The Relationship of Local Control to Distant Metastasis in Muscle Invasive Bladder Cancer

~22-5347/95/1546-2059$03.0010

vol. 154,2059-2064, December 1995

Tm:JOURNAL OF UROLOGY Copfight 0 1995 by AMERICAN UROLOGICAL ASSOCIATION, INC.

printed in U S A

THE RELATIONSHIP OF LOCAL CONTROL TO DISTANT METASTASIS IN MUSCLE INVASIVE BLADDER CANCER ALAN POLLACK, GUNAR K. ZAGARS, CHRISTOPHER J. COLE, COLIN P.N. DINNEY, DAVID A. SWANSON AND H.BARTON GROSSMAN From the Departments of Radiotherapy and Urology, University of Team, M.D. Anderson Cancer Center, Houston, Tercur ABSTRACT

Purpose: We examined the relationship of local failure to distant metastasis in patients with muscle invasive bladder cancer. Materials and Methods: This retrospective review included 240 patients treated with radical cystectomy with or without multiagent chemotherapy at our institution between 1984 and 1990 for clinical stage T2 to T4 transitional cell carcinoma of the bladder. The distribution of patients by clinical stage was 89 T2, 77 T3a, 51 T3b and 23 T4. Median followup was 55 months. Results: The actuarial 5-year local control, freedom from distant metastasis and overall survival rates were 80%, 68% and 52%, respectively. There was a profoundly significant relationship between local failure and distant metastasis with distant metastasis in 56% of those with local failure. The actuarial 5-year freedom from distant metastasis rate for those with local control was 77% compared to 29% for those with local failure (p <0.0001, log rank test). This relationship held when the group was subdivided by stage and when only cases of complete cystectomy were analyzed. The significance of this finding in light of the possible contribution of potential prognostic factors was examined. Univariate analyses revealed late clinical stage, abnormal pretreatment serum creatinine levels, the administration of chemotherapy, late pathological stage and lymph node involvementto correlate significantly with distant metastasis rates. Multivariate analyses using Cox proportional hazards models with freedom from distant metastasis as the end point revealed pathological stage, local failure and lymph node involvement to be the only significant covariates. Conclusions: Since local failure highly correlated with distant failure, treatment planning to optimize local control should be of foremost concern for those at high risk of failure by this mode (for example patients with T3b/4 disease). New treatment strategies, such as the use of preoperative radiotherapy as an adjunct to chemotherapy and radical surgery, should be considered in this high risk population. KEYWORDS: bladder, cystechmy, bladder neoplasms, neoplasm metastasis

Despite advances in the treatment of muscle invasive transitional cell carcinoma of the bladder with radical Cystectomy and multiagent chemotherapy approximately 4O?k of such patients die of metastatic disease. Data from pathologicallystaged casea suggest that local failure is minimal and, as a consequence,this end point is inhquently examined.l.2 The assumption is that local control is satisfhcto~and further improvements would be of little benefit since the weight of the problem lies in reduciug distant metastasis. However, recent studies by Grevens and Cole4 et al in which cases were categorized by clinical stage demonstrated that local failure is much more pronounced than previously believed. Patients with clinical evidence of extravesical extension who elected radical cystectomy with or without chemotherapy had local failure rates of greater than 30%. The salient issue concerning local control is whether the addition of more aggressive local measures, such as radiotherapy, would impact the course of the disease by reducing metastasis, thus enhancing survival. w e examined the strength of the relationship of local failure to distant metastasis and the independence of this relationship from other potential prognostic factors. MATERIALS AND METHODS

Patient and tumor characteristics. A total of 232 patients was treated with radical cystectomy with or without chemo-

therapy for muscle invasive disease at our institution between 1985 and 1990. These patients were compared to those treated with preoperative radiotherapy and cystectomy, as reported previously.' For the present study an additional 8 patients with T3W4 disease treated from 1984 to 1985 were included (total240). Patients age ranged from 40 to 86 years (mean 65, median 66). Median followup for the 126 survivors was 55 months (mean 57). There were 198 men and 42 women who elected radical cystectomy with or without chemotherapy. No patient received or was planned to receive radiotherapy. In fact, during the study period the only patients treated with primary radiotherapy for bladder cancer were those considered medically compromised for radical cystectomy. All patients were without distant metastasis before treatment initiation. Four patients had radiographic evidence of pelvic adenopathy, including 3 by computerized tomography and 1 by lymphangiography . Evaluation, staging and disease endpoints. The evaluation has been described previously.4 Tumors were graded on a scale of I to 111.6 Clinical stage was determined by pretreatment cystoscopy and bimanual examination using anesthesia. The 1983 American Joint Committee on Cancer Staging Manual criteria were used.6 The distribution of patients by clinical stage was 89 T2,77T3a, 51T3b and 23 T4 (2 had T4b disease). Local failure was defined as recurrence in the surgical bed. Pelvic failure was defined as pelvic lymph nodal and/or local recurrence. Distant metastasis included lymph

Accepted for publication April 13, 1995. Supported in part by Grants CA 06294 and CA 16692 from the National Cancer Institute, United States De artment of Health and Human Services, and an American Cancer L e t y Career Development Award. 2059

2060

LOCAL CONTROL AND DISTANT METASTASIS IN BLADDER CANCER

c w e s for those treated with cystectomy revealed only slight improvements in these outcome parameters (not shown). The actuarial 5-year local control rates for all patients and those treated with cystectomy were 87% and 93%for stage T2,86% and 91% for stage T3a, 68% and 71% for stage T3b, and 56% and 56% for stage T4, respectively. The actuarial 5-year freedom from distant metastasis rates for all patients and those treated with cystectomy were 75% and 78% for stage T2,76% and 77% for stage T3a, 51% and 52% for stage T3b, and 51% and 51% for stage T4, respectively. The data reveal little difference between stages T2 and T3a or T3b and T4 although the results were clearly poorer for those with T3bl4 disease compared to those with W3a disease for both end points. The local control and freedom from distant metastasis rates at 5 years were 87% and 75% for TW3a disease, and 64% and 51% for T3b/4 disease, respectively (p <0.001 for both end points). This prognostic division point is also reflected in the data in table 1,which shows the relationship of local control and local failure rates to the actuarial 4-year freedom from distant metastasis rates. Significantly higher actuarial freedom from distant metastasis rates were noted when local control was maintained. The strength of this relationship is illustrated in figure 2, which includes all patients. The exclusion of patients who did not complete cystectomy did not alter the significance of this finding. When subdivided by stage, local control was associated with significantly higher actuarial freedom from distant metastasis rates for all subgroups with adequate patient numbers. The practical implication of these findings is that improvements in local control impact on the rate of distant metastasis and possibly survival. In fact, the relationship of local control to overall survival rate was equally significant (p <0.001). Further support for this hypothesis was revealed by a detailed analysis of the distribution of prognostic factors by freedom from distant metastasis and the association of these prognostic factors to the actuarial h e d o m from distant metastasis rates. RESULTS Table 2 shows the distribution of prognostic factors by Local failure occurred in 41 of the 240 patients overall distant metastasis status. Of the patients with distant me(17%)and 30 of the 228 treated with cystectomy (13%).Of the tastasis a significantly greater proportion had clinical stage former patients local failure occurred more than 3 months T3b/4, pretreatment creatinine levels of greater than 1.5 before or in the absence of distant failure in 63%, within 3 mg./dl., pathological stage P2 to 4, lymph node involvement months of distant failure in 32% or more than 3 months after or treatment by chemotherapy. Distribution by age was of distant failure in 5%. Distant metastasis developed in 66 of borderline significance. Univariate actuarial analyses of the the 240 patients overall (27.5%) and 26% of those treated potential prognostic factors correlating with freedom from with cystectomy. The crude survival rate was 53% overall distant metastasis are shown in table 3. The only significant pretreatment factors were clinical stage and creatinine level. and 56% for those treated with cystectomy. The actuarial 5-year local control, freedom from distant In terms of postoperative factors pathological stage, lymph metastasis and overall survival rates for the entire cohort node status and chemotherapy (most often given postoperawere 80%, 68% and 52%. respectively. The rates for those tively) were significant. Trends for lower actuarial freedom treated with cystectomy were 84%, 69% and 548, respec- from distant metastasis rates were noted in patients older tively. Figure 1 shows the relationship of stage to actuarial than 65 years and those having tumors with lymphaticlocal control and distant metastasis for all patients. The vascular invasion.

nodal involvement above the b h t i o n of the iliac vessels or in the m g w d regions and metastases resulting from hematogenous spread. Mica1 cystectomy and multiagent chemotherapy. Of the 240 patients who elected radical cystectomy initially or after neoadjuvant chemotherapy the procedure was not done in 12 of whom 1 underwent salvage by chemotherapy and the remaining were considered as having local failure. Of the 58 patients who had evidence of lymph node involvement from lymphadenectomy (25% of the 229 in whom lymph node status was known) 46 underwent radical cystectomy. The incidence of lymph node metastasis by clinical stage for all patients in whom lymph node status was known was 18%(15 of 85 patients) for T2,30% (22 of 73) for T3a, 28% (14 of 50) for T3b and 33% (7 of 21) for T4. Multiagent chemotherapy was given to 67.5% of patients overall, including 7.5% preoperatively, 49.2% postoperatively, and 10.8% preoperatively and postoperatively. The 2 regimens used were cisplatin, cyclophosphamide and doxorubicin, and methotrexate, vincristine, doxombicin and cisplatin. Neoadjuvant or adjuvant chemotherapy was administered to 61% of those with stage W3a disease and 85% of those with stage T3bA disease. Statistical analyses. The chi-square test was used to examine the differences between proportions.7 Actuarial curves were calculated from the date of surgery (attempted or completed cystectomy) using the Berkson-Gage method and the log rank statistic was used to test for significant differences.8 Confidence intervals for the actuarial data were calculated using the method described by Rothman.9 The log linear relative hazard function of Cox was used for actuarial multivariate analyses.8 The analysis of time-dependent variables was performed as described by Cox and Oaks.10 A statistical software package was used for all calculations. The cause of death was unknown in 6 patients and, therefore, they were only included in calculations of the overall survival rate.

B

A

F R A C I

0

N

0.1

o.2{

- ] O . O

o

0.24

10

a0

aa

40

w

w

70

w

o

o

10

20

w

MONTHS AFTER SURGERY

.

~ 40

50

Q

70

]

Q

FIG. 1. Actuarial local control (A) and distant metastasis ( B )by stage for all patients. 0 ,stage T2.W, stage T3a. A,stage T3b. +, stage T4

~

2061

LOCAL CONTROL AND DISTANT METASTASIS IN BLADDER CANCER TABLE1. Relationship of local control to actuarial 4-year freedom

from distant metastasis according to stage C

Pts. (No.)

Local Control Planned cystectomy: Overall T3a T3b T4 TW3a T3W4

Completed cystectomy: Overall

Local Relapse' 23 (41) 29 (10) (47 (10) 12 (13) <33 (8) 35(20) 19(21)

77 (193) 82 (77) 81 (64) 67 (37) 62 (15) 81 (141) 66 (52)

T2

NO. without Distant Metastases (40)

p Value

T2

0.009

CO.OOO1 0.02 0.2 <0.0001 0.008

20(30)
0

Es$:

Metastases I"

77 (192)

k

T3a T3b T4 Grade: 2 3 Morphology: Papillary NonpapiUary Lymphatic-vascular invasion: No Yes No.tumors at c y s b WPY: 1 More than 1 Tumor size (cm.) Leas than 5 5 or Greater

0.02 68 (40) 58 (34) 30 (18) 13 (8)

19 (29) 16 (25) 20 (31) 10 (16)

0.48 11 (7) 152 (931

6 (10) 57 (90)

16 (17) 81 (83)

7 (16) 36 (84)

0.97

0.24

125 (74) 43 (26)

42 (67) 21 (33) 0.61

87 (62)

54 (38)

38 (66) 20 (34)

73 (79) 20 (21)

26 1741 9 (26)

143 (85) 26 (16)

61 (79) 14 (21)

84 (50) 85 (50)

24 (37) 41 (63)

26 (16) 139 (84)

10 (16) 52 (84)

0.61

Sex: M

F

F R.age (yrs.):

0.8 D-O-a-0

A

C T I

0.6-

0.4-

0

N

0.2-

L

4 LOCALCONTROL t LOCALFAILURE

0.0!

0

.

.

10

I

20

.

. 30

. 40

, 50

.

i

60

MONTHSAFTER SURGERY RG..2. Actuarial freedom f?om distant metastasis for all patients subdmded into those wth and without local control.

65 or Younger Older than 65 Pretreatment hem5 globin (gmfdl.): Less than 12 12 or Greater Retreatment blood m a nitrogen (mgfdl.): 20 or Less Greater than 20 Creatinine (mgfdl.): 1.6 or Leas Greater than 1.5 Excretory u r o p p h y : Normal Abnormal Pathological stage: wlalpis

m a P3W4 Lymphadenectomy: Neg. POS.

These results guided the construction of Cox proportional hazards models for multivariate analysis using freedom from distant metastasis as the end point. The independent pmgnostic value of clinical stage, creatinine level, pathological stage, pathological lymph node status, chemotherapy, patient age, lymphatic-vascular invasion and local failure was tested. The only significant covariates were pathological stage (p = 0.004), local failure (p = 0.007) and lymph node status (p = 0.03). When only pretreatment factors were included with local failure, the only significant covariate was local failure (p <0.0001).Cox proportional hazards analyses were also performed with local failure as a time-dependent variable. The most significant predictor of distant metastasis under these analytical conditions was local failure (p <0.0001).Pathological stage was also prognostic (p = 0.001), while lymph node status was not. The multivariate analyses demonstrate that the association of local failure with distant metastasis is independent of other known prognostic factors, including clinical stage. However, since local control for patients with clinical TU3a disease is already high (87% at 5 years), these patients are unlikely to benefit significantly from the addition of more aggtessive local measures, such as preoperative radiotherapy. In contrast, patients with clinical T3W4 disease had an actuarial 5-year local control rate of only 64% and, therefore,

Chemotherapy: No Yea

pValue

\

!mi

Clinical stage:

<0.0001

T2

1.

TABLE2. Distributwn of potential pmgnostic factors by distant metustrrses for all patients

0.26 0.08

0.96

0.74 131 (81) 30 (19)

50 (83) 10 (17)

145 (90) 16 (10)

49 (79) 13 (21)

0.03 0.99

46 (66) 24 (34)

19 (66) 10 (34)

38 (23) 49 (30) 76 (47)

1 (2) 14 (24) 44 (76)

132 (82) 30 (18)

34 (65) 28 (45)

60 (36)

14 (22) 51 (78)

o.OOO1


Chi-square teat waa used to determine p values.

should benefit from additional local therapy. The data in table 1 and the actuarial curves in figure 3 illustrate the profound impact of local control on freedom fmm distant metastasis in this subset. DISCUSSION

The most prominent pattern of failure in patients with muscle invasive transitional cell carcinoma of the bladder remains hematogenous spread despite the use of multiagent chemotherapy. While response rates in excess of 70% have been reported for patients with metastatic disease and encouraging results have been described using neoadjuvant and adjuvant chemotherapy, distant metastasis rates remain high.11-14 Our data reveal 5-year distant metastasis rates of 32% overall and 49% for those with clinical stage T3W4 disease. This high rate of distant metastasis was observed despite the fact that 85% of stage T3W4 patients received combination chemotherapy. Thus, our resulta and those of others in the combination chemotherapy era demonstrate

LOCAL CONTROL AND DISTANT METASTASIS IN BLADDER CANCER

2062

TABLE3. Relatwnship of potential prognostic factors to actuarial Ci-year fmeabm from distant metastases for all patients Faetors clinical stage: T2 T3a T3b T4 Grade: 2 3 Morphology: Papillary Nonpapillary Lymphatic-vascular invasion: No Yes No. tumors at cystoacopy: 1 More than 1 Tumor size (cm.): Less than 5 5 or Greater

%

Confidence Interval

74 76 51 52

62-83 64-85 41-72 31-72

55 68

29-79 61-75

67 64

45-82 53-73

D Value

0.004

0.50 0.92

70 63

61-77 50-75

62 71

52-71 59-80

-0- LOCALCONTROL t LOCALFAILURE

0.52

69 62

58-78 43-78

69 60

61-76 43-75

73 62

6281 53-7 1

0.0

0 0.47

10

20

30

40

50

MONTHS AFTER SURGERY FIG.3. Actuarial freedom fro? distant m e t a s e e s for patients with stage T3W4 disease subdiwded intn those w t h and without local control.

0.10

pt. age (yrs.):

65 or Younger Older than 65 Pretreatment hemoglobin

0.41

0.15

0.20

Sex: M F

I

0.59

(gmJdl.):

65 Less than 12 47-80 12 or Greater 69 61-76 Pretreatment blood urea nitrogen (mgJdl.): 20 or Less 68 60-75 70 Greater than 20 52-83 Creatinine (rngJd1.k 70 1.5 or Less 63-77 44 32-70 Greater than 1.5 Excretory urography: Normal 68 55-79 65 Abnormal 46-80 Pathological stage: Po/a/Pis 97 86-100 75 62-86 W3a P3w4 56 46-66 Lymphadeneetomy result Neg. 76 69-83 Pos. 41 28-47 Chemotherapy: 79 No 68-87 Yes 62 53-70 Log rank test was used to determine p values.

0.65

0.003 0.52 0.0001


somewhat disappointing cure rates for muscle invasive bladder cancer. Although new chemotherapeutic and biological agents are on the horizon, a reassessment of currently available methods may be equally rewarding. To this end we investigated whether such a rationale might support more aggressive local therapy with the goal of reducing distant metastasis and, ultimately, improving patient survival. Our results confirmed a strongly significant association between local failure and distant metastasis. This association was found for each clinical stage, and the groupings of TV3a and T3b/4 (table 1).Detailed univariate analyses were performed to assess if an identifiable prognostic factor could account for the association of local failure with distant metastasis. Of the pretreatment factors clinical stage and creatinine level correlated with distant metastasis (table 3). These factors were unevenly distributed between patients who were free of metastasis and those who were not (table 2). However, multivariate analyses using Cox proportional hazards models with these factors and local failure showed only local failure to be independently predictive of distant metastasis. These analyses were also performed on the subset of patients treated with cystectomy and the results were similar (data not shown). The emphasis on pretreatment factors

is relevant to the identification of conditions that warrant the addition of more aggressive local control measures, such as preoperative radiotherapy. The finding that no pretreatment factors correlated independently with distant metastasis when analyzed with local failure provides a strong argument in support of expanded efforts to reduce relapse locally. Preoperative radiotherapy is an effective method for reducing local failure: is associated with minimal morbidity and could be added to the currently fashionable cystectomy/chemotherapy protocols. A second multivariate analysis included posttreatment prognostic factors in addition to pretreatment factors. Pathological stage and lymph node involvement are well known correlates of distant metastasis, which was confirmed by our results. Multivariate analysis showed that these factors and local failure were prognostic for distant metastasis. The finding that local failure remained independent of pathological stage and lymph node status affirms the vigor of this association. Others have examined the relationship between local failure and distant metastasis for different tumor types, including those of the prostate, breast and r e ~ t u m . 1 6To ~ ~our knowledge our report represents the first establishment of such a relationship in patients with bladder cancer. It is striking that so few modem studies of bladder cancer address the issue of local control. The most common criticism of such analyses is that local failure is a marker of biological aggressiveness and that, regardless of whether local control is obtained, distant metastasis is imminent.16.m Although some clinical trials support this theory, others do not. Perhaps the best example is from the breast cancer literature. The National Surgical Adjuvant Breast and Bowel Project B-06trial showed that more aggressive local treatment in the form of breast irradiation significantly reduced the local failure rate yet had no effect on distant disease freedom or overall survival rates.Ig In contrast, Cuzick et a1 described a survival advantage with the addition of local radiotherapy in the more recent randomized trials involving simple mastectomy.20 The clinical significance of local control relative to distant metastasis and survival continues to be the subject of much debate.z1 The results are much clearer in animal studies in which local failure has been associated with significantly higher distant metastasis rates.22 Our data provide a sound rationale for maximizing local control, a result that may be achieved by reevaluating preoperative radiotherapy as an adjuvant to radical cystectomy and chemotherapy.

LOCAL CONTROL AND DISTANT METASTASIS IN BLADDER CANCER CONCLUSIONS

2063

long-term results of a controlled prospective study and further

clinical experience. J. Urol., 16%47,1995. In a recent comparison of preoperative radiotherapy plus cystectomy to cystectomy with or without chemotherapy we 15. Z W m , G. K., von Eschenbach, A C., Ayala, A G., Schdtheisa, T. E. and Sherman, N. E.:The influence of local control on noted a significant improvement in local control for patients metastatic dissemination of prostate cancer treated by extarwith clinical stage T3b disease treated with preoperative nal beam megavoltage radiation therapy. Cancer, 68: 2370, radi~therapy.~ There were too few patients with T4 disease to 1991. perfom a meaningful Of this subset* For those 16. Fisher, B., Andeman, S.,Fisher, E.R., Redmond, C., Wickerham, with clinical T3W4 disease our present study revealed a D. L., Wolmark, N., Mamounas, E. P., D e u k h , M. and &year local failure rate of 36%, a distant metastasis rate of Margolese, R.: Significance of ipsilateral breast tumour recur49% and an estimated distant metastasis rate of 81% for rence after hmpectomy. Lancet, ss8: 327, 1991. those with evidence of local failure. These high local and 17. Vigliotti, A, Rich, T. A., Romdahl, M. M.,Withers, H. R. and Oswald, M. J.: Postoperative adjuvant radiotherapy for adedistant failure rates, taken in the context of a strong associnocarcinoma of the rectum and recbsigmoid. Int. J. Rad. Onation between the 2 end points (fig. 31, indicate that patients col. Biol. Phys., 13: 999, 1987. in this subset may benefit from the addition of preoperative radiotherapy. Current treatment strategies for patients with 18. Walsh, p. c.:M u v a n t radiotherapy after radical prostatectomy: is it indicated? J. Urol., 138: 1427, 1987. clinical T3b/4 bladder cancer are inadequate. New treatment 19. B. and -Ond* '.: LumpectOmY for bresst cancer: an approaches, including the integration of preoperative radioupdate of the NSABP experience. National Surgical Adjuvant therapy, shouId be considered in the design of future clinical Breast and Bowel Project. Monogr. Natl. Cancer Inst., 11: 7. trials for this patient population. 1992. REFERENCES

1. Wishnow, K. I., Levinson, A. K., Johnson, D. E., Tenney, D. M., Grignon, D. J., Ro, J. Y., Ayala, A. J., Logothetis, C. J., Swanson, D. A., Babaian, R. J. and von Eschenbach, A. C.: Stage B (P2/3A/NO) transitional cell carcinoma of bladder highly curable by radical cystectomy. Urology, 39 12, 1992. 2. Brendler, C. B., Steinberg, G. D., Marshall, F. F.,Moatwin, J. L. and Walsh, P. C.: Local recurrence and survival following nerve-sparing radical cystoprostatectomy. J. Urol., 144: 1137, 1990. 3. Greven, K M., Spera, J. A., Solin, L. J., Morgan, T. and Hanks, G. E.: Local recurrence after cystectomy alone for bladder carcinoma. Cancer, 6 9 2767, 1992. 4. Cole, C. J., Pollack, A., Zagars, G. K, Dinney, C. P., Swanson, D. A. and von Eschenbach, A C.: Local control of muscleinvasive bladder cancer: preoperative radiotherapy and cystectomy versus cystectomy alone. Int. J. Rad. Oncol. Biol. Phys., 3 2 331, 1995. 5. Mostofi, F. J., Sobin, L. H. and Torlain, H.: Histological ?Lping of Urinary Bladder Tumors. Geneva: World Health Organization, 1973. 6. Beahrs, 0. H. and Myers, M. H.: Manual for Staging of Cancer, 2nd ed. Philadelphia: J. B. Lippincott Co., chapt. 29, pp. 171176, 1983. 7. Altman, D. G.: Practical Statistics for Medical Research. New Yo&: Chapman and Hall, chapb. 9-10> PP. 179-276,1991. 8. Hams, E. K. and Albert, A: Survivorship Analysis for Clinical Studies. New York: Marcel Dekker Inc., pp. 5-125, 1991. 9. Rothman, K J.: Estimation of confidence limits for the nunulative probability of survival in life table analysis. J. Chronic Dis., 31: 557, 1978. 10. Cox, D.R., and Oakes, D.: Timedependent covariates. In:Analysis of Survival Data. New York Chapman and Hall, chapt. 8, pp. 112-141, 1984. 11. Logothetis, C. J., Dexeus, F. H., Chong, C., Sella, A., Ayala, A G., &,, J.y. and mat, s.: Cisplatin, cyclophosph-de and doxe mbiein chemotherapy for mesectable UrOthefial tumors: the M. D. Anderson experience. J. Urol., 141: 33,1989. 12. Sternberg, C. N., Yagoda, A., Scher, H. I., Watson, R. C., Herr, H. W., Morse, M.J., Sogani, P. C., Vaughan, E. D., Jr., Bander, N., Weiselberg, L. R., Geller, N., Hollander, P. S., Lipperman, R., Fair, W. R. and Whitmore, W. F.,Jr.: M-VAC (methotrexate, vinblastine, doxorubicin and cisplatin) for advanced transitional cell carcinoma of the urotheliurn. J. Urol., 139 461, 1988. 13. Skinner, D, G., Daniels, J . R., Russell, C. A, Lieskovsky, G., Boyd, S. D., Nichols, P., Kern, W., Sakmoto, J., =do, M. and Groshen, S.: The role of adjuvant chemotherapy followingcystechrny for invasive bladder cancer: a prospective comparative trial. J. Urol., 145: 459, 1991. Wellek, S., Voges, G. E., Rossmann, 14. Sttickle, M., Meyenburg, W., M., Gertenbach, U., Thiiroff, J. W., Huber, C. and Hohenfellner, R.: Adjuvant polychemotherapy of nonorganconfined bladder cancer after radical cysteetomy revisited:

20. Cuzick, J., Stewart, H., RutqVist, L., Houghton, J., Edwards, R., Redmond, C., Peto, R., Baum, M., Fisher, B., Host, H., Lythgoe, J., Ribeiro, G. and Scheurlen, H.: Cause-specific mortality in long-term survivors of breast cancer who participated in trials of radiotherapy. J. Clin. Oncol., 12 447, 1994. 21. Lippman, M. E.: How should we manage breast cancer in the breast, or Buddy, can you paradigm? J. Natl. Cancer. Inst., 87: 3, 1995. 22. Ramsay, J., Suit, H. D. and Sedlacek, R.: Experimental studies on the incidence of metastases after failure of radiation treatment and the effect of salvage surgery. Int. J. Rad. Oncol. Biol. Phys., 14: 1165,1988. EDITORIAL COMMENT The authors reviewed a large seriea of patients who underwent radicalcystectomyat asingleinstitutionformuscleinvaeivebladdercancer. They identified a e&yi6-t relationship between local failure and distant metastasis, which is independent of preoperative &d stage independent of and grade and, mom significantly, stage and lymph node status. Therefore, they suggest that futuretreataer be merit strawes for patients with mwle hVasive directed at local failure and theyrefer to from institution in which a sigdicant improvement in local control was radiotherapy (referene 4 in article). 6th lt is a d tto the siwace of this report without a more detailed analysis of pathological stage. and its correlation to local control and metastasis. For example,it is not eta^ in the articlewhat percent ofpatienb with po to 3a tumorshad local recurrenee to thmwith p3b/4 tumors. since a third of invasive bladder cancer e88e8 are understaged clinically, it is easy to understand why local failure wss a better predietor of distant than cbcal stape. However, it had how many patients withpathologicauy

a~~~l~cTitiealimportancebecauseripreop extension Or lymph node metastases. since moat recent studies more found Preoperative radiotherapy to be iIK?ffeCtiWin talitY from bladder cancer, the authors cannot make a strong case for preoperative radiotherapy without a more wmplete analysis of Patholo@cal Charles B. Brendler Department of Surgery Sectwn of Urology University of Chicago Chicago, Illinois REPLY BY AUTHORS

we appreciate the Editorial Comment, which addresses the main concern that we did not perform a thorough investigation Of the relationship of pathological stage to local recurrence.Such analyses are not relevant to the main Point Of article. our Primary goal was to determine whether there is a strong relationship between local control and distant metastasis as a rationale for implementing more aggressive local treatment preopemtively in specific patient

2064

LOCAL CONTROL AND DISTANT METASTASIS IN BLADDER CANCER

eubeets. The decision to administer radiotherapy preoperatively

would be guided by preoperative prognostic factors and not pathological stage. As we reported previously (reference 4 in article), clinical stage T3W4 tumors have high local failure rates and, therefore, it is this group that stands to benefit the most from this

approach. Thus,a detailed analysis of the local control rates based on pathological stage would not affect our conclusions. We contend that the patients most likely to benefit from preoperative radiotherapy are those with perivesical extension andor limited pelvic lymph node metastases.