Echogenicity of Liver Metastases Is an Independent Prognostic Factor After Potentially Curative Treatment

Hypothesis  Intraoperative echogenic appearance of liver metastases from colorectal cancer is a prognostic factor of outcome after curative treatment.

Design  Retrospective analysis of prospectively collected data.

Setting  Department of Surgery at a university hospital.

Patients  One hundred forty-three consecutive patients with hepatic metastases from colorectal cancer who underwent liver resection with curative intent between 1992 and 1998.

Intervention  Curative treatment was achieved by liver resection alone, liver resection plus edge cryotherapy, or liver resection plus cryotherapy to lesions not amenable to further resection. In patients with more than 2 lesions, a hepatic artery catheter was placed for regional chemotherapy.

Main Outcome Measures  The echogenic appearance of the liver metastases was assessed by intraoperative ultrasound by a single person throughout the study using a 5-MHz ultrasound probe. The findings were prospectively entered into the database.

Results  Fifty-four percent of patients had hyperechoic metastases. This group had significantly longer overall (log rank, P<.001) and recurrence-free survival (log rank, P = .004) compared with patients who had hypoechoic metastases (36%). A significantly higher percentage of mucin-secreting tumors were found in the hypoechoic patient group (χ², P = .001). Dukes stage of the primary tumor (P = .02), echogenicity of the liver secondaries (P = .04), and diameter of the largest resected metastasis (P = .01) were independent prognostic factors for recurrence-free survival in the Cox regression model.

Conclusion  These results support the hypothesis that echogenicity of liver metastases from colorectal cancer is an independent prognostic factor of outcome after curative resection.

THE LIVER is a common site of metastatic disease from colorectal cancer with resection a well-established treatment option, offering a 5-year survival rate of approximately 25%.^1,2 In situ tumor destruction by means of cryotherapy has been used widely with evidence of prolonged survival^3-7 and has been shown recently to be of curative value in a prospective randomized trial.⁸

Primary tumor stage, size and number of metastases, disease-free interval, bilateral liver disease, preoperative carcinoembryonic antigen (CEA) level, and resection margin have been identified as prognostic factors of prolonged survival in patients with liver metastases from colorectal cancer.^2,9-13 We recently reported that patients with colorectal liver metastases who underwent cryotherapy had a significantly better survival rate if their metastases were hyperechoic.¹⁴ We have confirmed a survival benefit for patients with hyperechoic lesions treated with hepatic artery chemotherapy for unresectable colorectal liver metastases (D.L.M., unpublished data, 2000).

The underlying possible reason for the favorable survival outcome in patients with hyperechoic lesions is unknown. The purposes of this study were to determine the effect of echogenicity on survival in patients undergoing potentially curative surgical treatment (liver resection) and to determine the relationship between echogenicity and pathologic features of the resected specimens. To our knowledge, this is the first report to demonstrate the effect of echogenicity on survival after liver resection in patients with colorectal liver metastases.

All patients with liver diseases (N = 1470) treated in our unit from 1992 to 1998 were identified from our prospective database. There were 839 patients with liver metastases from colorectal cancer and in 143 of these, liver resection as a potentially curative treatment was possible. This 143-patient cohort formed the sample for this analysis. Patient selection for liver resection used standard criteria, including absence of severe cardiopulmonary disease, absence of extrahepatic disease (colonoscopy, abdominal and chest computed tomography, bone scan, and computed tomographic portography), and anatomical resectability. Data for 143 patients were extracted from the database, preoperative and postoperative clinical admission notes, and patient interviews. Data examined included demographics; pathologic features of the primary tumor and the liver metastases; operative details, including echogenicity of the liver lesion on intraoperative ultrasound; postoperative course until discharge; and overall and recurrence-free survival time.

The echogenicity of liver metastases was assessed intraoperatively in all patients by the same investigator (D. L. M.). Relevant data were available only from operative notes, as hard copy ultrasound images were not issued. An ultrasound machine (Aloka SSD-650 CL; Aloka Co Ltd, Tokyo, Japan) with a 5-MHz T-shaped linear array probe was used. The appearance of the metastases was graded as hyperechoic or hypoechoic depending on their acoustic density compared with the surrounding liver parenchyma. Patients with lesions containing both hyperechoic and hypoechoic areas were graded as having mixed echogenicity. Compared with our previous evaluation of the prognostic relevance of echogenicity in hepatic cryotherapy patients,¹⁴ in the current study we were able to analyze the "mixed echogenicity" group separately as we had a sufficient number of patients.

All patients included in this analysis underwent potentially curative treatment of their liver lesions. Curability was defined as complete resection, or resection plus cryotherapy, of all identifiable lesions. It is noteworthy that the overall survival rate was similar in all treatment types (resection, resection plus edge cryotherapy, resection plus cryotherapy to additional lesions, and resection plus cryotherapy to additional lesions plus edge cryotherapy) as shown in Figure 1. A hepatic artery catheter for adjuvant chemotherapy was placed in patients with more than 2 lesions. Cryotherapy was performed with the L. C. S. system (Cryotech LC system 3000; Spembly, Andover, England), which delivers liquid nitrogen to the tip of a triple-lumen, insulated, trocar-tipped probe. Cryotherapy (using partial double freeze-thaw cycles under ultrasound control) was used either for edge freezing of inadequate resection margins (edge cryotherapy) or for complete tumor destruction of additional lesions in the remaining liver not amenable to further liver resection. Adjuvant regional chemotherapy (delivered by an external ambulatory pump) was given through the hepatic artery catheter with 5-fluorouracil, 1 g daily for 4 days with oral folinic acid, 15 mg 3 times daily every 2 weeks. When an implanted pump was used, floxuridine, 0.1 to 0.3 mg/kg daily, was given for 14 days followed by 14 days' heparin infusion. Thereafter, chemotherapy was restarted. Patients treated with cryotherapy alone were not included in the current series as we already have reported the effect of echogenicity on survival in these patients.¹⁴ Patients had 3 monthly follow-up visits, which included clinical examination and serum CEA measurement. Abdominal computed tomographic scan, chest, bone, and brain investigations were performed only for marker rise or symptoms.

Survival estimates were calculated by the Kaplan-Meier method. Possible prognostic factors were assessed for their impact on survival time by the log-rank test (categorical variables) or Cox regression model (continuous variables) for unifactorial analysis (a significant difference was assumed for P<.05). The Cox regression model was used for multifactorial analysis. The χ² test was used to compare the distribution of categorical covariates and unpaired t test for continuous variables for patients with different types of echogenicity. Statistical analysis was performed using a computer program (SPSS version 8.0; SPSS Inc, Chicago, Ill).

There were 80 men and 63 women with a median age of 62 years (range, 31 to 83 years). Seventy-three patients (51%) had died at the time of last follow-up evaluation with median follow-up of 22 months. Distribution for Dukes stage of the primary tumor was A, 2%; B, 40%; and C, 58%. There were synchronous liver metastases diagnosed with the colorectal primary tumor in 55 patients (38.9%) and metachronous liver metastases in 31 patients (21.7%) within 12 months following the primary resection; the remaining metachronous metastases were diagnosed in 57 patients (39.4%) thereafter. A median number of 2 metastases (range, 1 to 8) with a median diameter of the largest lesion of 4.5 cm (range, 1.4-17.0 cm) were treated. Sixty-six patients (47%) had bilobar disease. Major resections, defined as resection of more than 2 segments, were performed in 53% of patients. Resection as the only curative treatment modality was used in 36% of patients; in all of these patients the resection margin was greater than 1 cm. Additional cryotherapy to the resection edge, to improve on margin clearance, was used for 33% of cases, and cryotherapy to destroy any remaining liver metastases was performed in 31% of patients. A hepatic artery catheter was inserted in 48% of patients.

There were no intraoperative deaths, but 2 patients (1.4%) died during the 30-day postoperative period; 1 patient died after myocardial infarction, and 1, from multiorgan failure following sepsis. Echogenicity data of the liver metastases were available for 95 patients (66%). There were 34 patients (36%) with hypoechoic lesions, 51 patients (54%) with hyperechoic lesions, and 10 patients (10%) with mixed echogenic lesions.

Mucin production in the liver metastases was found in 22 patients (15%). A significantly higher percentage of mucin production was observed in the group with hypoechoic lesions (χ²₂ analysis, 13.32; P = .001) (Table 1). Necrosis within the liver metastases was found in 84 patients (60%), and calcification of the liver metastases was observed in only 8 patients (7%). The 2 latter factors were not significantly correlated to echogenicity. None of the surrounding normal liver tissue specimens showed features of either severe fatty infiltration or cirrhosis on histological examination, which could have misled echogenicity categorization of lesions as this is based on normal parenchyma to tumor comparison. In 86% of patients, serum CEA levels were raised (>5 µg/L) prior to liver resection.

Median survival time for 143 patients was 35 months, with a 1-, 3-, and 5-year survival rate of 88%, 47%, and 19%, respectively. A significant survival benefit was observed for patients with hyperechoic liver metastases compared with those with hypoechoic lesions (median, 42 months, 95% confidence interval, 38.53-45.76 vs median, 27 months, 95% confidence interval, 12.92-40.24; log rank, P<.001). Patients with mixed echogenicity had a median survival of 35 months, which was not statistically different (log rank, P = .3) (Figure 2). This survival benefit was not only significant when analyzing all patients together, but also when analyzing patients within treatment subgroups. In patients for whom resection was the single treatment modality, hyperechoic metastases were associated with a significant survival benefit (log rank, P = .04). In patients for whom cryotherapy was added to liver resection, hyperechoic lesions again influenced survival positively (log-rank test, P = .01).

Mucin production within the metastases did not significantly influence median survival (positive for mucin at 31 months vs negative for mucin at 36 months, log rank, P = .6). In the mucin-negative patient group, echogenicity of liver metastases was found to influence survival significantly (log-rank test, P = .001, data not shown). The following potential prognostic factors were evaluated: age; sex; Dukes stage of primary tumor; grading of primary tumor and of liver metastases, unilobar vs bilobar liver metastases; metachronous vs synchronous occurrence of liver metastases; number of treated metastases; diameter of metastases and preoperative serum CEA level (Table 2). Of these, only Dukes stage of primary tumor and diameter of the largest lesion had a significant influence on survival in our study group. The patient group with Dukes A stage tumors had no death at last follow-up visit; patients with Dukes B stage had a 46 months' median survival rate vs 27 months in patients with Dukes C stage (log-rank test, P<.001).

The same analyses were performed for recurrence-free survival, and echogenicity was also found to be of prognostic relevance. Median time to recurrence was 31 months in patients with hyperechoic liver metastases compared with 9 months for patients with hypoechoic lesions (log-rank test, P = .004). Patients with mixed echogenicity of liver metastases had a median time to recurrence of 16 months (log-rank test, P = .3) (Figure 3). The diameter of the largest treated liver metastasis (P = .001), grading of the metastases (P = .03), and Dukes stage were found to be the other prognostic factors that influenced recurrence-free survival (Dukes B at 23 months vs Dukes C at 10 months, log-rank test, P = .001) (Table 3).

The analysis of prognostic factor distribution in patients with hyperechoic and hypoechoic colorectal liver metastases revealed no significant differences (Table 4). The Cox regression method analysis of prognostic factors for recurrence-free survival found Dukes stage (P = .02), echogenicity (P = .04), and diameter of the largest liver metastasis (P = .01) as independent factors. Dukes stage was the only independent predictor of overall survival (P = .001).

Liver resection for hepatic metastases from colorectal cancer is currently the best treatment option.^1,15 Unfortunately, only approximately 25% of patients with synchronous or metachronous hepatic metastases are suitable for resection.^16,17 Several studies of colorectal cancer patients have shown metastases in the liver to be one of the most important determinants of survival.^16,18 Several factors have been identified and described as being important predictors of survival after curative resection.^9-12 Intraoperative ultrasound has become an important part of intraoperative assessment of liver cancer, and differences in the echogenic appearance of liver metastases have been described.¹⁹

Our group has recently identified echogenicity as a prognostic factor affecting survival after hepatic cryotherapy.¹⁴ The present study demonstrates that echogenicity of colorectal liver metastases is an important prognostic factor for survival after curative liver resection. Both long-term and recurrence-free survival benefits were seen in patients whose liver metastases had a hyperechoic appearance, irrespective of treatment subgroups (resection alone or resection plus additional cryotherapy). Those patients with hyperechoic liver lesions had a median recurrence-free survival time 3 times longer than those patients who had hypoechoic lesions (31 months vs 9 months) (Figure 3). The median overall survival benefit of patients with hyperechoic lesions was also significantly increased (42 months vs 27 months) (Figure 2). Apart from echogenicity of liver metastases, only Dukes stage of primary tumor and the diameter of the largest treated liver metastasis were found to influence survival independently, as found in other studies.^2,10,13,20 Our results lend support to the theory that many prognostic factors recognized in univariate analysis, such as liver disease distribution or timing of metastatic diagnosis, may not be independent as previously believed.²¹ In our study, 47% of patients were treated for bilobar disease, and this group had the same survival rate as patients with unilobar disease. We found that patients with synchronous liver metastases had similar survival rates as those with metachronous disease. Our present study demonstrates that echogenicity influenced survival similarly in patients who underwent liver resection and in patients who underwent resection plus cryotherapy. Additional hepatic artery chemotherapy was administered to the same percentage of patients in both groups (hyperechoic vs hypoechoic lesions). Echogenicity is thus not likely to be simply a predictor of better response to adjuvant chemotherapy, as indicated in our previous study.¹⁴

We suggest that a tumor biological factor could be responsible for the different echogenic appearance of the metastases. Mucin secretion is an established negative prognostic factor for primary colorectal cancer.^22,23 The appearance of liver metastases on intraoperative ultrasound could be due to this biological behavior of the tumor, especially in hypoechoic liver metastases. While we found that almost all the mucin-secreting tumors (11 [79%] of 14) (Table 1) were hypoechoic, this feature only accounts for 32% of the hypoechoic tumors. Echogenicity still significantly altered survival when tumors without mucin were studied. The pathologic or structural differences, which are responsible for hypoechogenicity in tumors negative for mucin, are presently unknown. The vascularity and the stromal content of tumors could be further related to echogenicity, and the effect of immune cell response on echogenic appearance will be assessed in further studies.

We have identified an independent prognostic factor in the treatment of colorectal liver metastases, confirming the results reported in a smaller patient group.¹⁴ Echogenic appearance of metastases can easily be determined preoperatively, but we understand that it may differ from intraoperative findings. After confirmation of the prognostic relevance of echogenicity in this series, we are currently comparing preoperative and intraoperative echogenic findings. Their similarity has to be confirmed before preoperative appearance of colorectal metastases to the liver can be recommended as a simple and helpful predictor of outcome for various treatment strategies.

We are grateful to Health Care of Australia, Service of Mayne Nickless Ltd, Sydney, New South Wales, for supporting our unit.

Dr Gruenberger is a hepatobiliary fellow, supported by Novartis Nutrition Research AG, Neuenegg, Switzerland.

Reprints: David Lawson Morris, UNSW, Department of Surgery, St George Hospital, Pitney Clinical Science Building, Level 3, Kogarah, Sydney, NSW 2217, Australia (e-mail: David.Morris@unsw.edu.au).