- Email: [email protected]
Pelvic exenteration for locally advanced rectal carcinoma: Factors predicting improved survival
Volume 121
Number 5
SURGERY MAY
Original
1997
communications
Pelvic exenteration for locally advanced rectal carcinoma: Factors predicting improved surviva Sarkis H. Mete&Sian, Adel K. El-Naggar,
MD,
MD,a John M. PhD, Kenneth
Skibber, MD, Geoffrey G. Giacco, MS, R. Hess, PhD, and Tyvin A. Rich, MD,b
Houston, Texas
Background. The purpose of this retrospective review was to determine whether a number of clinicopatholo@c factors (age, gender, type of exenteration, tumor extent, adjuvant therapy, tumor DNA ploidy, and Sphase fraction) that could be determined before operation were useful in predicting survival in patients undergoing pelvic exenteration for rectal cancer. Methods. Between 1983 and 1992, 40 patients (15 male and 25 female) at our institution underwent pelvic exenteration for rectal adenocnrcinoma in which tumor-free pathologic marginswere obtained. Twenty-nine patients presented with primary tumors; 11 had recurrent disease. A total exenteration was performed in 20 patients, posterior exenterution in 18 patients, and an anterior exenteration in 2 patients. Results. By multivariate (Cox proportional hazards repession) analysis, age, preoperative chemoradiation therapy, and an S phase of IO 70 or greater were found to be significant predictors of survival. Age older than 55 years was associated with a relative risk for cancer-related death (RR) of 0.13 (p = 0.02), and ch,emoradiation had an RR of 0.05 (p = O.Ol), indicating their beneficial effect. An Sphase fraction of 10% or greater had an RR of 16.97 (p = 0.03), indicating a poor survival. The clinicopathologic factors listed above were used to derive a prognostic index (PI). A PI of less than 1.37 was associated with a 5-year survival rate of 65% (low risk), whereas patients with a PI of 1.37 or greater had a S-year survival rate of 20% (high risk) (p = 0.005). Conclusions.These results indicate that adjuvant chemoradiation may significantly improve survival in patients who requirepelvic exenteration for resection of locally advanced rectal carcinoma.An Sphase fraction of 10% or greater is also predictive of a poor outcome. Use of these factors allowed the generation of a PI that identifies high- and low-risk patients. Consideration of the ability to deliver chemoradiation and the determinates of the tumor Sphase fraction in patients requiring pelvic exenteration for rectal cancer may be helpful in predicting outcome and planning therapy. (Surgery 1997;121:479-87.) From the Departments of Surgical Oncology, Biomathmetics, Medical Infmrnatics and Radiotherapy, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
Accepted
for publication
Reprint requests: John cology-Box 106, The Center, 1515 Holcombe “Currently Quebec,
at McGill Canada.
Oct.
14, 1996.
M. Skibber, MD, Department of Surgical OnUniversity of Texas M. D. Anderson Cancer Blvd., Houston TX 77030. University,
Royal
Victoria
Hospital,
bCurrently lottestille,
at the University va.
Copyright
0 1997
by Mosby
of Virginia
Year Book,
Health
Sciences
Center,
Char-
Inc.
Montreal, 0039-6060/97/$5.00tO
11/56/79606
SURGERY
479
480
Metevissian
et al.
6% to 10% of rectal cancers are locally advanced at presentation’ and require extensive surgery for complete tumor extirpation. Pelvic exenteration involving en bloc removal of the rectum, bladder, distal ureters, and reproductive organs can be required to obtain negative margins of resection. The ability to accomplish tumor-free resection margins is the most important treatment factor in determining patient outcome. When tumor-free margins cannot be obtained, local failure rates are high and survival is short. The presence of residual tumor after resection is an overwhelming factor in a poor outcome. No matter how extensive, resections that leave known residual disease cannot be considered curative. This operation described by Brunshwig2 and Appleby3 for the treatment of advanced or recurrent cervical carcinoma and advanced rectal carcinoma, respectively, is associated with significant morbidity and mortality. Nevertheless, a number of studies’, 4s have demonstrated a 5-year survival rate ranging from 33% to 50% for these selected patients with rectal cancer. In addition, this operation decreases cancer-related pelvic morbidities such as fistulas and ureteral or bowel ob struction. These problems are difficult to manage and lead to an agonizing death for the patient. However, it is difficult to justify the performance of this procedure for palliation alone. A major development in the multidisciplinary management of locally advanced rectal cancers is the use of preoperative radiotherapy.7-10 This treatment has resulted in an improved resectability rate, with 55% to 75% of patients being converted to a resectable stat~s.~~ Despite this improvement, overall recurrence rates after pelvic exenteration remain unacceptably high even with pathologically negative margins. Currently, the only reliable markers of prognosis remain the depth of tumor penetration and the lymph node status. Therefore a search for biologic markers that may predict outcome and guide the use of preoperative adjuvant therapy is needed to aid in the selection of patients most likely to benefit from this extensive surgical procedure. The purpose of this retrospective review was to determine the value of pretreatment DNA flow cytometry data as a prognostic marker12* l3 and the impact on survival of using preoperative chemoradiation on patients requiring pelvic exenteration for locally advanced rectal carcinoma in whom tumor-free margins were obtained. These parameters were used to derive a prognostic index (PI) to identify high-risk patients. These high-risk patients may be candidates for more intense neoadjuvant management, novel treatment strategies, or the selection of palliative nonsurgical therapy. APPROXIMATELY
MATERIAL
AND
METHODS
Patients. Between 1983 and 1992,40 patients underwent pelvic exenteration for rectal adenocarcinoma at
su%FY May 1997
The University of Texas M. D. Anderson Cancer Center. The clinical characteristics of these patients are summarized in Table I. Patients were operated on by multiple staff surgeons during this period. Twenty-nine patients presented with primary tumors, and 11 presented with recurrent tumors. A total pelvic exenteration, defined as the removal of the rectum with the bladder and prostate in men or the gynecologic organs in women, was performed in 20 patients. A posterior pelvic exenteration, defined as removal of only the female gynecologic organs in addition to proctectomy, was performed in 18 patients. In two patients intestinal continuity was reestablished by means of performance of a low rectal anastomosis; thus by definition an anterior exenteration was performed. All patients in this study had tumor-free final pathologic margins. Patients with microscopic or gross tumor at resection margins are excluded. Charts were reviewed to obtain the following information: age, gender, tumor size, stage (TNM), and adjuvant therapy. The tumor size and extent of intramural penetration were derived from the pathology report. Although all patients were deemed to have locally advanced disease at operation as a result of adherence to pelvic organs and pretreatment examination and imaging, 10 were found to have T3 tumors on final pathology examination after preoperative chemoradiation. Adjuvant therapy. Ten patients received only radiotherapy (three before operation and seven after operation), with doses ranging from 45 to 58 Gy (mean, 49 Gy) , Seventeen patients received concurrent preoperative chemoradiation treatments. In all cases 5-fluorouracil was used as a radiosensitizer at a dose of 300 mg/ m2. Preoperative radiotherapy doses ranged from 45 to 60 Gy (mean, 52.25 Gy). The use of only preoperative radiation in 10 patients occurred between 1981 and 1988. Although the use of chemoradiation occurred in 17 patients between 1989 and 1992,12 patients who did not receive adjuvant therapy were seen throughout the study period. Flow cytometric technique. Pretreatment biopsy tissue was fixed in 10% formaldehyde for 2 to 8 hours and embedded in paraf!dn (Paraplast Monoject Scientific, St. Louis, MO) with a melting point of 56” C. Nuclear suspensions of the neoplasms were prepared by using a modified version of the method of Hedley et all4 and McLemore et a1.15 In each block selected for analysis at least 75% of the tissue was neoplastic. In a given neoplasm, tissue blocks with more than 10% necrosis were avoided when possible. Otherwise, the necrotic areas were mapped and nonnecrotic tissue was reembedded for flow cytometric analyses. Two or more 50 pm sections were cut on a Reichet-Jung 2030 microtome (Cambridge Instruments, Heidelberg, Germany) and
Surge y Volume 121, Number 5
Meterissian
et al.
481
0
20
40
60
80
100
120
140
180
180
200
Months
Fig. 1. Overall survival, deaths from colorectal
placed into 15 x 100 mm glass culture tubes (Corning Glass Works, Corning, NY). The sections were deparaffinized three times for 15 minutes each in tubes containing 3, 4, and 5 ml Histo-Solve x (Biochemical Sciences, Bridgeport, NJ). Rehydration was performed by exposing sections to graded ethyl alcohol: 100% twice, 95% twice, 80% once, 70% once, and 50% overnight. Final hydration was achieved by placing specimens into Dulbecco s phosphate-buffered saline solution twice (Gibco Laboratories, Grand Island, NY). The specimens were incubated for 30 minutes at 37” C in prewarmed 0.5 % pepsin at pH of 1.5. The enzymatic digestion was stopped with 175 pl of cold 250 mg/ml pepstatin A (Sigma Chemical Co., St. Louis, MO). It was then disaggregated with 3 ml syringes and 18-gauge needles and filtered through 37 pm nylon mesh filters (Small Parts Inc., Miami, FL). DNA staining. The concentration of nuclei per milliliterwas established by using an automated cell counter (Coulter 2131; Coulter Electronics, Hialeah, FL), and dilutions were made to achieve a lo6 cells/ml concentration. A 900 ~1 aliquot of the final dilution was treated with 100 p1 0.25 mg/ml ribonuclease A (Worthington, Biochemical, Freehold, NJ) for 30 minutes at 37” C. The tubes containing these cells were then frozen and stained with 50 ~10.25% propidium iodide. Staining was performed at least 20 minutes before analysis, and the entire run was completed within 3 hours. Specimens were analyzed on an Epics Profile I (Epics Division, Coulter Electronics) equipped with an argon laser operating at 488 nm, with a 610 long-pass filter and a 488 band pass. Flow rates were adjusted to count approximately 75 nuclei/set. Peak versus integral signals were used to gate out doublets. CelI cycle and DNA ploidy analyses. Cytospin slides were prepared from a lo6 nuclei/ml concentration (unexposed to ribonuclease) and stained with WrightGiemsa. The slides were reviewed to confirm the pres-
Table
carcinoma.
I. Clinical
characteristics
Charactetistics
n
%
22 18
55 45
17
20
46 54
15 25
37.15 62.5
29 11
72.5 27.5
2 18 20
5 45
10
25 75
Age W
94 >55 Tumor size” (cm)
55 >5 Gender Male Female Tumor status Primary Recurrent Type of exenteration Anterior Posterior Total T stage T3
T4
30
Adjuvant therapy None Radiotherapy alone Chemoradiation DNA ploidy Diploid Aneuploid S-phase fraction 110 210 *Size
unknown
in three
13 10
17 19
21 12 28
50
32.5 25 42.5 47.5 52.5 30 70
patients.
ence of tumor nuclei and monitor the quality of processing. At least 5000 nuclei were evaluated in each sample. The DNA histograms were analyzed by using the sonogram gating procedure of Johnston et all6 Coulter Cytologic software (Version 2.2; Coulter Elec-
SU%eTY
482 Meterissian et al. May
Table
II. Results of univariate
and multivariate
analyses of survival*
Univariade
>55vs
p Value
RR
95%
analysis
554
0.88
0.31, 2.48
0.80
0.13
0.02,
0.96
0.02t
0.52
0.19, 1.38
0.19
3.94
0.44,
35.5
0.22
2.30
0.78, 6.78
0.13
6.36
0.62,
65.3
0.11
0.72
0.22,
2.41
0.60
0.23
0.03,
2.01
0.15
0.35
0.09, 1.35
0.13
0.05
0.01, 0.83
0.01t
1.04
0.37,
2.95
0.93
0.26
0.04,
0.16
2.40
0.54,
10.62
0.25
16.97
1.57
0.53,
4.64
0.41
0.95
regression
CI
Multivariate
RR
*Results from Cm proportional hazards tp < 0.05. fTota1 includes two anterior exenterations.
95%
analysis
Variables
Gender Female vs male Exenteration type Total: vs posterior Treatment Irradiation + surgery vs surgery alone Chemotherapy, irradiation + surgery vs surgery alone DNA ploidy Aneuploid vs diploid S-phase fraction >lO% vs
1997
CI
1.77
p Value
0.93,3.09
0.03t
0.16,
0.95
5.48
analysis.
tronics) was used only in cases exhibiting excessive debris and no evidence of doublets. All tumors were analyzed with cytologic software by using the same debris subtraction model. Nonneoplastic tissue was used as the biologic diploid standard, and the first Go/G1 population was used to denote the diploid stem line. The DNA index was defined as the ratio of the peak channel number of the test sample to the peak channel number of the normal diploid control.1’ By definition, the DNA index of a diploid population is equal to 1.0. A tumor was considered aneuploidwhen a distinct second Go/G1 peak, accounting for at least 10% of the cells analyzed, was present (or any number of such separate peaks). A tetraploid was defined as a distinct and sharp peak at the GYM position with at least 20% of the total events of the histogram. The proliferative fraction was determined as the percentage of cells in the S phase (after debris subtraction, when necessary) according to the method of Baisch et al.i8 The cell volume of the Go/i diploid peaks ranged from 2.5 to 6.8 with a mean of 4.3 2 1.5. Follow-up. Follow-up information was obtained either from the chart or by direct correspondence with the patient or the immediate family. The median follow-up of the survivors was 42 months, and survival was determined by including only disease-specific mortality (i.e., deaths from other causes were censored). No patients were lost to follow-up. Statistical analyses. Overall survival was computed by using the Kaplan-Meier method.’ The Cox-Mantel logrank test was used to compare survival time distribu-
tions. Univariate and multivariate relative risk estimates and corresponding 95% confidence intervals (CIs) were computed by using the Cox proportional hazards regression model.* Analyses were performed with BMDP/ 386 Dynamic, Release 3.0 (BMDP Statistical Software, Inc., Los Angeles, CA) and SPSS for Windows, Release 6.0 (SPSS Inc., Chicago, IL). RESULTS Influence of clinicopathologic parameters on survival. The cumulative survival of the 40 patients in this study is summarized in Fig. 1. The 5-year overall survival was 49% with a median survival of 56 months. Univariate and multivariate Cox proportional hazards regression analyses were performed with a number of clinicopathologic parameters including age, gender, type of exenteration, adjuvant therapy, tumor DNA ploidy, tumor S-phase fraction, and tumor extent at presentation (Table II). By univariate analysis none of these factors significantly influenced survival. Both the type of exenteration (total-anterior versus posterior) and an S phase fraction of 10% or greater were associated with the highest relative risk (RR) of death from rectal cancer (2.3 and 2.4, respectively). In contrast, adjuvant chemoradiation provided a protective effect, decreasing the risk by 35% (RR value). These effects were not statistically significant on univariate analysis, but important effects emerged on multivariate analysis. Age older than 55 years and the administration of adjuvant preoperative chemoradiation were associated
suw? Volume 121, Number 5
Mettissian
et al.
483
Combined Failure Distant Only
* Local
Local
Recurrence
7 Distant Recurrence l Death from Disease
0 Only
0 Censored
31111 8.
No Failure
l
. l
0
50 Time
l
150
100 in Months
from
200
Exenteration
Fig. 2. Patterns of failure after pelvic exenteration.
with a statistically significant beneficial effect (p = 0.02 and p = 0.01, respectively). Our data indicate that age older than 55 years actually appeared to be a protective factor (RR = 0.13; CI = 0.02 to 0.96). Adjuvant preoperative chemoradiation reduced the death rate 20-fold compared with surgery alone (RR = 0.05; CI = 0.01 to 0.83), whereas adjuvant postoperative external-beam radiotherapy produced a fivefold improvement compared with surgery alone (RR = 0.23; CI = 0.03 to 2.01), but this was not statistically significant (p = 0.15). Tumor stage at presentation did not influence survival (RR = 0.95). The need for a total exenteration increased the risk of death sixfold versus posterior exenteration (p = 0.11). Although one might expect women to fare better than men because they are more likely to undergo a posterior exenteration, this was not the case because women had a RR of 3.94 versus that for men (p = 0.22). Comparison of DNA aneuploidy versus diploidy in predicting deaths from rectal cancer was not statistically significant. The apparent univariate effects of gender and DNA ploidy on survival are confounded by the influence of other factors such as the S-phase fraction and adjuvant therapy. Only the S-phase fraction had a significant negative effect on survival, with an Sphase fraction of 10% or greater being associated with a RR of 16.97 (CI = 0.93 to 3.09; p = 0.03). Patterns of failure after pelvic exenteration. The patterns of failure of the 40 patients in this study are summarized graphically in Fig. 2. Nineteen patients had no evidence of local or distant failure at follow-up. Five patients had local failure alone, and one was still alive after nearly 50 months of follow-up. Four patients had both local and distant failure, and all died of disease. Ten patients experienced a distant recurrence without evidence of local failure, and six died of their disease. Four of the patients with only distant recurrence were still alive at the time of this study. One patient, who received preoperative chemoradiation and who had no
Table
*All
III. Survival from pelvic exenteration
Study
No. of cases
5.yr Suroival rate C%)
Lopez et aL4 Estes et aL5 Pearlman et aL6 Kraybill et al.lg Boey et al.” Williams et al.”
24 16” 20 11 49 10
42 49 50 44 38 43
patients
had locoregional
recurrence
before
exenteration.
evidence of residual disease on final pathology examinations and thus qualified as a pathologic complete response, remains free of disease. Prognostic index. With the increasing use of preoperative chemoradiation in the treatment of T3 rectal adenocarcinomas and its observed beneficial effect in this study, an attempt was made to develop a PI to simplify the identification of so-called high-risk patients who would be candidates for additional adjuvant therapy. With the Cox model, the following PI was formulated: PI = 1.37 (gender) - 2.05 (age) - 0.06 (tumor status) + 1.85 (type of exenteration) - 1.46 (treatment) - 2.91 (chemoradiation) + 2.83 (S-phasefraction) - 1.34 (DNA ploidy) , where the parameter values are gender: 0 for male, 1 for female; age: 0 for 554 years, 1 for 254 years; tumor status: 0 for primary, 1 for recurrent; type of exenteration: 0 for posterior, 1 for total or anterior; treatment: 0 for other, 1 for irradiation and surgery; chemoradiation: 0 for other, 1 for irradiation, chemotherapy, and surgery; S-phase fraction: 0 for 9%, 1 for 10%; and DNA ploidy: 0 for diploid, 1 for aneuploid. With this formula, 14 patients with a PI greater than 1.37 had a significantly (p= 0.005) decreased survival rate. Low-risk patients (Pl of 1.37 or less; n = 26) had a 5-year survival rate of 68%, whereas high-risk patients
SUVY
484 Meterissian et al.
May 1997
.80ii .2 5 .80rn _ % ‘3 .40z 3 .20-
0
N=40 Deaths=1
8
12
24 lime
in Months
36 from
48
60
Exenteration
Fig. 3. PI based on COX model. Low-risk patients (PI 5 1.37): 6 deaths in 26 patients; high-risk (PI > 1.37): 9 deaths in 14 patients (p = 0.005). (PI of 1.37 or greater) had a 5-year survival rate of 24% (Fig. 3). Thus this PI formula identified patients at high risk for treatment failure.
DISCUSSION Pelvic exenteration, although a radical surgical procedure, is often the most appropriate treatment available for locally advanced primary and recurrent rectal adenocarcinoma.46, 1g-21 The number of exenterations specifically for rectal cancer included in this study is comparable with the 49 cases reported by Boey et a1.2o The overall 5-year survival rate of the 40 patients in this studywas 49%, which is comparable to the data reported in the literature (Table III). Long-term local control was obtained in 73% of the patients studied. An important issue in the care of these patients is the selection of those who will benefit most from this extensive procedure. Clearly, the first criterion for those who will obtain benefit is the ability to obtain negative margins. The use of preoperative irradiation has been established to improve resectability in locally advanced rectal cancer and to improve outcomes. However, local and distant failures persist. This is a retrospective study whose purpose was to find potential pretreatment factors to predict long-term survival in patients with advanced rectal cancer who required multivisceral resection to obtain tumor-free margins. The purpose is not to address the morbidity or mortality of such resections, although these problems can be substantial. In addition, it does not address the outcome for the patients in whom tumor-free margins are not obtained. Patients with residual tumor in the pelvis have a poor outcome in terms of survival and local progression despite additional radiotherapy. Such resections cannot be considered curative. They were excluded from our study because data from patients with residual disease would add little to our analysis for factors predicting
patients
long-term survival and would confound the analysis of biologic factors. The conclusions drawn from these data are limited by the retrospective nature of the study. Unseen bias in the selection of treatment for individual patients can impact on the observed outcomes. The major factor in selecting the use of preoperative or postoperative radiation appeared to be previous exposure to pelvic irradiation either for a previous rectal primary tumor in those patients with recurrence or for another pelvic malignancy. An uncontrolled study such as this does not allow for the evaluation of the efficacy of treatment plans. By using univariate and multivariate Cox proportional hazards regression analyses, the effect of age, gender, type of exenteration, adjuvant therapy, DNA ploidy, S-phase fraction, and tumor status at presentation (i.e., primary versus recurrent, tumor size) as risk factors for dying of rectal cancer were analyzed. Tumor size was not a predictor of recurrence or survival in our study. Also it makes it unlikely that a bias for giving multimodality therapy to subgroups of patients on the basis of tumor size affected our study. This was previously shown by Ledesma et a1.22 in a series of patients undergoing pelvic exenteration. This was contradicted by Shingleton et al., 23 who found that tumor size was an independent predictor. Tumor size alone has not been accepted as an independent factor predicting survival in patients with rectal cancer. Although older age may be thought of as a contraindication to such radical surgery, the results of this study indicate otherwise. Patients older than 55 years actually fared better than those who were 54 years old or younger (RR = 0.13; p = 0.02)) indicating that older patients can undergo a pelvic exenteration for long-term benefit. The decreased survival of the younger patients is intriguing because it may indicate more aggressive tumor biologic factors.
sv@T Volume
Mete-rksian 121. Number
et al.
485
5
Patients undergoing a posterior exenteration fared better than those undergoing either a total or anterior exenteration, probably because the more advanced tumors required more extensive surgery. The type of exenteration also influenced survival, but the effect was not statistically significant. In a small study Estes et a1.5 also showed that patients who had posterior exenterations achieved a better mean survival than those who had complete exenterations (46 versus 29 months). Clearly the most appropriate operative procedure is one that obtains tumor-free margins. This is related to tumor size, location, and invasion into other organs at presentation. Pelvic exenteration for recurrent disease did not predict a poorer survival, which is discordant with data reported by others. 24 For example, Yeung et a1.25 found long-term survival in only two of 43 patients who underwent pelvic exenteration for recurrent rectal adenocarcinema. Our overall survival results in patients with recurrence could be explained because the 11 patients with recurrent disease in our study were a carefully selected group who were all able to undergo a complete resection with tumor-free margins. Obtaining tumorfree margins can be more difficult in patients operated on for recurrent rectal cancer. Another reason for the survival results could be that the patients with completely resectable recurrent tumors had anastomotic or perianastomotic recurrences, which carry a better prognosis than these involving the pelvic sidewall or sacrum. Vasselopoulos et a1.26 reported a 5-year survival rate of 49% in 15 patients who underwent a complete resection for such a perianastomotic recurrence. On the basis of our data, pelvic exenteration for recurrent disease can be reasonable treatment if negative margins are achieved, especially when patients can receive preoperative chemoradiation. Because most of the clinicopathologic variables studied are weak predictors of survival and the tumor stage is obtained only after operation, there is a need to identify new prognostic factors that can be determined before operation and help guide management. Such factors include DNA ploidy and the S-phase fraction, whi.ch can be measured easily on an intraluminal biopsy specimen before therapy is instituted. Harlow et a1.13 showed that the tumor proliferative activity measured by the Sphase fraction in Dukes class C colon carcinoma was an important biologic factor in determining patient prdgnosis. Other studies have shown that DNA ploidy may be an independent predictor of patient survival in colorectal carcinomas,2”2g particularly diploid colorectal cancers. To date, there have been no published studies looking at the predictive value of tumor DNA ploidy and the tumor S-phase fraction specifically in patients requiring pelvic exenteration for rectal cancer in whom neoadju-
vant therapy has been used. This study shows that with multivariate Cox proportional hazards regression analysis the S-phase fraction is a statistically significant predictor of survival in such patients. Patients with an S-phase fraction of 10% or greater had 16.97 times (CI = 0.93 to 3.09) greater chance of dying of rectal cancer (p = 0.03) than patients with an S-phase fraction less than 10%. Because the S-phase fraction can be determined before operation on biopsy specimens obtained by endoscopy or needle biopsy, it could be used to evaluate risk before treatment. The role of S-phase fraction determination in estimating prognosis has been examined in two other studies in which irradiation has been used. Centeus et a1.30 found that the S-phase fraction predicted the local control of well or moderately differentiated prostate cancers. In another study of head and neck cancers treated by irradiation alone it was reported that tumors with a low S-phase fraction exhibited a higher local recurrence rate. This was postulated to result from the relative resistance to radiotherapy of slowly proliferating cells. These results appear to be at odds with our findings. However, it should be noted that we are dealing with different tumor histologic characteristics, the addition of surgical therapy, and the use ofchemoradiation therapy compared with radiotherapy alone. Studies of the role of the S-phase fraction in the outcome of patients subjected to neoadjuvant therapy for rectal cancer are needed to validate our findings. Preoperative radiotherapy has been used in a number of studies in patients with locally advanced rectal adenocarcinoma with improved resectability and arguably improved survival and recurrence data.7-10 Preop erative chemoradiation is currently administered in our institution to selected patients with T3 rectal carcinomas with high rates of complete pathologic responses and excellent local tumor contro13r As many as 37% of the patients have a complete pathologic response, and an additional 46% experience appreciable downstaging of their tumors. Chemoradiation has been used in locally advanced tumors in hopes of improving the long-term recurrence and survival results. It is reasonable to consider that further expansion of this multidisciplinary neoadjuvant approach should be used in high-risk patients before performing pelvic exenteration for recurrent rectal cancer. In this study preoperative chemoradiation was used in 1’7 patients, and 10 patients received radiotherapy alone. Radiotherapy alone improved survival fivefold, but this effect was not statistically significant compared with the survival of patients receiving surgery alone (Table II). In contrast, preoperative chemoradiation had a dramatic effect. All 17 patients received concurrent intravenous 5-fluorouracil (300 mg/m*) and external-beam radiotherapy (mean dose, 52 Gy) . Early on,
486
Meterissian
SUPY
et al.
May 1997
patients received bolus of 5-fluorouracil, whereas in the latter part of this study, it was given by continuous infirsion. Because of an insufficient number of patients, these two methods of administration could not be directly compared. As seen in Table II, Cox multivariate analysis revealed a 20-fold protection from rectal cancer death in patients who received preoperative chemoradiation (CI = 0.01 to 0.83; p= 0.01) compared with the result for those who received surgery alone. Use of preoperative chemoradiation proved to be a significant prognostic factor in our predictive model. A number of studies”rO have shown improvement in survival in patients with locally advanced rectal carcinoma who received only preoperative radiotherapy. At Massachusetts General Hospital’ the administration of 40 to 50 Gy followed by surgery resulted in a 5-year actuarial survival rate of 28%. In the patients in whom a complete resection was achieved, the 5-year actuarial rate for survival was 43%. Similarly, the groups at Tufts Univers@ and the University of Virginia9 reported 3and 5-year actuarial rates for survival of 54% and 52%, respectively. Mendenhall et al.” found that a complete resection improved the 5-year survival rate from 9% to 20%. The survival data reported in our study can be attributed to the selection of patients for performance of negative margin resections. This study does imply a beneficial effect from preoperative chemoradiation on survival in locally advanced rectal carcinoma; however, the retrospective and uncontrolled nature of the study does not allow definitive comparisons of treatments. Confirmation of this beneficial effect will await reporting data from prospective trials with patients requiring multivis ceral resections for locally advanced rectal cancer. Whether recurrence after exenteration was local or distant, survival was poor (Fig. 2). Distant metastases were documented in 11 of the 15 patients who died of disease after pelvic exenteration, and 10 of these 11 had an S-phase fraction greater than 10% (Fig. 2). Four patients had only a local recurrence at time of death, but because none had an autopsy, the possibility of concomitant distant disease was not ruled out. S-phase fraction of greater than 10% did not predict the likelihood of distant versus local failure. Those at high risk of failure appear unlikely to benefit in terms of survival from pelvic exenteration. The use of a PI may allow patients to be stratified into low- (PI, 1.37 or less) and high-risk groups. Of the 15 deaths, two thirds of the deaths occurred in the highrisk group (PI, greater than 1.37). The use of such an index may allow us to design better studies of treatment plans for such patients. It must be noted, however, that this PI was derived in a study of 40 patients, and to extend our findings, larger groups of patients undergoing pelvic exenteration for rectal cancer must be analyzed.
In summary, this study attempted to identify important pretreatment prognostic factors to define the value of adjuvant therapy in patients requiring pelvic exenteration for rectal cancer. On the basis of a multivariate analysis, it appears that a high proliferative fraction (Sphase, 10% or greater), age older than 55 years, and adjuvant preoperative chemoradiation significantly affected survival in patients undergoing pelvic exenteration. With the multivariate Cox model, a PI was derived to identify low- and high-risk groups, thereby enabling an estimate of prognosis before initiating treatment. Identification of high-risk groups may indicate which patients are candidates for novel treatment approaches or for nonresectable palliation. This study is only a start in addressing the biologic factors of such advanced tumors and how they interact with treatment planning for optimization of outcomes. REFERENCES 1. Lopez
MJ, Monafo
treatment
W. Role
of locally
of extended
advanced
resection
colorectal
in the initial
carcinoma.
Surgery
1993;113:365-72. 2. Brunshwig
A. Complete excision Cancer 1948;1:177-83.
carcinoma.
3. Appleby LH. with ureteral 4. Lopez
viscera
Proctocystectomy: the management transplants. Am J Surg 1950$9:5X60.
MJ, Rraybill
Exenterative it worthwhile?
of pelvic
WG,
Bouney
RS, Johnston
surgey for locally advanced Surgery 1987;102:64451.
for advanced of colostomy
WD,
Bricker
rectosigmoid
EM.
cancers:
Is
5. Estes N, Thomas JH, Jewel1 WR, Beggs D, Harden C. Pelvic exenteration: treatment for failed rectal cancer surgery. Am Surg 1993;59:420-2. 6. Pearlman N, Stiegmann flied
rectal
7. Dosoretz radiation Cancer
cancer.
GV, Donohue Cancer
DE, Gunderson for unresectable 1983;523:8148.
R. Extended
resection
of
1989;63:243841. LL, Hedberg rectal and
S, et al. Preoperative rectosigmoid carcinoma.
ir-
8. Emami B, Willet C, Dilepich M, et al. Effect of preoperative irradiation on resectability of colorectal carcinomas. Int J Radiat Oncol Biol Phys 1989;8:1295-9. 9. Fortier GA, Rrochak RJ, Jung AK, Constable to preoperative irradiation in rectal cancer: cal control
and
complications
surgery and abdominoperineal Phys 1986;12:155963. 10. Mendenhall WM, Bland rectal adenocarcinoma
WC. Dose implications
associated
with
resection.
IntJ Radiat
response for lo-
sphincter-sparing Oncol
Biol
RI, PfaffWW, et al. Initially unresectable treated with preoperative irradiation and
surgery. Ann Surg 1986;205:414. 11. Cohen AM, Minsky BD. Aggressive surgical management of locally advanced primary and recurrent rectal cancer. Dis Colon Rectum 1990;32:43248. 12. Scivetti P, Donava M, Riecordi A, Fioccia R, Dianigi P, Mazini G. Prognostic significance of DNA content in large bowel carcinoma: a retrospective flow cytometric study. Cancer Lett 1989;46:21%9. 13. Harlow SP, Ericksen BL, Poggensee L, et al. Prognostic implications of proliferative activity and DNA aneuploidy in Astler-Coller Dukes stage C colonic adenocarcinomas. Cancer Res 1991; 51:2403-g. 14. Hedley DW, Friedlander ML, Taylor IW, et al. Method for analysis of cellular DNA content of paraffin-embedded pathological materials using flow cytometry. J Histochem Cytochem 1983;31: 1333-5.
SU%@Y Volume
MeterSan
121, Number
15. McLemore
DD, El-Naggar
methodology paraffin-embedded
AK, Stephens
processing of several
option
Surg Gynecol 20. BoeyJ, Wang colorectal
in advanced
carcinoma.
Ann
for
EJ, Bnmo
malignant
Surg
rectal
25. Yeung RS, Moffat FL, Falk RE. Pelvic exenteration and extensive primary colorectal adenocarcinoma. 72:1853-8. 26. Vasselopoulos
tumors
487
as
28. Kokal W, Sheibani in the prognosis
advanced
A. Total
pelvic
Surg 1981;194:701-3. SJ, Gelder MS, Hatch
and histopathologic after pelvic exenteration
KD, Baker
factors predicting for cancer
30. Centeus analysis
1988;lOZ:
exenteration
Ledesma
for recurrent Cancer 1993;
VOLUMES
liferative survival J Radiat
in V, Aus-
in colorectal BA, Zietman of DNA ploidy,
tion 81.
for
cancer.
Br J Surg
1985;72:828-30.
AL, Shipley WV, et al. Flow cytometric percent S phase fraction, and total pro-
fraction as a prognostic indicator for local control following radiation therapy for prostatic carcinoma. Oncol Phys 1994;30:309-15.
31. Meterissian D. Patterns
recurrence of the cervix.
AVAILABLE
to
K, Terz J, Harade JR. Tumor DNA content of colorectal carcinoma. JAMA 1986;255:
S, Skibber of residual rectal
and Int
J, Anjani J, Cleary K, Ruibein L, Ota disease after preoperative chemoradia-
adenocarcinoma.
Ann
Surg
Oncol
1989;73:1027-34.
BOUND
of recur-
EJ. Treatment
NC, Robins RA, Evans DF, Tivner DR, Baldwin RW, JD. The influence of tumor cell DNA abnormalities
on survival
Surg
JM,
recur-
3123-7. 29. Armitoge Hardcastle
1982;195:513-8.
Ann
PP, Yuon
of treatment of perineal Cancer 1979;43:9525.
distribution in human colon carcinomas and its relationship clinical behavior. J Natl Cancer Inst 1982;69:15-22.
of the pelvis.
for locally
JS. The results of the rectum.
rence of adenocarcinoma of the colon and rectum at the anastomotic site. Surg Gynecol Obstet 1981;152:777-80. 27. Wolley RC, Schreiber K, Koss LG, Kovas M, Sherman A. DNA
by pulse
exenteration
disease
carcinoma?
S, Mittleman
disease. Ann HM, Soang
Gynecol
and tech-
Huddleston CB, Sawyers JL, Potts JR, Sharp SW. Is total pelvic exenteration reasonable
treatment
tin M. Clinical and survival
24. Polk HC, Spratt rence of cancer
of ploidy
pelvic
Obstet 1988;106:259-63. J, Ong GB. Pelvic exenteration
21. Williams LF Jr, KW, McDougal
colorectal 23. Shingleton
solid
from
W, Linden WA. Analysis of PCPdata to deterof cells in the various phases of the cell cycle.
Environ Biophys 1975;12:31-9. WG, Lopez MJ, Brecker EM. Total
a therapeutic
Obstet
B. Automated comparison
characteristics of human Cancer Res 19X$38:3333-9.
18. Baisch H, Gohde mine the fraction
primary 670-6. 22. Ledesma
J. Modified
DNA histograms 1990;65:279-91.
Comput Biomed Res 1978;11:393404. B, Gohde W, Johnston DA. Determination
and proliferative cytophotometry.
Radiat 19. Kraybill
CC, Jardine
to improve flow cytometric material. Stain Tech
16. Johnston DA White RA, Barologie interpretation of DNA distributions: niques. 17. Barologie
et al.
5
TO SUBSCRIBERS
Bound volumes of Surgery are available to subscribers (only) for the 1997 issues from the publisher at a cost of $81.50 for domestic, $108.61 for Canadian, and $101.50 for international subscribers for Vol. 121 (January to June) and Vol. 122 (July to December). Shipping charges are included. Each bound volume contains a subject index and an author index, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular journal subscription. Payment must accompany all orders. Contact Mosby-Year Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 631463318; phone (800)325-4177, ext. 4351, or (314)453-4351.
1994;2:173-
Number 5
SURGERY MAY
Original
1997
communications
Pelvic exenteration for locally advanced rectal carcinoma: Factors predicting improved surviva Sarkis H. Mete&Sian, Adel K. El-Naggar,
MD,
MD,a John M. PhD, Kenneth
Skibber, MD, Geoffrey G. Giacco, MS, R. Hess, PhD, and Tyvin A. Rich, MD,b
Houston, Texas
Background. The purpose of this retrospective review was to determine whether a number of clinicopatholo@c factors (age, gender, type of exenteration, tumor extent, adjuvant therapy, tumor DNA ploidy, and Sphase fraction) that could be determined before operation were useful in predicting survival in patients undergoing pelvic exenteration for rectal cancer. Methods. Between 1983 and 1992, 40 patients (15 male and 25 female) at our institution underwent pelvic exenteration for rectal adenocnrcinoma in which tumor-free pathologic marginswere obtained. Twenty-nine patients presented with primary tumors; 11 had recurrent disease. A total exenteration was performed in 20 patients, posterior exenterution in 18 patients, and an anterior exenteration in 2 patients. Results. By multivariate (Cox proportional hazards repession) analysis, age, preoperative chemoradiation therapy, and an S phase of IO 70 or greater were found to be significant predictors of survival. Age older than 55 years was associated with a relative risk for cancer-related death (RR) of 0.13 (p = 0.02), and ch,emoradiation had an RR of 0.05 (p = O.Ol), indicating their beneficial effect. An Sphase fraction of 10% or greater had an RR of 16.97 (p = 0.03), indicating a poor survival. The clinicopathologic factors listed above were used to derive a prognostic index (PI). A PI of less than 1.37 was associated with a 5-year survival rate of 65% (low risk), whereas patients with a PI of 1.37 or greater had a S-year survival rate of 20% (high risk) (p = 0.005). Conclusions.These results indicate that adjuvant chemoradiation may significantly improve survival in patients who requirepelvic exenteration for resection of locally advanced rectal carcinoma.An Sphase fraction of 10% or greater is also predictive of a poor outcome. Use of these factors allowed the generation of a PI that identifies high- and low-risk patients. Consideration of the ability to deliver chemoradiation and the determinates of the tumor Sphase fraction in patients requiring pelvic exenteration for rectal cancer may be helpful in predicting outcome and planning therapy. (Surgery 1997;121:479-87.) From the Departments of Surgical Oncology, Biomathmetics, Medical Infmrnatics and Radiotherapy, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
Accepted
for publication
Reprint requests: John cology-Box 106, The Center, 1515 Holcombe “Currently Quebec,
at McGill Canada.
Oct.
14, 1996.
M. Skibber, MD, Department of Surgical OnUniversity of Texas M. D. Anderson Cancer Blvd., Houston TX 77030. University,
Royal
Victoria
Hospital,
bCurrently lottestille,
at the University va.
Copyright
0 1997
by Mosby
of Virginia
Year Book,
Health
Sciences
Center,
Char-
Inc.
Montreal, 0039-6060/97/$5.00tO
11/56/79606
SURGERY
479
480
Metevissian
et al.
6% to 10% of rectal cancers are locally advanced at presentation’ and require extensive surgery for complete tumor extirpation. Pelvic exenteration involving en bloc removal of the rectum, bladder, distal ureters, and reproductive organs can be required to obtain negative margins of resection. The ability to accomplish tumor-free resection margins is the most important treatment factor in determining patient outcome. When tumor-free margins cannot be obtained, local failure rates are high and survival is short. The presence of residual tumor after resection is an overwhelming factor in a poor outcome. No matter how extensive, resections that leave known residual disease cannot be considered curative. This operation described by Brunshwig2 and Appleby3 for the treatment of advanced or recurrent cervical carcinoma and advanced rectal carcinoma, respectively, is associated with significant morbidity and mortality. Nevertheless, a number of studies’, 4s have demonstrated a 5-year survival rate ranging from 33% to 50% for these selected patients with rectal cancer. In addition, this operation decreases cancer-related pelvic morbidities such as fistulas and ureteral or bowel ob struction. These problems are difficult to manage and lead to an agonizing death for the patient. However, it is difficult to justify the performance of this procedure for palliation alone. A major development in the multidisciplinary management of locally advanced rectal cancers is the use of preoperative radiotherapy.7-10 This treatment has resulted in an improved resectability rate, with 55% to 75% of patients being converted to a resectable stat~s.~~ Despite this improvement, overall recurrence rates after pelvic exenteration remain unacceptably high even with pathologically negative margins. Currently, the only reliable markers of prognosis remain the depth of tumor penetration and the lymph node status. Therefore a search for biologic markers that may predict outcome and guide the use of preoperative adjuvant therapy is needed to aid in the selection of patients most likely to benefit from this extensive surgical procedure. The purpose of this retrospective review was to determine the value of pretreatment DNA flow cytometry data as a prognostic marker12* l3 and the impact on survival of using preoperative chemoradiation on patients requiring pelvic exenteration for locally advanced rectal carcinoma in whom tumor-free margins were obtained. These parameters were used to derive a prognostic index (PI) to identify high-risk patients. These high-risk patients may be candidates for more intense neoadjuvant management, novel treatment strategies, or the selection of palliative nonsurgical therapy. APPROXIMATELY
MATERIAL
AND
METHODS
Patients. Between 1983 and 1992,40 patients underwent pelvic exenteration for rectal adenocarcinoma at
su%FY May 1997
The University of Texas M. D. Anderson Cancer Center. The clinical characteristics of these patients are summarized in Table I. Patients were operated on by multiple staff surgeons during this period. Twenty-nine patients presented with primary tumors, and 11 presented with recurrent tumors. A total pelvic exenteration, defined as the removal of the rectum with the bladder and prostate in men or the gynecologic organs in women, was performed in 20 patients. A posterior pelvic exenteration, defined as removal of only the female gynecologic organs in addition to proctectomy, was performed in 18 patients. In two patients intestinal continuity was reestablished by means of performance of a low rectal anastomosis; thus by definition an anterior exenteration was performed. All patients in this study had tumor-free final pathologic margins. Patients with microscopic or gross tumor at resection margins are excluded. Charts were reviewed to obtain the following information: age, gender, tumor size, stage (TNM), and adjuvant therapy. The tumor size and extent of intramural penetration were derived from the pathology report. Although all patients were deemed to have locally advanced disease at operation as a result of adherence to pelvic organs and pretreatment examination and imaging, 10 were found to have T3 tumors on final pathology examination after preoperative chemoradiation. Adjuvant therapy. Ten patients received only radiotherapy (three before operation and seven after operation), with doses ranging from 45 to 58 Gy (mean, 49 Gy) , Seventeen patients received concurrent preoperative chemoradiation treatments. In all cases 5-fluorouracil was used as a radiosensitizer at a dose of 300 mg/ m2. Preoperative radiotherapy doses ranged from 45 to 60 Gy (mean, 52.25 Gy). The use of only preoperative radiation in 10 patients occurred between 1981 and 1988. Although the use of chemoradiation occurred in 17 patients between 1989 and 1992,12 patients who did not receive adjuvant therapy were seen throughout the study period. Flow cytometric technique. Pretreatment biopsy tissue was fixed in 10% formaldehyde for 2 to 8 hours and embedded in paraf!dn (Paraplast Monoject Scientific, St. Louis, MO) with a melting point of 56” C. Nuclear suspensions of the neoplasms were prepared by using a modified version of the method of Hedley et all4 and McLemore et a1.15 In each block selected for analysis at least 75% of the tissue was neoplastic. In a given neoplasm, tissue blocks with more than 10% necrosis were avoided when possible. Otherwise, the necrotic areas were mapped and nonnecrotic tissue was reembedded for flow cytometric analyses. Two or more 50 pm sections were cut on a Reichet-Jung 2030 microtome (Cambridge Instruments, Heidelberg, Germany) and
Surge y Volume 121, Number 5
Meterissian
et al.
481
0
20
40
60
80
100
120
140
180
180
200
Months
Fig. 1. Overall survival, deaths from colorectal
placed into 15 x 100 mm glass culture tubes (Corning Glass Works, Corning, NY). The sections were deparaffinized three times for 15 minutes each in tubes containing 3, 4, and 5 ml Histo-Solve x (Biochemical Sciences, Bridgeport, NJ). Rehydration was performed by exposing sections to graded ethyl alcohol: 100% twice, 95% twice, 80% once, 70% once, and 50% overnight. Final hydration was achieved by placing specimens into Dulbecco s phosphate-buffered saline solution twice (Gibco Laboratories, Grand Island, NY). The specimens were incubated for 30 minutes at 37” C in prewarmed 0.5 % pepsin at pH of 1.5. The enzymatic digestion was stopped with 175 pl of cold 250 mg/ml pepstatin A (Sigma Chemical Co., St. Louis, MO). It was then disaggregated with 3 ml syringes and 18-gauge needles and filtered through 37 pm nylon mesh filters (Small Parts Inc., Miami, FL). DNA staining. The concentration of nuclei per milliliterwas established by using an automated cell counter (Coulter 2131; Coulter Electronics, Hialeah, FL), and dilutions were made to achieve a lo6 cells/ml concentration. A 900 ~1 aliquot of the final dilution was treated with 100 p1 0.25 mg/ml ribonuclease A (Worthington, Biochemical, Freehold, NJ) for 30 minutes at 37” C. The tubes containing these cells were then frozen and stained with 50 ~10.25% propidium iodide. Staining was performed at least 20 minutes before analysis, and the entire run was completed within 3 hours. Specimens were analyzed on an Epics Profile I (Epics Division, Coulter Electronics) equipped with an argon laser operating at 488 nm, with a 610 long-pass filter and a 488 band pass. Flow rates were adjusted to count approximately 75 nuclei/set. Peak versus integral signals were used to gate out doublets. CelI cycle and DNA ploidy analyses. Cytospin slides were prepared from a lo6 nuclei/ml concentration (unexposed to ribonuclease) and stained with WrightGiemsa. The slides were reviewed to confirm the pres-
Table
carcinoma.
I. Clinical
characteristics
Charactetistics
n
%
22 18
55 45
17
20
46 54
15 25
37.15 62.5
29 11
72.5 27.5
2 18 20
5 45
10
25 75
Age W
94 >55 Tumor size” (cm)
55 >5 Gender Male Female Tumor status Primary Recurrent Type of exenteration Anterior Posterior Total T stage T3
T4
30
Adjuvant therapy None Radiotherapy alone Chemoradiation DNA ploidy Diploid Aneuploid S-phase fraction 110 210 *Size
unknown
in three
13 10
17 19
21 12 28
50
32.5 25 42.5 47.5 52.5 30 70
patients.
ence of tumor nuclei and monitor the quality of processing. At least 5000 nuclei were evaluated in each sample. The DNA histograms were analyzed by using the sonogram gating procedure of Johnston et all6 Coulter Cytologic software (Version 2.2; Coulter Elec-
SU%eTY
482 Meterissian et al. May
Table
II. Results of univariate
and multivariate
analyses of survival*
Univariade
>55vs
p Value
RR
95%
analysis
554
0.88
0.31, 2.48
0.80
0.13
0.02,
0.96
0.02t
0.52
0.19, 1.38
0.19
3.94
0.44,
35.5
0.22
2.30
0.78, 6.78
0.13
6.36
0.62,
65.3
0.11
0.72
0.22,
2.41
0.60
0.23
0.03,
2.01
0.15
0.35
0.09, 1.35
0.13
0.05
0.01, 0.83
0.01t
1.04
0.37,
2.95
0.93
0.26
0.04,
0.16
2.40
0.54,
10.62
0.25
16.97
1.57
0.53,
4.64
0.41
0.95
regression
CI
Multivariate
RR
*Results from Cm proportional hazards tp < 0.05. fTota1 includes two anterior exenterations.
95%
analysis
Variables
Gender Female vs male Exenteration type Total: vs posterior Treatment Irradiation + surgery vs surgery alone Chemotherapy, irradiation + surgery vs surgery alone DNA ploidy Aneuploid vs diploid S-phase fraction >lO% vs
1997
CI
1.77
p Value
0.93,3.09
0.03t
0.16,
0.95
5.48
analysis.
tronics) was used only in cases exhibiting excessive debris and no evidence of doublets. All tumors were analyzed with cytologic software by using the same debris subtraction model. Nonneoplastic tissue was used as the biologic diploid standard, and the first Go/G1 population was used to denote the diploid stem line. The DNA index was defined as the ratio of the peak channel number of the test sample to the peak channel number of the normal diploid control.1’ By definition, the DNA index of a diploid population is equal to 1.0. A tumor was considered aneuploidwhen a distinct second Go/G1 peak, accounting for at least 10% of the cells analyzed, was present (or any number of such separate peaks). A tetraploid was defined as a distinct and sharp peak at the GYM position with at least 20% of the total events of the histogram. The proliferative fraction was determined as the percentage of cells in the S phase (after debris subtraction, when necessary) according to the method of Baisch et al.i8 The cell volume of the Go/i diploid peaks ranged from 2.5 to 6.8 with a mean of 4.3 2 1.5. Follow-up. Follow-up information was obtained either from the chart or by direct correspondence with the patient or the immediate family. The median follow-up of the survivors was 42 months, and survival was determined by including only disease-specific mortality (i.e., deaths from other causes were censored). No patients were lost to follow-up. Statistical analyses. Overall survival was computed by using the Kaplan-Meier method.’ The Cox-Mantel logrank test was used to compare survival time distribu-
tions. Univariate and multivariate relative risk estimates and corresponding 95% confidence intervals (CIs) were computed by using the Cox proportional hazards regression model.* Analyses were performed with BMDP/ 386 Dynamic, Release 3.0 (BMDP Statistical Software, Inc., Los Angeles, CA) and SPSS for Windows, Release 6.0 (SPSS Inc., Chicago, IL). RESULTS Influence of clinicopathologic parameters on survival. The cumulative survival of the 40 patients in this study is summarized in Fig. 1. The 5-year overall survival was 49% with a median survival of 56 months. Univariate and multivariate Cox proportional hazards regression analyses were performed with a number of clinicopathologic parameters including age, gender, type of exenteration, adjuvant therapy, tumor DNA ploidy, tumor S-phase fraction, and tumor extent at presentation (Table II). By univariate analysis none of these factors significantly influenced survival. Both the type of exenteration (total-anterior versus posterior) and an S phase fraction of 10% or greater were associated with the highest relative risk (RR) of death from rectal cancer (2.3 and 2.4, respectively). In contrast, adjuvant chemoradiation provided a protective effect, decreasing the risk by 35% (RR value). These effects were not statistically significant on univariate analysis, but important effects emerged on multivariate analysis. Age older than 55 years and the administration of adjuvant preoperative chemoradiation were associated
suw? Volume 121, Number 5
Mettissian
et al.
483
Combined Failure Distant Only
* Local
Local
Recurrence
7 Distant Recurrence l Death from Disease
0 Only
0 Censored
31111 8.
No Failure
l
. l
0
50 Time
l
150
100 in Months
from
200
Exenteration
Fig. 2. Patterns of failure after pelvic exenteration.
with a statistically significant beneficial effect (p = 0.02 and p = 0.01, respectively). Our data indicate that age older than 55 years actually appeared to be a protective factor (RR = 0.13; CI = 0.02 to 0.96). Adjuvant preoperative chemoradiation reduced the death rate 20-fold compared with surgery alone (RR = 0.05; CI = 0.01 to 0.83), whereas adjuvant postoperative external-beam radiotherapy produced a fivefold improvement compared with surgery alone (RR = 0.23; CI = 0.03 to 2.01), but this was not statistically significant (p = 0.15). Tumor stage at presentation did not influence survival (RR = 0.95). The need for a total exenteration increased the risk of death sixfold versus posterior exenteration (p = 0.11). Although one might expect women to fare better than men because they are more likely to undergo a posterior exenteration, this was not the case because women had a RR of 3.94 versus that for men (p = 0.22). Comparison of DNA aneuploidy versus diploidy in predicting deaths from rectal cancer was not statistically significant. The apparent univariate effects of gender and DNA ploidy on survival are confounded by the influence of other factors such as the S-phase fraction and adjuvant therapy. Only the S-phase fraction had a significant negative effect on survival, with an Sphase fraction of 10% or greater being associated with a RR of 16.97 (CI = 0.93 to 3.09; p = 0.03). Patterns of failure after pelvic exenteration. The patterns of failure of the 40 patients in this study are summarized graphically in Fig. 2. Nineteen patients had no evidence of local or distant failure at follow-up. Five patients had local failure alone, and one was still alive after nearly 50 months of follow-up. Four patients had both local and distant failure, and all died of disease. Ten patients experienced a distant recurrence without evidence of local failure, and six died of their disease. Four of the patients with only distant recurrence were still alive at the time of this study. One patient, who received preoperative chemoradiation and who had no
Table
*All
III. Survival from pelvic exenteration
Study
No. of cases
5.yr Suroival rate C%)
Lopez et aL4 Estes et aL5 Pearlman et aL6 Kraybill et al.lg Boey et al.” Williams et al.”
24 16” 20 11 49 10
42 49 50 44 38 43
patients
had locoregional
recurrence
before
exenteration.
evidence of residual disease on final pathology examinations and thus qualified as a pathologic complete response, remains free of disease. Prognostic index. With the increasing use of preoperative chemoradiation in the treatment of T3 rectal adenocarcinomas and its observed beneficial effect in this study, an attempt was made to develop a PI to simplify the identification of so-called high-risk patients who would be candidates for additional adjuvant therapy. With the Cox model, the following PI was formulated: PI = 1.37 (gender) - 2.05 (age) - 0.06 (tumor status) + 1.85 (type of exenteration) - 1.46 (treatment) - 2.91 (chemoradiation) + 2.83 (S-phasefraction) - 1.34 (DNA ploidy) , where the parameter values are gender: 0 for male, 1 for female; age: 0 for 554 years, 1 for 254 years; tumor status: 0 for primary, 1 for recurrent; type of exenteration: 0 for posterior, 1 for total or anterior; treatment: 0 for other, 1 for irradiation and surgery; chemoradiation: 0 for other, 1 for irradiation, chemotherapy, and surgery; S-phase fraction: 0 for 9%, 1 for 10%; and DNA ploidy: 0 for diploid, 1 for aneuploid. With this formula, 14 patients with a PI greater than 1.37 had a significantly (p= 0.005) decreased survival rate. Low-risk patients (Pl of 1.37 or less; n = 26) had a 5-year survival rate of 68%, whereas high-risk patients
SUVY
484 Meterissian et al.
May 1997
.80ii .2 5 .80rn _ % ‘3 .40z 3 .20-
0
N=40 Deaths=1
8
12
24 lime
in Months
36 from
48
60
Exenteration
Fig. 3. PI based on COX model. Low-risk patients (PI 5 1.37): 6 deaths in 26 patients; high-risk (PI > 1.37): 9 deaths in 14 patients (p = 0.005). (PI of 1.37 or greater) had a 5-year survival rate of 24% (Fig. 3). Thus this PI formula identified patients at high risk for treatment failure.
DISCUSSION Pelvic exenteration, although a radical surgical procedure, is often the most appropriate treatment available for locally advanced primary and recurrent rectal adenocarcinoma.46, 1g-21 The number of exenterations specifically for rectal cancer included in this study is comparable with the 49 cases reported by Boey et a1.2o The overall 5-year survival rate of the 40 patients in this studywas 49%, which is comparable to the data reported in the literature (Table III). Long-term local control was obtained in 73% of the patients studied. An important issue in the care of these patients is the selection of those who will benefit most from this extensive procedure. Clearly, the first criterion for those who will obtain benefit is the ability to obtain negative margins. The use of preoperative irradiation has been established to improve resectability in locally advanced rectal cancer and to improve outcomes. However, local and distant failures persist. This is a retrospective study whose purpose was to find potential pretreatment factors to predict long-term survival in patients with advanced rectal cancer who required multivisceral resection to obtain tumor-free margins. The purpose is not to address the morbidity or mortality of such resections, although these problems can be substantial. In addition, it does not address the outcome for the patients in whom tumor-free margins are not obtained. Patients with residual tumor in the pelvis have a poor outcome in terms of survival and local progression despite additional radiotherapy. Such resections cannot be considered curative. They were excluded from our study because data from patients with residual disease would add little to our analysis for factors predicting
patients
long-term survival and would confound the analysis of biologic factors. The conclusions drawn from these data are limited by the retrospective nature of the study. Unseen bias in the selection of treatment for individual patients can impact on the observed outcomes. The major factor in selecting the use of preoperative or postoperative radiation appeared to be previous exposure to pelvic irradiation either for a previous rectal primary tumor in those patients with recurrence or for another pelvic malignancy. An uncontrolled study such as this does not allow for the evaluation of the efficacy of treatment plans. By using univariate and multivariate Cox proportional hazards regression analyses, the effect of age, gender, type of exenteration, adjuvant therapy, DNA ploidy, S-phase fraction, and tumor status at presentation (i.e., primary versus recurrent, tumor size) as risk factors for dying of rectal cancer were analyzed. Tumor size was not a predictor of recurrence or survival in our study. Also it makes it unlikely that a bias for giving multimodality therapy to subgroups of patients on the basis of tumor size affected our study. This was previously shown by Ledesma et a1.22 in a series of patients undergoing pelvic exenteration. This was contradicted by Shingleton et al., 23 who found that tumor size was an independent predictor. Tumor size alone has not been accepted as an independent factor predicting survival in patients with rectal cancer. Although older age may be thought of as a contraindication to such radical surgery, the results of this study indicate otherwise. Patients older than 55 years actually fared better than those who were 54 years old or younger (RR = 0.13; p = 0.02)) indicating that older patients can undergo a pelvic exenteration for long-term benefit. The decreased survival of the younger patients is intriguing because it may indicate more aggressive tumor biologic factors.
sv@T Volume
Mete-rksian 121. Number
et al.
485
5
Patients undergoing a posterior exenteration fared better than those undergoing either a total or anterior exenteration, probably because the more advanced tumors required more extensive surgery. The type of exenteration also influenced survival, but the effect was not statistically significant. In a small study Estes et a1.5 also showed that patients who had posterior exenterations achieved a better mean survival than those who had complete exenterations (46 versus 29 months). Clearly the most appropriate operative procedure is one that obtains tumor-free margins. This is related to tumor size, location, and invasion into other organs at presentation. Pelvic exenteration for recurrent disease did not predict a poorer survival, which is discordant with data reported by others. 24 For example, Yeung et a1.25 found long-term survival in only two of 43 patients who underwent pelvic exenteration for recurrent rectal adenocarcinema. Our overall survival results in patients with recurrence could be explained because the 11 patients with recurrent disease in our study were a carefully selected group who were all able to undergo a complete resection with tumor-free margins. Obtaining tumorfree margins can be more difficult in patients operated on for recurrent rectal cancer. Another reason for the survival results could be that the patients with completely resectable recurrent tumors had anastomotic or perianastomotic recurrences, which carry a better prognosis than these involving the pelvic sidewall or sacrum. Vasselopoulos et a1.26 reported a 5-year survival rate of 49% in 15 patients who underwent a complete resection for such a perianastomotic recurrence. On the basis of our data, pelvic exenteration for recurrent disease can be reasonable treatment if negative margins are achieved, especially when patients can receive preoperative chemoradiation. Because most of the clinicopathologic variables studied are weak predictors of survival and the tumor stage is obtained only after operation, there is a need to identify new prognostic factors that can be determined before operation and help guide management. Such factors include DNA ploidy and the S-phase fraction, whi.ch can be measured easily on an intraluminal biopsy specimen before therapy is instituted. Harlow et a1.13 showed that the tumor proliferative activity measured by the Sphase fraction in Dukes class C colon carcinoma was an important biologic factor in determining patient prdgnosis. Other studies have shown that DNA ploidy may be an independent predictor of patient survival in colorectal carcinomas,2”2g particularly diploid colorectal cancers. To date, there have been no published studies looking at the predictive value of tumor DNA ploidy and the tumor S-phase fraction specifically in patients requiring pelvic exenteration for rectal cancer in whom neoadju-
vant therapy has been used. This study shows that with multivariate Cox proportional hazards regression analysis the S-phase fraction is a statistically significant predictor of survival in such patients. Patients with an S-phase fraction of 10% or greater had 16.97 times (CI = 0.93 to 3.09) greater chance of dying of rectal cancer (p = 0.03) than patients with an S-phase fraction less than 10%. Because the S-phase fraction can be determined before operation on biopsy specimens obtained by endoscopy or needle biopsy, it could be used to evaluate risk before treatment. The role of S-phase fraction determination in estimating prognosis has been examined in two other studies in which irradiation has been used. Centeus et a1.30 found that the S-phase fraction predicted the local control of well or moderately differentiated prostate cancers. In another study of head and neck cancers treated by irradiation alone it was reported that tumors with a low S-phase fraction exhibited a higher local recurrence rate. This was postulated to result from the relative resistance to radiotherapy of slowly proliferating cells. These results appear to be at odds with our findings. However, it should be noted that we are dealing with different tumor histologic characteristics, the addition of surgical therapy, and the use ofchemoradiation therapy compared with radiotherapy alone. Studies of the role of the S-phase fraction in the outcome of patients subjected to neoadjuvant therapy for rectal cancer are needed to validate our findings. Preoperative radiotherapy has been used in a number of studies in patients with locally advanced rectal adenocarcinoma with improved resectability and arguably improved survival and recurrence data.7-10 Preop erative chemoradiation is currently administered in our institution to selected patients with T3 rectal carcinomas with high rates of complete pathologic responses and excellent local tumor contro13r As many as 37% of the patients have a complete pathologic response, and an additional 46% experience appreciable downstaging of their tumors. Chemoradiation has been used in locally advanced tumors in hopes of improving the long-term recurrence and survival results. It is reasonable to consider that further expansion of this multidisciplinary neoadjuvant approach should be used in high-risk patients before performing pelvic exenteration for recurrent rectal cancer. In this study preoperative chemoradiation was used in 1’7 patients, and 10 patients received radiotherapy alone. Radiotherapy alone improved survival fivefold, but this effect was not statistically significant compared with the survival of patients receiving surgery alone (Table II). In contrast, preoperative chemoradiation had a dramatic effect. All 17 patients received concurrent intravenous 5-fluorouracil (300 mg/m*) and external-beam radiotherapy (mean dose, 52 Gy) . Early on,
486
Meterissian
SUPY
et al.
May 1997
patients received bolus of 5-fluorouracil, whereas in the latter part of this study, it was given by continuous infirsion. Because of an insufficient number of patients, these two methods of administration could not be directly compared. As seen in Table II, Cox multivariate analysis revealed a 20-fold protection from rectal cancer death in patients who received preoperative chemoradiation (CI = 0.01 to 0.83; p= 0.01) compared with the result for those who received surgery alone. Use of preoperative chemoradiation proved to be a significant prognostic factor in our predictive model. A number of studies”rO have shown improvement in survival in patients with locally advanced rectal carcinoma who received only preoperative radiotherapy. At Massachusetts General Hospital’ the administration of 40 to 50 Gy followed by surgery resulted in a 5-year actuarial survival rate of 28%. In the patients in whom a complete resection was achieved, the 5-year actuarial rate for survival was 43%. Similarly, the groups at Tufts Univers@ and the University of Virginia9 reported 3and 5-year actuarial rates for survival of 54% and 52%, respectively. Mendenhall et al.” found that a complete resection improved the 5-year survival rate from 9% to 20%. The survival data reported in our study can be attributed to the selection of patients for performance of negative margin resections. This study does imply a beneficial effect from preoperative chemoradiation on survival in locally advanced rectal carcinoma; however, the retrospective and uncontrolled nature of the study does not allow definitive comparisons of treatments. Confirmation of this beneficial effect will await reporting data from prospective trials with patients requiring multivis ceral resections for locally advanced rectal cancer. Whether recurrence after exenteration was local or distant, survival was poor (Fig. 2). Distant metastases were documented in 11 of the 15 patients who died of disease after pelvic exenteration, and 10 of these 11 had an S-phase fraction greater than 10% (Fig. 2). Four patients had only a local recurrence at time of death, but because none had an autopsy, the possibility of concomitant distant disease was not ruled out. S-phase fraction of greater than 10% did not predict the likelihood of distant versus local failure. Those at high risk of failure appear unlikely to benefit in terms of survival from pelvic exenteration. The use of a PI may allow patients to be stratified into low- (PI, 1.37 or less) and high-risk groups. Of the 15 deaths, two thirds of the deaths occurred in the highrisk group (PI, greater than 1.37). The use of such an index may allow us to design better studies of treatment plans for such patients. It must be noted, however, that this PI was derived in a study of 40 patients, and to extend our findings, larger groups of patients undergoing pelvic exenteration for rectal cancer must be analyzed.
In summary, this study attempted to identify important pretreatment prognostic factors to define the value of adjuvant therapy in patients requiring pelvic exenteration for rectal cancer. On the basis of a multivariate analysis, it appears that a high proliferative fraction (Sphase, 10% or greater), age older than 55 years, and adjuvant preoperative chemoradiation significantly affected survival in patients undergoing pelvic exenteration. With the multivariate Cox model, a PI was derived to identify low- and high-risk groups, thereby enabling an estimate of prognosis before initiating treatment. Identification of high-risk groups may indicate which patients are candidates for novel treatment approaches or for nonresectable palliation. This study is only a start in addressing the biologic factors of such advanced tumors and how they interact with treatment planning for optimization of outcomes. REFERENCES 1. Lopez
MJ, Monafo
treatment
W. Role
of locally
of extended
advanced
resection
colorectal
in the initial
carcinoma.
Surgery
1993;113:365-72. 2. Brunshwig
A. Complete excision Cancer 1948;1:177-83.
carcinoma.
3. Appleby LH. with ureteral 4. Lopez
viscera
Proctocystectomy: the management transplants. Am J Surg 1950$9:5X60.
MJ, Rraybill
Exenterative it worthwhile?
of pelvic
WG,
Bouney
RS, Johnston
surgey for locally advanced Surgery 1987;102:64451.
for advanced of colostomy
WD,
Bricker
rectosigmoid
EM.
cancers:
Is
5. Estes N, Thomas JH, Jewel1 WR, Beggs D, Harden C. Pelvic exenteration: treatment for failed rectal cancer surgery. Am Surg 1993;59:420-2. 6. Pearlman N, Stiegmann flied
rectal
7. Dosoretz radiation Cancer
cancer.
GV, Donohue Cancer
DE, Gunderson for unresectable 1983;523:8148.
R. Extended
resection
of
1989;63:243841. LL, Hedberg rectal and
S, et al. Preoperative rectosigmoid carcinoma.
ir-
8. Emami B, Willet C, Dilepich M, et al. Effect of preoperative irradiation on resectability of colorectal carcinomas. Int J Radiat Oncol Biol Phys 1989;8:1295-9. 9. Fortier GA, Rrochak RJ, Jung AK, Constable to preoperative irradiation in rectal cancer: cal control
and
complications
surgery and abdominoperineal Phys 1986;12:155963. 10. Mendenhall WM, Bland rectal adenocarcinoma
WC. Dose implications
associated
with
resection.
IntJ Radiat
response for lo-
sphincter-sparing Oncol
Biol
RI, PfaffWW, et al. Initially unresectable treated with preoperative irradiation and
surgery. Ann Surg 1986;205:414. 11. Cohen AM, Minsky BD. Aggressive surgical management of locally advanced primary and recurrent rectal cancer. Dis Colon Rectum 1990;32:43248. 12. Scivetti P, Donava M, Riecordi A, Fioccia R, Dianigi P, Mazini G. Prognostic significance of DNA content in large bowel carcinoma: a retrospective flow cytometric study. Cancer Lett 1989;46:21%9. 13. Harlow SP, Ericksen BL, Poggensee L, et al. Prognostic implications of proliferative activity and DNA aneuploidy in Astler-Coller Dukes stage C colonic adenocarcinomas. Cancer Res 1991; 51:2403-g. 14. Hedley DW, Friedlander ML, Taylor IW, et al. Method for analysis of cellular DNA content of paraffin-embedded pathological materials using flow cytometry. J Histochem Cytochem 1983;31: 1333-5.
SU%@Y Volume
MeterSan
121, Number
15. McLemore
DD, El-Naggar
methodology paraffin-embedded
AK, Stephens
processing of several
option
Surg Gynecol 20. BoeyJ, Wang colorectal
in advanced
carcinoma.
Ann
for
EJ, Bnmo
malignant
Surg
rectal
25. Yeung RS, Moffat FL, Falk RE. Pelvic exenteration and extensive primary colorectal adenocarcinoma. 72:1853-8. 26. Vasselopoulos
tumors
487
as
28. Kokal W, Sheibani in the prognosis
advanced
A. Total
pelvic
Surg 1981;194:701-3. SJ, Gelder MS, Hatch
and histopathologic after pelvic exenteration
KD, Baker
factors predicting for cancer
30. Centeus analysis
1988;lOZ:
exenteration
Ledesma
for recurrent Cancer 1993;
VOLUMES
liferative survival J Radiat
in V, Aus-
in colorectal BA, Zietman of DNA ploidy,
tion 81.
for
cancer.
Br J Surg
1985;72:828-30.
AL, Shipley WV, et al. Flow cytometric percent S phase fraction, and total pro-
fraction as a prognostic indicator for local control following radiation therapy for prostatic carcinoma. Oncol Phys 1994;30:309-15.
31. Meterissian D. Patterns
recurrence of the cervix.
AVAILABLE
to
K, Terz J, Harade JR. Tumor DNA content of colorectal carcinoma. JAMA 1986;255:
S, Skibber of residual rectal
and Int
J, Anjani J, Cleary K, Ruibein L, Ota disease after preoperative chemoradia-
adenocarcinoma.
Ann
Surg
Oncol
1989;73:1027-34.
BOUND
of recur-
EJ. Treatment
NC, Robins RA, Evans DF, Tivner DR, Baldwin RW, JD. The influence of tumor cell DNA abnormalities
on survival
Surg
JM,
recur-
3123-7. 29. Armitoge Hardcastle
1982;195:513-8.
Ann
PP, Yuon
of treatment of perineal Cancer 1979;43:9525.
distribution in human colon carcinomas and its relationship clinical behavior. J Natl Cancer Inst 1982;69:15-22.
of the pelvis.
for locally
JS. The results of the rectum.
rence of adenocarcinoma of the colon and rectum at the anastomotic site. Surg Gynecol Obstet 1981;152:777-80. 27. Wolley RC, Schreiber K, Koss LG, Kovas M, Sherman A. DNA
by pulse
exenteration
disease
carcinoma?
S, Mittleman
disease. Ann HM, Soang
Gynecol
and tech-
Huddleston CB, Sawyers JL, Potts JR, Sharp SW. Is total pelvic exenteration reasonable
treatment
tin M. Clinical and survival
24. Polk HC, Spratt rence of cancer
of ploidy
pelvic
Obstet 1988;106:259-63. J, Ong GB. Pelvic exenteration
21. Williams LF Jr, KW, McDougal
colorectal 23. Shingleton
solid
from
W, Linden WA. Analysis of PCPdata to deterof cells in the various phases of the cell cycle.
Environ Biophys 1975;12:31-9. WG, Lopez MJ, Brecker EM. Total
a therapeutic
Obstet
B. Automated comparison
characteristics of human Cancer Res 19X$38:3333-9.
18. Baisch H, Gohde mine the fraction
primary 670-6. 22. Ledesma
J. Modified
DNA histograms 1990;65:279-91.
Comput Biomed Res 1978;11:393404. B, Gohde W, Johnston DA. Determination
and proliferative cytophotometry.
Radiat 19. Kraybill
CC, Jardine
to improve flow cytometric material. Stain Tech
16. Johnston DA White RA, Barologie interpretation of DNA distributions: niques. 17. Barologie
et al.
5
TO SUBSCRIBERS
Bound volumes of Surgery are available to subscribers (only) for the 1997 issues from the publisher at a cost of $81.50 for domestic, $108.61 for Canadian, and $101.50 for international subscribers for Vol. 121 (January to June) and Vol. 122 (July to December). Shipping charges are included. Each bound volume contains a subject index and an author index, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular journal subscription. Payment must accompany all orders. Contact Mosby-Year Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 631463318; phone (800)325-4177, ext. 4351, or (314)453-4351.
1994;2:173-