Prognostic factors in male urethral cancer

Authors


  • See editorial on pages 2361–3, this issue.

Abstract

BACKGROUND:

Male urethral cancer is a rare neoplasm, with the published literature consisting of small single-institution retrospective series. As such, there is no objective analysis of prognostic factors and treatment outcome. The author sought to use the population-based Surveillance, Epidemiology, and End Results (SEER) database to evaluate prognostic factors in male urethral cancer.

METHODS:

From 1988 to 2006, 2065 men were identified in the SEER database as having primary urethral cancer. Median follow-up was 2.5 years. Cancer-specific and overall survival was computed using the Kaplan-Meier method, and Cox proportional hazards analysis was used to evaluate patient age at diagnosis, year of diagnosis, race, histologic type, grade, T stage, nodal status, M stage, extent of surgery, and type of radiation as potential significant independent predictors of survival.

RESULTS:

Overall survival at 5 and 10 years was 46.2% (95% confidence interval [CI], 43.9-48.6%) and 29.3% (95% CI, 26.6-32.0%), respectively, whereas cancer-specific survival at 5 and 10 years was 68.0% (95% CI, 65.5-70.5%) and 60.1% (95% CI, 57.0-63.2%), respectively. Advanced age, higher grade, higher T stage, systemic metastases, other histology versus transitional cell carcinoma (TCC), and no surgery versus radical resection were predictors of death and death from disease, whereas adenocarcinoma was associated with a lower likelihood of death and death from disease as compared with TCC. In addition, nodal metastasis was a predictor of death. Surgery had a better outcome than radiation for stage T2-T4 nonmetastatic disease.

CONCLUSIONS:

Age, grade, TNM stage, histology, and extent of surgery were predictive of overall and cancer-specific survival. Cancer 2011. © 2010 American Cancer Society.

Primary male urethral neoplasms are rare. With the largest single-institution series consisting of only 46 patients,1 assessment of prognostic factors and treatment outcome are difficult. Given the rarity of the disease and heterogeneity of treatment strategies in reported series, defining the role of surgery, radiation, and multimodality treatment has been a challenge. To overcome the sample size limitation of single-center reports, the author used the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) database2 for 1973 to 2006 to evaluate prognostic factors in male urethral cancer and assess treatment outcome.

MATERIALS AND METHODS

Between January 1973 and December 2006, 2724 men were identified in the SEER database2 from all 17 registries as having primary urethral cancer with International Classification of Disease-Oncology (ICD-O) urethra site code C68.0. ICD-O code C68.0 excludes bladder tumors and tumors at the bladder neck or arising from the paraurethral gland, but includes tumors of Cowper gland, the prostatic utricle, and urethral glands. Detailed extent of disease coding was available in SEER only after 1988. Thus, 659 men diagnosed from 1973 to 1987 were excluded, including 2 men with autopsy-only cancer. The remaining 2065 men diagnosed from 1988 to 2006 comprise the study population. None of these patients was coded as having a primary bladder tumor in the SEER database; all of these records were cross-linked with those of patients having primary bladder cancer, with none identified as having both. The histologic type was classified as transitional cell carcinoma (TCC), squamous cell carcinoma (SCC), adenocarcinoma, and other (including sarcoma and melanoma) as determined by the ICD-O-3 histologic code. TCC consisted of ICD-O-3 codes 8050, 8120, 8123-8124, and 8130; SCC consisted of 8051-8052, 8070-8072, 8074, 8076, and 8083; adenocarcinoma consisted of 8140, 8144, 8255, 8260-8262, 8310, 8323, 8380, 8460, 8480-8481, 8490, 8500, 8550, 8560, and 8570; sarcoma consisted of 8004, 8800-8801, 8900, and 8910; melanoma consisted of 8720-8721, 8730, and 8770-8772; and other consisted of 8000, 8010, 8012, 8020-8021, 8033, 8041, 8046, 8082, 8240, 8246, 8951, 8980, 8990, 9085, 9100, and 9110. The extent of surgery of the primary tumor was classified as none/biopsy, ablation, simple excision, radical resection, not otherwise specified (NOS), and unknown. The type of radiation treatment was classified as none, external beam radiation alone, brachytherapy, combination of external beam and brachytherapy, NOS, and unknown. The characteristics of the study population are presented in Table 1. Median follow-up was 2.5 (interquartile range, 0.9-5.5) years. At last follow-up, 1146 (55.5%) men were dead with 540 (26.2%) dead of disease.

Table 1. Characteristics of the Study Population
ParameterAll Men Diagnosed 1988-2006, N = 2065, No. (%)
  • IQR indicate interquartile range; NOS, not otherwise specified.

  • a

    Tumor size available in only 293 patients.

Age, median (IQR)73 (64-80)
Year of diagnosis 
 1988-1995549 (26.6%)
 1996-2000559 (27.1%)
 2001-2006957 (46.3%)
Race 
 White1813 (87.8%)
 Black166 (8.0%)
 Other79 (0.3%)
 Unknown7 (0.3%)
Histologic type 
 Transitional cell carcinoma1603 (77.6%)
 Squamous cell carcinoma245 (11.9%)
 Adenocarcinoma103 (5.0%)
 Other114 (5.5%)
T stage 
 Ta333 (16.1%)
 Tis538 (26.1%)
 T1517 (25.0%)
 T2289 (14.0%)
 T399 (4.8%)
 T485 (4.1%)
 TX204 (9.9%)
Nodal status 
 Negative1480 (71.7%)
 Positive129 (22.1%)
 NX456 (22.1%)
M stage 
 M0/MX1966 (95.2%)
 M199 (4.8%)
Grade 
 1221 (10.7%)
 2472 (22.9%)
 3549 (26.6%)
 4283 (13.7%)
 Unknown540 (26.2%)
Tumor size, median mm (IQR)a15 (1-30)
Tumor size by stage, median mm (IQR)a
 Ta9 (3-20)
 Tis1 (1-3)
 T110 (2-20)
 T222.5 (15-30)
 T340 (18-50)
 T450 (30-70)
 TX60 (30-80)
Extent of surgery 
 None/biopsy393 (19.0%)
 Ablation180 (8.7%)
 Simple excision1260 (61.0%)
 Radical resection208 (10.1%)
 Surgery NOS15 (0.7%)
 Unknown if surgery performed9 (0.4%)
Type of radiation 
 None1823 (88.3%)
 External beam radiation alone199 (9.6%)
 Brachytherapy alone3 (0.1%)
 Combination external beam and brachytherapy7 (0.3%)
 Radiation NOS4 (0.2%)
 Unknown if radiation delivered29 (1.4%)
Follow-up, median y (IQR)2.5 (0.9-5.5)

Statistical Analysis

Cancer-specific survival and overall survival were computed using the Kaplan-Meier method, with the log-rank test used to compare groups. Cox proportional hazards analysis was performed using forward stepwise variable selection to obtain maximum likelihood estimates of the hazard ratios and 95% confidence intervals (CIs). Parameters evaluated as predictors of cancer-specific and overall survival included patient age at diagnosis, year of diagnosis, race, histologic type, grade, T stage, nodal status, M stage, extent of surgery, and type of radiation treatment. A missing variable category was used for missing data in the regression analysis. The proportional hazards assumption was tested with partial residual plots for the covariates. The Cox proportional hazards analysis was repeated with tumor size as a predictor in the subset of 293 men with the tumor size recorded. The SEER data use policies have been adhered to for this study. Statistical analyses were performed using the SPSS statistical package (SPSS Inc., Chicago, Ill).

RESULTS

The treatment rendered by stage is presented in Table 2. Radical resection and external beam radiation were more commonly used in those with stage T2-T4 cancer, ranging from 21% and 19%, respectively, for T2 cancer to 39% and 33%,respectively, for T4 cancer.

Table 2. Treatment Rendered by Stage
StageNo.SurgeryRadiation
None/ BiopsyAblationExcisionRadicalNOSUnknownNoneExternal BeamBrachy therapyCombinationNOSUnknown
  1. NOS indicates not otherwise specified.

Ta33328 (8%)38 (11%)256 (77%)10 (3%)0 (0%)1 (0.3%)332 (100%)0 (0%)0 (0%)0 (0%)0 (0%)1 (0.3%)
Tis53890 (17%)52 (10%)348 (65%)42 (8%)2 (0.4%)4 (0.7%)521 (97%)8 (1.5%)0 (0%)2 (0.4%)0 (0%)7 (1.3%)
T151776 (15%)61 (12%)349 (68%)28 (5%)2 (0.4%)1 (0.2%)456 (88%)51 (10%)1 (0.2%)1 (0.2%)1 (0.2%)7 (1.4%)
T228944 (15%)17 (6%)166 (57%)60 (21%)2 (0.7%)0 (0%)228 (79%)56 (19%)1 (0.3%)2 (0.7%)0 (0%)2 (0.7%)
T39915 (15%)3 (3%)56 (57%)23 (23%)2 (2%)0 (0%)65 (66%)29 (29%)0 (0%)0 (0%)0 (0%)5 (5%)
T48525 (29%)2 (2%)25 (29%)33 (39%)0 (0%)0 (0%)53 (62%)28 (33%)0 (0%)0 (0%)1 (1.2%)3 (3%)
TX204115 (56%)7 (3%)60 (29%)12 (6%)7 (3%)3 (1.5%)168 (82%)27 (13%)1 (0.5%)2 (1%)2 (1%)4 (2%)
Total2065393 (19%)180 (9%)1260 (61%)208 (10%)15 (0.7%)9 (0.4%)1823 (88%)199 (10%)3 (0.1%)7 (0.3%)4 (0.2%)29 (1.4%)

Overall survival at 5 and 10 years was 46.2% (95% CI, 43.9-48.6%) and 29.3% (95% CI, 26.6-32.0%), respectively (Fig. 1). The significant independent predictors of death on multivariate Cox proportional hazards analysis are presented in Table 3. Advanced age, higher grade, higher T stage, nodal metastases, systemic metastases, other histology versus TCC, and no surgery versus radical resection were predictive of increased likelihood of death, whereas adenocarcinoma was associated with a lower likelihood of death as compared with TCC. With the analysis limited to the 293 with tumor size recorded, advanced age, higher grade, higher T stage, nodal metastases, and earlier year of diagnosis were predictive of increased likelihood of death.

Figure 1.

Overall survival is shown.

Table 3. Predictors of Death on Multivariate Cox Proportional Hazards Analysis
PredictorHazard Ratio (95% CI)P
  1. CI indicates confidence interval; SCC, squamous cell carcinoma; TCC, transitional cell carcinoma.

Multivariate analysis on all patients, n = 2065 
 Age at diagnosis1.05 (1.04-1.06)<.001
 Grade  
  2 vs 11.21 (0.94-1.55).14
  3 vs 11.61 (1.25-2.06)<.001
  4 vs 11.45 (1.10-1.92).008
 T stage  
  Ta vs T10.89 (0.72-1.11).30
  Tis vs T10.74 (0.60-0.91).005
  T2 vs T11.10 (0.91-1.34).33
  T3 vs T11.30 (0.98-1.72).067
  T4 vs T11.98 (1.46-2.69)<.001
 M stage: M1 vs M0/MX3.04 (2.28-4.05)<.001
 Extent of surgery  
  Ablation vs radical resection1.10 (0.83-1.45).51
  None/biopsy vs radical resection1.56 (1.22-2.00)<.001
  Excision vs radical resection1.14 (0.91-1.42).27
 Nodal status: positive vs negative1.36 (1.04-1.77).026
 Histology  
  SCC vs TCC1.07 (0.87-1.30).53
  Adenocarcinoma vs TCC0.69 (0.51-0.91).010
  Other vs TCC1.43 (1.11-1.84).006
Multivariate analysis on subset with tumor size recorded, n = 293 
 Age at diagnosis1.05 (1.03-1.07)<.001
 Year of diagnosis0.95 (0.91-0.98).003
 Grade  
  2 vs 11.19 (0.59-2.38).63
  3 vs 11.77 (0.87-3.59).11
  4 vs 12.62 (1.17-5.87).020
 T stage  
  Ta vs T11.09 (0.59-2.02).79
  Tis vs T11.11 (0.67-1.85).68
  T2 vs T11.84 (1.14-2.97).012
  T3 vs T11.01 (0.56-1.81).98
  T4 vs T12.35 (1.17-4.70).016
 Nodal status: positive vs negative2.07 (1.17-3.67).013

Cancer-specific survival at 5 and 10 years was 68.0% (95% CI, 65.5-70.5%) and 60.1% (95% CI, 57.0-63.2%), respectively (Fig. 2A). The Kaplan-Meier curves for cancer-specific survival stratified by T stage, nodal status, M stage, grade, histology, extent of surgery, and radiation delivered are presented in Figures 2B-H. The significant independent predictors of death from disease on multivariate Cox proportional hazards analysis are presented in Table 4. Advanced age, higher grade, higher T stage, systemic metastases, other histology versus TCC, and no surgery versus radical resection were predictive of increased likelihood of death from disease, whereas adenocarcinoma was associated with a lower likelihood of death from disease as compared with TCC. With the analysis limited to the 293 men with tumor size recorded, increased tumor size and higher grade were predictive of increased likelihood of death from disease.

Figure 2.

(A) Cancer-specific survival is shown. (B) Cancer-specific survival is stratified by T stage. (C) Cancer-specific survival is stratified by nodal status. (D) Cancer-specific survival is stratified by M stage. (E) Cancer-specific survival is stratified by grade. (F) Cancer-specific survival is stratified by histology. (G) Cancer-specific survival is stratified by extent of surgery. (H) Cancer-specific survival is stratified by radiation delivered. (I) Cancer-specific survival is stratified by treatment type (radical surgery, radiation, both, or neither) in men with T2-T4 nonmetastatic urethral cancer. (J) Cancer-specific survival is stratified by treatment type (excision/radical surgery, radiation, both, or neither) in men with T2-T4 nonmetastatic urethral cancer. TCC indicates transitional cell carcinoma; SCC, squamous cell carcinoma.

Table 4. Predictors of Death From Disease on Multivariate Cox Proportional Hazards Analysis
PredictorHazard Ratio (95% CI)P
  1. CI indicates confidence interval; SCC, squamous cell carcinoma; TCC, transitional cell carcinoma.

Multivariate analysis on all patients (n = 2065) 
 Age at diagnosis1.03 (1.02-1.03)<.001
 Grade  
  2 vs 12.06 (1.27-3.34).004
  3 vs 13.38 (2.10-5.43)<.001
  4 vs 13.16 (1.91-5.21)<.001
 T stage  
  Ta vs T10.60 (0.43-0.84).003
  Tis vs T10.67 (0.50-0.89).006
  T2 vs T11.39 (1.06-1.81).015
  T3 vs T11.29 (0.86-1.95).22
  T4 vs T12.49 (1.69-3.67)<.001
 M stage: M1 vs M0/MX3.61 (2.52-5.16)<.001
 Extent of surgery  
  Ablation vs radical resection0.96 (0.65-1.43).83
  None/biopsy vs radical resection1.66 (1.20-2.29).002
  Excision vs radical resection1.13 (0.85-1.52).40
 Histology  
  SCC vs TCC0.98 (0.75-1.30).91
  Adenocarcinoma vs TCC0.54 (0.34-0.86).010
  Other vs TCC1.52 (1.07-2.16).020
Multivariate analysis on subset with tumor size recorded, n = 293  
 Tumor size, cm1.11 (1.01-1.22).030
 Grade  
  2 vs 12.42 (0.54-10.83).25
  3 vs 16.10 (1.44-25.84).014
  4 vs 19.68 (2.16-43.28).003

In the subset of 453 men with stage T2-T4 nonmetastatic urethral cancer, 98 had radical surgery alone without radiation, 97 had radiation alone (external beam, brachytherapy, or a combination) without radical surgery, 15 underwent both radical surgery and radiation, and 243 had neither radical surgery nor radiation. In the 15 men undergoing both radical surgery and radiation, 2 had radiation before surgery and 13 had radiation after surgery. The cancer-specific survival curves in these 453 men by treatment type (radical surgery, radiation, both, or neither) are presented in Figure 2I. There was a significantly higher cancer-specific survival with radical surgery alone compared with radiation alone (P = .018) or with neither radical surgery nor radiation (P = .049).

Figure 2J presents the cancer-specific survival curves in these 453 men with stage T2-T4 nonmetastatic urethral cancer by treatment type when surgery includes excision as well as radical surgery. Excision or radical surgery alone was undertaken in 277 men, radiation alone in 34, both surgery and radiation in 78, and neither in 64. In the 78 men undergoing both excision or radical surgery and radiation, 3 had radiation before surgery, and 75 had radiation after surgery. There was a significantly higher cancer-specific survival with excision or radical surgery alone compared with radiation alone (P = .001) or with neither surgery nor radiation (P < .001). There was a trend toward a higher cancer-specific survival with the combination of surgery and radiation compared with radiation alone (P = .053) and a significantly higher cancer-specific survival with the combination of surgery and radiation compared with neither surgery nor radiation (P = .017).

DISCUSSION

Primary urethral cancer is a rare malignancy, comprising <1% of all malignancies in male patients. As such, much of the published literature consists of case reports and limited case series from single institutions where patients have been managed with a variety of therapeutic modalities, making it difficult to arrive at any consensus on optimal therapeutic strategy. Early reports have identified tumor stage and site as the predominant prognostic factors, with higher-stage tumors having a worse prognosis3, 4 as well as those arising in the posterior as compared with the anterior urethra.3-6 More recent studies have shown that in addition to stage1, 7-9 and site,1, 8, 9 grade is also a prognostic factor.9 Indeed, conservative management with local excision may be an option for low-grade neoplasms.10

Whereas neoplasms of the anterior urethra can be managed successfully with conservative surgical options,3, 5, 6, 9, 11 the successful management of posterior urethral cancers, which often present later,4 is more challenging.9 Before the introduction of radical surgical principles in the 1950s for locally extensive posterior urethral cancers,12-14 the prognosis was dismal with more conservative approaches. Marshall reported that 4 of 5 patients with bulbar urethral cancer having radical resection had no evidence of disease 5 to 15 years after surgery.12 Subsequently, others have even advocated radical resection with inferior pubic rami resection with or without preoperative radiation for high-stage invasive proximal urethral cancers.15-17 The key role played by radical surgery in the management of posterior urethral cancers after Marshall's report12 is highlighted by Farrer and Lupu's18 compilation of the major series published between 1957 and 1983, which reported a 30% 5-year survival in 30 patients treated with radical surgery versus a 3% 5-year survival in 33 patients treated without radical surgery. However, others have advocated a more central role for radiation therapy in the management of posterior urethral tumors.19

The results with radiotherapy alone have not been encouraging, as summarized by Zeidman et al.20 Bracken et al reported that definitive radiation therapy failed to control the tumor in all 3 patients with posterior tumors, all of whom died of disease.11 Raghavaiah reported that although 2 patients with distal urethral tumors responded well to external beam radiation, 2 others with extensive bulbomembranous lesions died at 4 and 9 months from metastatic disease.21 Kaplan et al reported that none of 36 patients with bulbomembranous urethral tumors treated with radiotherapy alone survived 5 years, whereas 10 of 99 patients with bulbomembranous urethral tumors treated surgically survived 5 years.22 However, the results were comparable to radiation and surgery for distal urethral tumors; there were two 5-year survivors of 11 patients treated with radiotherapy alone versus fourteen 5-year survivors of 68 patients with distal urethral tumors treated with some form of surgical treatment.22

Although the extensive nature of urethral tumors has prompted many to recommend radical surgery as the treatment of choice in most cases,3 the suboptimal results with surgery alone has led to more recent recommendations of a combined modality approach with chemotherapy, radiotherapy, and surgical resection.1, 8 Gheiler et al reported that 1 of 8 patients undergoing surgery after chemotherapy and radiation had residual tumor in the surgical specimen and that 6 of these 8 patients remained disease free at 25 to 96 months follow-up.8 Similar down-staging with chemotherapy and radiation for urethral squamous cell carcinoma has been reported by others.23-26

Dalbagni et al identified nodal status and histologic type to be predictive of overall and cancer-specific survival on univariate analysis; however, the small study size of 46 patients precluded multivariate analysis.1 The present study, which is the only study large enough to evaluate prognostic factors on multivariate analysis, identified age, grade, stage, absence of systemic metastases, TCC versus other histology, adenocarcinoma versus TCC, and radical resection versus no surgery as predictors of overall and cancer-specific survival and in addition, nodal metastases as an adverse predictor of overall survival. These findings are consistent with earlier limited case series that have anecdotally reported stage3, 4, 7-9 and grade9 as prognostic factors without formal statistical analysis because of limited case numbers. In the SEER database, urethral tumors are not classified by site as anterior or posterior urethral in origin, and thus this variable has not been included in the analysis. Age was found to be prognostic of both overall survival and cancer-specific survival, as advanced age may limit the ability to administer chemotherapy or to perform radical surgery.

The finding of decreased overall and cancer-specific survival with other histology (non-TCC, non-SCC, nonadenocarcinoma) versus TCC and improved overall and cancer-specific survival with adenocarcinoma versus TCC is a novel finding. None of the published studies has been large enough to adequately assess the impact of histology on outcome. The only study to assess this was that of Dalbagni et al, which identified histologic type as a predictor of overall and cancer-specific survival (P = .0001).1 In the study by Dalbagni et al, there were 16 cancer-specific deaths in the 29 men with SCC/epidermoid carcinoma and 4 in 15 men with TCC; however, there was only 1 patient each with adenocarcinoma or other histology.1 In our study, 1603 (77.6%) of tumors were TCC, 245 (11.9%) were SCC, 103 (5.0%) were adenocarcinoma, and 114 (5.5%) were other histology. This distribution contrasts with that in the published series from referral centers; Swartz et al reviewed the published data, consisting of 25 series from North America from 1966 to 2006, tabulating 132 (68.0%) men with SCC, 34 (17.5%) men with TCC, and 9 (4.6%) men with adenocarcinoma.27 The discrepancy may be because of the population-based nature of the SEER database versus the tertiary referral base of the published literature as well as the inclusion of patients with carcinoma in situ, who are often excluded from the case series in the published literature. Although misclassification of bladder tumors involving the prostatic urethra as a urethral TCC is a possibility, this analysis specifically excluded patients with a diagnosis of bladder cancer. The small number of adenocarcinomas precludes the likelihood of misclassification of prostatic adenocarcinoma as urethral adenocarcinoma.

In the analysis on the subset of 293 men with tumor size recorded, year of diagnosis was predictive of overall survival but not cancer-specific survival on multivariate analysis. Patients with a later year of diagnosis had improved overall survival. Although this may be encouraging in reflecting improvements in treatment such as chemotherapy that are not captured in this study, one would then expect an increase in cancer-specific survival as well. However, the small sample size may also limit the accuracy of the actuarial estimates.28 In the only study to evaluate year of diagnosis as a prognostic factor, Dalbagni et al did not find it to be predictive of overall or cancer-specific survival.1 In the analysis of cancer-specific survival, there was an 11% increase in likelihood of death from disease for each 1 cm increase in tumor size. Tumor size has not been previously evaluated as a prognostic factor in any published study. However, tumor size is correlated with the T stage (Table 1) and, as such, the finding of tumor size being predictive of cancer-specific survival is consistent with tumor stage being a known prognostic factor.

To better assess the impact of treatment modality on outcome, analysis was also performed limited to the 453 men with stage T2-T4 nonmetastatic urethral cancer. There was a significantly higher cancer-specific survival with radical surgery alone compared with radiation alone (P = .018) or with neither radical surgery nor radiation (P = .049) (Fig. 2I). The small number of patients having both radical surgery and radiation precludes comparison of the outcome with that of either treatment modality alone. On multivariate analysis, there was a survival benefit to radical resection as compared with no surgery. To our knowledge, the present study is the first to report a statistically significant impact of treatment modality on outcome, albeit in a registry study where selection biases are unknown. Dalbagni et al have reported 26 deaths in 40 men undergoing primary surgery versus 5 deaths in 6 men undergoing salvage surgery, a difference not achieving statistical significance, likely because of small numbers (univariate log-rank P = .065); no difference in cancer-specific survival was noted for primary versus salvage surgery (univariate log rank P = .15).1

In the SEER database, specific cases cannot be reviewed to determine the selection bias in choosing operative candidates as it pertains to comorbidities or factors other than disease characteristics that go into treatment selection. As such, there are likely biases in selection of more favorable cases for surgery or combined modality therapy where a complete resection is more likely. Furthermore, it is possible that surgery has been performed at referral centers of excellence, with more extensive experience in treatment of urethral cancer, whereas other modalities were used at less experienced centers.

Comparison of outcome by extent of surgery in the present study is difficult, as more conservative options may be reasonable for anterior urethral tumors or low-stage, low-grade tumors; the inability to stratify by tumor site (anterior vs posterior) makes this comparison difficult. To account for the finding that excision may be appropriate for low-stage, low-grade tumors or those in the anterior urethra, the analysis in the 453 men with stage T2-T4 nonmetastatic urethral cancer was repeated comparing modalities when excision was included along with radical resection as surgery. There was a similar finding of a significantly higher cancer-specific survival with excision or radical surgery alone compared with radiation alone (P = .001) or with neither surgery nor radiation (P < .001) (Fig. 2J). Although our study identified a better outcome for surgery as compared with radiation as a single modality, this may be because of selection bias, as stated previously. The potential incremental benefit of both surgery and radiation is difficult to demonstrate in this study because of the limited number of patients undergoing both treatment modalities.

This study has several limitations. In the SEER database, there is no coding for the site of the urethral tumors (anterior vs posterior). This is a significant limitation of the SEER database, as tumor site is a known prognostic factor in urethral cancer and determines the extent of surgery performed. As such, comparison of outcomes by extent of surgery is somewhat limited without adjusting for site; furthermore, comparison with the published literature is difficult without stratifying by tumor site. Another limitation is that data on chemotherapy are not included in the SEER database, precluding assessment of the impact of multimodality therapy with chemotherapy on treatment outcome. Another limitation of the SEER database is the lack of data on health-related quality of life outcomes, which is a significant consideration in the treatment of this disease. Nevertheless, the population-based nature of the SEER database may eliminate the regional variances in treatment that single-center studies are subject to. Furthermore, the large sample size permits multivariate analysis, which is not possible in single-center studies. Despite the large number of patients overall, the small number of patients in some subsets, especially those undergoing radical surgery and radiation, makes it difficult to evaluate the incremental benefit of combined modality therapy. Although this study is a retrospective analysis of a prospective national database, all published studies are retrospective case series that often span a larger number of years than the 19 years in the present study.

In summary, the present study is the first large-scale study to permit evaluation of prognostic factors in men with urethral cancer using multivariate statistical analysis. Advanced age, higher grade, higher stage, systemic metastases, other histology versus TCC, and no surgery versus radical resection were predictive of increased likelihood of death as well as death from disease, whereas adenocarcinoma was associated with a lower likelihood of death and death from disease as compared with TCC. In addition, nodal metastasis was predictive of increased likelihood of death. Further study in the setting of prospective multicenter trials is necessary to evaluate the respective roles of surgery, radiation, and chemotherapy in multimodality therapy of this rare malignancy, as conclusions regarding the role of various therapeutic modalities are difficult to draw in a retrospective study fraught with selection biases.

CONFLICT OF INTEREST DISCLOSURES

The author made no disclosures.

Ancillary