- Email: [email protected]
Contemporary survival trends in penile cancer: Results from the National Cancer Database
Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Original article
Contemporary survival trends in penile cancer: Results from the National Cancer Database Minh N. Pham, B.S.a, Allison M. Deal, M.S.b, Jed E. Ferguson III, M.D., Ph.D.c, Yue Wang, B.S.d, Angela B. Smith, M.D., M.S.e, Matthew E. Nielsen, M.D., M.S.e,f, Raj S. Pruthi, M.D.e, Michael E. Woods, M.D.e,⁎ a University of North Carolina School of Medicine, Chapel Hill, NC Lineberger Comprehensive Cancer Center, Biostatistics and Clinical Data Management Core, University of North Carolina, Chapel Hill, NC c Department of Urology, University of North Carolina, Chapel Hill, NC d Department of Biostatistics, University of North Carolina, Chapel Hill, NC e Department of Urology, Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC f Department of Epidemiology and Health Policy & Management, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, NC b
Received 30 November 2016; received in revised form 6 June 2017; accepted 9 August 2017
Abstract Objective: To investigate contemporary survival trends in penile cancer. Methods: The National Cancer Database was queried for men with penile cancer diagnosed between 1998 and 2009. Patient, tumor, treatment, and facility characteristics were obtained. Overall survival (OS) was analyzed using the Kaplan-Meier method and multivariable Cox proportional hazards regression model for all cases and stratified by stage. Results: A total of 8,122 cases of penile cancer were reported from 1998 to 2009 in the National Cancer Database. Complete staging, survival, and covariate data were available for 5,043 cases. The estimated crude 5-year OS for the entire cohort was 61.0%. For all patients, no significant differences in crude 5-year OS were detected between 2006 to 2009 and 2002 to 2005 compared to 1998 to 2001. On multivariable analysis, OS did not significantly differ across all eras. Regional lymph node dissection was associated with improved OS (hazard ratio [HR] ¼ 0.777, P ≤ 0.0001). In patients who underwent lymph node dissection, dissection of ≥8 nodes significantly improved survival (HR ¼ 0.672; P ¼ 0.0011). Additional modeling stratified by stage revealed that OS for stage II cancers increased significantly in 2006 to 2009 compared to 1998 to 2001 (HR ¼ 0.714; P ¼ 0.0034). Conclusions: Survival in penile cancer has remained unchanged as a whole and for each stage, except for stage II disease. An improved survival trend was detected in stage II penile cancer. Performing a lymph node dissection, especially extensive dissections, may benefit longterm survival. r 2017 Elsevier Inc. All rights reserved.
Keywords: Penile cancer; Survival trends; United States; National Cancer Database
1. Introduction Penile cancer is a rare malignancy in the developed world. The American Cancer Society estimates that penile cancer will account for 2,030 new cases and 340 deaths in the United States in 2016 [1]. The prognosis is excellent when diagnosed early but dramatically worsens with nodal metastasis [2]. One report recently estimated the U.S. 5-year ⁎
Corresponding author. Tel.: þ1-919-966-2574. E-mail address: [email protected] (M.E. Woods). http://dx.doi.org/10.1016/j.urolonc.2017.08.009 1078-1439/r 2017 Elsevier Inc. All rights reserved.
overall survival (OS) as 66% [3]. Recently, evidence has emerged supporting that survival and quality of life can be improved through organ-sparing techniques [4,5], earlier detection of inguinal adenopathy [6–9], and postchemotherapy surgical consolidation of metastases [10–12]. The distribution of care has also changed at a facility-level. Community hospitals and academic hospitals are comparably likely to diagnose the disease, yet the responsibility of treatment is centralizing toward academic centers [13]. Despite these changes, little data exists examining mortality trends in the past 2 decades. One report of 1,800 patients
2
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
found that U.S. 5-year survival declined from 72% in cases from 1990 to 1995 compared to 63% in 2002 to 2007 [14]. Using a larger sample, we further evaluated OS trends in penile cancer cases diagnosed between 1998 and 2009 through the National Cancer Database (NCDB).
2. Materials and methods The NCDB is a national cancer registry jointly supported by the American College of Surgeons Commission on Cancer and the American Cancer Society. Over 1,500 accredited hospital-based cancer programs contribute to the NCDB, capturing 470% of newly diagnosed cancers in the United States and Puerto Rico. All data are entered by participating institutions and standardized according to coding guidelines published in the Facility Oncology Registry Data Standards (https://www.facs.org/quality-pro grams/cancer/ncdb/registrymanuals/cocmanuals/fordsma nual). Data include the following: patient and hospital characteristics, staging, histology, anatomic location, firstline treatment, and outcomes. All patient data were deidentified before the date of query. This study was approved by the University of North Carolina’s Institutional Review Board. The NCDB was queried for cases of penile cancer. Cases with the histologic International Classification of Diseases for Oncology (ICD)-O-3 codes pertaining to the following histologies were included: squamous cell carcinoma (8070, 8071, 8072, 8073, 8074, 8075, 8076), erythroplasia of Queyrat (8080), and Bowen disease (8081). From 1998 to 2009, the American Joint Committee on Cancer staging definitions for penile cancer remained unchanged [15,16]. Data were available for patients diagnosed in 1998 through 2012, but only cases between 1998 and 2009 were included to maintain consistent staging definitions and owing to the absence of 5-year OS data available from 2010 and later. Cases diagnosed at autopsy were excluded. The analysis was limited to NCDB sequence numbers 00, corresponding to cases with only 1 lifetime cancer diagnosis, and 01, representing tumors that were the first of multiple cancer diagnoses. Cases with the NCDB class code 00, which were those diagnosed but not treated by the reporting facility and did not require follow-up per Commission on Cancer guidelines, were excluded. We used NCDB analytic stage groups, which corresponded to pathologic stage group when available, and clinical stage group when pathologic data was unavailable. Cases without complete analytic staging and survival data were excluded. Our final sample was limited to patients with complete covariate data within the multivariable analysis with the exception of unknown grade for stage IV. Demographic variables included age, race/ethnicity, insurance status, median household income of the patient’s area of residence, and travel distance to facility. The NCDB began recording Charlson/Deyo scores in 2003, and owing
to the large volume of missing scores within our sample, Charlson/Deyo scores were described but excluded from the primary multivariable model. A sensitivity analysis including comorbidity in the multivariable model was then performed with patients diagnosed during 2003 to 2009. Facility-level characteristics included academic vs. community centers and facility region. Tumor characteristics included analytic stage group, grade, and the presence of invasive or in-situ disease. Primary site was described but excluded from the multivariable analysis owing to a substantial proportion of cases with missing data. Cases recorded as erythroplasia of Queyrat (ICD-O-3 8080), Bowen disease (ICD-O-3 8081), and analytic stage group 0 with pathologic stage Tis were analyzed as having welldifferentiated grade. Treatment characteristics analyzed included regional lymph node dissection (LND), chemotherapy, and radiation therapy. When a case met 3 criteria of having an unclear history of LND owing to missing data, was described to be pN0, and was reported to have “no lymph nodes examined” within the data, the case was regarded in the model as not having undergone LND. The survival impact of lymph node yield on survival was evaluated through a separate multivariable model limited to patients that underwent LND. Descriptive statistics are reported overall and by era, with comparisons between era made by chi-squared tests. The Kaplan-Meier method was used to estimate crude median survival and 5-year OS rates. The log-rank test was used for comparisons of survival curves. Multivariable Cox proportional hazards regression models were used to evaluate the combined effect of patient and clinical characteristics on survival, for the entire cohort, and stratified by stage. Chemotherapy and radiation therapy were included in the model as time-varying covariates [17]. For the stratified analysis for stage IV patients, grade was excluded from the model owing to the large amount of missing data and its relatively low clinical importance in systemic disease. Estimates from the adjusted Cox model were used to create the adjusted Kaplan-Meier curves [18]. Analyses were performed using SAS statistical software v.9.4 (Cary, NC).
3. Results Between 1998 and 2009, a total of 8,122 cases, in whom penile cancer was the only cancer diagnosis or the first of multiple cancer diagnoses, were reported to the NCDB. Of these, 5,043 had complete analytic staging, survival, and covariate information (Fig. 1). Pathologic staging was not reported for 1,743 (34.6%) cases, requiring the usage of clinical staging. Patient, tumor, treatment, and facility characteristics are summarized in Table 1. The median age of diagnosis was 65 years. Non-Hispanic White, nonHispanic African-American, and Hispanic patients accounted for 79.0%, 9.2%, and 9.4% of cases, respectively, 93.6% were insured. Stages were distributed, in
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Penile cancer is first diagnosis or first of multiple diagnoses (N = 8122) Missing staging data (n = 929) (n = 7193) Missing survival data (n = 32) (n = 7161) Missing covariate data (n = 2118) Final cohort analyzed (n = 5043) Fig. 1. Process of exclusion from the final cohort.
descending order of frequency, as I (34.4%), 0 (25.7%), II (21.0%), III (13.0%), and IV (5.9%). The most common sites for tumor localization were the glans followed by the prepuce. Most tumors were invasive (75.7%) and of moderately differentiated (37.1%) or well-differentiated grade (45.9%). More patients were treated at community (61.0%) than academic programs (39.0%); however, the proportion of patients managed at academic centers increased from 36.4% in 1998 to 2001, to 36.7% in 2002 to 2005, and 43.5% in 2006 to 2009 (P o 0.0001). Chemotherapy and radiotherapy were administered in 7.3% and 7.3%, respectively. Whereas radiotherapy rates remained stable, chemotherapy rates significantly increased from 5.5% in 1998 to 2001, to 7.3% in 2002 to 2005, and 8.8% in 2006 to 2009 (P ¼ 0.0012). The relative rates of penile cancer stages for each year are shown in Fig. 2 (n ¼ 7,193). Stage I cases were most common, ranging from 29% to 40%. The proportion of stage 0 cases ranged from 25% to 33%. Stage II cases varied from 16% to 23%. For stage III, the proportion fluctuated between 9% and 14%. Stage IV accounted for 4% to 8% of cases. Overall, the 5-year OS rate was 61.0%. No significant differences in crude 5-year OS were found among 1998 to 2001 (60.4%), 2002 to 2005 (61.3%), and 2006 to 2009 (61.2%) (P ¼ 0.99). Median OS during each era was 7.4 years for 1998 to 2001, 7.7 years for 2002 to 2005, and unavailable for 2006 to 2009 (owing to survival being greater than 50% by the time the data was queried) with mean follow-up times for survivors of 8.4 years, 7.0 years, 4.2 years, respectively. Fig. 3A demonstrates the adjusted Kaplan-Meier curves for the overall sample stratified by era. After adjusting for patient, tumor, and facility characteristics in a multivariable analysis, we found no significant differences in OS across all eras for penile cancer as a whole (Table 2). Negative prognostic factors were as follows: age of diagnosis at 60 years or older, non-Hispanic
3
African-American race, lower median household income, higher stage, less differentiated grade, and chemotherapy or radiotherapy at any point following diagnosis. Age of diagnosis younger than 40 years, carcinoma in-situ, and regional LND were associated with improved OS. Additional modeling stratified by stage included the same patient, tumor, treatment and facility characteristics, excluding grade for the stage IV model. This model revealed improved survival in stage II patients in 2006 to 2009 compared to 1998 and 2001. For stages 0, I, III, and IV, no significant survival changes were detected across all eras. A total of 1,059 cases of stage II disease were analyzed. Of note, the proportion of stage II cases treated at academic centers significantly increased from 35.3% in 1998 to 2001, to 41.3% in 2002 to 2005, and 44.8% in 2006 to 2009 (P ¼ 0.0375). A 5-year OS for stage II patients was 53.5% with estimates for each era as follows: 1998 to 2001 (50.7%), 2002 to 2005 (51.0%), 2006 to 2009 (57.6%) (P ¼ 0.12). Median OS for stage II patients was 5.3 years, 5.3 years, and 6.5 years during 1998 to 2001, 2002 to 2005, and 2006 to 2009, respectively. Fig. 3B demonstrates the adjusted Kaplan-Meier curves for stage II disease stratified by era. On multivariable analysis of stage II cases, OS significantly improved in 2006 to 2009 compared to 1998 to 2001 [hazard ratio (HR) ¼ 0.714; P ¼ 0.0034]. After controlling for demographic, tumor, facility, and treatment characteristics, regional LND was a significant prognostic factor for the combined cohort of penile cancer (HR ¼ 0.777; P o 0001) and the stage IV disease group (HR ¼ 0.556; P o 0.0001). LND was not significantly associated with improved OS in stage II (HR ¼ 0.810; P ¼ 0.0603). In a multivariable analysis of patients who underwent LND (n ¼ 744), a yield of ≥8 nodes significantly benefited survival for all stages combined (HR ¼ 0.672; P ¼ 0.0011) and for stage III cases (HR ¼ 0.545; P ¼ 0.009). The impact of comorbidity on survival was evaluated by including Charlson/Deyo scores in a separate multivariable analysis of cases diagnosed from 2003 to 2009 (data not shown). Increased comorbidity was associated with poorer OS. For penile cancer as a whole, OS remained unchanged between 2003 to 2005 and 2006 to 2009. Younger age and regional LND, but not carcinoma in-situ, remained positive prognostic factors. Older age, lower income, higher stage, less differentiated grade, and treatment with chemotherapy or radiotherapy at any point following diagnosis remained negative prognostic factors. After accounting for comorbidity, non-Hispanic African-American race was no longer significantly associated with negative survival outcomes. Compared to 2003 to 2005, survival in 2006 to 2009 improved in stage II. LND continued to be significantly associated with improved OS in penile cancer as a whole, stage IV, and additionally stage III. Among those treated with regional lymphadenectomy, yields of 8 nodes or greater remained significantly associated with improved OS for penile cancer as a whole, stage III, and also stage IV.
4
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Table 1 Demographic, tumor, treatment, and facility characteristics for penile cancer cases diagnosed from 1998 to 2009 Characteristics Age at diagnosis, y, n (%) o40 40–59 60–79 ≥80 Race/ethnicity, n (%) White, non-Hispanic African-American, non-Hispanic Hispanic Other Stage group, n (%) 0 I II III IV Grade, n (%) Well differentiated Moderately differentiated Poorly differentiated Undifferentiated Behavior, n (%) Invasive In Situ Insurance status, n (%) Yes No Primary site, n (%)b Unknown Prepuce Glans Body Overlapping lesion Chemotherapy, n (%) Yes No Radiotherapy, n (%) Yes No Lymph node dissection Yes No Charlson/Deyo scoreb,c Unknown 0 1 ≥2 Region, n (%) South Midwest Northeast West Facility, n (%) Academic Community Median income ($),d n (%) o38,000 38,000–47,999 48,000–62,999 463,000
1998–2001 (n ¼ 1,662)
2002–2005 (n ¼ 1,566)
2006–2009 (n ¼ 1,815)
Overall (n ¼ 5,043)
P valuea
100 498 826 238
(6.0) (30.0) (49.7) (14.3)
97 494 764 211
(6.2) (31.6) (48.8) (13.5)
95 577 893 250
(5.2) (31.8) (49.2) (13.8)
292 1,569 2,483 699
(5.8) (31.1) (49.2) (13.9)
0.7841
1,337 143 143 39
(80.5) (8.6) (8.6) (2.4)
1,254 134 144 34
(80.1) (8.6) (9.2) (2.2)
1,395 185 188 47
(76.9) (10.2) (10.4) (2.6)
3,986 462 475 120
(79.0) (9.2) (9.4) (2.4)
0.1974
445 637 303 196 81
(26.8) (38.3) (18.2) (11.8) (4.9)
384 526 334 224 98
(24.5) (33.6) (21.3) (14.3) (6.3)
465 572 422 237 119
(25.6) (31.5) (23.3) (13.1) (6.6)
1,294 1,735 1,059 657 298
(25.7) (34.4) (21.0) (13.0) (5.9)
o0.0001
803 616 232 11
(48.3) (37.1) (14.0) (0.7)
699 584 280 3
(44.6) (37.3) (17.9) (0.2)
815 671 319 10
(44.9) (37.0) (17.6) (0.6)
2,317 1,871 831 24
(45.9) (37.1) (16.5) (0.5)
0.0103
1,236 (74.4) 426 (25.6)
1,203 (76.8) 363 (23.2)
1,378 (75.9) 437 (24.1)
3,817 (75.7) 1,226 (24.3)
0.2569
1,567 (94.3) 95 (5.7)
1,474 (94.1) 92 (5.9)
1,677 (92.4) 138 (7.6)
4,718 (93.6) 325 (6.4)
0.0418
768 208 517 97 72
(46.2) (12.5) (31.1) (5.8) (4.3)
791 143 456 100 76
(50.5) (9.1) (29.1) (6.4) (4.9)
976 161 483 128 67
(53.8) (8.9) (26.6) (7.1) (3.7)
2,535 512 1,456 325 215
(50.3) (10.2) (28.9) (6.4) (4.3)
o0.0001
92 (5.5) 1,570 (94.5)
115 (7.3) 1,451 (92.7)
159 (8.8) 1,656 (91.2)
366 (7.3) 4,677 (92.7)
0.0012
118 (7.1) 1,544 (92.9)
119 (7.6) 1,447 (92.4)
131 (7.2) 1,684 (92.8)
368 (7.3) 4,675 (92.7)
0.8507
259 (15.6) 1,403 (84.4)
255 (16.3) 1,311 (83.7)
328 (18.1) 1,487 (81.9)
842 (16.7) 4,201 (83.3)
0.1262
1,662 0 0 0
405 918 (79.1) 178 (15.3) 65 (5.6)
0 1,322 (72.8) 377 (20.8) 116 (6.4)
2,067 2,240 (75.3) 555 (18.6) 181 (6.1)
o0.0001
485 469 299 409
(29.2) (28.2) (18.0) (24.6)
605 (36.4) 1,057 (63.6) 392 454 447 369
(23.6) (27.3) (26.9) (22.2)
454 433 322 357
(29.0) (27.7) (20.6) (22.8)
574 (36.7) 992 (63.4) 331 433 441 361
(21.1) (27.7) (28.2) (23.1)
547 456 410 402
(30.1) (25.1) (22.6) (22.2)
790 (43.5) 1,025 (56.5) 441 467 473 434
(24.3) (25.7) (26.1) (22.9)
1,486 1,358 1,031 1,168
(29.5) (26.9) (20.4) (23.2)
0.0177
1,969 (39.0) 3,074 (61.0)
o0.0001
1,164 1,354 1,361 1,164
(23.1) (26.9) (27.0) (23.1)
0.2570
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
5
Table 1 Continued Characteristics Distance to facility (miles), n (%) o10 10–50 450
1998–2001 (n ¼ 1,662) 878 (52.8) 562 (33.8) 222 (13.4)
2002–2005 (n ¼ 1,566) 749 (47.8) 597 (38.1) 220 (14.1)
2006–2009 (n ¼ 1,815) 837 (46.1) 710 (39.1) 268 (14.8)
Overall (n ¼ 5,043) 2,464 (48.9) 1,869 (37.1) 710 (14.1)
P valuea 0.0019
a
Chi-squared tests were performed to compare differences between eras. Not included in primary multivariate analysis. Charlson/Deyo scores were included in a sensitivity analysis using a separate multivariable model. c Not recorded in the National Cancer Database until January 1, 2003 and later. Cases with missing scores are not accounted for in percentages. d Representing median household income of the patient’s residential area measured by the 2000 U.S. census data. b
4. Discussion In this study, we used a large, representative national database to evaluate OS trends for penile cancer diagnosed between 1998 and 2009. We analyzed outcomes for all cases and stratified by NCDB analytic stages, corresponding to American Joint Committee on Cancer-defined pathologic stage when available or clinical stage when the former was missing. Survival remained largely unchanged, and changes in staging distribution over time were small. Notably, multivariable analysis of stage II disease found that OS improved significantly in 2006 to 2009 compared to 1998 to 2001. To our knowledge, this is the largest analysis of survival trends in penile cancer and the first to report an improvement in stage II disease. Previous studies demonstrate inconclusive evidence about penile cancer survival trends in the industrialized world, complicated by differences in cancer registries and health care systems. In the Netherlands, 5-year cancerspecific survival (CSS) for node-negative disease improved from 82% during 1956 to 1993 to 91% during 1994 to 2012 after dynamic sentinel node biopsies were implemented to replace prophylactic inguinal LNDs [19]. For the United States, Verhoeven et al. [14] evaluated 1,820 penile cancer cases from the Surveillance, Epidemiology and End Results program and found a 5-year survival decline from 72% to
Fig. 2. Temporal trends in penile cancer staging distributions (n ¼ 7,193).
63% between 1990 and 2007. The contrasting results between our study and those of Verhoeven et al. may be explained by different periods of study. After disease management centralized in the United Kingdom, 5-year CSS improved by 7% to 12% [20]. Our data shows management shifting toward academic facilities. Compared to community programs, academic centers see more than twice the amount of cases per year, manage more advanced-staged disease, perform more LNDs in eligible patients, and have higher lymph node yields during their dissections [13]. We detected no survival difference between academic and community programs. That being said, a survival change may require more time before being detected. Alternatively, the benefit may be restricted to stages most affected by centralization. Stage II survival improved significantly in 2006 to 2009 on multivariable analysis. Stage II penile cancer was defined in 1998 to 2009 as T1N1, T2N0, or T2N1 disease [15,16]. Several changes in T1-, T2-, and N1-staged management related to centralization may contribute to this trend. Increased rates of stage II patients were seen at academic centers in 2002 to 2005 (41.3%) and 2006 to 2009 (44.8%) compared to 1998 to 2001 (35.3%). Among patients with T1b or greater T-stage, node-positive disease, or high-grade penile cancer (grade 3 or 4), academic centers delivered higher rates of guideline-concordant management and were 2.29 times more likely to perform LNDs compared to community programs [13]. The greater likelihood of LNDs at academic centers is accompanied by an increasing trend in its frequency at these facilities: the proportion of these procedures performed among eligible patients at academic centers, over community centers, increased by 40% between 1998 and 2012 [13]. Academic centers were also more likely to perform extensive LNDs and on average yielded more nodes [13], which have been associated with improved survival [21,22]. Whereas we did not detect a significant survival benefit from LNDs for the stage II group, potentially because of an underpowered analysis, our results for all stages combined confirmed these findings, showing that LND and high lymph node yields (≥8 nodes) were associated with improved survival. Thus, the management of stage II disease may be benefiting from its centralization toward academic centers, where the care
6
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Fig. 3. Adjusted Kaplan-Meier plot for overall survival stratified by era for (A) overall disease and (B) stage 2 penile cancer.
delivered is more likely to be consistent with guidelines and thorough in technique. We found that OS in stages 0 and I remained stable in the backdrop of an increasing acceptance of organ-sparing intervention [13,23]. The usage of partial penectomy among patients with localized disease (clinical Ta-T2), for instance, has increased in the last 2 decades [23]. Although clinical practice has transitioned, there has been little data describing the impact of this paradigm shift in the United States. In the Netherlands, penile preservation techniques, such as laser therapy, local excision, and glans resection, were associated with increased recurrence rates compared to partial or total amputation but had comparable CSS [24]. Nonetheless, management in the Netherlands is highly centralized, limiting the study’s generalizability. Further validation of this data in the United States may be difficult owing to the rarity of the malignancy spread across a large geographic area. This study could not directly compare U.S. and Dutch outcomes because recurrence rates and CSS were not reported by the NCDB; however, the stability of OS among early stage disease in the real-world setting is reassuring. Survival in stages III and IV also remained unchanged. The lack of progress in survival outcomes in these stages is consistent with the absence of improved systemic therapies until 2008 [25]. Bleomycin, methotrexate plus cisplatin was introduced in 1999 and remained the most common regimen throughout our time-points [26]. More recently, studies evaluating cisplatin-based chemotherapy with surgical consolidation have suggested a more promising efficacy and tolerability over bleomycin-containing regimens [10–12]. Additionally, chemotherapy usage in metastatic disease is increasing [23]. Together, these changes may potentially impact penile cancer mortality in the future.
We estimate the crude 5-year OS rate for penile cancer as a whole to be 61%. In comparison, the estimated Swedish 5-year relative OS rate was 82% [27]. Several hypotheses may contribute to this disparity. Relative OS, compared to the standard OS in our study, represents cancer survival accounting for competing mortality risks, defined as the ratio between the observed cancer survivor rate and expected survivor rate in cancer-free individuals [28]. Thus, for a given cancer cohort, relative OS is expectedly higher than standard OS. Additionally, given the rarity of penile cancer, adherence to treatment guidelines is variable. United States centers performed LNDs in only 36.3% of eligible patients compared to 50% to 74% in Sweden [13,27]. Guideline-discordant care is associated with poorer outcomes in other cancers [29,30]. The rate of invasive cancer is also higher in the United States (76%) than Sweden (45%) [27]. Finally, because the NCDB measured survival outcomes through OS, the gap between Swedish and U.S. survival may partially be influenced by deaths not attributable to penile cancer. Racial and socioeconomic differences also contribute to outcomes in the United States. African-American race has been associated with poorer OS whereas private insurance and residence in areas with higher median income were associated with improved survival [3]. In our study, nonHispanic African-American race and residence in regions with lower income status were similarly predictors of poorer survival. Comorbidities affecting African Americans may partially contribute to the observed difference in survival. After accounting for Charlson/Deyo scores in our multivariable model, non-Hispanic African-American race was no longer associated with poorer survival, supporting the hypothesis that comorbidities may have confounded our primary findings related to race and ethnicity. Racial
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
7
Table 2 Multivariable Cox proportional hazards regression models for overall survival Characteristic
Levels
HR
Era
1998–2001 2002–2005 2006–2009 o40 40–59 60–79 ≥80 White, non-Hispanic African-American, non-Hispanic Hispanic Other 0 I II III IV Well differentiated Moderately differentiated Poorly differentiated Undifferentiated, anaplastic Invasive In situ Yes No No Yes No Yes No Yes South Midwest Northeast West Academic Community o38,000 38,000–47,999 48,000–62,999 463,000 o10 10–50 450
Ref 0.940 0.914 0.635 Ref 1.811 4.220 Ref 1.286 1.045 0.810 1.007 Ref 1.388 1.893 4.523 Ref 1.246 1.515 2.712 Ref 0.691 Ref 1.097 Ref 1.616 Ref 1.217 Ref 0.777 Ref 1.002 0.962 0.903 Ref 1.039 1.246 1.142 1.070 Ref Ref 1.000 0.985
Age at diagnosis, y
Race/ethnicity
Stage group
Grade
Behavior Insurance status Chemotherapya Radiation therapya Regional lymph node dissection Region
Facility type Median income ($)
Median distance to facility (miles)
95% CI
P value
0.851–1.037 0.822–1.017 0.473–0.851
0.2161 0.0983 0.0024
1.623–2.021 3.697–4.817
o0.0001 o0.0001
1.115–1.483 0.884–1.235 0.594–1.105 0.707–1.433
0.0006 0.6064 0.1837 0.9707
1.239–1.555 1.661–2.158 3.826–5.347
o0.0001 o0.0001 o0.0001
1.117–1.391 1.333–1.722 1.585–4.641
o0.0001 o0.0001 0.0003
0.480–0.995
0.0469
0.904–1.330
0.3484
1.377–1.896
o0.0001
1.044–1.419
0.0122
0.688–0.877
o0.0001
0.897–1.119 0.848–1.091 0.801–1.018
0.9738 0.5459 0.0966
0.948–1.140 1.092–1.421 1.008–1.294 0.948–1.209
0.4118 0.0011 0.0370 0.2741
0.913–1.096 0.861–1.128
0.9953 0.8320
Ref ¼ reference. a Reflects chemotherapy or radiation therapy at any point in following diagnosis. Variables were analyzed as time-varying covariates.
disparities in treatments also exist. For instance, AfricanAmericans are less likely to undergo chemotherapy for metastatic disease or any surgery for penile cancer [23,31]. Our study has limitations inherent in the NCDB. The NCDB is limited to measuring OS and does not report CSS. One approach to address this limitation is to adjust for comorbidity. Unfortunately, the database did not record Charlson/Deyo scores until 2003. We chose to exclude these scores in our main multivariable analysis to preserve a substantial number of patients; however, a subgroup analysis including comorbidity for patients diagnosed between 2003 and 2009 showed largely the same findings. Therefore, the results of this sensitivity analysis supported the use
of our primary model without Charlson scores to evaluate differences in eras. If comorbidity rates had increased between 1998 and 2009, our ability to detect significant survival improvement across eras would be inhibited by competing mortality risks; however, with the data available, this would be difficult to validate. To appropriately account for the impact of chemotherapy and radiation therapy on survival, we had to address the potential for guarantee time bias [17]. Although chemotherapy rates had significantly increased during the period of study, these rates reflected treatment with chemotherapy at any point following diagnosis. Thus, we chose to include chemotherapy and radiation therapy through an extended Cox model with
8
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
time-varying covariates rather than through a conditional landmark analysis. Our model subsequently found that undergoing chemotherapy and radiation therapy at any point following diagnosis were associated with poorer outcomes for the overall disease group. Although initially counterintuitive, one explanation to this phenomenon may be that these covariates also represent patients who recur and require chemotherapy or radiotherapy following the initial treatment phase, thus biasing the variable to become a negative prognostic factor.
[12]
[13]
[14]
5. Conclusions [15]
Our study provides evidence that survival in penile cancer has remained similar since 1998, except for stage II cancers. Centralization of stage II disease at academic centers and improved care at these centers, particularly through their expertise in LNDs, may potentially explain the observed improvement in stage II survival.
[16] [17]
[18]
References [1] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30, http://dx.doi.org/10.3322/caac.21332. [2] Ficarra V, Akduman B, Bouchot O, Palou J, Tobias-Machado M. Prognostic factors in penile cancer. Urology 2010;76:S66–73, http: //dx.doi.org/10.1016/j.urology.2010.04.008. [3] Sharma P, Ashouri K, Zargar-Shoshtari K, Luchey AM, Spiess PE. Racial and economic disparities in the treatment of penile squamous cell carcinoma: results from the National Cancer Database. Urol Oncol 2016;34:122.e9–122.e15, http://dx.doi.org/10.1016/j.urolonc. 2015.10.001. [4] Minhas S, Kayes O, Hegarty P, Kumar P, Freeman A, Ralph D. What surgical resection margins are required to achieve oncological control in men with primary penile cancer? BJU Int 2005;96:1040–3, http: //dx.doi.org/10.1111/j.1464-410X.2005.05769.x. [5] Pietrzak P, Corbishley C, Watkin N. Organ-sparing surgery for invasive penile cancer: early follow-up data. BJU Int 2004;94: 1253–7, http://dx.doi.org/10.1111/j.1464-410X.2004.05153.x. [6] Lont AP, Horenblas S, Tanis PJ, Gallee MPW, Tinteren H, van, Nieweg OE. Management of clinically nude negative penile carcinoma: improved survival after the introduction of dynamic sentinel node biopsy. J Urol 2003;170:783–6, http://dx.doi.org/ 10.1097/01.ju.0000081201.40365.75. [7] Tanis PJ, Lont AP, Meinhardt W, Olmos RAV, Nieweg OE, Horenblas S. Dynamic sentinel node biopsy for penile cancer: reliability of a staging technique. J Urol 2002;168:76–80. [8] Leijte JAP, Hughes B, Graafland NM, Kroon BK, Olmos RAV, Nieweg OE, et al. Two-center evaluation of dynamic sentinel node biopsy for squamous cell carcinoma of the penis. J Clin Oncol 2009;27:3325–9, http://dx.doi.org/10.1200/JCO.2008.20.6870. [9] Kroon BK, Horenblas S, Lont AP, Tanis PJ, Gallee MPW, Nieweg OE. Patients with penile carcinoma benefit from immediate resection of clinically occult lymph node metastases. J Urol 2005;173:816–9, http://dx.doi.org/10.1097/01.ju.0000154565.37397.4d. [10] Pagliaro LC, Williams DL, Daliani D, Williams MB, Osai W, Kincaid M, et al. Neoadjuvant paclitaxel, ifosfamide, and cisplatin chemotherapy for metastatic penile cancer: a phase II study. J Clin Oncol 2010;28:3851–7, http://dx.doi.org/10.1200/JCO.2010. 29.5477. [11] Pettaway CA, Pagliaro L, Theodore C, Haas G. Treatment of visceral, unresectable, or bulky/unresectable regional metastases of penile
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
cancer. Urology 2010;76:S58–65, http://dx.doi.org/10.1016/j.urology. 2010.03.082. Leijte JAP, Kerst JM, Bais E, Antonini N, Horenblas S. Neoadjuvant chemotherapy in advanced penile carcinoma. Eur Urol 2007;52: 488–94, http://dx.doi.org/10.1016/j.eururo.2007.02.006. Matulewicz RS, Flum AS, Helenowski I, Jovanovic B, Palis B, Bilimoria KY, et al. Centralization of penile cancer management in the United States: a combined analysis of the American Board of Urology and National Cancer Data Base. Urology 2016;90:82–8, http: //dx.doi.org/10.1016/j.urology.2015.12.058. Verhoeven RHA, Janssen-Heijnen MLG, Saum KU, Zanetti R, Caldarella A, Holleczek B, et al. Population-based survival of penile cancer patients in Europe and the United States of America: no improvement since 1990. Eur J Cancer 2013;49:1414–21, http: //dx.doi.org/10.1016/j.ejca.2012.10.029. American Joint Commission on Cancer: AJCC cancer staging manual, 5th ed. Philadelphia: Lippincott-Raven, 1997. American Joint Commission on Cancer: AJCC cancer staging manual. Springer Science & Business Media: 2002. Giobbie-Hurder A, Gelber RD, Regan MM. Challenges of guaranteetime bias. J Clin Oncol 2013;31:2963–9, http://dx.doi.org/10.1200/ JCO.2013.49.5283. Thakkar B, Hur K, Henderson WG, Oprian C. A method to generate Kaplan-Meier and adjusted survival curves using SAS SUGI 23, Paper 226, March 22-25. Nashville, USA. 1998. Available at: http://www2.sas.com/proceedings/sugi23/Stats/p226.pdf Accessed 02.05.2011. Djajadiningrat RS, Graafland NM, van Werkhoven E, Meinhardt W, Bex A, van der Poel HG, et al. Contemporary management of regional nodes in penile cancer-improvement of survival? J Urol 2014;191:68–73, http://dx.doi.org/10.1016/j.juro.2013. 07.088. Ayres BE, Hounsome L, Alnajjar H, Sharma D, Verne J, Watkin NA. Has centralisation of penile cancer services in the United Kingdom improved survival? Eur Urol Suppl 2014;13:e50, http://dx.doi.org/ 10.1016/S1569-9056(14)60052-X. Johnson TV, Hsiao W, Delman KA, Jani AB, Brawley OW, Master VA. Extensive inguinal lymphadenectomy improves overall 5-year survival in penile cancer patients. Cancer 2010;116:2960–6, http://dx. doi.org/10.1002/cncr.25091. Zargar-Shoshtari K, Sharma P, Djajadiningrat R, Catanzaro M, Ye D-W, Zhu Y, et al. Extent of pelvic lymph node dissection in penile cancer may impact survival. World J Urol 2016;34:353–9, http: //dx.doi.org/10.1007/s00345-015-1593-5. Mossanen M, Holt S, Gore JL, Lin DW, Wright JL. 15 Years of penile cancer management in the United States: an analysis of the use of partial penectomy for localized disease and chemotherapy in the metastatic setting. Urol Oncol 2016 http://dx.doi.org/10.1016/j. urolonc.2016.06.019. Djajadiningrat RS, van Werkhoven E, Meinhardt W, van Rhijn BWG, Bex A, van der Poel HG, et al. Penile sparing surgery for penile cancer—does it affect survival? J Urol 2014;192:120–5, http://dx.doi. org/10.1016/j.juro.2013.12.038. Theodore C, Skoneczna I, Bodrogi I, Leahy M, Kerst JM, Collette L, et al. A phase II multicentre study of irinotecan (CPT 11) in combination with cisplatin (CDDP) in metastatic or locally advanced penile carcinoma (EORTC PROTOCOL 30992). Ann Oncol 2008;19:1304–7, http://dx.doi.org/10.1093/annonc/mdn149. Haas GP, Blumenstein BA, Gagliano RG, Russell CA, Rivkin SE, Culkin DJ, et al. Cisplatin, methotrexate and bleomycin for the treatment of carcinoma of the penis: a Southwest Oncology Group study. J Urol 1999;161:1823–5. Kirrander P, Sherif A, Friedrich B, Lambe M, Håkansson U, Steering Committee of the Swedish National Penile Cancer Register: Swedish National Penile Cancer Register: incidence, tumour characteristics, management and survival. BJU Int 2016;117:287–92, http://dx.doi. org/10.1111/bju.12993.
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎ [28] Cho H, Howlader N, Mariotto AB, Cronin KA. Estimating relative survival for cancer patients from the SEER Program using expected rates based on Ederer I versus Ederer II method. Surveillance Research Program NCI Technical Report #2011-01. [29] Stitzenberg KB, Sanoff HK, Penn DC, Meyers MO, Tepper JE. Practice patterns and long-term survival for early-stage rectal cancer. J Clin Oncol 2013;31:4276–82, http://dx.doi.org/10.1200/JCO.2013. 49.1860.
9
[30] Lee J-Y, Kim TH, Suh DH, Kim JW, Kim HS, Chung HH, et al. Impact of guideline adherence on patient outcomes in early-stage epithelial ovarian cancer. Eur J Surg Oncol 2015;41:585–91, http: //dx.doi.org/10.1016/j.ejso.2015.01.006. [31] Slopnick EA, Kim SP, Kiechle JE, Gonzalez CM, Zhu H, Abouassaly R. Racial disparities differ for African Americans and Hispanics in the diagnosis and treatment of penile cancer. Urology 2016 http://dx.doi. org/10.1016/j.urology.2016.06.048.
Original article
Contemporary survival trends in penile cancer: Results from the National Cancer Database Minh N. Pham, B.S.a, Allison M. Deal, M.S.b, Jed E. Ferguson III, M.D., Ph.D.c, Yue Wang, B.S.d, Angela B. Smith, M.D., M.S.e, Matthew E. Nielsen, M.D., M.S.e,f, Raj S. Pruthi, M.D.e, Michael E. Woods, M.D.e,⁎ a University of North Carolina School of Medicine, Chapel Hill, NC Lineberger Comprehensive Cancer Center, Biostatistics and Clinical Data Management Core, University of North Carolina, Chapel Hill, NC c Department of Urology, University of North Carolina, Chapel Hill, NC d Department of Biostatistics, University of North Carolina, Chapel Hill, NC e Department of Urology, Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC f Department of Epidemiology and Health Policy & Management, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, NC b
Received 30 November 2016; received in revised form 6 June 2017; accepted 9 August 2017
Abstract Objective: To investigate contemporary survival trends in penile cancer. Methods: The National Cancer Database was queried for men with penile cancer diagnosed between 1998 and 2009. Patient, tumor, treatment, and facility characteristics were obtained. Overall survival (OS) was analyzed using the Kaplan-Meier method and multivariable Cox proportional hazards regression model for all cases and stratified by stage. Results: A total of 8,122 cases of penile cancer were reported from 1998 to 2009 in the National Cancer Database. Complete staging, survival, and covariate data were available for 5,043 cases. The estimated crude 5-year OS for the entire cohort was 61.0%. For all patients, no significant differences in crude 5-year OS were detected between 2006 to 2009 and 2002 to 2005 compared to 1998 to 2001. On multivariable analysis, OS did not significantly differ across all eras. Regional lymph node dissection was associated with improved OS (hazard ratio [HR] ¼ 0.777, P ≤ 0.0001). In patients who underwent lymph node dissection, dissection of ≥8 nodes significantly improved survival (HR ¼ 0.672; P ¼ 0.0011). Additional modeling stratified by stage revealed that OS for stage II cancers increased significantly in 2006 to 2009 compared to 1998 to 2001 (HR ¼ 0.714; P ¼ 0.0034). Conclusions: Survival in penile cancer has remained unchanged as a whole and for each stage, except for stage II disease. An improved survival trend was detected in stage II penile cancer. Performing a lymph node dissection, especially extensive dissections, may benefit longterm survival. r 2017 Elsevier Inc. All rights reserved.
Keywords: Penile cancer; Survival trends; United States; National Cancer Database
1. Introduction Penile cancer is a rare malignancy in the developed world. The American Cancer Society estimates that penile cancer will account for 2,030 new cases and 340 deaths in the United States in 2016 [1]. The prognosis is excellent when diagnosed early but dramatically worsens with nodal metastasis [2]. One report recently estimated the U.S. 5-year ⁎
Corresponding author. Tel.: þ1-919-966-2574. E-mail address: [email protected] (M.E. Woods). http://dx.doi.org/10.1016/j.urolonc.2017.08.009 1078-1439/r 2017 Elsevier Inc. All rights reserved.
overall survival (OS) as 66% [3]. Recently, evidence has emerged supporting that survival and quality of life can be improved through organ-sparing techniques [4,5], earlier detection of inguinal adenopathy [6–9], and postchemotherapy surgical consolidation of metastases [10–12]. The distribution of care has also changed at a facility-level. Community hospitals and academic hospitals are comparably likely to diagnose the disease, yet the responsibility of treatment is centralizing toward academic centers [13]. Despite these changes, little data exists examining mortality trends in the past 2 decades. One report of 1,800 patients
2
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
found that U.S. 5-year survival declined from 72% in cases from 1990 to 1995 compared to 63% in 2002 to 2007 [14]. Using a larger sample, we further evaluated OS trends in penile cancer cases diagnosed between 1998 and 2009 through the National Cancer Database (NCDB).
2. Materials and methods The NCDB is a national cancer registry jointly supported by the American College of Surgeons Commission on Cancer and the American Cancer Society. Over 1,500 accredited hospital-based cancer programs contribute to the NCDB, capturing 470% of newly diagnosed cancers in the United States and Puerto Rico. All data are entered by participating institutions and standardized according to coding guidelines published in the Facility Oncology Registry Data Standards (https://www.facs.org/quality-pro grams/cancer/ncdb/registrymanuals/cocmanuals/fordsma nual). Data include the following: patient and hospital characteristics, staging, histology, anatomic location, firstline treatment, and outcomes. All patient data were deidentified before the date of query. This study was approved by the University of North Carolina’s Institutional Review Board. The NCDB was queried for cases of penile cancer. Cases with the histologic International Classification of Diseases for Oncology (ICD)-O-3 codes pertaining to the following histologies were included: squamous cell carcinoma (8070, 8071, 8072, 8073, 8074, 8075, 8076), erythroplasia of Queyrat (8080), and Bowen disease (8081). From 1998 to 2009, the American Joint Committee on Cancer staging definitions for penile cancer remained unchanged [15,16]. Data were available for patients diagnosed in 1998 through 2012, but only cases between 1998 and 2009 were included to maintain consistent staging definitions and owing to the absence of 5-year OS data available from 2010 and later. Cases diagnosed at autopsy were excluded. The analysis was limited to NCDB sequence numbers 00, corresponding to cases with only 1 lifetime cancer diagnosis, and 01, representing tumors that were the first of multiple cancer diagnoses. Cases with the NCDB class code 00, which were those diagnosed but not treated by the reporting facility and did not require follow-up per Commission on Cancer guidelines, were excluded. We used NCDB analytic stage groups, which corresponded to pathologic stage group when available, and clinical stage group when pathologic data was unavailable. Cases without complete analytic staging and survival data were excluded. Our final sample was limited to patients with complete covariate data within the multivariable analysis with the exception of unknown grade for stage IV. Demographic variables included age, race/ethnicity, insurance status, median household income of the patient’s area of residence, and travel distance to facility. The NCDB began recording Charlson/Deyo scores in 2003, and owing
to the large volume of missing scores within our sample, Charlson/Deyo scores were described but excluded from the primary multivariable model. A sensitivity analysis including comorbidity in the multivariable model was then performed with patients diagnosed during 2003 to 2009. Facility-level characteristics included academic vs. community centers and facility region. Tumor characteristics included analytic stage group, grade, and the presence of invasive or in-situ disease. Primary site was described but excluded from the multivariable analysis owing to a substantial proportion of cases with missing data. Cases recorded as erythroplasia of Queyrat (ICD-O-3 8080), Bowen disease (ICD-O-3 8081), and analytic stage group 0 with pathologic stage Tis were analyzed as having welldifferentiated grade. Treatment characteristics analyzed included regional lymph node dissection (LND), chemotherapy, and radiation therapy. When a case met 3 criteria of having an unclear history of LND owing to missing data, was described to be pN0, and was reported to have “no lymph nodes examined” within the data, the case was regarded in the model as not having undergone LND. The survival impact of lymph node yield on survival was evaluated through a separate multivariable model limited to patients that underwent LND. Descriptive statistics are reported overall and by era, with comparisons between era made by chi-squared tests. The Kaplan-Meier method was used to estimate crude median survival and 5-year OS rates. The log-rank test was used for comparisons of survival curves. Multivariable Cox proportional hazards regression models were used to evaluate the combined effect of patient and clinical characteristics on survival, for the entire cohort, and stratified by stage. Chemotherapy and radiation therapy were included in the model as time-varying covariates [17]. For the stratified analysis for stage IV patients, grade was excluded from the model owing to the large amount of missing data and its relatively low clinical importance in systemic disease. Estimates from the adjusted Cox model were used to create the adjusted Kaplan-Meier curves [18]. Analyses were performed using SAS statistical software v.9.4 (Cary, NC).
3. Results Between 1998 and 2009, a total of 8,122 cases, in whom penile cancer was the only cancer diagnosis or the first of multiple cancer diagnoses, were reported to the NCDB. Of these, 5,043 had complete analytic staging, survival, and covariate information (Fig. 1). Pathologic staging was not reported for 1,743 (34.6%) cases, requiring the usage of clinical staging. Patient, tumor, treatment, and facility characteristics are summarized in Table 1. The median age of diagnosis was 65 years. Non-Hispanic White, nonHispanic African-American, and Hispanic patients accounted for 79.0%, 9.2%, and 9.4% of cases, respectively, 93.6% were insured. Stages were distributed, in
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Penile cancer is first diagnosis or first of multiple diagnoses (N = 8122) Missing staging data (n = 929) (n = 7193) Missing survival data (n = 32) (n = 7161) Missing covariate data (n = 2118) Final cohort analyzed (n = 5043) Fig. 1. Process of exclusion from the final cohort.
descending order of frequency, as I (34.4%), 0 (25.7%), II (21.0%), III (13.0%), and IV (5.9%). The most common sites for tumor localization were the glans followed by the prepuce. Most tumors were invasive (75.7%) and of moderately differentiated (37.1%) or well-differentiated grade (45.9%). More patients were treated at community (61.0%) than academic programs (39.0%); however, the proportion of patients managed at academic centers increased from 36.4% in 1998 to 2001, to 36.7% in 2002 to 2005, and 43.5% in 2006 to 2009 (P o 0.0001). Chemotherapy and radiotherapy were administered in 7.3% and 7.3%, respectively. Whereas radiotherapy rates remained stable, chemotherapy rates significantly increased from 5.5% in 1998 to 2001, to 7.3% in 2002 to 2005, and 8.8% in 2006 to 2009 (P ¼ 0.0012). The relative rates of penile cancer stages for each year are shown in Fig. 2 (n ¼ 7,193). Stage I cases were most common, ranging from 29% to 40%. The proportion of stage 0 cases ranged from 25% to 33%. Stage II cases varied from 16% to 23%. For stage III, the proportion fluctuated between 9% and 14%. Stage IV accounted for 4% to 8% of cases. Overall, the 5-year OS rate was 61.0%. No significant differences in crude 5-year OS were found among 1998 to 2001 (60.4%), 2002 to 2005 (61.3%), and 2006 to 2009 (61.2%) (P ¼ 0.99). Median OS during each era was 7.4 years for 1998 to 2001, 7.7 years for 2002 to 2005, and unavailable for 2006 to 2009 (owing to survival being greater than 50% by the time the data was queried) with mean follow-up times for survivors of 8.4 years, 7.0 years, 4.2 years, respectively. Fig. 3A demonstrates the adjusted Kaplan-Meier curves for the overall sample stratified by era. After adjusting for patient, tumor, and facility characteristics in a multivariable analysis, we found no significant differences in OS across all eras for penile cancer as a whole (Table 2). Negative prognostic factors were as follows: age of diagnosis at 60 years or older, non-Hispanic
3
African-American race, lower median household income, higher stage, less differentiated grade, and chemotherapy or radiotherapy at any point following diagnosis. Age of diagnosis younger than 40 years, carcinoma in-situ, and regional LND were associated with improved OS. Additional modeling stratified by stage included the same patient, tumor, treatment and facility characteristics, excluding grade for the stage IV model. This model revealed improved survival in stage II patients in 2006 to 2009 compared to 1998 and 2001. For stages 0, I, III, and IV, no significant survival changes were detected across all eras. A total of 1,059 cases of stage II disease were analyzed. Of note, the proportion of stage II cases treated at academic centers significantly increased from 35.3% in 1998 to 2001, to 41.3% in 2002 to 2005, and 44.8% in 2006 to 2009 (P ¼ 0.0375). A 5-year OS for stage II patients was 53.5% with estimates for each era as follows: 1998 to 2001 (50.7%), 2002 to 2005 (51.0%), 2006 to 2009 (57.6%) (P ¼ 0.12). Median OS for stage II patients was 5.3 years, 5.3 years, and 6.5 years during 1998 to 2001, 2002 to 2005, and 2006 to 2009, respectively. Fig. 3B demonstrates the adjusted Kaplan-Meier curves for stage II disease stratified by era. On multivariable analysis of stage II cases, OS significantly improved in 2006 to 2009 compared to 1998 to 2001 [hazard ratio (HR) ¼ 0.714; P ¼ 0.0034]. After controlling for demographic, tumor, facility, and treatment characteristics, regional LND was a significant prognostic factor for the combined cohort of penile cancer (HR ¼ 0.777; P o 0001) and the stage IV disease group (HR ¼ 0.556; P o 0.0001). LND was not significantly associated with improved OS in stage II (HR ¼ 0.810; P ¼ 0.0603). In a multivariable analysis of patients who underwent LND (n ¼ 744), a yield of ≥8 nodes significantly benefited survival for all stages combined (HR ¼ 0.672; P ¼ 0.0011) and for stage III cases (HR ¼ 0.545; P ¼ 0.009). The impact of comorbidity on survival was evaluated by including Charlson/Deyo scores in a separate multivariable analysis of cases diagnosed from 2003 to 2009 (data not shown). Increased comorbidity was associated with poorer OS. For penile cancer as a whole, OS remained unchanged between 2003 to 2005 and 2006 to 2009. Younger age and regional LND, but not carcinoma in-situ, remained positive prognostic factors. Older age, lower income, higher stage, less differentiated grade, and treatment with chemotherapy or radiotherapy at any point following diagnosis remained negative prognostic factors. After accounting for comorbidity, non-Hispanic African-American race was no longer significantly associated with negative survival outcomes. Compared to 2003 to 2005, survival in 2006 to 2009 improved in stage II. LND continued to be significantly associated with improved OS in penile cancer as a whole, stage IV, and additionally stage III. Among those treated with regional lymphadenectomy, yields of 8 nodes or greater remained significantly associated with improved OS for penile cancer as a whole, stage III, and also stage IV.
4
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Table 1 Demographic, tumor, treatment, and facility characteristics for penile cancer cases diagnosed from 1998 to 2009 Characteristics Age at diagnosis, y, n (%) o40 40–59 60–79 ≥80 Race/ethnicity, n (%) White, non-Hispanic African-American, non-Hispanic Hispanic Other Stage group, n (%) 0 I II III IV Grade, n (%) Well differentiated Moderately differentiated Poorly differentiated Undifferentiated Behavior, n (%) Invasive In Situ Insurance status, n (%) Yes No Primary site, n (%)b Unknown Prepuce Glans Body Overlapping lesion Chemotherapy, n (%) Yes No Radiotherapy, n (%) Yes No Lymph node dissection Yes No Charlson/Deyo scoreb,c Unknown 0 1 ≥2 Region, n (%) South Midwest Northeast West Facility, n (%) Academic Community Median income ($),d n (%) o38,000 38,000–47,999 48,000–62,999 463,000
1998–2001 (n ¼ 1,662)
2002–2005 (n ¼ 1,566)
2006–2009 (n ¼ 1,815)
Overall (n ¼ 5,043)
P valuea
100 498 826 238
(6.0) (30.0) (49.7) (14.3)
97 494 764 211
(6.2) (31.6) (48.8) (13.5)
95 577 893 250
(5.2) (31.8) (49.2) (13.8)
292 1,569 2,483 699
(5.8) (31.1) (49.2) (13.9)
0.7841
1,337 143 143 39
(80.5) (8.6) (8.6) (2.4)
1,254 134 144 34
(80.1) (8.6) (9.2) (2.2)
1,395 185 188 47
(76.9) (10.2) (10.4) (2.6)
3,986 462 475 120
(79.0) (9.2) (9.4) (2.4)
0.1974
445 637 303 196 81
(26.8) (38.3) (18.2) (11.8) (4.9)
384 526 334 224 98
(24.5) (33.6) (21.3) (14.3) (6.3)
465 572 422 237 119
(25.6) (31.5) (23.3) (13.1) (6.6)
1,294 1,735 1,059 657 298
(25.7) (34.4) (21.0) (13.0) (5.9)
o0.0001
803 616 232 11
(48.3) (37.1) (14.0) (0.7)
699 584 280 3
(44.6) (37.3) (17.9) (0.2)
815 671 319 10
(44.9) (37.0) (17.6) (0.6)
2,317 1,871 831 24
(45.9) (37.1) (16.5) (0.5)
0.0103
1,236 (74.4) 426 (25.6)
1,203 (76.8) 363 (23.2)
1,378 (75.9) 437 (24.1)
3,817 (75.7) 1,226 (24.3)
0.2569
1,567 (94.3) 95 (5.7)
1,474 (94.1) 92 (5.9)
1,677 (92.4) 138 (7.6)
4,718 (93.6) 325 (6.4)
0.0418
768 208 517 97 72
(46.2) (12.5) (31.1) (5.8) (4.3)
791 143 456 100 76
(50.5) (9.1) (29.1) (6.4) (4.9)
976 161 483 128 67
(53.8) (8.9) (26.6) (7.1) (3.7)
2,535 512 1,456 325 215
(50.3) (10.2) (28.9) (6.4) (4.3)
o0.0001
92 (5.5) 1,570 (94.5)
115 (7.3) 1,451 (92.7)
159 (8.8) 1,656 (91.2)
366 (7.3) 4,677 (92.7)
0.0012
118 (7.1) 1,544 (92.9)
119 (7.6) 1,447 (92.4)
131 (7.2) 1,684 (92.8)
368 (7.3) 4,675 (92.7)
0.8507
259 (15.6) 1,403 (84.4)
255 (16.3) 1,311 (83.7)
328 (18.1) 1,487 (81.9)
842 (16.7) 4,201 (83.3)
0.1262
1,662 0 0 0
405 918 (79.1) 178 (15.3) 65 (5.6)
0 1,322 (72.8) 377 (20.8) 116 (6.4)
2,067 2,240 (75.3) 555 (18.6) 181 (6.1)
o0.0001
485 469 299 409
(29.2) (28.2) (18.0) (24.6)
605 (36.4) 1,057 (63.6) 392 454 447 369
(23.6) (27.3) (26.9) (22.2)
454 433 322 357
(29.0) (27.7) (20.6) (22.8)
574 (36.7) 992 (63.4) 331 433 441 361
(21.1) (27.7) (28.2) (23.1)
547 456 410 402
(30.1) (25.1) (22.6) (22.2)
790 (43.5) 1,025 (56.5) 441 467 473 434
(24.3) (25.7) (26.1) (22.9)
1,486 1,358 1,031 1,168
(29.5) (26.9) (20.4) (23.2)
0.0177
1,969 (39.0) 3,074 (61.0)
o0.0001
1,164 1,354 1,361 1,164
(23.1) (26.9) (27.0) (23.1)
0.2570
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
5
Table 1 Continued Characteristics Distance to facility (miles), n (%) o10 10–50 450
1998–2001 (n ¼ 1,662) 878 (52.8) 562 (33.8) 222 (13.4)
2002–2005 (n ¼ 1,566) 749 (47.8) 597 (38.1) 220 (14.1)
2006–2009 (n ¼ 1,815) 837 (46.1) 710 (39.1) 268 (14.8)
Overall (n ¼ 5,043) 2,464 (48.9) 1,869 (37.1) 710 (14.1)
P valuea 0.0019
a
Chi-squared tests were performed to compare differences between eras. Not included in primary multivariate analysis. Charlson/Deyo scores were included in a sensitivity analysis using a separate multivariable model. c Not recorded in the National Cancer Database until January 1, 2003 and later. Cases with missing scores are not accounted for in percentages. d Representing median household income of the patient’s residential area measured by the 2000 U.S. census data. b
4. Discussion In this study, we used a large, representative national database to evaluate OS trends for penile cancer diagnosed between 1998 and 2009. We analyzed outcomes for all cases and stratified by NCDB analytic stages, corresponding to American Joint Committee on Cancer-defined pathologic stage when available or clinical stage when the former was missing. Survival remained largely unchanged, and changes in staging distribution over time were small. Notably, multivariable analysis of stage II disease found that OS improved significantly in 2006 to 2009 compared to 1998 to 2001. To our knowledge, this is the largest analysis of survival trends in penile cancer and the first to report an improvement in stage II disease. Previous studies demonstrate inconclusive evidence about penile cancer survival trends in the industrialized world, complicated by differences in cancer registries and health care systems. In the Netherlands, 5-year cancerspecific survival (CSS) for node-negative disease improved from 82% during 1956 to 1993 to 91% during 1994 to 2012 after dynamic sentinel node biopsies were implemented to replace prophylactic inguinal LNDs [19]. For the United States, Verhoeven et al. [14] evaluated 1,820 penile cancer cases from the Surveillance, Epidemiology and End Results program and found a 5-year survival decline from 72% to
Fig. 2. Temporal trends in penile cancer staging distributions (n ¼ 7,193).
63% between 1990 and 2007. The contrasting results between our study and those of Verhoeven et al. may be explained by different periods of study. After disease management centralized in the United Kingdom, 5-year CSS improved by 7% to 12% [20]. Our data shows management shifting toward academic facilities. Compared to community programs, academic centers see more than twice the amount of cases per year, manage more advanced-staged disease, perform more LNDs in eligible patients, and have higher lymph node yields during their dissections [13]. We detected no survival difference between academic and community programs. That being said, a survival change may require more time before being detected. Alternatively, the benefit may be restricted to stages most affected by centralization. Stage II survival improved significantly in 2006 to 2009 on multivariable analysis. Stage II penile cancer was defined in 1998 to 2009 as T1N1, T2N0, or T2N1 disease [15,16]. Several changes in T1-, T2-, and N1-staged management related to centralization may contribute to this trend. Increased rates of stage II patients were seen at academic centers in 2002 to 2005 (41.3%) and 2006 to 2009 (44.8%) compared to 1998 to 2001 (35.3%). Among patients with T1b or greater T-stage, node-positive disease, or high-grade penile cancer (grade 3 or 4), academic centers delivered higher rates of guideline-concordant management and were 2.29 times more likely to perform LNDs compared to community programs [13]. The greater likelihood of LNDs at academic centers is accompanied by an increasing trend in its frequency at these facilities: the proportion of these procedures performed among eligible patients at academic centers, over community centers, increased by 40% between 1998 and 2012 [13]. Academic centers were also more likely to perform extensive LNDs and on average yielded more nodes [13], which have been associated with improved survival [21,22]. Whereas we did not detect a significant survival benefit from LNDs for the stage II group, potentially because of an underpowered analysis, our results for all stages combined confirmed these findings, showing that LND and high lymph node yields (≥8 nodes) were associated with improved survival. Thus, the management of stage II disease may be benefiting from its centralization toward academic centers, where the care
6
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Fig. 3. Adjusted Kaplan-Meier plot for overall survival stratified by era for (A) overall disease and (B) stage 2 penile cancer.
delivered is more likely to be consistent with guidelines and thorough in technique. We found that OS in stages 0 and I remained stable in the backdrop of an increasing acceptance of organ-sparing intervention [13,23]. The usage of partial penectomy among patients with localized disease (clinical Ta-T2), for instance, has increased in the last 2 decades [23]. Although clinical practice has transitioned, there has been little data describing the impact of this paradigm shift in the United States. In the Netherlands, penile preservation techniques, such as laser therapy, local excision, and glans resection, were associated with increased recurrence rates compared to partial or total amputation but had comparable CSS [24]. Nonetheless, management in the Netherlands is highly centralized, limiting the study’s generalizability. Further validation of this data in the United States may be difficult owing to the rarity of the malignancy spread across a large geographic area. This study could not directly compare U.S. and Dutch outcomes because recurrence rates and CSS were not reported by the NCDB; however, the stability of OS among early stage disease in the real-world setting is reassuring. Survival in stages III and IV also remained unchanged. The lack of progress in survival outcomes in these stages is consistent with the absence of improved systemic therapies until 2008 [25]. Bleomycin, methotrexate plus cisplatin was introduced in 1999 and remained the most common regimen throughout our time-points [26]. More recently, studies evaluating cisplatin-based chemotherapy with surgical consolidation have suggested a more promising efficacy and tolerability over bleomycin-containing regimens [10–12]. Additionally, chemotherapy usage in metastatic disease is increasing [23]. Together, these changes may potentially impact penile cancer mortality in the future.
We estimate the crude 5-year OS rate for penile cancer as a whole to be 61%. In comparison, the estimated Swedish 5-year relative OS rate was 82% [27]. Several hypotheses may contribute to this disparity. Relative OS, compared to the standard OS in our study, represents cancer survival accounting for competing mortality risks, defined as the ratio between the observed cancer survivor rate and expected survivor rate in cancer-free individuals [28]. Thus, for a given cancer cohort, relative OS is expectedly higher than standard OS. Additionally, given the rarity of penile cancer, adherence to treatment guidelines is variable. United States centers performed LNDs in only 36.3% of eligible patients compared to 50% to 74% in Sweden [13,27]. Guideline-discordant care is associated with poorer outcomes in other cancers [29,30]. The rate of invasive cancer is also higher in the United States (76%) than Sweden (45%) [27]. Finally, because the NCDB measured survival outcomes through OS, the gap between Swedish and U.S. survival may partially be influenced by deaths not attributable to penile cancer. Racial and socioeconomic differences also contribute to outcomes in the United States. African-American race has been associated with poorer OS whereas private insurance and residence in areas with higher median income were associated with improved survival [3]. In our study, nonHispanic African-American race and residence in regions with lower income status were similarly predictors of poorer survival. Comorbidities affecting African Americans may partially contribute to the observed difference in survival. After accounting for Charlson/Deyo scores in our multivariable model, non-Hispanic African-American race was no longer associated with poorer survival, supporting the hypothesis that comorbidities may have confounded our primary findings related to race and ethnicity. Racial
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
7
Table 2 Multivariable Cox proportional hazards regression models for overall survival Characteristic
Levels
HR
Era
1998–2001 2002–2005 2006–2009 o40 40–59 60–79 ≥80 White, non-Hispanic African-American, non-Hispanic Hispanic Other 0 I II III IV Well differentiated Moderately differentiated Poorly differentiated Undifferentiated, anaplastic Invasive In situ Yes No No Yes No Yes No Yes South Midwest Northeast West Academic Community o38,000 38,000–47,999 48,000–62,999 463,000 o10 10–50 450
Ref 0.940 0.914 0.635 Ref 1.811 4.220 Ref 1.286 1.045 0.810 1.007 Ref 1.388 1.893 4.523 Ref 1.246 1.515 2.712 Ref 0.691 Ref 1.097 Ref 1.616 Ref 1.217 Ref 0.777 Ref 1.002 0.962 0.903 Ref 1.039 1.246 1.142 1.070 Ref Ref 1.000 0.985
Age at diagnosis, y
Race/ethnicity
Stage group
Grade
Behavior Insurance status Chemotherapya Radiation therapya Regional lymph node dissection Region
Facility type Median income ($)
Median distance to facility (miles)
95% CI
P value
0.851–1.037 0.822–1.017 0.473–0.851
0.2161 0.0983 0.0024
1.623–2.021 3.697–4.817
o0.0001 o0.0001
1.115–1.483 0.884–1.235 0.594–1.105 0.707–1.433
0.0006 0.6064 0.1837 0.9707
1.239–1.555 1.661–2.158 3.826–5.347
o0.0001 o0.0001 o0.0001
1.117–1.391 1.333–1.722 1.585–4.641
o0.0001 o0.0001 0.0003
0.480–0.995
0.0469
0.904–1.330
0.3484
1.377–1.896
o0.0001
1.044–1.419
0.0122
0.688–0.877
o0.0001
0.897–1.119 0.848–1.091 0.801–1.018
0.9738 0.5459 0.0966
0.948–1.140 1.092–1.421 1.008–1.294 0.948–1.209
0.4118 0.0011 0.0370 0.2741
0.913–1.096 0.861–1.128
0.9953 0.8320
Ref ¼ reference. a Reflects chemotherapy or radiation therapy at any point in following diagnosis. Variables were analyzed as time-varying covariates.
disparities in treatments also exist. For instance, AfricanAmericans are less likely to undergo chemotherapy for metastatic disease or any surgery for penile cancer [23,31]. Our study has limitations inherent in the NCDB. The NCDB is limited to measuring OS and does not report CSS. One approach to address this limitation is to adjust for comorbidity. Unfortunately, the database did not record Charlson/Deyo scores until 2003. We chose to exclude these scores in our main multivariable analysis to preserve a substantial number of patients; however, a subgroup analysis including comorbidity for patients diagnosed between 2003 and 2009 showed largely the same findings. Therefore, the results of this sensitivity analysis supported the use
of our primary model without Charlson scores to evaluate differences in eras. If comorbidity rates had increased between 1998 and 2009, our ability to detect significant survival improvement across eras would be inhibited by competing mortality risks; however, with the data available, this would be difficult to validate. To appropriately account for the impact of chemotherapy and radiation therapy on survival, we had to address the potential for guarantee time bias [17]. Although chemotherapy rates had significantly increased during the period of study, these rates reflected treatment with chemotherapy at any point following diagnosis. Thus, we chose to include chemotherapy and radiation therapy through an extended Cox model with
8
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
time-varying covariates rather than through a conditional landmark analysis. Our model subsequently found that undergoing chemotherapy and radiation therapy at any point following diagnosis were associated with poorer outcomes for the overall disease group. Although initially counterintuitive, one explanation to this phenomenon may be that these covariates also represent patients who recur and require chemotherapy or radiotherapy following the initial treatment phase, thus biasing the variable to become a negative prognostic factor.
[12]
[13]
[14]
5. Conclusions [15]
Our study provides evidence that survival in penile cancer has remained similar since 1998, except for stage II cancers. Centralization of stage II disease at academic centers and improved care at these centers, particularly through their expertise in LNDs, may potentially explain the observed improvement in stage II survival.
[16] [17]
[18]
References [1] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016;66:7–30, http://dx.doi.org/10.3322/caac.21332. [2] Ficarra V, Akduman B, Bouchot O, Palou J, Tobias-Machado M. Prognostic factors in penile cancer. Urology 2010;76:S66–73, http: //dx.doi.org/10.1016/j.urology.2010.04.008. [3] Sharma P, Ashouri K, Zargar-Shoshtari K, Luchey AM, Spiess PE. Racial and economic disparities in the treatment of penile squamous cell carcinoma: results from the National Cancer Database. Urol Oncol 2016;34:122.e9–122.e15, http://dx.doi.org/10.1016/j.urolonc. 2015.10.001. [4] Minhas S, Kayes O, Hegarty P, Kumar P, Freeman A, Ralph D. What surgical resection margins are required to achieve oncological control in men with primary penile cancer? BJU Int 2005;96:1040–3, http: //dx.doi.org/10.1111/j.1464-410X.2005.05769.x. [5] Pietrzak P, Corbishley C, Watkin N. Organ-sparing surgery for invasive penile cancer: early follow-up data. BJU Int 2004;94: 1253–7, http://dx.doi.org/10.1111/j.1464-410X.2004.05153.x. [6] Lont AP, Horenblas S, Tanis PJ, Gallee MPW, Tinteren H, van, Nieweg OE. Management of clinically nude negative penile carcinoma: improved survival after the introduction of dynamic sentinel node biopsy. J Urol 2003;170:783–6, http://dx.doi.org/ 10.1097/01.ju.0000081201.40365.75. [7] Tanis PJ, Lont AP, Meinhardt W, Olmos RAV, Nieweg OE, Horenblas S. Dynamic sentinel node biopsy for penile cancer: reliability of a staging technique. J Urol 2002;168:76–80. [8] Leijte JAP, Hughes B, Graafland NM, Kroon BK, Olmos RAV, Nieweg OE, et al. Two-center evaluation of dynamic sentinel node biopsy for squamous cell carcinoma of the penis. J Clin Oncol 2009;27:3325–9, http://dx.doi.org/10.1200/JCO.2008.20.6870. [9] Kroon BK, Horenblas S, Lont AP, Tanis PJ, Gallee MPW, Nieweg OE. Patients with penile carcinoma benefit from immediate resection of clinically occult lymph node metastases. J Urol 2005;173:816–9, http://dx.doi.org/10.1097/01.ju.0000154565.37397.4d. [10] Pagliaro LC, Williams DL, Daliani D, Williams MB, Osai W, Kincaid M, et al. Neoadjuvant paclitaxel, ifosfamide, and cisplatin chemotherapy for metastatic penile cancer: a phase II study. J Clin Oncol 2010;28:3851–7, http://dx.doi.org/10.1200/JCO.2010. 29.5477. [11] Pettaway CA, Pagliaro L, Theodore C, Haas G. Treatment of visceral, unresectable, or bulky/unresectable regional metastases of penile
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
cancer. Urology 2010;76:S58–65, http://dx.doi.org/10.1016/j.urology. 2010.03.082. Leijte JAP, Kerst JM, Bais E, Antonini N, Horenblas S. Neoadjuvant chemotherapy in advanced penile carcinoma. Eur Urol 2007;52: 488–94, http://dx.doi.org/10.1016/j.eururo.2007.02.006. Matulewicz RS, Flum AS, Helenowski I, Jovanovic B, Palis B, Bilimoria KY, et al. Centralization of penile cancer management in the United States: a combined analysis of the American Board of Urology and National Cancer Data Base. Urology 2016;90:82–8, http: //dx.doi.org/10.1016/j.urology.2015.12.058. Verhoeven RHA, Janssen-Heijnen MLG, Saum KU, Zanetti R, Caldarella A, Holleczek B, et al. Population-based survival of penile cancer patients in Europe and the United States of America: no improvement since 1990. Eur J Cancer 2013;49:1414–21, http: //dx.doi.org/10.1016/j.ejca.2012.10.029. American Joint Commission on Cancer: AJCC cancer staging manual, 5th ed. Philadelphia: Lippincott-Raven, 1997. American Joint Commission on Cancer: AJCC cancer staging manual. Springer Science & Business Media: 2002. Giobbie-Hurder A, Gelber RD, Regan MM. Challenges of guaranteetime bias. J Clin Oncol 2013;31:2963–9, http://dx.doi.org/10.1200/ JCO.2013.49.5283. Thakkar B, Hur K, Henderson WG, Oprian C. A method to generate Kaplan-Meier and adjusted survival curves using SAS SUGI 23, Paper 226, March 22-25. Nashville, USA. 1998. Available at: http://www2.sas.com/proceedings/sugi23/Stats/p226.pdf Accessed 02.05.2011. Djajadiningrat RS, Graafland NM, van Werkhoven E, Meinhardt W, Bex A, van der Poel HG, et al. Contemporary management of regional nodes in penile cancer-improvement of survival? J Urol 2014;191:68–73, http://dx.doi.org/10.1016/j.juro.2013. 07.088. Ayres BE, Hounsome L, Alnajjar H, Sharma D, Verne J, Watkin NA. Has centralisation of penile cancer services in the United Kingdom improved survival? Eur Urol Suppl 2014;13:e50, http://dx.doi.org/ 10.1016/S1569-9056(14)60052-X. Johnson TV, Hsiao W, Delman KA, Jani AB, Brawley OW, Master VA. Extensive inguinal lymphadenectomy improves overall 5-year survival in penile cancer patients. Cancer 2010;116:2960–6, http://dx. doi.org/10.1002/cncr.25091. Zargar-Shoshtari K, Sharma P, Djajadiningrat R, Catanzaro M, Ye D-W, Zhu Y, et al. Extent of pelvic lymph node dissection in penile cancer may impact survival. World J Urol 2016;34:353–9, http: //dx.doi.org/10.1007/s00345-015-1593-5. Mossanen M, Holt S, Gore JL, Lin DW, Wright JL. 15 Years of penile cancer management in the United States: an analysis of the use of partial penectomy for localized disease and chemotherapy in the metastatic setting. Urol Oncol 2016 http://dx.doi.org/10.1016/j. urolonc.2016.06.019. Djajadiningrat RS, van Werkhoven E, Meinhardt W, van Rhijn BWG, Bex A, van der Poel HG, et al. Penile sparing surgery for penile cancer—does it affect survival? J Urol 2014;192:120–5, http://dx.doi. org/10.1016/j.juro.2013.12.038. Theodore C, Skoneczna I, Bodrogi I, Leahy M, Kerst JM, Collette L, et al. A phase II multicentre study of irinotecan (CPT 11) in combination with cisplatin (CDDP) in metastatic or locally advanced penile carcinoma (EORTC PROTOCOL 30992). Ann Oncol 2008;19:1304–7, http://dx.doi.org/10.1093/annonc/mdn149. Haas GP, Blumenstein BA, Gagliano RG, Russell CA, Rivkin SE, Culkin DJ, et al. Cisplatin, methotrexate and bleomycin for the treatment of carcinoma of the penis: a Southwest Oncology Group study. J Urol 1999;161:1823–5. Kirrander P, Sherif A, Friedrich B, Lambe M, Håkansson U, Steering Committee of the Swedish National Penile Cancer Register: Swedish National Penile Cancer Register: incidence, tumour characteristics, management and survival. BJU Int 2016;117:287–92, http://dx.doi. org/10.1111/bju.12993.
M.N. Pham et al. / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎ [28] Cho H, Howlader N, Mariotto AB, Cronin KA. Estimating relative survival for cancer patients from the SEER Program using expected rates based on Ederer I versus Ederer II method. Surveillance Research Program NCI Technical Report #2011-01. [29] Stitzenberg KB, Sanoff HK, Penn DC, Meyers MO, Tepper JE. Practice patterns and long-term survival for early-stage rectal cancer. J Clin Oncol 2013;31:4276–82, http://dx.doi.org/10.1200/JCO.2013. 49.1860.
9
[30] Lee J-Y, Kim TH, Suh DH, Kim JW, Kim HS, Chung HH, et al. Impact of guideline adherence on patient outcomes in early-stage epithelial ovarian cancer. Eur J Surg Oncol 2015;41:585–91, http: //dx.doi.org/10.1016/j.ejso.2015.01.006. [31] Slopnick EA, Kim SP, Kiechle JE, Gonzalez CM, Zhu H, Abouassaly R. Racial disparities differ for African Americans and Hispanics in the diagnosis and treatment of penile cancer. Urology 2016 http://dx.doi. org/10.1016/j.urology.2016.06.048.