Surgical therapy for anorectal melanoma

BACKGROUND: Anorectal melanoma is a rare but highly lethal malignancy. Historically, radical resection was considered the "gold standard" for treatment of potentially curable anorectal melanoma. The dismal prognosis of this disease has prompted us to recommend wide local excision as the initial therapeutic approach. The purpose of this study was to review our results in patients who underwent wide local excision or radical surgery (abdominoperineal resection [APR]) for localized anorectal melanoma. STUDY DESIGN: We reviewed the charts of all patients referred for resection of anorectal melanoma between 1988 and 2002. Endpoints included overall survival, disease-free survival, and local, regional, or systemic recurrence. RESULTS: Fifteen patients underwent curative-intent surgery; four underwent APR and 11 underwent wide local excision. Eight patients (53%) are alive; 7 (47%) are disease-free (followup 6 months to 13 years). Of 12 patients who have been followed for more than 2 years, 4 are alive (33%) and 3 are disease-free (25%). Seven patients have been followed for more than 5 years and two are alive and disease-free (29%). All of the longterm survivors underwent local excision as the initial operation. There were no differences in local recurrence, systemic recurrence, disease-free survival, or overall survival between the APR group and the local excision group. Local recurrence occurred in 50% of the APR group and 18% of the local excision group; regional recurrence occurred in 25% versus 27%. Distant metastases were common (75% versus 36%). CONCLUSION: In patients who have undergone resection with curative intent for anorectal melanoma, most recurrences occur systemically regardless of the initial surgical procedure. Local resection does not increase the risk of local or regional recurrence. APR offers no survival advantage over local excision. We advocate wide local excision as primary therapy for anorectal melanoma when technically feasible.     

Late recurrence of malignant melanoma. Analysis of 168 patients.

Analysis of 7104 patients with melanoma seen at Duke University identified 168 who experienced their first recurrence 10 or more years after diagnosis, for an incidence of 2.4%. This included patients with all stages of disease. There was no sex, age, or primary site predominance. The mean disease-free interval for cutaneous melanomas was 14.3 years versus 22.3 years for ocular primary melanomas. The prognosis following relapse was related to the site of recurrence. Survival after local or regional node recurrence was often prolonged; survival after distant metastases was usually limited. Patients with ocular primaries had the highest incidence of distant metastases, and the shortest subsequent survival. An additional 483 patients were identified who survived 10 or more years without evidence of recurrence; of these 651 patients with long disease-free intervals, 25% (168 of 651) developed recurrent disease. This demonstrates that a 10-year disease-free interval cannot be considered a cure, and emphasizes the importance of continued annual follow-up.     

Selected Patients with Metastatic Melanoma May Benefit from Liver Resection

Background In the last few years there has been expanding use of hepatic resection for non-colorectal metastases. The purpose of this study is to evaluate the experience of liver resection for patients with metastatic melanoma. Methods Eighteen patients with metastatic melanoma were explored for possible surgical resection. All patients fitted the following criteria: absence of extra-hepatic disease after evaluation with CT/MRI and FDG-PET scans; disease-free interval longer than 24 months after the resection of the primary melanoma; presumed completely resectable lesions; absence of clinical co-morbidities. Results Liver resection was performed in 10 patients; 8 out of 18 presented with irresectable tumors and/or peritoneal metastases and were not operated. One patient presented with postoperative biliary fistula and was conservatively managed. No other complications or postoperative mortality were observed. After a mean follow-up of 25.4 months, 5 patients are alive and without evidence of recurrence. Overall median survival was 22 months; overall survival and disease-free survival were 70% and 50% respectively. Conclusions Resection of liver metastases from melanoma in a selected group of patients may increase survival. Exploratory laparoscopy should be included in the preoperative armamentarium of diagnostic tools.     

Resection of pulmonary metastases from malignant melanoma: results of a 16-year experience

Between 1970 and 1986, 49 patients had resection of presumed pulmonary metastases from malignant melanoma. Sixteen patients were found to have benign disease only despite the appearance of a new nodule in 13. Patients with benign disease had a significantly longer mean survival (169 months) compared with the group with malignant disease (22 months). Median survival for all patients with malignant disease was 13 months. Survival after resection did not correlate with the Clark level of the original lesion, lymph node status, disease-free interval, or number of nodules on preoperative tomograms. Two of 10 patients with 1 nodule resected are long-term survivors (88 and 120 months). Exploration in patients with presumed pulmonary metastases from melanoma is justified to rule out benign disease even if a new solitary nodule is detected. There are no prognostic indicators predicting survival after resection of melanoma metastases, and a significant number of patients will have benign disease.     

Recurrent malignant melanoma: the identification of prognostic factors to predict survival.

The prognostic factors for stage 1, 2 melanoma have been elucidated. Tumor thickness, ulceration of the primary melanoma, and perhaps, primary site may be used to predict the percentage of patients with regional nodal disease or systemic metastases and the prognosis of patients who have only cutaneous disease at diagnosis. Very little is known about prognosis once there is a recurrence. A retrospective, computer-aided chart review identified 4,185 patients registered at the Duke University Melanoma Database who had stage 1, 2 disease at diagnosis. During a mean follow-up period of 7 years, 35.9% experienced a recurrence. Local regional recurrences explained 62.5% to 85.5% of the recurrences. Even after elective node dissections, local regional recurrences explained most relapses (58.1%). Sixty-five percent of the recurrences occurred within the first 3 years of of follow-up. There was a pronounced difference in 5-year survival in those patients who suffered a recurrence sometime during their clinical course compared with those who never relapsed (p = 0.00001, for trunk primary melanoma). Patients with local or regional recurrence have a better prognosis than patients who relapse systemically, with 5-year survivals from the time of recurrence of 55% for a patient with a local recurrence, 51% for a patient with a regional nodal recurrence, and 20% for a patient with a systemic recurrence. A multivariate regression analysis identified thickness, ulceration of the primary melanoma, and age and location of the primary melanoma on the extremity as variables that predicted prognosis. The only factors concerning the recurrent state that added prognostic information was the disease-free interval and the presence of systemic metastases as the initial recurrence.     

Pulmonary metastasectomy in a patient with malignant melanoma after a disease-free interval of 15 years

Prognosis in melanoma patients after resection of pulmonary metastases is poor. Little information is available about the relationship between long-term disease-free interval and survival rate after pulmonary metastasectomy. We report a patient treated for malignant melanoma of the forehead who developed a pulmonary metastasis after a disease-free interval of 15 years. Lobectomy was performed, but unfortunately 12 months later the patient died of brain metastases.     

Partial Penectomy for Patients With Squamous Cell Carcinoma of the Penis: The Memorial Sloan-Kettering Experience

Background To present our institution’s experience with squamous cell carcinoma (SCC) of the penis, with analysis of oncologic efficacy and survival. Methods Between 1989 and 2005, we identified 32 consecutive patients (median age, 61 years) with SCC of the penis managed with partial penectomy. Clinicopathologic variables were examined, and overall and disease-specific survival were determined. Results Pathologic stage of the primary tumor was pTis in 1 patient (3%), pT1 in 11 (34%), pT2 in 16 (50%), and pT3 in 4 (13%). Pathologic grade was well differentiated in 9 patients (28%), moderately differentiated in 20 (63%), and poorly differentiated in 2 (6%). Twenty-five patients (78%) underwent inguinal lymph node dissection, with 15 (60%) demonstrating nodal metastases. Twenty-two patients (69%) underwent pelvic lymph node dissection; 21 were negative for pelvic nodal metastases, and 1 had grossly positive nodes. One patient developed local recurrence. After a mean follow-up of 34 months, overall survival was 56%. Numbers of patients alive and disease-free were 9 and 11 in the low-stage and advanced-stage groups, and 8 and 12 in the well and moderately differentiated groups, respectively. Both patients with poorly differentiated disease died of disease within 12 months from presentation. Conclusions Partial penectomy for SCC of the penis provides excellent local control, with low recurrence rate, and acceptable maintenance of urinary and sexual function. Outcomes are generally poor, however, for patients with regional metastases, even in moderately differentiated disease. Future studies are needed to identify a reliable method of predicting regional metastases.     

Resection of Noncolorectal and Nonneuroendocrine Liver Metastases: Late Metastases Are the Only Chance of Cure

Resection of liver colorectal metastases allows a 5-year survival in 25% to 35% of patients. The outcome of patients with noncolorectal metastases is unknown because of the heterogeneity of this group. The aim of this retrospective study was to evaluate predictive factors of survival in patients who underwent resection of noncolorectal and nonneuroendocrine (NCRNE) liver metastases. From 1980 to 1997, 284 patients underwent hepatectomy for liver metastases of whom 39 (25 men and 14 women, mean age 55 years) had curative resection for NCRNE liver metastases. No patients had extrahepatic disease. The primary tumors were gastrointestinal (n = 15), genitourinary (n = 12) and miscellaneous (n = 12). The mean number of metastases was 1.8, and the mean size of the lesions was 51 mm. The median disease-free interval was 27 months. Twenty patients had a major hepatectomy and 19 a minor resection, with simultaneous resection of the primary in 6 cases. Overall survival was evaluated using the Kaplan-Meier method. There was no operative mortality, and 8% morbidity. The survival at 1, 3, and 5 years was 81, 40, and 35%, respectively. Patients with a disease-free interval higher than 24 months had a greater survival rate than those with a disease-free interval of less than 24 months (100% vs. 10%; p = 0.0004). Survival was not significantly influenced by age, sex, type of primary tumor, number, size and localization of metastases, type of hepatectomy, or blood transfusion. Resection of NCRNE liver metastases should be justified for patients without extrahepatic disease and resectable metastases, especially for those who have a disease-free interval of more than 24 months.     

Isolated regional perfusion in the treatment of subungual melanoma

Subungual melanoma is rare and represents only 1% to 3% of all diagnosed melanomas in Western countries. The tumor is frequently mistaken for a benign lesion and the delay in diagnosis and final treatment may be responsible for the high local recurrence rate and the low disease-free survival rate. From 1965 to 1982 the combined-modality therapy of amputation and adjuvant isolated regional perfusion with melphalan with or without dactinomycin was used in the treatment of 22 patients with subungual melanoma. Disease was staged according to the M. D. Anderson classification, as follows: stage I (primary melanoma), 11 patients; stage IIIA (in-transit metastases and/or satellitosis), three patients; stage IIIB (regional lymph nodes), seven patients; and stage IIIAB (in-transit metastases and/or satellitosis and regional lymph nodes), one patient. There were no cardiovascular complications and no treatment mortality. During a follow-up of at least 4.5 years, 12 patients (55%) developed distant metastases, including four patients with stage I disease (36%) and eight patients with stage III disease (73%). There were no locoregional recurrences. The median survival was three years (range, 0.5 to 12.5 years) and the overall five-year survival was 40%, with 56% of patients having stage I disease and 27% having stage III disease. The prognosis of subungual melanoma is determined by the stage of the disease. Isolated regional perfusion may prolong disease-free survival in patients with subungual melanoma compared with previously published data.     

Melanoma of the penis, scrotum and male urethra: a 40-year single institution experience

PURPOSE: Genitourinary melanoma is rare and classically associated with a poor prognosis. We describe our experience with 10 patients with penile or urethral involvement. In addition, we present what is to our knowledge the largest reported series of melanoma of the scrotum (6 cases). MATERIALS AND METHODS: We reviewed the records of 16 men who presented consecutively to our institution with genitourinary melanoma between 1962 and 2000. Clinical and pathological characteristics were assessed, including Breslow thickness, primary surgical intervention and clinical course. RESULTS: Of 10 patients with penile or urethral melanoma 1997 American Joint Committee on Cancer melanoma pathological stage was T1 (depth less than 0.75 mm) in 4, T2 (0.75 to 1.5 mm) in 3 and T3 (1.51 to 4 mm) in 3. Only 1 of 4 patients with clinically palpable inguinal nodes had inguinal metastases at lymphadenectomy (BILND) and 3 who underwent prophylactic superficial BILND had negative findings. In 7 patients with T1-2N0M0 disease there were no local recurrences after wide local excision (WLE) or partial penectomy at a median followup of 35 months. Six of 7 men were rendered disease-free. One patient died of melanoma that developed at a second primary site. The 3 patients with T3 tumors who underwent partial (2) or radical (1) penectomy with or without BILND died of disease (2) or had progression (1). In all patients with penile melanoma the 5-year actuarial disease specific and recurrence-free survival rates were 80% and 60%, respectively, at a median followup of 39 months (range 20 to 210). Six patients with scrotal melanoma were treated with WLE without local recurrences. Three of the 6 patients had palpable inguinal nodes, of whom 2 died after chemotherapy for unresectable disease and 1 died of other causes 51 months after negative BILND. The 3 men with clinically negative groins who did not undergo prophylactic BILND had distant (1) or regional (2) metastases and died of disease. In patients with scrotal melanoma the 5-year actuarial disease specific and recurrence-free survival rates were 33.3% and 33.3%, respectively, at a median followup of 36 months. CONCLUSIONS: Partial penectomy or WLE provided effective local control for low stage penile or urethral melanomas and all scrotal lesions. Patients showing clinically positive, proven metastasis died despite appropriate surgical procedures and multi-agent chemotherapy. Prophylactic modified inguinal lymphadenectomy should be considered in select patients with penile, scrotal and anterior urethral melanoma.     

Anorectal melanoma: a large clinicopathologic study from India

Anorectal melanoma is a rare tumor. It has dismal prognosis; only 10% of patients live longer than 5 years. It commonly presents as bleeding from the rectum or as a hemorrhoidal mass. The management of this tumor is controversial. Some authors believe that a palliative local excision is the treatment of choice, whereas others recommend radical surgery in localized disease. We present a retrospective analysis of 72 patients who were managed at our center between 1990 and 2001. This is the largest series from India. Most patients (48/72; 66%) presented with distant metastases. Twenty-four patients (24/72; 33%) underwent an abdomino-perineal resection. Nineteen patients (19/24) had positive lymph node disease, and the mean disease-free survival in these patients was 10.3 months. Disease-free survival in the node-negative patients was 26.5 months. A subset of patients with localized disease can benefit from radical surgery. In patients with large bulky localized disease, radical surgery provides better palliation than local excision.     

Malignant melanoma and pregnancy.

Among 100 patients diagnosed with melanoma during pregnancy and followed a mean of 6.8 years, when compared with a nonpregnant female population, there was a significantly shorter disease-free interval for the pregnant group. Median disease-free intervals were 5.8 and 11.9 years, respectively. The time to development of lymph node metastases was shorter in the pregnant patients (p = 0.015). Nodal metastases developed in 48% of the pregnant patients and only 26% of the nonpregnant patients, at 10 years. Multivariate analysis demonstrated that pregnancy at diagnosis was significantly associated with the development of metastatic disease (p = 0.008), when controlling for tumor site, thickness, and Clark level. Pregnancy, however, was not a risk factor for patient mortality. The literature continues to be split on the role of pregnancy in melanoma; however, most recent series show no difference in survival. Multiple studies have failed to show significant effects of female hormones on melanoma cells or on the incidence or progression of melanoma.     

Resection of lung metastases: long-term results and prognostic analysis based on 5206 cases--the International Registry of Lung Metastases

The International Registry of Lung Metastases was established in 1991 to asses the long-term results of pulmonary metastasectomy. The Registry has accrued 5206 cases of lung metastasectomy, from 18 departments of thoracic surgery in Europe (n = 13), USA (n = 4) and Canada (n = 1). Of these patients 4572 (88%) underwent complete surgical resection. The primary tumor was epithelial in 2260 (43%), sarcoma in 2173 (42%), germ cell in 363 (7%), and melanoma in 328 (6%) patients. The disease-free interval was 0 to 11 months in 1729 (33%) cases, 12 to 35 months in 1857 (36%) and more than 36 months in 1620 (31%). Single metastases accounted for 2383 (46%) cases and multiple lesions for 2726 (52%). Mean follow up was 46 months. Analysis was performed by Kaplan-Meier estimates of survival, relative risk of death and multivariate Cox model. The actuarial survival after complete metastasectomy was 36% at 5 years, 26% at 10 years and 22% at 15 years (median 35 months); the corresponding values for incomplete resection were 13% at 5 years and 7% at 10 years (median 15 months). Among complete resections, the 5-year survival was 33% for patients with a disease free-interval of 0 to 11 months and 45% for those with a disease-free interval of more than 36 months; 43% for single lesions and 27 for four or more lesions. Multivariate analysis showed a better prognosis for patients with germ cell tumors, disease-free interval of 36 months and more and single metastases. These results confirm that lung metastasectomy is a safe and potentially curative procedure.     

Thoracotomy for metastatic malignant melanoma of the lung.

The outcome of 29 patients who underwent lung resection for treatment of metastatic malignant melanoma from January 1976 to November 1988 was studied. Twenty-two patients underwent total resection for cure of all apparent metastatic disease, whereas seven patients did not undergo total resection. Of the 22 patients who underwent curative resection, the median survival was 11 months, with a 2-year survival of 13.6% and a 5-year survival of 4.5%. Four patients who underwent curative resection are currently alive and free of disease, with one patient surviving more than 10 years. The patients who underwent palliative resection had a median survival of 5 months, only one patient living longer than 10 months. The difference in survival of the patients who underwent curative resection compared with palliative resection was statistically significant. The thickness of the primary cutaneous malignant melanoma, the presence of regional lymph node metastases, the disease-free interval from primary diagnosis to metastatic pulmonary disease, and whether one or two metastatic nodules were removed during curative lung resection were not statistically significant in altering survival. These results demonstrate that although prolonged survival for metastatic melanoma is rare, lung resection in selected patients may be associated with long-term survival.     

Adjuvant chemotherapy in malignant melanoma using dacarbazine, carmustine, cisplatin, and tamoxifen: a University of South Florida and H. Lee Moffitt Melanoma Center Study.

A prospective, nonrandomized trial was performed of the four-drug chemotherapy protocol consisting of dacarbazine, carmustine, cisplatin, and tamoxifen citrate given to high-risk patients for recurrence of melanoma after local regional treatment. The treated patients were consecutively registered and 6 patients who did not elect to be treated served as the control population. Criteria for inclusion in the trial were the presence of four or more lymph nodes positive for metastatic melanoma on regional modal dissection, the presence of metastatic disease in second station lymph node areas such as the iliac basin, greater than 5 cm in maximal diameter tumor burden in the nodal basin, and patients who had resected stage 4 (systemic metastases) disease with clear margins and were rendered free of disease. Actuarial survival curves for the treated group and the control subjects were similar (p = 0.91). There was a definite trend toward an increased disease-free survival for the group receiving adjuvant chemotherapy (p = 0.09). The mean disease-free survival for the control population was 200 days and for the treated group, 600 days. The study suggests a therapeutic benefit for adjuvant chemotherapy treatment of patients with metastatic melanoma who have been rendered free of disease but are at high risk for recurrence.     

Patterns of recurrence after sentinel lymph node biopsy for cutaneous melanoma

BACKGROUND: Previous sentinel lymph node (SLN) studies for cutaneous melanoma have shown that the SLN accurately reflects the nodal status of the corresponding nodal basin. However, there are few long-term studies that describe recurrence site patterns, predictors for recurrence, and overall survival and disease-free survival after SLN biopsy. METHODS: A retrospective review of patients over a 6-year period was performed to determine patient outcomes and the patterns of recurrence. In all cases, Tc-99 sulfur colloid along with isosulfan blue dye was injected at the primary melanoma site. After resection, the SLN was serially sectioned and evaluated by hematoxylin and eosin staining and immunohistochemistry. RESULTS: One hundred ninety-eight patients were identified who underwent SLN biopsy for cutaneous melanoma including T1 (n = 21), T2 (n = 88), T3 (n = 75), and T4 (n = 14) primary tumors. Of these patients, 38 had a positive SLN. Of the 38 patients with a positive SLN (mean follow-up 38 months), recurrent disease was identified in 10 (26.3%) at a mean interval of 14.2 months. The site of first recurrence was distant (n = 4) and local (n = 6). Regional lymphatic basin recurrence was not identified. Of the 160 patients with a negative SLN (mean follow-up 50 months), recurrent disease was identified in 16 (10.0%) at a mean interval of 31.3 months. The site of first recurrence was systemic (n = 11), local (n = 4), and nodal (n = 1). Overall survival and disease-free survival for patients with a positive SLN at 55 months was 53.3% and 47.7% respectively, while overall survival and disease-free survival for patients with a negative SLN at 53 months was 92.2% and 87.7% respectively (P <0.01). Univariate and multivariate analysis of the entire cohort (n = 198) identified primary tumor depth and positive SLN status as significant predictors of recurrence. CONCLUSIONS: The incidence of nodal basin recurrence after SLN biopsy was found to be 0.6%. Primary tumor depth and pathological status of the SLN are significant predictors of local and systemic recurrence. Long-term follow-up indicates that patients with a positive SLN clearly recur sooner and have decreased overall survival than those with a negative SLN.     

Results of ilioinguinal dissection for stage II melanoma.

Eighty-two Stage II melanoma patients with inguinal lymph node metastases have undergone ilioinguinal node dissections at UCLA during the past 10 years. Twenty-four (29.3%) patients had involvement of both inguinal and iliac nodes, whereas 58 (70.7%) patients had only inguinal metastases. The frequency of iliac metastases did not relate to location, Clark's level or thickness of the primary tumor or interval from diagnosis of primary tumor to lymphadenectomy, but was related to the number of inguinal nodes involved with metastases, rising from 14.6% with one positive inguinal node to 50% with four or more inguinal node metastases. Twenty of 24 (83.3%) patients with inguinal and iliac node metastases developed recurrent disease, whereas 32/58 (55.2%) patients with only inguinal node metastases and no tumor in the iliac nodes recurred. The time to recurrence was much shorter if iliac nodes were diseased (median disease-free interval 5.8 months versus 25.6 months). Three of five patients with clinically negative but histologically positive inguinal and iliac nodes survived 5 years, while only 1/18 patients with clinically positive inguinal nodes and diseased iliac nodes lived 5 years. Those with clinically negative but histologically positive inguinal nodes and iliac metastases had recurrence and survival rates similar to those with clinically negative but histologically positive inguinal nodes and no iliac metastases. Ilioinguinal lymphadenectomy provides significant prognostic information for Stage II patients with inguinal metastases and may be therapeutic for those with iliac metastases. Therefore, ilioinguinal dissection is the operation of choice for melanoma patients with regional metastases to the inguinal area.     

Radical prostatectomy for carcinoma of the prostate: long-term follow-up of 115 patients

From July 1969 to September 1990, 370 patients with prostatic cancer underwent radical prostatectomy at our institution. Of these 370 patients, 115 consecutive patients could be followed for more than 10 years (mean 12.5). Patients with stage pT1-pT3 tumors received no further treatment until progression occurred. Patients with regional lymph node metastases (stages pT2-3pN1-2M0) were treated by either an immediate orchiectomy or an adjuvant hormonal therapy. No radiotherapy was applied prior to radical prostatectomy or thereafter. Of the 115 patients followed for more than 10 years, 84 had stage pT1-2, 22 had stage pT3, and 9 had stage pT2-3pN1-2 tumors. The observed 10-year survival rate of all 115 patients (including those with regional lymph node metastases) was found to be 67.0%. The 10-year disease-free survival rate was 58.3% and the tumor-related survival rate was 83.5%. Considering only patients with locally confined (stage pT1-2) tumors, the 10-year survival rate was 75.0%. This observed survival rate equals the 10-year survival expectancy of a male age-matched control population (69.9%). Progression (local recurrence or distant metastatic spread) was noted in 27.8% of patients within the 10-year interval after radical prostatectomy. Within this time interval, 16.5% of the patients died from their disease.     

Long-term results of surgical resection of lung metastases

Between 1990 and 2000, 56 consecutive patients underwent lung resection for removal of metastatic disease. Mortality, disease-free interval, and overall survival were studied. Only patients with a complete follow-up were included and data were collected conform the protocol of the International Registry of Lung Metastases. The primary tumour in our series was an epithelial tumour in 25 patients (45%), sarcoma in 15 (27%), germ cell tumours in 11 (19%) and melanoma in 5. Operative mortality was 1.4% (1 out of 73 procedures). Germ cell tumours had the best survival (76% at 5 years), and melanoma the worst (0% at 5 years). Multivariate analysis showed that survival for patients who underwent 2 or more metastasectomies was surprisingly good with a 5-year survival rate of 46%. Survival was not related to disease-free interval, multiple lung metastases, or pneumonectomy. It is in accordance with some reports that a short disease-free interval, numerous lung metastases, or recurrence after the first metastasectomy should not preclude patients from operation.     

Results of surgery following radical radiotherapy for invasive bladder cancer

Five hundred and ninety-one of 889 patients with T1 to T4 transitional cell carcinoma of the bladder had persistent or recurrent cancer after radical radiotherapy. Durable local control was significantly poorer for patients with grade 1 or T4 cancer before radiotherapy. Three hundred and twenty-two patients received additional surgical treatment: 211 were endoscopically managed and 111 had secondary cystectomy. The survival of patients with residual or recurrent cancer after radiotherapy was significantly improved by secondary local treatment (P less than 0.0001). A comparison was made between endoscopic treatment and cystectomy after radiotherapy. Patients having secondary cystectomy were younger (mean age 60.0 years) than those managed endoscopically (66.8 years). The 5-year actuarial survival rate (from the date of radiotherapy) for patients who had endoscopic treatment was 47.1% compared with 62.5% for those who had cystectomy (P = 0.16). After both treatments survival was significantly correlated with the T category of the tumour before radiotherapy. Local tumour control was better after cystectomy; 85.6% of patients were locally tumour-free at the end of follow-up compared with 44.5% of those managed endoscopically. There was no overall difference in the subsequent risk of metastases between the two forms of surgery. However, seven of 12 patients managed endoscopically prior to secondary cystectomy died of their cancer. Five of these patients died from metastases even though they were locally disease-free. There was a significantly increased risk of metastases in patients managed endoscopically who were not locally disease-free after treatment (P = 0.0003). Caution is advised in persisting with endoscopic treatment after radiotherapy if local control is not readily achieved.