Formal Hepatic Resection of Colorectal Liver Metastases Ploidy and Prognosis
D. SCOTT LIND, M.D.,* GEORGE A. PARKER, M.D.,* J. SHELTON HORSLEY, III, M.D.,-t MICHAEL J. KORNSTEIN, M.D.,4 JAMES P. NEIFELD, M.D.,* HARRY D. BEAR, M.D., PH.D.,* and WALTER LAWRENCE, JR., M.D.*
Fifty consecutive patients who underwent 52 formal hepatic resections (excluding isolated wedge resections) for metastatic colorectal cancer were analyzed to determine whether DNA content was of prognostic significance. The Dukes' stages of the colorectal primaries were: A (10%), B (20%), C (40%), D (28%), and unknown in 2%. Four patients whose liver metastases were discovered at the time of resection of the primary bowel cancer underwent concomitant liver resection, and the remaining patients underwent delayed resections. The hepatic resections performed were right lobectomy (50%), extended right lobectomy (19%), left lobectomy (13%), left lateral segmentectomy (6%), left lobectomy and right wedge (6%), extended left lobectomy (4%), and right lobectomy and left wedge (2%). The overall morbidity rate was 29%. The in-hospital mortality rate was 9%. As of November 1991, 36 patients have recurred. The 5-year actuarial survival was 28%. Flow cytometry could be performed on 37 archival specimens, 15 of which were found to be diploid whereas 22 were aneuploid. All metastases from Dukes A colorectal primaries demonstrated a diploid DNA content. In addition, there was no difference in actuarial survival between diploid and aneuploid tumors. These data suggest that in selected patients, formal hepatic resection of colorectal liver metastases can be performed with an acceptable morbidity rate, mortality rate, and survival, but ploidy of the resected tumor is not of prognostic significance.
T| s HE LIVER IS a frequent and often the only site of metastases from colorectal cancer. As many as 10% of patients with colorectal cancer present with concomitant liver metastases at the time of detection oftheir primary tumor and, in many patients who undergo
presumed curative resection of their colorectal primary, the liver is a major site of recurrence. In the last decade, the number of resections ofcolorectal liver metastases has increased dramatically, and yet there remains no consenPresented at the 103rd Annual Scientific Session of the Southern Surgical Association, Hot Springs, Virginia, December 1-5, 1991. Address reprint requests to George A. Parker, M.D., Division of Surgical Oncology, Department of Surgery, Medical College of Virginia, Virginia Commonwealth University, Box 1 1, MCV Station, Richmond, VA 23298-001 1. Accepted for publication January 30, 1992.
677
From the Division of Surgical Oncology of the Department of Surgery,* The Ella Gordon Valentine American Cancer Society Professor of Clinical Oncology,t and the Department of Pathology,f Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia
sus among surgeons regarding the appropriate indications for resection.
Evidence is accumulating that flow cytometric analysis of DNA content has prognostic significance in many malignancies.' Relatively few studies have examined the prognostic significance of ploidy in colorectal liver metastases.2'3 Ploidy may provide a reliable, objective indicator of tumor biology that may identify patients with neoplasms who are less likely to benefit from surgical intervention. In addition, now that chemotherapy has been shown to have a modest effect on the natural history of colorectal cancer,4 future protocols of hepatic resection will undoubtedly include chemotherapy arms. Properly designed clinical trials require knowledge about important prognostic variables. Therefore, we sought to review our results with formal hepatic resection of colorectal liver metastases and to determine if DNA content of the resected metastases was of prognostic significance. Materials and Methods Patients The records of all patients who underwent formal hepatic resection ofcolorectal liver metastases at the Medical College of Virginia from January 1980 to January 1991 were reviewed. Isolated wedge resections were excluded from analysis. There were a total of 50 patients who underwent 52 formal resections with curative intent. The female:male ratio was 1.2:1, and the ages ranged from 37 to 81 years, with a median of 56 years. The Dukes' stages of the colorectal primaries were: A in 5 patients, B in 10
LIND AND OTHERS
678
patients, C in 20 patients, D in 14 patients, and unknown in 1 patient. Hepatic recurrence was detected by a rising carcinoembryonic antigen level in 30 patients, whereas four patients were symptomatic, prompting computed tomography (CT scan). One patient presented with jaundice, whereas in another patient hepatic recurrence was detected at the time of laparotomy for a small bowel obstruction. Preoperative evaluation consisted of physical examination and standard laboratory and radiologic assessment. Hepatic imaging consisted primarily of contrastenhanced CT scan and, more recently, dynamic CT scan. If the CT scan suggested the need for a major hepatic resection (i.e., lobectomy or extended lobectomy), hepatic angiography was performed to delineate hepatic arterial anatomy. In the absence of severe comorbid disease, obvious extrahepatic disease, or more than three hepatic metastases, laparotomy was advised. Intraoperative sonography was routinely performed at the time of exploration to identify lesions not seen on CT scan as well as to demonstrate hepatic venous anatomy. The extent of hepatic resection performed appears in Table 1. Of the 14 patients whose liver metastases were discovered at the time of resection of their primary bowel cancer (i.e., synchronous metastases), four underwent concomitant liver resection. Two patients with Dukes stage D disease at presentation had wedge resections of liver metastases performed concomitant with resection of the colorectal primary at an outside institution. They subsequently developed hepatic recurrence, prompting referral for formal hepatic resection. In the remaining eight patients with synchronous metastases, formal hepatic resection was delayed 1 to 2 months after their bowel surgery. In those patients presenting with metachronous liver metastases, the interval from bowel surgery to liver resection ranged from 2 to 88 months, with a median of20 months. Patient follow-up after hepatic resection was determined by the individual surgeon. Only two patients received adjuvant chemotherapy after hepatic resection. Preparation of Single-cell Suspension All resected liver specimens were previously fixed in formalin. Approximately four to five 50-tm sections were prepared from the paraffin-embedded blocks (Before TABLE 1. Extent of Hepatic Resection
Hepatic Resection Right lobectomy Extended right lobectomy Left lobectomy Left lateral segmentectomy Left lobectomy and right wedge Extended left lobectomy Right lobectomy and left wedge
No. of Patients 26 (50%) 10 (19%) 7 (13%) 3 (6%) 3 (6%) 2 (4%) 1 (2%)
Ann. Surg. * June 1992
preparation, a section of each block was examined pathologically to verify the presence of tumor.). The tissue then was deparaffinized in xylene followed by progressive rehydration in ethanol at decreasing concentrations. The tissue was then minced into small pieces, resuspended in 0.5% pepsin solution, and incubated for 2 hours at 37 C. After centrifugation at 1200 rpm for 5 minutes, the pellet was resuspended in RPMI 1640 culture media (Whittaker Labs, Walkersville, MD), filtered through 40-,um nylon mesh to remove clumps, and counted to assure recovery of> 1.0 X 105 total cells. The cells were again centrifuged, washed, and stored overnight in RPMI- 1640 at 2 to 8 C. Cell Staining and DNA Analysis After storage, the cells were first centrifuged at 1200 rpm for 5 minutes, then the pellet was resuspended in phosphate-buffered saline (1.0 X 106 cells per 0.4 mL) and incubated in a 0.003% trypsin solution for 10 minutes. This was followed by the addition of a mixture of 2.5% trypsin inhibitor and 0.5% ribonuclease A and incubation for an additional 10 minutes. A solution of 2.5% propidium iodide and 5.8% spermine tetrachloride (Sigma Chemical Co., St. Louis, MO) then was added directly to the cells on ice in light-free conditions. After 10 minutes, the cells were centrifuged at 1200 rpm for 5 minutes and resuspended in 1 mL RPMI. Cellular DNA was measured using a flow cytometer (FACSCAN, Becton Dickinson, Mountainveiw, CA). Data were analyzed using RFIT DNA software (Becton Dickinson). When possible, flow cytometric analysis also was run on normal liver as a control. Definitions Flow cytometric analysis of specimens was performed by a single pathologist who was unaware of the clinical outcome of the patients. Tumors were considered diploid if the DNA histogram displayed a single GO/GI peak, whereas aneuploid tumors demonstrated more than one GO/GI peak. Statistical Analysis Statistical analysis was done using the SAS statistical software package (SAS Institution, Cary, NC). Survival curves were generated using the Kaplan-Meier method.5 Statistical significance between survival curves was determined by the Wilcoxon log-rank test.6 Results Morbidity and Mortality Rates There were 16 complications, for a 29% morbidity rate. Not unexpectedly, the most frequent complication was a collection of bile (biloma). The in-hospital mortality rate
Vol. 215 * NO. 6
RESECTION OF COLORECTAL LIVER METASTASES
9%. The complications and causes of death appear in Table 2. The median estimated blood loss was 3000 mL (range, 700 to 15,000 mL) and the median hospital stay was 16 days (range, 7 to 38 days). Patients who underwent liver resections concomitant with removal of the primary tumor and those in whom the estimated blood loss exceeded 3 L tended to have more complications, although this did not reach statistical significance (p = 0.051). There was no relationship between age or extent of hepatic resection and frequency of complications.
TABLE 3. Pathologic Characteristics
was
Pathology of Hepatic Metastases
The pathologic characteristics ofthe resected specimens in Table 3. Sixty-two per cent of patients had a single metastasis after pathologic examination of the resected specimen. Grossly, no patient had more than three metastases, but three resected specimens did exhibit tumor satellites surrounding a single metastasis. The size of lesions ranged from 0.5 to 15.0 cm, with a median of 6.2 cm. The pathologic margins were free of tumor in 81% of patients, microscopically involved in 17%, and unknown in 2%.
679
Characteristic
No. of Patients
Microscopic margins Negative Positive Unknown No. of metastases Solitary Two Three Size of metastases 6 cm Distribution of metastases Unilobar Bilobar
42 (81%) 9 (17%) 1 (2%) 32 (62%) 12 (23%) 8 (15%) 27 (52%) 25 (48%) 46 (88%) 6 (12%)
appears
Pattern and Treatment of Recurrence As of November 1991, 36 patients have recurred. Recurrence was determined by clinical, radiologic, and pathologic means. The locations of recurrences appear in
Table 4. Most of those patients in whom the margins of resection were microscopically involved had recurrences in the liver. Of those 15 patients in whom the liver alone was the site of first recurrence, 10 were re-explored and five were reresected. Table 5 illustrates the characteristics of the reresected patients. One patient in whom the brain was the sole site of first recurrence, and another patient in whom the lung was the sole site offirst recurrence, were treated surgically. One patient who recurred in the liver, but was unresectable at re-exploration, was treated with cryotherapy, whereas another was treated with combined
No. of Patients
Nonfatal
Biloma/abscess
8
Pleural effusion Pneumothorax Deep vein thrombosis
4
Splenectomy Coagulopathy Liver failure Myocardial infarction Necrotizing fasciitis Total
Survival As of November 1991, 36 patients have died: five were operative deaths and one patient died of pneumonia 3 months after operation with no evidence of disease. The remaining 30 patients died of disease. Fourteen patients are alive: five with disease, and the remaining nine have no evidence of disease. The median follow-up of those patients alive without any evidence of disease is 41.5 months. The actuarial survival curve appears in Figure 1.
Ploidy of Resected Hepatic Metastases
Paraffin-embedded blocks were recovered for 37 patients. DNA histograms produced from tissue fixed before 1983 were excluded secondary to fixative-induced artifact. DNA histograms representative of diploid and aneuploid specimens appear in Figure 2. The median coefficient of variance of all DNA histograms was 4.3 (range, 0.5 to 9.3). The effect of DNA content on actuarial survival appears in Figure 3. Discussion The natural history of colorectal liver metastases is not one of uniform demise within 6 months, but one of vary-
TABLE 2. Morbidity and Mortality Rates
Complication
interstitial and external beam radiotherapy and chemotherapy. An additional 10 patients who had recurrences were treated with chemotherapy.
Fatal TABLE 4. Pattern of Recurrence
Site
1 I I
1
Liver I 2 I
-
16/55 (29%)
5/55 (9%)
Lung Brain Multiple sites (including liver) Multiple sites (excluding liver) Total
No. of Patients 15 7 2 7
5 36
680
LIND AND OTHERS
Ann. Surg. * June 1992
TABLE 5. Hepatic Reresections
First Resection
Second Resection
Patient
Type of Resection
Disease-free Interval
Type of Resection
Status
I 2 3
Left lateral segment Left lateral segment Extended right lobectomy Left lobectomy Extended left
19 mo 11 mo 26 mo
Right lobectomy Left lobectomy Wedge left lobe
DOD 26 mo DOD 22 mo (liver negative) DOD 34 mo (liver negative)
13 mo 31 mo
Wedge right lobe Wedge right lobe
NED 18 mo NED 46 mo
4 5
lobectomy DOD, dead of disease; NED, no evidence of disease.
ing survival depending primarily on tumor biology and extent of liver involvement. Better knowledge of liver anatomy as well as improved surgical techniques and perioperative care have resulted in a marked increase in the number of hepatic resections performed for colorectal liver metastases. Many argue that the results attributed to surgery merely reflect tumor biology rather than surgical success.7 Certainly those patients who undergo resection represent a select group. Our experience echoes the fact that preoperative staging is notoriously inaccurate. Imaging studies tend to grossly underestimate the number of liver metastases and are even less sensitive at detecting intra-abdominal extrahepatic disease.8 More sophisticated and sensitive imaging studies are becoming available, but their specificity is not yet sufficient to deny patients laparotomy.9 Exploration allows more accurate staging of the extent of disease. Therefore, survival comparisons made between operative and nonoperative groups are inappropriate. In addition, those patients considered as candidates for exploration inherently have a better prognosis simply because of less comorbid disease. Undoubtedly, surgical selection plays a role in ultimate outcome,
but in the absence of any other effective therapy or a controlled randomized trial, resection remains the treatment of choice for selected patients with hepatic colorectal metastases.
rD a
DIPLOID
3
a.
1.0 INB Z >0.9 0.8
a
> 0.7
a
0
'a Z 0.5 2 o.s 0 10.4o 0.30- 0 CL
I
ANEUPLOID
3 Lb a
0.2-
0.1 *
0.01 0
J
5
10 15 20 25 30 35 40 45 50 55 60
MONTHS FIG. 1. Actuarial survival of patients undergoing formal hepatic resection of colorectal liver metastases.
I
a
209
400
Clio
rL2-R
I
I
999
FIG. 2. Representative DNA histograms of resected metastases.
Vol. 215 * No. 6
RESECTION OF COLORIECTAL LIVER METASTASES
1.0
0.9
DIPLOIID (N-1S)
z 0.8
5;D U)
---- ANZUPILOID (N=22)
:~~~~~~~I - -
0.7
II
0.6
z 0.5
Ir 0 a. 0.4 .
0~
0.3
0~
0.2
. * ...
.
.
35
40
.~~~~~~~~~
0.1 o.o
0
5
10
15 20 25
30
45
50
55
60
MONTHS FIG. 3. Actuarial survival of patients undergoing formal hepatic resection of colorectal liver metastases: diploid versus aneuploid.
The selection criteria for resection of hepatic metastases from colorectal carcinoma remain controversial. At one extreme, some advocate a "swiss cheese" approach, resecting up to 13 separate liver metastases, and claim improved survival.'0 Others prefer to "watch and wait," allowing those who may not benefit from resection to declare themselves." This biologic selection process may improve survival statistics, but also may deny some patients a potentially curative resection. Many of the generally accepted criteria for resection have been generated from pooled data.'2 Typically, we favor exploration in patients without evidence of extrahepatic disease and in whom there are less than four liver metastases documented by preoperative imaging. Occasionally a more aggressive surgical approach is justified. For example, a patient with solitary liver and lung metastases may benefit from staged resections. The operative technique of hepatic resection varies widely. Recently a variety of sophisticated devices have been brought into the operating room to aid the surgeon performing hepatic resection.13 No method has been proven superior, and we prefer finger fracture with meticulous hemostasis. One technique we have found useful is intraoperative ultrasound. Previously we have reported that intraoperative ultrasound was the most sensitive indicator of the number of lesions present in the liver and affected operative decision making in 49% of liver resections for both primary and metastatic tumors.'4 The amount of liver to resect is also controversial. Some advocate wedge resection, others formal hepatic resection, and still others favor segmentectomy.'5 Often it is safer to perform a formal resection rather then a large wedge resection, but the paramount principle is to remove the lesion with a adequate margin.
681
Although multi-institutional studies are valuable, it is impossible to standardize surgery and therefore it is imperative for single institutions to periodically review their results. Our morbidity rate of 29% is comparable to that in published reports,12 and when one considers that this study excluded isolated wedge resections, a 9% mortality rate is not unreasonable. In addition, most of the in-hospital deaths were in the first few years of the study, and there have been no operative deaths in the last 4 years. There is evidence to suggest that patients with synchronous metastases who undergo liver resection concomitant with resection of the colorectal primary have impaired survival compared with those who undergo delayed resection.'6 Although it is difficult to sort out selection factors with respect to survival in these patients, our data at least suggest an increased morbidity rate with concomitant formal hepatic resection. Most of these patients were in the first few years of the study. We now favor a policy of concomitant wedge resection or left lateral segmentectomy, but delay formal hepatic resection until the patient has recovered from his or her bowel surgery. The finding that intraoperative blood loss of more than 3 L correlates with postoperative complications has been reported by others'7 and may be related simply to the extent of resection. This assumes greater importance if one considers other reports of a detrimental effect of perioperative transfusion,'8 and more recently hypotension, on ultimate survival. '9 The patterns of recurrence in this study are similar to those reported by others.20 Some of these recurrences represent occult disease not detectable at the time of liver resection. The liver alone was a site of recurrence in 42% of cases. Liver recurrence either alone or as a part of disseminated disease was more common when the pathologic margins of the hepatic resection were positive and in the presence of bilobar metastases. Those patients with recurrence confined to the liver and believed to be suitable candidates for reresection were reexplored. Five of these patients were reresected, and two patients are alive without evidence of disease 18 and 48 months after the hepatic reresection. Although these numbers are small, combined with other reports21'22 they suggest that reresection of colorectal liver metastases is also beneficial in selected patients. The frequency of isolated liver recurrences raises the question whether regional adjuvant therapy might have any impact. Although it suffers from small numbers, one prospective randomized study recently reported an increased time to treatment failure in patients with solitary metastases that were resected and received intra-arterial FUdR after operation. Furthermore, none of those patients had recurrences in the liver.23 This, together with the data presented here in which the lung and multiple sites (excluding liver) were frequent sites of recurrence,
682
LIND AND OTHERS
emphasize the need for effective systemic adjuvant therapy. The 5-year survival rate of 28% is similar to that in other published results. Some data suggest a plateau in the survival curve at 5 years,'2 whereas others argue that these patients are never cured in that their survival curves never approach those of age-matched controls.24 Although our data appeared to plateau, we did have a few late recurrences, emphasizing the need for long-term follow-up. Several studies have found flow cytometric analysis of tumor DNA ploidy to be of prognostic significance in a number of malignancies.25 Kokal et al.26 found ploidy to be the single most important variable in predicting both recurrence and death from disease in primary colorectal cancer. Others argue that DNA content provides little prognostic information over and above routine clinical and pathologic parameters.27 Although there is good correlation between fresh and archival samples with respect to ploidy,28 many studies fail to retrieve material on all patients, and therefore the data may not be representative of the entire population. Small sample size, limited followup, and methodologic variation between institutions represent additional potential sources of error. In spite of these shortcomings, ploidy may offer an objective measurement of tumor biology. There are few data concerning the prognostic significance of DNA content of hepatic metastases from colorectal cancer. In a study of 65 hepatic metastases from large bowel cancer, Yamaguchi et al.3 found DNA ploidy patterns to be a useful indicator of prognosis when subjected to multivariate analysis. Kokal et al.,2 however, found that DNA content of 53 liver metastases from colorectal cancer bore no relationship to survival or response to therapy. Ten per cent of patients in this study had Dukes A colorectal primary tumors, and, in spite of this favorable pathology, developed hepatic metastases. All of these colorectal primaries were diploid by flow cytometry. This small number together with a failure to analyze a similar cohort of patients with Dukes A lesions who failed to develop metastases make any valid statistical analysis impossible. When we examined the relationship between ploidy and disease-free and overall survival, our data are consistent with those of Kokal et al.2 There was no statistically significant difference in actuarial survival with respect to DNA content of the resected hepatic colorectal metastases. Although we were unable to accurately analyze the DNA histograms of specimens fixed before 1983, overall survival of these patients was representative of the original cohort. In addition, therapy of treatment failures (i.e., those that recurred after hepatic resection) was similar between diploid and aneuploid groups. Although this study suffers from small numbers, it suggests that once distant metastases are present, ploidy has no effect on survival. Other studies support the contention that in more advanced
Ann.
Surg. * June 1992
stages of disease tumor ploidy loses any prognostic significance.29 Our experience suggests that in selected patients, formal hepatic resection of colorectal liver metastases can be performed with acceptable morbidity, mortality, and survival rates, but DNA content of the resected tumor is not of prognostic with respect to survival. Unfortunately, patterns of recurrence stress the fact that surgery remains a local therapeutic modality, and more effective systemic therapy is desperately needed.
Acknowledgments The authors thank Judith Luck, C.T. (ASCP) for technical assistance, and Jeretha Jones and Luke Wolfe for their valuable contributions in the preparation of this manuscript.
References 1. Merkel DE, McGuire WL. Ploidy, proliferative activity and prognosis. Cancer 1990; 65:1194-1205. 2. Kokal WA, Duda RB, Azumi N, et al. Tumor DNA content in primary and metastatic colorectal carcinoma. Arch Surg 1986; 121: 1434-1439. 3. Yamaguchi A, Ishida T, Takegawa S, et al. Flow cytometric analysis of colorectal cancer with hepatic metastases and its relationship to metastatic characteristics and prognosis. Oncology 1990; 47: 478-482 4. NIH Consensus Conference. Adjuvant therapy for patients with colon and rectal cancer. JAMA 1990; 264:1444-1450. 5. Kaplan EL, Meier P. Nonparametric estimation from incomplete observation. J Am Stat Assoc 1958; 53:457-481. 6. Gehan E. A generalized Wilcoxon test for comparing arbitrarily single censored samples. Biometrika 1965; 52:203. 7. Silen W. Hepatic resection of for metastases from colorectal carcinoma is of dubious value. Arch Surg 1989; 124:1021-1022. 8. Steele G Jr, Bleday R, Mayer RJ, et al. A prospective evaluation of hepatic resection for colorectal carcinoma metastases to the liver: Gastrointestinal Tumor Study Group Protocol 6584. J Clin Oncol 1991; 9:1105-1112. 9. Steele G Jr, Ravikumar TS, Benotti PN. New surgical treatments for recurrent colorectal cancer. Cancer 1990; 65:723-730. 10. Minton JP, Hamilton WB, Sardi A, et al. Results of surgical excision of one to 13 hepatic metastases in 98 consecutive patients. Arch Surg 1989; 124:46-48. 11. Cady B, Stone MD. The role of surgical resection of liver metastases in colorectal carcinoma. Semin Oncol 1991; 18:399-406. 12. Hughes KS, Simon R, Songhorabodi S, et al. Resection of the liver for colorectal carcinoma metastases: a multi-institutional study of indications for resection. Surgery 1988; 103:278-287. 13. Schoder T, Hasselgren PO, Brackett K, et al. Techniques of liver resection. Arch Surg 1987; 122:1166-1171. 14. Parker GA, Lawrence W Jr, Horsley JS, et al. Intraoperative ultrasound of the liver affects operative decision making. Ann Surg 1989; 209:569-577. 15. Franco D, Smadja C, Kahwaji F, et al. Segmentectomies in the management of liver tumors. Arch Surg 1988; 123:519-522. 16. Schlag P, Hohenberger P, Herfarth C. Resection of liver metastases in colorectal cancer: competitive analysis of treatment results in synchronous versus metachronous metastases. Eur J Surg Oncol 1990; 16:360-365. 17. Vetto JT, Hughes KS, Rosenstein R, Sugarbaker PH. Morbidity and mortality of hepatic resection for metastastic colorectal cancer. Dis Colon Rectum 1990; 33:408-413. 18. Stephenson KR, Steinberg SM, Hughes KS, et al. Perioperative blood transfusions are associated with decreased time to recurrence and
Vol. 215 * No. 6
RESECTION OF COLORECTAL LIVER METASTASES
683
decreased survival after resection of colorectal liver metastases. Ann Surg 1988; 208:679-687. Younes RN, Rogatko A, Brennan MF. The influence of intraoperative hypotension and perioperative blood transfusion on diseasefree survival in patients with complete resection of colorectal liver metastases. Ann Surg 1991; 214:107-113. Hohenberger P, Schlag P, Schwarz V, Herfarth C. Tumor recurrence and options for further treatment after resection of liver metastases in patients with colorectal cancer. J Surg Oncol 1990; 44:245251. Griffith KD, Sugarbaker PH, Chang AE. Repeat hepatic resections for colorectal metastases. Surgery 1990; 107:101-104. Stone MD, Cady B, Jenkins RL, et al. Surgical therapy for recurrent liver metastases from colorectal cancer. Arch Surg 1990; 125: 718-722. Wagman LD, Kemeny MM, Leong L, et al. A randomized evaluation of the treatment of colorectal cancer metastatic to the liver. J Clin Oncol 1990; 8:1885-1893.
24. Adson MA. Resection of liver metastases: when is it worthwhile? World J Surg 1987; 11:511-520. 25. Seckinger D, Sugarbaker E, Franfort 0. DNA content in human cancer: application in pathology and clinical medicine. Arch Pathol Lab Med 1989; 113:619-626. 26. Kokal WA, Gardine RL, Sheibani K, et al. Tumor DNA content in resectable primary colorectal carcinoma. Ann Surg 1989; 209: 188-193. 27. Armitage NC, Robins RA, Evans DF, et al. Tumor cell DNA content in colorectal cancer and its relationship to survival. Br J Surg 1985; 72:828-830. 28. Kallioniemi OP. Comparison of fresh and paraffin-embedded tissue as starting material for DNA flow cytometry and evaluation of intratumoral heterogeneity. Cytometry 1988; 9:164-169. 29. Finan PJ, Quirke P, Dixon MF, et al. Is DNA aneuploidy a good prognostic indicator in patients with advanced colorectal cancer. Br J Cancer 1986; 54:327-330.
DISCUSSION DR. R. SCOTT JONES (Charlottesville, Virginia): I want to compliment Dr. Parker and his colleagues for an excellent piece of work. This obviously is a fairly large series of patients accumulated over a short period and treated quite expertly. And I think that Dr. Parker should be congratulated. I think much of the information that he presented would certainly support clearly the information that has accumulated from a number of institutions and is currently in the literature. And I think the new thing that he looks at that really deserves comment is the issue of what is the role of looking at DNA as a prognostic indicator. And the data that he showed us would suggest it is not particularly helpful, but I would like to suggest that there are some different ways of looking at this. Particularly, with the data that Dr. Parker has, he could look at the nature of the primary tumors. He has the opportunity to look at the tumors that were metastatic in the liver, which I think was information he showed us today. And he also has an opportunity to look at recurrent tumors after liver metastases. Now the reason I make that point is that it is theoretically possible that that information may change from one of those populations to the other. And it would be interesting to know, and you have the resources available to take a look at that. I think it would be especially interesting to know what the tumors looked like in those patients who were reresected. It might be interesting that those patients constitute a special subpopulation. I wanted to make one other point about flow cytometry and liver tumors. It was just a coincidence I happened to have on my desk some information when I talked to Dr. Parker about his work. There is a publication from Japan by Fujimoto. They studied a large number, 203 patients, who had hepatocellular carcinoma, in which all were resected. They did the same kind of study on these hepatocellular carcinomas. And they found that the flow cytometry was a significant prognostic indicator in that population of patients. I think it is interesting to draw the contrast between that series of patients who had primary hepatocellular carcinomas and in whom DNA content was significantly predictive of survival and Dr. Parker's population of patients of metastatic tumors in whom it was not. So I think that there is probably some room for further investigation in this issue of the biologic nature of the material in terms of prognosis. In any event, I want to support Dr. Parker and his colleagues in their work. I think they are really doing excellent work in this field, and it was a pleasure to hear your presentation, Dr. Parker, and also to read your manuscript. Thank you very much.
parallels what has been reported in the literature. And also, I think you point out that there is a failure rate of 70%, and this is an area that we have to address. I have some questions. First of all, George, in your examination of the margin, you had 80% clear margins, but you also had about 20% positive margins. Do you take any steps to avoid getting a positive margin? I have been dismayed myself to find that after resection when I give it to the pathologist, he will frequently say, "Oh, this margin is close." And I will go back and redo it. So now I ink all our specimens, and I just wondered is this of value. You mentioned that there was not a survival difference in the group that had positive margins, but I just wondered if you look at that more carefully whether or not there is some subtle change. I think Kevin Hughes has shown that the ones who did the best in their natural history survival registry were the ones that did have clear margins. The second question I would ask is, are there some negative selection factors? Are there some patients on whom we actually should not operate? For example, the patient who has a short free interval whose primary tumor was an aggressive Dukes C. I am personally always concerned about that group and would wonder if it is of value to try to resect those patients. Thirdly, there is a problem sometimes with blood loss in these patients. Is there value in using something like the cell saver? Again, I want to thank the Association for the privilege of the floor and the opportunity to hear this very nice paper.
19.
20.
21.
22. 23.
DR. HAROLD J. WANEBO (Providence, Rhode Island): President Ochsner, Dr. Jones, I also want to compliment Dr. Parker and his group for both an excellent presentation and a nice review and their effort to look at ploidy. It was interesting that he showed that some patients with Dukes' A cancer can metastasize. I think we are all aware of that, but sometimes we forget that point. I think your 5-year survival rate basically
DR. LESTER WILLIAMS (Nashville, Tennessee): Dr. Ochsner, Dr. Jones, I tried to decline the request to discuss this paper on the basis that I was a reluctant supporter of the general concept. Dr. Parker convinced me that I should make that point even though it would be less than supportive. Thus, I have seen the manuscript and I did agree to discuss it. Before going to Nashville, I was exposed to the philosophy of Blake Cady, Bill McDermott and Bill Silen and that is the basis for saying I might be less than supportive. Their philosophy was that the patients who fulfill the usually accepted criteria to have the liver resected usually have a small bulk of tumor confined to the liver. By that observation alone in almost all cases, they have a biologically favorable tumor. And the issue for us is not that the patient lived longer than appropriate controls, but lived longer than the "expected death within months," derived from historic controls. And so they have been suggesting that the best criterion for a biologically favorable tumor was the fact that it could be resected. In fact, in this manuscript that is true for most of these patients, as it is for most other reports. We now know that ploidy does not help us much. This question is, can we pick out favorable patients? And if we cannot, should we begin to look at adjuvant therapy and accept that the patient who becomes a candidate for this kind of surgery is by definition-although we do not understand why-biologically different than most. And as such, are there routine adjuvant ways that we should be treating these patients? Thank you for the opportunity to discuss it.
D. SCOTT LIND, M.D.,* GEORGE A. PARKER, M.D.,* J. SHELTON HORSLEY, III, M.D.,-t MICHAEL J. KORNSTEIN, M.D.,4 JAMES P. NEIFELD, M.D.,* HARRY D. BEAR, M.D., PH.D.,* and WALTER LAWRENCE, JR., M.D.*
Fifty consecutive patients who underwent 52 formal hepatic resections (excluding isolated wedge resections) for metastatic colorectal cancer were analyzed to determine whether DNA content was of prognostic significance. The Dukes' stages of the colorectal primaries were: A (10%), B (20%), C (40%), D (28%), and unknown in 2%. Four patients whose liver metastases were discovered at the time of resection of the primary bowel cancer underwent concomitant liver resection, and the remaining patients underwent delayed resections. The hepatic resections performed were right lobectomy (50%), extended right lobectomy (19%), left lobectomy (13%), left lateral segmentectomy (6%), left lobectomy and right wedge (6%), extended left lobectomy (4%), and right lobectomy and left wedge (2%). The overall morbidity rate was 29%. The in-hospital mortality rate was 9%. As of November 1991, 36 patients have recurred. The 5-year actuarial survival was 28%. Flow cytometry could be performed on 37 archival specimens, 15 of which were found to be diploid whereas 22 were aneuploid. All metastases from Dukes A colorectal primaries demonstrated a diploid DNA content. In addition, there was no difference in actuarial survival between diploid and aneuploid tumors. These data suggest that in selected patients, formal hepatic resection of colorectal liver metastases can be performed with an acceptable morbidity rate, mortality rate, and survival, but ploidy of the resected tumor is not of prognostic significance.
T| s HE LIVER IS a frequent and often the only site of metastases from colorectal cancer. As many as 10% of patients with colorectal cancer present with concomitant liver metastases at the time of detection oftheir primary tumor and, in many patients who undergo
presumed curative resection of their colorectal primary, the liver is a major site of recurrence. In the last decade, the number of resections ofcolorectal liver metastases has increased dramatically, and yet there remains no consenPresented at the 103rd Annual Scientific Session of the Southern Surgical Association, Hot Springs, Virginia, December 1-5, 1991. Address reprint requests to George A. Parker, M.D., Division of Surgical Oncology, Department of Surgery, Medical College of Virginia, Virginia Commonwealth University, Box 1 1, MCV Station, Richmond, VA 23298-001 1. Accepted for publication January 30, 1992.
677
From the Division of Surgical Oncology of the Department of Surgery,* The Ella Gordon Valentine American Cancer Society Professor of Clinical Oncology,t and the Department of Pathology,f Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia
sus among surgeons regarding the appropriate indications for resection.
Evidence is accumulating that flow cytometric analysis of DNA content has prognostic significance in many malignancies.' Relatively few studies have examined the prognostic significance of ploidy in colorectal liver metastases.2'3 Ploidy may provide a reliable, objective indicator of tumor biology that may identify patients with neoplasms who are less likely to benefit from surgical intervention. In addition, now that chemotherapy has been shown to have a modest effect on the natural history of colorectal cancer,4 future protocols of hepatic resection will undoubtedly include chemotherapy arms. Properly designed clinical trials require knowledge about important prognostic variables. Therefore, we sought to review our results with formal hepatic resection of colorectal liver metastases and to determine if DNA content of the resected metastases was of prognostic significance. Materials and Methods Patients The records of all patients who underwent formal hepatic resection ofcolorectal liver metastases at the Medical College of Virginia from January 1980 to January 1991 were reviewed. Isolated wedge resections were excluded from analysis. There were a total of 50 patients who underwent 52 formal resections with curative intent. The female:male ratio was 1.2:1, and the ages ranged from 37 to 81 years, with a median of 56 years. The Dukes' stages of the colorectal primaries were: A in 5 patients, B in 10
LIND AND OTHERS
678
patients, C in 20 patients, D in 14 patients, and unknown in 1 patient. Hepatic recurrence was detected by a rising carcinoembryonic antigen level in 30 patients, whereas four patients were symptomatic, prompting computed tomography (CT scan). One patient presented with jaundice, whereas in another patient hepatic recurrence was detected at the time of laparotomy for a small bowel obstruction. Preoperative evaluation consisted of physical examination and standard laboratory and radiologic assessment. Hepatic imaging consisted primarily of contrastenhanced CT scan and, more recently, dynamic CT scan. If the CT scan suggested the need for a major hepatic resection (i.e., lobectomy or extended lobectomy), hepatic angiography was performed to delineate hepatic arterial anatomy. In the absence of severe comorbid disease, obvious extrahepatic disease, or more than three hepatic metastases, laparotomy was advised. Intraoperative sonography was routinely performed at the time of exploration to identify lesions not seen on CT scan as well as to demonstrate hepatic venous anatomy. The extent of hepatic resection performed appears in Table 1. Of the 14 patients whose liver metastases were discovered at the time of resection of their primary bowel cancer (i.e., synchronous metastases), four underwent concomitant liver resection. Two patients with Dukes stage D disease at presentation had wedge resections of liver metastases performed concomitant with resection of the colorectal primary at an outside institution. They subsequently developed hepatic recurrence, prompting referral for formal hepatic resection. In the remaining eight patients with synchronous metastases, formal hepatic resection was delayed 1 to 2 months after their bowel surgery. In those patients presenting with metachronous liver metastases, the interval from bowel surgery to liver resection ranged from 2 to 88 months, with a median of20 months. Patient follow-up after hepatic resection was determined by the individual surgeon. Only two patients received adjuvant chemotherapy after hepatic resection. Preparation of Single-cell Suspension All resected liver specimens were previously fixed in formalin. Approximately four to five 50-tm sections were prepared from the paraffin-embedded blocks (Before TABLE 1. Extent of Hepatic Resection
Hepatic Resection Right lobectomy Extended right lobectomy Left lobectomy Left lateral segmentectomy Left lobectomy and right wedge Extended left lobectomy Right lobectomy and left wedge
No. of Patients 26 (50%) 10 (19%) 7 (13%) 3 (6%) 3 (6%) 2 (4%) 1 (2%)
Ann. Surg. * June 1992
preparation, a section of each block was examined pathologically to verify the presence of tumor.). The tissue then was deparaffinized in xylene followed by progressive rehydration in ethanol at decreasing concentrations. The tissue was then minced into small pieces, resuspended in 0.5% pepsin solution, and incubated for 2 hours at 37 C. After centrifugation at 1200 rpm for 5 minutes, the pellet was resuspended in RPMI 1640 culture media (Whittaker Labs, Walkersville, MD), filtered through 40-,um nylon mesh to remove clumps, and counted to assure recovery of> 1.0 X 105 total cells. The cells were again centrifuged, washed, and stored overnight in RPMI- 1640 at 2 to 8 C. Cell Staining and DNA Analysis After storage, the cells were first centrifuged at 1200 rpm for 5 minutes, then the pellet was resuspended in phosphate-buffered saline (1.0 X 106 cells per 0.4 mL) and incubated in a 0.003% trypsin solution for 10 minutes. This was followed by the addition of a mixture of 2.5% trypsin inhibitor and 0.5% ribonuclease A and incubation for an additional 10 minutes. A solution of 2.5% propidium iodide and 5.8% spermine tetrachloride (Sigma Chemical Co., St. Louis, MO) then was added directly to the cells on ice in light-free conditions. After 10 minutes, the cells were centrifuged at 1200 rpm for 5 minutes and resuspended in 1 mL RPMI. Cellular DNA was measured using a flow cytometer (FACSCAN, Becton Dickinson, Mountainveiw, CA). Data were analyzed using RFIT DNA software (Becton Dickinson). When possible, flow cytometric analysis also was run on normal liver as a control. Definitions Flow cytometric analysis of specimens was performed by a single pathologist who was unaware of the clinical outcome of the patients. Tumors were considered diploid if the DNA histogram displayed a single GO/GI peak, whereas aneuploid tumors demonstrated more than one GO/GI peak. Statistical Analysis Statistical analysis was done using the SAS statistical software package (SAS Institution, Cary, NC). Survival curves were generated using the Kaplan-Meier method.5 Statistical significance between survival curves was determined by the Wilcoxon log-rank test.6 Results Morbidity and Mortality Rates There were 16 complications, for a 29% morbidity rate. Not unexpectedly, the most frequent complication was a collection of bile (biloma). The in-hospital mortality rate
Vol. 215 * NO. 6
RESECTION OF COLORECTAL LIVER METASTASES
9%. The complications and causes of death appear in Table 2. The median estimated blood loss was 3000 mL (range, 700 to 15,000 mL) and the median hospital stay was 16 days (range, 7 to 38 days). Patients who underwent liver resections concomitant with removal of the primary tumor and those in whom the estimated blood loss exceeded 3 L tended to have more complications, although this did not reach statistical significance (p = 0.051). There was no relationship between age or extent of hepatic resection and frequency of complications.
TABLE 3. Pathologic Characteristics
was
Pathology of Hepatic Metastases
The pathologic characteristics ofthe resected specimens in Table 3. Sixty-two per cent of patients had a single metastasis after pathologic examination of the resected specimen. Grossly, no patient had more than three metastases, but three resected specimens did exhibit tumor satellites surrounding a single metastasis. The size of lesions ranged from 0.5 to 15.0 cm, with a median of 6.2 cm. The pathologic margins were free of tumor in 81% of patients, microscopically involved in 17%, and unknown in 2%.
679
Characteristic
No. of Patients
Microscopic margins Negative Positive Unknown No. of metastases Solitary Two Three Size of metastases 6 cm Distribution of metastases Unilobar Bilobar
42 (81%) 9 (17%) 1 (2%) 32 (62%) 12 (23%) 8 (15%) 27 (52%) 25 (48%) 46 (88%) 6 (12%)
appears
Pattern and Treatment of Recurrence As of November 1991, 36 patients have recurred. Recurrence was determined by clinical, radiologic, and pathologic means. The locations of recurrences appear in
Table 4. Most of those patients in whom the margins of resection were microscopically involved had recurrences in the liver. Of those 15 patients in whom the liver alone was the site of first recurrence, 10 were re-explored and five were reresected. Table 5 illustrates the characteristics of the reresected patients. One patient in whom the brain was the sole site of first recurrence, and another patient in whom the lung was the sole site offirst recurrence, were treated surgically. One patient who recurred in the liver, but was unresectable at re-exploration, was treated with cryotherapy, whereas another was treated with combined
No. of Patients
Nonfatal
Biloma/abscess
8
Pleural effusion Pneumothorax Deep vein thrombosis
4
Splenectomy Coagulopathy Liver failure Myocardial infarction Necrotizing fasciitis Total
Survival As of November 1991, 36 patients have died: five were operative deaths and one patient died of pneumonia 3 months after operation with no evidence of disease. The remaining 30 patients died of disease. Fourteen patients are alive: five with disease, and the remaining nine have no evidence of disease. The median follow-up of those patients alive without any evidence of disease is 41.5 months. The actuarial survival curve appears in Figure 1.
Ploidy of Resected Hepatic Metastases
Paraffin-embedded blocks were recovered for 37 patients. DNA histograms produced from tissue fixed before 1983 were excluded secondary to fixative-induced artifact. DNA histograms representative of diploid and aneuploid specimens appear in Figure 2. The median coefficient of variance of all DNA histograms was 4.3 (range, 0.5 to 9.3). The effect of DNA content on actuarial survival appears in Figure 3. Discussion The natural history of colorectal liver metastases is not one of uniform demise within 6 months, but one of vary-
TABLE 2. Morbidity and Mortality Rates
Complication
interstitial and external beam radiotherapy and chemotherapy. An additional 10 patients who had recurrences were treated with chemotherapy.
Fatal TABLE 4. Pattern of Recurrence
Site
1 I I
1
Liver I 2 I
-
16/55 (29%)
5/55 (9%)
Lung Brain Multiple sites (including liver) Multiple sites (excluding liver) Total
No. of Patients 15 7 2 7
5 36
680
LIND AND OTHERS
Ann. Surg. * June 1992
TABLE 5. Hepatic Reresections
First Resection
Second Resection
Patient
Type of Resection
Disease-free Interval
Type of Resection
Status
I 2 3
Left lateral segment Left lateral segment Extended right lobectomy Left lobectomy Extended left
19 mo 11 mo 26 mo
Right lobectomy Left lobectomy Wedge left lobe
DOD 26 mo DOD 22 mo (liver negative) DOD 34 mo (liver negative)
13 mo 31 mo
Wedge right lobe Wedge right lobe
NED 18 mo NED 46 mo
4 5
lobectomy DOD, dead of disease; NED, no evidence of disease.
ing survival depending primarily on tumor biology and extent of liver involvement. Better knowledge of liver anatomy as well as improved surgical techniques and perioperative care have resulted in a marked increase in the number of hepatic resections performed for colorectal liver metastases. Many argue that the results attributed to surgery merely reflect tumor biology rather than surgical success.7 Certainly those patients who undergo resection represent a select group. Our experience echoes the fact that preoperative staging is notoriously inaccurate. Imaging studies tend to grossly underestimate the number of liver metastases and are even less sensitive at detecting intra-abdominal extrahepatic disease.8 More sophisticated and sensitive imaging studies are becoming available, but their specificity is not yet sufficient to deny patients laparotomy.9 Exploration allows more accurate staging of the extent of disease. Therefore, survival comparisons made between operative and nonoperative groups are inappropriate. In addition, those patients considered as candidates for exploration inherently have a better prognosis simply because of less comorbid disease. Undoubtedly, surgical selection plays a role in ultimate outcome,
but in the absence of any other effective therapy or a controlled randomized trial, resection remains the treatment of choice for selected patients with hepatic colorectal metastases.
rD a
DIPLOID
3
a.
1.0 INB Z >0.9 0.8
a
> 0.7
a
0
'a Z 0.5 2 o.s 0 10.4o 0.30- 0 CL
I
ANEUPLOID
3 Lb a
0.2-
0.1 *
0.01 0
J
5
10 15 20 25 30 35 40 45 50 55 60
MONTHS FIG. 1. Actuarial survival of patients undergoing formal hepatic resection of colorectal liver metastases.
I
a
209
400
Clio
rL2-R
I
I
999
FIG. 2. Representative DNA histograms of resected metastases.
Vol. 215 * No. 6
RESECTION OF COLORIECTAL LIVER METASTASES
1.0
0.9
DIPLOIID (N-1S)
z 0.8
5;D U)
---- ANZUPILOID (N=22)
:~~~~~~~I - -
0.7
II
0.6
z 0.5
Ir 0 a. 0.4 .
0~
0.3
0~
0.2
. * ...
.
.
35
40
.~~~~~~~~~
0.1 o.o
0
5
10
15 20 25
30
45
50
55
60
MONTHS FIG. 3. Actuarial survival of patients undergoing formal hepatic resection of colorectal liver metastases: diploid versus aneuploid.
The selection criteria for resection of hepatic metastases from colorectal carcinoma remain controversial. At one extreme, some advocate a "swiss cheese" approach, resecting up to 13 separate liver metastases, and claim improved survival.'0 Others prefer to "watch and wait," allowing those who may not benefit from resection to declare themselves." This biologic selection process may improve survival statistics, but also may deny some patients a potentially curative resection. Many of the generally accepted criteria for resection have been generated from pooled data.'2 Typically, we favor exploration in patients without evidence of extrahepatic disease and in whom there are less than four liver metastases documented by preoperative imaging. Occasionally a more aggressive surgical approach is justified. For example, a patient with solitary liver and lung metastases may benefit from staged resections. The operative technique of hepatic resection varies widely. Recently a variety of sophisticated devices have been brought into the operating room to aid the surgeon performing hepatic resection.13 No method has been proven superior, and we prefer finger fracture with meticulous hemostasis. One technique we have found useful is intraoperative ultrasound. Previously we have reported that intraoperative ultrasound was the most sensitive indicator of the number of lesions present in the liver and affected operative decision making in 49% of liver resections for both primary and metastatic tumors.'4 The amount of liver to resect is also controversial. Some advocate wedge resection, others formal hepatic resection, and still others favor segmentectomy.'5 Often it is safer to perform a formal resection rather then a large wedge resection, but the paramount principle is to remove the lesion with a adequate margin.
681
Although multi-institutional studies are valuable, it is impossible to standardize surgery and therefore it is imperative for single institutions to periodically review their results. Our morbidity rate of 29% is comparable to that in published reports,12 and when one considers that this study excluded isolated wedge resections, a 9% mortality rate is not unreasonable. In addition, most of the in-hospital deaths were in the first few years of the study, and there have been no operative deaths in the last 4 years. There is evidence to suggest that patients with synchronous metastases who undergo liver resection concomitant with resection of the colorectal primary have impaired survival compared with those who undergo delayed resection.'6 Although it is difficult to sort out selection factors with respect to survival in these patients, our data at least suggest an increased morbidity rate with concomitant formal hepatic resection. Most of these patients were in the first few years of the study. We now favor a policy of concomitant wedge resection or left lateral segmentectomy, but delay formal hepatic resection until the patient has recovered from his or her bowel surgery. The finding that intraoperative blood loss of more than 3 L correlates with postoperative complications has been reported by others'7 and may be related simply to the extent of resection. This assumes greater importance if one considers other reports of a detrimental effect of perioperative transfusion,'8 and more recently hypotension, on ultimate survival. '9 The patterns of recurrence in this study are similar to those reported by others.20 Some of these recurrences represent occult disease not detectable at the time of liver resection. The liver alone was a site of recurrence in 42% of cases. Liver recurrence either alone or as a part of disseminated disease was more common when the pathologic margins of the hepatic resection were positive and in the presence of bilobar metastases. Those patients with recurrence confined to the liver and believed to be suitable candidates for reresection were reexplored. Five of these patients were reresected, and two patients are alive without evidence of disease 18 and 48 months after the hepatic reresection. Although these numbers are small, combined with other reports21'22 they suggest that reresection of colorectal liver metastases is also beneficial in selected patients. The frequency of isolated liver recurrences raises the question whether regional adjuvant therapy might have any impact. Although it suffers from small numbers, one prospective randomized study recently reported an increased time to treatment failure in patients with solitary metastases that were resected and received intra-arterial FUdR after operation. Furthermore, none of those patients had recurrences in the liver.23 This, together with the data presented here in which the lung and multiple sites (excluding liver) were frequent sites of recurrence,
682
LIND AND OTHERS
emphasize the need for effective systemic adjuvant therapy. The 5-year survival rate of 28% is similar to that in other published results. Some data suggest a plateau in the survival curve at 5 years,'2 whereas others argue that these patients are never cured in that their survival curves never approach those of age-matched controls.24 Although our data appeared to plateau, we did have a few late recurrences, emphasizing the need for long-term follow-up. Several studies have found flow cytometric analysis of tumor DNA ploidy to be of prognostic significance in a number of malignancies.25 Kokal et al.26 found ploidy to be the single most important variable in predicting both recurrence and death from disease in primary colorectal cancer. Others argue that DNA content provides little prognostic information over and above routine clinical and pathologic parameters.27 Although there is good correlation between fresh and archival samples with respect to ploidy,28 many studies fail to retrieve material on all patients, and therefore the data may not be representative of the entire population. Small sample size, limited followup, and methodologic variation between institutions represent additional potential sources of error. In spite of these shortcomings, ploidy may offer an objective measurement of tumor biology. There are few data concerning the prognostic significance of DNA content of hepatic metastases from colorectal cancer. In a study of 65 hepatic metastases from large bowel cancer, Yamaguchi et al.3 found DNA ploidy patterns to be a useful indicator of prognosis when subjected to multivariate analysis. Kokal et al.,2 however, found that DNA content of 53 liver metastases from colorectal cancer bore no relationship to survival or response to therapy. Ten per cent of patients in this study had Dukes A colorectal primary tumors, and, in spite of this favorable pathology, developed hepatic metastases. All of these colorectal primaries were diploid by flow cytometry. This small number together with a failure to analyze a similar cohort of patients with Dukes A lesions who failed to develop metastases make any valid statistical analysis impossible. When we examined the relationship between ploidy and disease-free and overall survival, our data are consistent with those of Kokal et al.2 There was no statistically significant difference in actuarial survival with respect to DNA content of the resected hepatic colorectal metastases. Although we were unable to accurately analyze the DNA histograms of specimens fixed before 1983, overall survival of these patients was representative of the original cohort. In addition, therapy of treatment failures (i.e., those that recurred after hepatic resection) was similar between diploid and aneuploid groups. Although this study suffers from small numbers, it suggests that once distant metastases are present, ploidy has no effect on survival. Other studies support the contention that in more advanced
Ann.
Surg. * June 1992
stages of disease tumor ploidy loses any prognostic significance.29 Our experience suggests that in selected patients, formal hepatic resection of colorectal liver metastases can be performed with acceptable morbidity, mortality, and survival rates, but DNA content of the resected tumor is not of prognostic with respect to survival. Unfortunately, patterns of recurrence stress the fact that surgery remains a local therapeutic modality, and more effective systemic therapy is desperately needed.
Acknowledgments The authors thank Judith Luck, C.T. (ASCP) for technical assistance, and Jeretha Jones and Luke Wolfe for their valuable contributions in the preparation of this manuscript.
References 1. Merkel DE, McGuire WL. Ploidy, proliferative activity and prognosis. Cancer 1990; 65:1194-1205. 2. Kokal WA, Duda RB, Azumi N, et al. Tumor DNA content in primary and metastatic colorectal carcinoma. Arch Surg 1986; 121: 1434-1439. 3. Yamaguchi A, Ishida T, Takegawa S, et al. Flow cytometric analysis of colorectal cancer with hepatic metastases and its relationship to metastatic characteristics and prognosis. Oncology 1990; 47: 478-482 4. NIH Consensus Conference. Adjuvant therapy for patients with colon and rectal cancer. JAMA 1990; 264:1444-1450. 5. Kaplan EL, Meier P. Nonparametric estimation from incomplete observation. J Am Stat Assoc 1958; 53:457-481. 6. Gehan E. A generalized Wilcoxon test for comparing arbitrarily single censored samples. Biometrika 1965; 52:203. 7. Silen W. Hepatic resection of for metastases from colorectal carcinoma is of dubious value. Arch Surg 1989; 124:1021-1022. 8. Steele G Jr, Bleday R, Mayer RJ, et al. A prospective evaluation of hepatic resection for colorectal carcinoma metastases to the liver: Gastrointestinal Tumor Study Group Protocol 6584. J Clin Oncol 1991; 9:1105-1112. 9. Steele G Jr, Ravikumar TS, Benotti PN. New surgical treatments for recurrent colorectal cancer. Cancer 1990; 65:723-730. 10. Minton JP, Hamilton WB, Sardi A, et al. Results of surgical excision of one to 13 hepatic metastases in 98 consecutive patients. Arch Surg 1989; 124:46-48. 11. Cady B, Stone MD. The role of surgical resection of liver metastases in colorectal carcinoma. Semin Oncol 1991; 18:399-406. 12. Hughes KS, Simon R, Songhorabodi S, et al. Resection of the liver for colorectal carcinoma metastases: a multi-institutional study of indications for resection. Surgery 1988; 103:278-287. 13. Schoder T, Hasselgren PO, Brackett K, et al. Techniques of liver resection. Arch Surg 1987; 122:1166-1171. 14. Parker GA, Lawrence W Jr, Horsley JS, et al. Intraoperative ultrasound of the liver affects operative decision making. Ann Surg 1989; 209:569-577. 15. Franco D, Smadja C, Kahwaji F, et al. Segmentectomies in the management of liver tumors. Arch Surg 1988; 123:519-522. 16. Schlag P, Hohenberger P, Herfarth C. Resection of liver metastases in colorectal cancer: competitive analysis of treatment results in synchronous versus metachronous metastases. Eur J Surg Oncol 1990; 16:360-365. 17. Vetto JT, Hughes KS, Rosenstein R, Sugarbaker PH. Morbidity and mortality of hepatic resection for metastastic colorectal cancer. Dis Colon Rectum 1990; 33:408-413. 18. Stephenson KR, Steinberg SM, Hughes KS, et al. Perioperative blood transfusions are associated with decreased time to recurrence and
Vol. 215 * No. 6
RESECTION OF COLORECTAL LIVER METASTASES
683
decreased survival after resection of colorectal liver metastases. Ann Surg 1988; 208:679-687. Younes RN, Rogatko A, Brennan MF. The influence of intraoperative hypotension and perioperative blood transfusion on diseasefree survival in patients with complete resection of colorectal liver metastases. Ann Surg 1991; 214:107-113. Hohenberger P, Schlag P, Schwarz V, Herfarth C. Tumor recurrence and options for further treatment after resection of liver metastases in patients with colorectal cancer. J Surg Oncol 1990; 44:245251. Griffith KD, Sugarbaker PH, Chang AE. Repeat hepatic resections for colorectal metastases. Surgery 1990; 107:101-104. Stone MD, Cady B, Jenkins RL, et al. Surgical therapy for recurrent liver metastases from colorectal cancer. Arch Surg 1990; 125: 718-722. Wagman LD, Kemeny MM, Leong L, et al. A randomized evaluation of the treatment of colorectal cancer metastatic to the liver. J Clin Oncol 1990; 8:1885-1893.
24. Adson MA. Resection of liver metastases: when is it worthwhile? World J Surg 1987; 11:511-520. 25. Seckinger D, Sugarbaker E, Franfort 0. DNA content in human cancer: application in pathology and clinical medicine. Arch Pathol Lab Med 1989; 113:619-626. 26. Kokal WA, Gardine RL, Sheibani K, et al. Tumor DNA content in resectable primary colorectal carcinoma. Ann Surg 1989; 209: 188-193. 27. Armitage NC, Robins RA, Evans DF, et al. Tumor cell DNA content in colorectal cancer and its relationship to survival. Br J Surg 1985; 72:828-830. 28. Kallioniemi OP. Comparison of fresh and paraffin-embedded tissue as starting material for DNA flow cytometry and evaluation of intratumoral heterogeneity. Cytometry 1988; 9:164-169. 29. Finan PJ, Quirke P, Dixon MF, et al. Is DNA aneuploidy a good prognostic indicator in patients with advanced colorectal cancer. Br J Cancer 1986; 54:327-330.
DISCUSSION DR. R. SCOTT JONES (Charlottesville, Virginia): I want to compliment Dr. Parker and his colleagues for an excellent piece of work. This obviously is a fairly large series of patients accumulated over a short period and treated quite expertly. And I think that Dr. Parker should be congratulated. I think much of the information that he presented would certainly support clearly the information that has accumulated from a number of institutions and is currently in the literature. And I think the new thing that he looks at that really deserves comment is the issue of what is the role of looking at DNA as a prognostic indicator. And the data that he showed us would suggest it is not particularly helpful, but I would like to suggest that there are some different ways of looking at this. Particularly, with the data that Dr. Parker has, he could look at the nature of the primary tumors. He has the opportunity to look at the tumors that were metastatic in the liver, which I think was information he showed us today. And he also has an opportunity to look at recurrent tumors after liver metastases. Now the reason I make that point is that it is theoretically possible that that information may change from one of those populations to the other. And it would be interesting to know, and you have the resources available to take a look at that. I think it would be especially interesting to know what the tumors looked like in those patients who were reresected. It might be interesting that those patients constitute a special subpopulation. I wanted to make one other point about flow cytometry and liver tumors. It was just a coincidence I happened to have on my desk some information when I talked to Dr. Parker about his work. There is a publication from Japan by Fujimoto. They studied a large number, 203 patients, who had hepatocellular carcinoma, in which all were resected. They did the same kind of study on these hepatocellular carcinomas. And they found that the flow cytometry was a significant prognostic indicator in that population of patients. I think it is interesting to draw the contrast between that series of patients who had primary hepatocellular carcinomas and in whom DNA content was significantly predictive of survival and Dr. Parker's population of patients of metastatic tumors in whom it was not. So I think that there is probably some room for further investigation in this issue of the biologic nature of the material in terms of prognosis. In any event, I want to support Dr. Parker and his colleagues in their work. I think they are really doing excellent work in this field, and it was a pleasure to hear your presentation, Dr. Parker, and also to read your manuscript. Thank you very much.
parallels what has been reported in the literature. And also, I think you point out that there is a failure rate of 70%, and this is an area that we have to address. I have some questions. First of all, George, in your examination of the margin, you had 80% clear margins, but you also had about 20% positive margins. Do you take any steps to avoid getting a positive margin? I have been dismayed myself to find that after resection when I give it to the pathologist, he will frequently say, "Oh, this margin is close." And I will go back and redo it. So now I ink all our specimens, and I just wondered is this of value. You mentioned that there was not a survival difference in the group that had positive margins, but I just wondered if you look at that more carefully whether or not there is some subtle change. I think Kevin Hughes has shown that the ones who did the best in their natural history survival registry were the ones that did have clear margins. The second question I would ask is, are there some negative selection factors? Are there some patients on whom we actually should not operate? For example, the patient who has a short free interval whose primary tumor was an aggressive Dukes C. I am personally always concerned about that group and would wonder if it is of value to try to resect those patients. Thirdly, there is a problem sometimes with blood loss in these patients. Is there value in using something like the cell saver? Again, I want to thank the Association for the privilege of the floor and the opportunity to hear this very nice paper.
19.
20.
21.
22. 23.
DR. HAROLD J. WANEBO (Providence, Rhode Island): President Ochsner, Dr. Jones, I also want to compliment Dr. Parker and his group for both an excellent presentation and a nice review and their effort to look at ploidy. It was interesting that he showed that some patients with Dukes' A cancer can metastasize. I think we are all aware of that, but sometimes we forget that point. I think your 5-year survival rate basically
DR. LESTER WILLIAMS (Nashville, Tennessee): Dr. Ochsner, Dr. Jones, I tried to decline the request to discuss this paper on the basis that I was a reluctant supporter of the general concept. Dr. Parker convinced me that I should make that point even though it would be less than supportive. Thus, I have seen the manuscript and I did agree to discuss it. Before going to Nashville, I was exposed to the philosophy of Blake Cady, Bill McDermott and Bill Silen and that is the basis for saying I might be less than supportive. Their philosophy was that the patients who fulfill the usually accepted criteria to have the liver resected usually have a small bulk of tumor confined to the liver. By that observation alone in almost all cases, they have a biologically favorable tumor. And the issue for us is not that the patient lived longer than appropriate controls, but lived longer than the "expected death within months," derived from historic controls. And so they have been suggesting that the best criterion for a biologically favorable tumor was the fact that it could be resected. In fact, in this manuscript that is true for most of these patients, as it is for most other reports. We now know that ploidy does not help us much. This question is, can we pick out favorable patients? And if we cannot, should we begin to look at adjuvant therapy and accept that the patient who becomes a candidate for this kind of surgery is by definition-although we do not understand why-biologically different than most. And as such, are there routine adjuvant ways that we should be treating these patients? Thank you for the opportunity to discuss it.
