allows intraarterial chemotherapy to reach a much higher local concentration with reduced systemic side effects. The usage of drugs with a high first pass effect leads to higher tumor response rates of up to 80%.16, 17 In addition to the higher local tumor concentration, continuous infusion allows prolonged drug exposure to the tumor cells and thus potentially increases the cell kill fraction.18, 19
To prove the impact of hepatic infusional chemotherapy on the natural history of hepatic metastases, randomized studies were initiated over the past 15 years comparing regional versus systemic therapy.20–26 All these studies showed a significantly different response rate favoring the intraarterial approach but were unable to show a significant survival benefit. The inability to transform the three times higher response rates into survival advantages was due to the fact that all these studies allowed the crossover of patients to HAI at the time of failure of intravenous chemotherapy. Subgroup analyses showed a significant survival advantage in some of these trials,21, 24 and a metaanalysis including all randomized trials showed a significantly different survival after one year.27 Both systemic and intraarterial chemotherapy have been proven to achieve significant survival benefits over best supportive care in patients with CRCLM.11, 28
We have recently described the prognostic factors in our series of patients with CRCLM treated by regional chemotherapy.29 We found that a less than 50% liver involvement, an absence of extrahepatic disease, and a greater than 25% serum CEA decrease were associated with improved survival.
The rationale of hepatic arterial infusion is supported by the arterial supply of hepatic metastases.15 Therefore, a higher drug concentration at the tumor site can be achieved, thus lowering systemic side effects. Depending on the chemotherapeutic agent used (FUDR or 5-FU), a first pass effect of over 90 % was defined,18, 30 lowering the systemic side effects and leading to higher quality of life while the patients were on treatment.22
Our results show the known high rate of 50% responders to the treatment achieving a median survival of 15 months. We included patients with localized extrahepatic disease as long as we were able to macroscopically remove it during surgery. This influenced the overall survival rate, as the presence of extrahepatic disease was found to be statistically significant in the multivariate analysis. The other prognostic factors, the differentiation of the primary tumor and percentage of hepatic replacement, are well-established indicators for survival in CRCLM.31
The echogenic appearance of CRCLM during intraoperative ultrasound is a new prognostic indicator in this specific patient group. We were previously able to show a significant survival advantage in patients with hyperechoic lesions both in those treated with cryotherapy32 and in those undergoing liver resection for CRCLM.10 The percentage of patients showing a complete response in regard to CEA decrease was significantly higher in the group of hyperechoic CRCLM in the current study.
The confirmation of echogenicity of CRCLM as an independent prognostic survival indicator raises the question of the underlying cause for the differences in appearance. We were able to show in the current series of liver resection specimens that the difference is not correlated to the surrounding liver tissue or an intralesional difference in blood vessels. The only histologic finding was the presence of significantly more mucin producing tumors in the hypoechoic group. We were able to confirm the prognostic impact of echogenicity in other tumor entities inside the liver as well (unpublished data). These findings led to the suggestion that an intratumoral biologic factoror a tumor in the surrounding tissue might be responsible for the underlying difference. We have performed several immunohistochemic investigations with established biologic indicators but have been so far unable to find any significant correlation.
The confirmation of the independent influence of echogenicity on prognosis in an additional patient group should lead to the inclusion of echogenicity in the decision-making process regarding best treatment options for patients suffering from liver tumors.
The authors thank Mayne Nickless for unit support.
- 12Modulation of fluorouracil by leucovorin in patients with advanced colorectal cancer: evidence in terms of response rate. Advanced Colorectal Cancer Meta-Analysis Project. J Clin Oncol. 1992; 10: 896-903.
- 13Meta-analysis of randomized trials testing the biochemical modulation of fluorouracil by methotrexate in metastatic colorectal cancer. Advanced Colorectal Cancer Meta Analysis Project. J Clin Oncol. 1994; 12: 960-969.
- 28Expectancy or primary chemotherapy in patients with advanced asymptomatic colorectal cancer: a randomized trial. Nordic Gastrointestinal Tumor Adjuvant Therapy Group. J Clin Oncol. 1992; 10: 904-911.
- 29Prognostic factors in patients with colorectal liver metastases undergoing regional chemotherapy. HPB. in press., , , , , .