- Email: [email protected]
Evaluation of current surveillance guidelines following radical cystectomy and proposal of a novel risk-based approach
Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
Original article
Evaluation of current surveillance guidelines following radical cystectomy and proposal of a novel risk-based approach Suzanne B. Stewart-Merrill, M.D.a, Stephen A. Boorjian, M.D.a, Robert Houston Thompson, M.D.a, Sarah P. Psutka, M.D.a, John C. Cheville, M.D.b, Prabin Thapac, Eric J. Bergstrahl, M.S.c, Matthew K. Tollefson, M.D.a, Igor Frank, M.D.a,* a
Department of Urology, Mayo Clinic, Rochester, MN Department of Pathology, Mayo Clinic, Rochester, MN c Department of Health Sciences Research, Mayo Clinic, Rochester, MN b
Received 1 February 2015; received in revised form 8 April 2015; accepted 29 April 2015
Abstract Background: Evidence supporting surveillance guidelines after radical cystectomy (RC) are lacking. Herein, we evaluate the ability of the National Comprehensive Cancer Network (NCCN) guidelines and the European Association of Urology (EAU) guidelines to capture recurrences and provide an alternative approach that balances risks of recurrence with non–bladder cancer death. Methods: We identified 1,797 patients who had M0 urothelial carcinoma who underwent RC at our institution between 1980 and 2007. The success of current guidelines to capture recurrences was assessed by calculating the percentage of recurrences detected during the recommended follow-up time: the NCCN—2 years and the EAU—5 years. An alternative protocol was created using Weibull distributions, which estimate when a patient's risk of non–bladder cancer death exceeds their risk of recurrence. Results: At a median follow-up of 10.6 years (interquartile range : 6.8–15.2), a total of 714 patients recurred. Of these, 491 (68.7%) would have been detected by the NCCN guidelines and 642 (89.8%) by the EAU guidelines. Using a risk-adapted approach, vastly different surveillance durations were appreciated. For example, for patients older than 80 years with pT0Nx-0 or pTa/CIS/1Nx-0 disease, recurrence risk to any location never exceeded their risk of non–bladder cancer death, whereas for patients aged 60 years and younger with pT3/4Nx-0 or pTanyNþ disease, risk of abdominal/pelvis recurrence remained greater than their risk of non–bladder cancer death for 420 years. Conclusions: The duration of post-RC follow-up recommended by the NCCN and the EAU does not comprehensively capture recurrences. A surveillance algorithm based on the interaction between recurrence risk and competing health factors individualizes recommendations and may improve capture of recurrences and resource allocation. r 2015 Elsevier Inc. All rights reserved.
Keywords: Bladder cancer; Surveillance; Cystectomy; Guidelines; Recurrence
1. Introduction Currently, the National Comprehensive Cancer Network (NCCN) and the European Association of Urology (EAU) This publication was made possible by CTSA Grant: UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH), USA. The contents of the article are solely the responsibility of the authors and do not necessarily represent the official view of the NIH. * Corresponding author. Tel.:þ1-507-266-0191; fax:þ1-507-284-4951. E-mail address: [email protected] (I. Frank). http://dx.doi.org/10.1016/j.urolonc.2015.04.017 1078-1439/r 2015 Elsevier Inc. All rights reserved.
provide the most recognized oncologic surveillance guidelines following radical cystectomy (RC). However, these recommendations are not well supported by existing data and are disparate regarding total duration of follow-up. Specifically, the NCCN recommends that patients be followed up according to the same protocol, regardless of tumor stage, for 2 years [1]. Extension of surveillance beyond this time is then left up to clinical discretion. The EAU uses a stage-based approach to surveillance, recommending that patients with advanced-category disease be followed up more frequently. However, despite this
339.e2
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
risk-stratified approach, total follow-up for all stage groups is for 5 years [2]. The effect of stopping surveillance at 2 or 5 years has not been assessed, thereby making it unclear if this time frame is sufficient to comprehensively capture recurrences. Additionally, leaving the option of extending follow-up to the provider potentiates the existent heterogeneity in surveillance care [3]. The stopping points for surveillance are based on the appreciation that most of the recurrences following RC occur within 2 to 5 years. Such determinations are based on calculations of cumulative incidence of recurrences over time [4–10]. However, this approach does not account for how a patient's disease recurrence risk following RC changes with time and how competing comorbid conditions may influence this risk. Incorporating these features into guideline strategies may allow for development of more meaningful surveillance stopping points. Additionally, as health care transitions to a value-based system, and quality assessment and reimbursement become connected with adherence to standardized guidelines, improvement on surveillance strategies becomes imperative. Based on the earlier description, our objective was to first evaluate how many patients with recurrence following RC would have been detected if surveillance were strictly performed for the total time recommended by the NCCN or the EAU. Second, we present a novel approach to surveillance that dynamically models the relationship between a patient's risk of non–bladder cancer death and the risk of recurrence and propose individualized follow-up recommendations following RC.
2. Methods
outcomes annually through patient or treating physician correspondence. 2.2. Classification of disease recurrence and non–bladder cancer–related death Disease recurrence was defined as metastasis on imaging or biopsy at least 30 days after RC. Recurrences were classified according to location as urethral, upper urinary tract, abdomen/pelvis, thorax, and other (i.e., bone, central nervous system, and skin). For each patient, the first recurrence in time was counted as an event, and all subsequent recurrences were censored to prevent double counting. Recurrences occurring simultaneously in multiple locations were counted as individual events; however, it cannot be inferred if such events were independent. Non– bladder cancer death was defined as mortality at least 30 days after RC attributed to any cause except that related to the primary bladder cancer diagnosis, RC surgery, or postRC recurrence. 2.3. Evaluation of current guidelines We evaluated the NCCN and the EAU surveillance guidelines by determining the total number of recurrences that would have been captured if patients after RC were strictly followed up for the time prescribed: the EAU—5 years and the NCCN—2 years. The option of extending surveillance based on clinical discretion, as stated in the NCCN guidelines, was not examined owing to subjectivity. The total number of recurrences captured by these guidelines was also assessed according to pathologic stage (pT0Nx-0, pTa/CIS/1Nx-0, pT2Nx-0, pT3/4Nx-0, and pTanyNþ) and recurrence location.
2.1. Patient sample 2.4. Statistical analysis After obtaining institutional review board approval, we reviewed our cystectomy registry to identify 2,401 patients who underwent RC for pTanyNanyM0 urothelial carcinoma between 1980 and 2007. A total of 604 patients were excluded owing to nonurothelial histology (n = 251), fewer than 90 days of follow-up (n = 77), development of metastatic disease within 30 days of surgery (n = 69), receipt of neoadjuvant/adjuvant radiation therapy (n = 204), and unidentifiable site of recurrence (n = 3). Patients who had received perioperative chemotherapy (n = 228) were included. Given the retrospective nature of our study, postoperative oncologic surveillance regimens were not standardized. However, follow-up after RC generally consisted of a history/physical examination, urine cytology, and imaging of the chest/abdomen/pelvis every 3 months for the first 2 years, every 6 months for the next 2 years, and annually thereafter. Bone scan and brain imaging were performed when clinically indicated. For patients who were followed up outside our institution, registry coordinators captured overall health and bladder cancer–specific
The risks of both post-RC recurrence and non–bladder cancer death were estimated using Weibull models. Weibull models are an alternative parametric approach to Cox proportional hazard regressions [11]. Uniquely, Weibull distributions allow for simultaneous estimation of the hazard ratio and the relative increase or decrease in the risk of the event with time. This type of modeling provides a more reliable and precise method by which to predict how a particular risk of an event may mature over time [11]. For estimation of the risk of post-RC recurrence over time, patients were stratified according to both pathologic stage and relapse location, as described earlier. The risk of non– bladder cancer–related death over time was estimated after stratifying patients according to age: 60 years and younger, 61 to 70 years, 71 to 80 years, and older than 80 years. The Weibull models were then graphed, allowing us to identify the point in time where the risk of non–bladder cancer death exceeded the risk of recurrence. Statistical analysis was done using SAS version 9.2.
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8 Table 1 Clinicopathologic features of patients following radical cystectomy (n ¼ 1,797). Feature
n (%)
Age, y r60 61–70 71–80 4 80
348 650 624 175
(19.4) (36.2) (34.7) (9.7)
Sex Male Female
1,474 (82.0) 323 (18.0)
ECOG performance status 0 1 2 3 4
1,464 253 69 10 1
(81.5) (14.1) (3.8) (0.6) (0.1)
268 662 268 365 234
(14.9) (36.8) (14.9) (20.3) (13.0)
Pathologic category pT0Nx-0 pTa/CIS/1Nx-0 pT2Nx-0 pT3/4Nx-0 pTanyNþ Positive surgical margin
23 (1.5)
Perioperative chemotherapy
228 (12.7)
3. Results Table 1 summarizes patient clinicopathologic features. The median age for the cohort was 68.8 years (interquartile range [IQR]: 62.2–75.2), whereas the median number of lymph nodes dissected at RC was 11 (IQR: 7–18). Median postoperative follow-up was 10.6 years (IQR: 6.8–15.2), during which time 714 patients developed disease recurrence at a median of 1.1 years (IQR: 0.5–2.5) following surgery. Non–bladder cancer deaths occurred in 634 patients at a median of 6.9 years (IQR: 3.0–12.3) after surgery. Of those patients who died of non–bladder cancer causes, 57 (16.4%) were 60 years and younger, 222 (34.2%) were 61 to 70 years, 279 (44.7%) were 71 to 80 years, and 76 (43.4%) were older than 80 years.
339.e3
Of the 714 patients in whom disease recurred, 98 (10.5%) had recurrence in the urethra, 87 (9.3%) in the upper urinary tract, 448 (47.9%) in the abdomen/pelvis, 122 (13.1%) in the thorax, and 179 (19.2%) at other sites. Simultaneous recurrence in 2 or more locations was found in 195 (27.1%) patients. Location-specific recurrence patterns, stratified by pathologic stage from RC, are shown in Table 2. We then evaluated the number of recurrences captured by the NCCN and the EAU surveillance guidelines after RC (Table 3). Here, we found that a surveillance period of 2 years, as outlined by the NCCN, would capture 68.7% of all recurrences. Alternatively, monitoring patients for a period of 5 years, as recommended by the EAU, would allow detection of 89.9% of recurrences. Both the guidelines were most limited in their ability to detect recurrences among patients with lower-stage disease: pT0Nx-0 (NCCN, 60.7%; EAU, 80.4%) and pTa/CIS/1Nx-0 (NCCN, 47.9%; EAU, 79.9%) and among patients who developed a secondary upper urinary tract tumor (NCCN, 36.8%; EAU, 77.0%). The graphical interaction between Weibull models for the risk of recurrence stratified by stage and relapse location and the risk of non–bladder cancer death stratified by age group is shown in the Fig. Age-, stage, and relapse location–specific surveillance durations were then estimated from the Weibull models where the risk of non–bladder cancer death exceeded the risk of recurrence (Table 4). For example, in patients with pT0Nx-0 disease across all age groups, the risk of recurrence in the urethra, upper urinary tract, and thorax was never found to be greater than the risk of non–bladder cancer death following RC. For recurrences to the abdomen/pelvis among patients with pT0Nx-0 disease, the risk of non–bladder cancer–related death began to exceed their risk of recurrence at 2 years for those 60 years and younger, at 1 year for those aged 61 to 70 years, and at 6 months for those aged 71 to 80 years. In contrast, the risk of abdominal/pelvic recurrence for patients 60 years and younger, with either pT3/4Nx-0 or pTanyNþ disease, remained greater than their risk of non–bladder cancer death for more than 20 years. Based on these time points, when risk of non–bladder cancer death exceeds the risk of recurrence, we provide in Table 5 post-RC individualized surveillance duration options according to age, stage, and relapse location.
Table 2 Location-specific recurrence patterns stratified by pathologic stage. Stage group
pT0Nx-0 (n ¼ 51) pTa/CIS/1Nx-0 (n ¼ 219) pT2Nx-0 (n ¼ 106) pT3/4Nx-0 (n ¼ 186) pTanyNþ (n ¼ 152)
Number of recurrences by location (%) Urethra
Upper urinary tract
Abdomen/pelvis
Thorax
Other
7 60 12 12 7
10 54 7 10 6
27 90 72 139 120
8 23 19 40 32
9 34 28 64 44
(11.5) (22.9) (8.7) (4.5) (3.3)
(16.4) (20.7) (5.1) (3.8) (2.9)
(44.3) (34.5) (52.2) (52.4) (57.4)
(13.1) (8.8) (13.8) (15.1) (15.3)
(14.7) (13.0) (20.3) (24.2) (21.1)
144 (80.4) 174 (97.2) 87 (71.3) 113 (92.6) 314 (70.1) 408 (91.1) 32 (36.8) 67 (77.0) 76 (77.5) 91 (92.8) 122 (80.3) 148 (97.4) 158 (84.9) 181 (97.3) 75 (70.7) 97 (91.5) 105 (47.9) 175 (79.9)
pT3/4Nx-0 (n ¼ 186)
491 (68.7) 642 (89.9) NCCN (2 y) EAU (5 y)
pTa/CIS/1Nx-0 pT2Nx-0 (n ¼ 219) (n ¼ 106)
Number of recurrences captured (%) Total (n ¼ 714) Guideline
31 (60.7) 41 (80.4)
Urethral (n ¼ 98) pT0Nx-0 (n ¼ 51)
pTanyNþ (n ¼ 152)
By recurrence location By stage group
Upper tract (n ¼ 87)
Abdomen/Pelvis (n ¼ 448)
Thorax (n ¼ 122)
Other (n ¼ 179)
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
Table 3 Number of recurrences captured by the NCCN- and the EAU-prescribed surveillance periods.
339.e4
4. Discussion To our knowledge, this is the first study to both evaluate the accuracy of the NCCN and the EAU surveillance guidelines in capturing recurrences following cystectomy and provide a novel approach to surveillance by modeling competing risks. Overall, we found that the protocols prescribed by the NCCN and the EAU fail to comprehensively capture recurrences. The lowest number of recurrences captured by the NCCN and the EAU occurred among patients with either pT0Nx-0 or pTa/ CIS/1Nx-0 disease or recurrences to the upper urinary tract. Based on these shortcomings, we used a unique methodology to better predict a patient's bladder cancer course, as it naturally interacts with other competing health factors. Specifically, by identifying when the competing risk of non–bladder cancer death exceeds the risk of tumor recurrence, this approach allowed us to develop individualized surveillance recommendations and eliminate the oversimplified stopping points used by the current guidelines. The oncologic surveillance protocols from the NCCN and the EAU, as well as those from independent investigations [4,5,7,10], stop surveillance between 2 and 5 years. This time frame originates from the appreciation that most recurrences present within 5 years [4,5,7,10,12], and it is derived by calculating the cumulative incidence of recurrence. This oversimplified approach fails to incorporate known risk factors and patterns of recurrence [8,10, 13–15] and lacks the dynamic ability to capture how a patient's chance of recurrence changes over time. As exemplified in prostate cancer, a patient's recurrence-free interval correlates with his or her likelihood of recurrence in the future [16]. In bladder cancer, the concept of how a patient's risk profile changes with time, as it relates to cancer-specific and overall survival, has been demonstrated using an analysis termed conditional survival [17]. Weibull modeling has the capacity to predict how recurrence risk matures over time and interacts with competing health factors, thereby allowing guidelines to be better formulated to an individual's natural course of disease. Personalizing guidelines may provide a better balance between the derived benefit from surveillance and medical resource allocation. Currently, oncologic surveillance testing has been criticized as being costly from a health care standpoint [18–20]. However, at the same time, patients' value follow-up of their cancer. As exemplified in breast cancer surveillance, patients preferred to be seen as frequently as deemed necessary, even if recurrences were not identified [21]. Tailoring surveillance recommendations to the individual may allow a better balance to be achieved between follow-up that is desired and that which is medically necessary. One way to achieve more individualized recommendations is by incorporating a patient's competing risks for recurrence into surveillance decision making. The simplified approach used by the current
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
A)
B)
C)
D)
339.e5
E)
Fig. Weibull models illustrating the time point at which risk of non–bladder cancer death exceeds risk of recurrence. Decreasing hazard rates of risk of recurrence over time stratified by both stage and relapse location; (A) urethra, (B) upper urinary tract, (C) abdomen/pelvis, (D) thorax, and (E) other. Increasing hazard rates of risk of non–bladder cancer death over time stratified by age groups. Age-, stage, and relapse site–specific time points where risk of non–bladder cancer death exceeds risk of recurrence.
339.e6
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
Table 4 Age-, stage, and relapse location–specific time points when risk of non–bladder cancer death exceeds risk of recurrence after radical cystectomy in years. Stage group
Relapse location
Age (y) r60 (n ¼ 348)
61-70 (n ¼ 650)
71-80 (n ¼ 624)
480 (n ¼ 175)
pT0Nx-0 (n ¼ 268)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
– – 2 – 0.5
– – 1 – –
– – 0.5 – –
– – – – –
pTa/CIS/1Nx-0 (n ¼ 662)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
2 1.5 3 – 1
1 0.5 1 – 0.5
0.5 – 0.5 –
– – – –
pT2Nx-0 (n ¼ 268)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
1 – 9 1.5 2
0.5 – 3 1 1.5
– – 2 0.5 1
– – 1 – 0.5
pT3/4Nx-0 (n ¼ 365)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
1 1 420 4 6
0.5 – 8 1.5 2.5
– – 3.5 1 1.5
– – 2 0.5 1
pTanyNþ (n ¼ 234)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
1 – 420 6 8
0.5 – 16 2 3
– – 6 1 1.5
– – 3.5 0.5 1
– ¼ Dash mark indicates that risk of non–bladder cancer death exceeded the risk of recurrence starting at 30 days following radical cystectomy, suggesting surveillance may not be necessary.
guidelines is unable to capture the interaction that occurs between a patient's natural course of disease with bladder cancer and his or her competing health risks. Although the influence of competing risks on decision making is yet to be used in bladder cancer research, this concept has been explored in surgical management decisions for renal cell carcinoma [22–24]. For example, Sun et al. [22] showed that in patients who had a high competing risk of mortality compared with that of renal
cell carcinoma, such as in the elderly and in those with low-risk disease, the benefit from surgical intervention became muted. Using our novel statistical approach, we found a similar result in which both elderly and low-risk patients had a higher competing risk of non–bladder cancer death when compared with their level of recurrence risk. However, in contrast, for patients with pT3/4Nx-0 and pTanyNþ disease, their level of recurrence risk in particular locations remained higher than that for their
Table 5 Duration of age- and stage-specific oncologic surveillance testing in years needed before risk of non–bladder cancer death exceeds risk of recurrence.a,b Stage group
pT0Nx-0 pTa/CIS/1Nx-0 pT2Nx-0 pT3/4Nx-0 pTanyNþ
Abdominal/pelvic imaging
Chest imaging
Urethral washing/cytology
r60
61–70
71–80
480
r60
61–70
71–80
480
r60
61–70
71–80
480
2 3 9 420 420
1 1 3 8 16
0.5 0.5 2 3.5 6
– – 1 2 3.5
– – 1.5 4 6
– – 1 1.5 2
– – 0.5 1 1
– – – 0.5 0.5
– 2 1 1 1
– 1 0.5 0.5 0.5
– 0.5 – – –
– – – – –
– ¼ Dash mark indicates that risk of non–bladder cancer death exceeded the risk of recurrence starting at 30 days following radical cystectomy, suggesting surveillance may not be necessary. a Patient symptoms and clinical suspicion should direct non–routinely scheduled testing and testing beyond the recommended stopping points. b Routine physical examination and pertinent laboratory testing are recommended throughout the duration of surveillance testing.
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
competing risk of non–bladder cancer death for more than 20 years. In these situations, where risk of recurrence remains higher than competing factors, the medical utility of continuing oncologic surveillance becomes more substantiated. Weibull modeling of both recurrence risk and risk of non–bladder cancer death enables patients and clinicians to individualize surveillance regimens that optimize capturing recurrences, minimize unnecessary testing, and direct medical resources to those patients who are remaining at risk. In developing these individualized oncologic surveillance durations following RC, it was not our intention to provide recommendations regarding the frequency or the type of surveillance testing used. As independent series [4,25] have shown a survival benefit to detecting asymptomatic recurrences with standard surveillance testing such as urine cytology and abdominal/chest imaging, we do not recommend deviating from these tests when following the proposed surveillance durations (Table 5). Furthermore, it was also not our intention to refute the benefit of continued follow-up pertaining to urinary reservoir function or nononcologic factors. After understanding the limitations of using the NCCN and the EAU guidelines, we believe it was first important to demonstrate how dynamic modeling of a patient's recurrence risk alongside his or her competing risks of mortality produces vastly different surveillance durations when compared with current practice. Future studies using alternative methods are being planned to provide recommendations for both the frequency and the type of testing when following these individualized surveillance durations. We recognize that by using a retrospective study design, our findings may be subject to inherent biases. For example, our oncologic follow-up was not standardized. However, the protocol used was consistent among all patients, and only 3% of patients were lost to follow-up. Additionally, both imaging modalities and practice patterns varied over the course of the study period. We also acknowledge that the poor survival seen in patients who develop recurrence following RC [26–29] has called into question the value of using oncologic surveillance. However, as mentioned previously, a survival benefit has been demonstrated with the detection of asymptomatic recurrences [4,25] as well as from elements of post-RC surveillance [14]. Although these findings lack prospective validation, they are promising and substantiate surveillance. We realize that the surveillance strategy presented could incorporate other prognostic factors to stratify risk of recurrence or non–bladder cancer death. However, with the addition of more variables, the model would become more complex and cumbersome to navigate. Thus, when proposing a new strategy, we felt it was important to provide a balance in both specificity and simplicity. A balanced algorithm may also be important if health care transitions into a more value-based system in which the surveillance of bladder cancer is being primarily provided
339.e7
by the general practitioner. Nevertheless, despite the degree of specificity, clinical judgment remains critical in defining if follow-up needs to be shortened or extended for the individual. 5. Conclusions The NCCN and the EAU bladder cancer surveillance guidelines do not comprehensively capture disease recurrence following RC. Both the guidelines underperform in patients with no evidence of disease or non–muscleinvasive disease at time of cystectomy and in those with secondary upper urinary tract recurrences. Using a novel statistical approach that dynamically models when a patient's risk of non–bladder cancer death exceeds his or her own risk of recurrence, we provide individualized recommendations for oncologic surveillance following RC. We believe this approach is an improvement over the current guidelines, as it eliminates oversimplified stopping points and improves the balance between the benefit derived from surveillance and medical resource allocation. References [1] Clark PE, NCCN Clinical Practice Guidelins in Oncology: Bladder Cancer, version.1.2013;2014. [2] Stenzl A, Cowan NC, De Santis M, Jakse G, Kuczyk MA, Merseburger AS, et al. The updated EAU guidelines on muscle-invasive and metastatic bladder cancer. Eur Urol 2009;55:815–25. [3] Dalbagni G, Bochner BH, Cronin A, Herr HW, Donat SM. A plea for a uniform surveillance schedule after radical cystectomy. J Urol 2011; 185:2091–6. [4] Giannarini G, Kessler TM, Thoeny HC, Nguyen DP, Meissner C, Studer UE. Do patients benefit from routine follow-up to detect recurrences after radical cystectomy and ileal orthotopic bladder substitution? Eur Urol 2010;58:486–94. [5] Kuroda M, Meguro N, Maeda O, Saiki S, Kinouchi T, Usami M, et al. Stage specific follow-up strategy after cystectomy for carcinoma of the bladder. Int J Urol 2002;9:129–33. [6] Montie JE. Follow-up after cystectomy for carcinoma of the bladder. Urol Clin North Am 1994;21:639–43. [7] Slaton JW, Swanson DA, Grossman HB, Dinney CP. A stage specific approach to tumor surveillance after radical cystectomy for transitional cell carcinoma of the bladder. J Urol 1999;162:710–4. [8] Soukup V, Babjuk M, Bellmunt J, Dalbagni G, Giannarini G, Hakenberg OW, et al. Follow-up after surgical treatment of bladder cancer: a critical analysis of the literature. Eur Urol 2012;62:290–302. [9] Vrooman OP, Witjes JA. Follow-up of patients after curative bladder cancer treatment: guidelines vs. practice. Curr Opin Urol 2010;20: 437–42. [10] Yafi FA, Aprikian AG, Fradet Y, Chin JL, Izawa J, Rendon R, et al. Surveillance guidelines based on recurrence patterns after radical cystectomy for bladder cancer: the Canadian Bladder Cancer Network experience. BJU Int 2012;110:1317–23. [11] Carroll KJ. On the use and utility of the Weibull model in the analysis of survival data. Control Clin Trials 2003;24:682–701. [12] Witjes JA, European Association of Urology: Guidelines on muscleinvasive and metastatic bladder cancer; 2013;77–9. [13] Huguet J. Follow-up after radical cystectomy based on patterns of tumour recurrence and its risk factors. Acta Urologica Espanol 2013;37:376–82.
339.e8
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
[14] Strope SA, Chang SH, Chen L, Sandhu G, Piccirillo JF, Schootman M. Survival impact of followup care after radical cystectomy for bladder cancer. J Urol 2013;190:1698–703. [15] Umbreit EC, Crispen PL, Shimko MS, Farmer SA, Blute ML, Frank I. Multifactorial site-specific recurrence model after radical cystectomy for urothelial carcinoma. Cancer 2010;116:3399–407. [16] Tollefson MK, Blute ML, Rangel LJ, Karnes RJ, Frank I. Lifelong yearly prostate specific antigen surveillance is not necessary for low risk prostate cancer treated with radical prostatectomy. J Urol 2010;184:925–9. [17] Ploussard G, Shariat SF, Dragomir A, Kluth LA, Xylinas E, MassonLecomte A, et al. Conditional survival after radical cystectomy for bladder cancer: evidence for a patient changing risk profile over time. Eur Urol 2013;66:361–70. [18] Kaplan AL, Litwin MS, Chamie K. The future of bladder cancer care in the USA. Nat Rev Urol 2013;66:59–62. [19] Riley GF, Potosky AL, Lubitz JD, Kessler LG. Medicare payments from diagnosis to death for elderly cancer patients by stage at diagnosis. Med Care 1995;33:828–41. [20] Svatek RS, Hollenbeck BK, Holmang S, Lee R, Kim SP, Stenzl A, et al. The economics of bladder cancer: costs and considerations of caring for this disease. Eur Urol 2014;66:253–62. [21] The GIVIO Investigators. Impact of follow-up testing on survival and health-related quality of life in breast cancer patients. A multicenter randomized controlled trial. J Am Med Assoc 1994;271:1587–92. [22] Sun M, Becker A, Tian Z, Roghmann F, Abdollah F, Larouche A, et al. Management of localized kidney cancer: calculating cancerspecific mortality and competing risks of death for surgery and nonsurgical management. Eur Urol 2014;65:235–41.
[23] Chen DY, Uzzo RG, Viterbo R. Thinking beyond surgery in the management of renal cell carcinoma: the risk to die from renal cell carcinoma and competing risks of death. World J Urol. 2014;32:607–13. [24] Kutikov A, Egleston BL, Wong YN, Uzzo RG. Evaluating overall survival and competing risks of death in patients with localized renal cell carcinoma using a comprehensive nomogram. J Clin Oncol 2010;28:311–7. [25] Boorjian SA, Tollefson MK, Cheville JC, Costello BA, Thapa P, Frank I. Detection of asymptomatic recurrence during routine oncological followup after radical cystectomy is associated with improved patient survival. J Urol 2011;186:1796–802. [26] Hassan JM, Cookson MS, Smith JA Jr., Chang SS. Patterns of initial transitional cell recurrence in patients after cystectomy. J Urol 2006;175:2054–7. [27] International Bladder Cancer Nomogram Consortium, Bochner BH, Kattan MW, Vora KC. Postoperative nomogram predicting risk of recurrence after radical cystectomy for bladder cancer. J Clin Oncol 2006;24:3967–72. [28] Shariat SF, Karakiewicz PI, Palapattu GS, Lotan Y, Rogers CG, Amiel GE, et al. Outcomes of radical cystectomy for transitional cell carcinoma of the bladder: a contemporary series from the Bladder Cancer Research Consortium. J Urol 2006;176:2414–22: [discussion 22]. [29] Stein JP, Lieskovsky G, Cote R, Groshen S, Feng AC, Boyd S, et al. Radical cystectomy in the treatment of invasive bladder cancer: long-term results in 1,054 patients. J Clin Oncol 2001;19: 666–75.
Original article
Evaluation of current surveillance guidelines following radical cystectomy and proposal of a novel risk-based approach Suzanne B. Stewart-Merrill, M.D.a, Stephen A. Boorjian, M.D.a, Robert Houston Thompson, M.D.a, Sarah P. Psutka, M.D.a, John C. Cheville, M.D.b, Prabin Thapac, Eric J. Bergstrahl, M.S.c, Matthew K. Tollefson, M.D.a, Igor Frank, M.D.a,* a
Department of Urology, Mayo Clinic, Rochester, MN Department of Pathology, Mayo Clinic, Rochester, MN c Department of Health Sciences Research, Mayo Clinic, Rochester, MN b
Received 1 February 2015; received in revised form 8 April 2015; accepted 29 April 2015
Abstract Background: Evidence supporting surveillance guidelines after radical cystectomy (RC) are lacking. Herein, we evaluate the ability of the National Comprehensive Cancer Network (NCCN) guidelines and the European Association of Urology (EAU) guidelines to capture recurrences and provide an alternative approach that balances risks of recurrence with non–bladder cancer death. Methods: We identified 1,797 patients who had M0 urothelial carcinoma who underwent RC at our institution between 1980 and 2007. The success of current guidelines to capture recurrences was assessed by calculating the percentage of recurrences detected during the recommended follow-up time: the NCCN—2 years and the EAU—5 years. An alternative protocol was created using Weibull distributions, which estimate when a patient's risk of non–bladder cancer death exceeds their risk of recurrence. Results: At a median follow-up of 10.6 years (interquartile range : 6.8–15.2), a total of 714 patients recurred. Of these, 491 (68.7%) would have been detected by the NCCN guidelines and 642 (89.8%) by the EAU guidelines. Using a risk-adapted approach, vastly different surveillance durations were appreciated. For example, for patients older than 80 years with pT0Nx-0 or pTa/CIS/1Nx-0 disease, recurrence risk to any location never exceeded their risk of non–bladder cancer death, whereas for patients aged 60 years and younger with pT3/4Nx-0 or pTanyNþ disease, risk of abdominal/pelvis recurrence remained greater than their risk of non–bladder cancer death for 420 years. Conclusions: The duration of post-RC follow-up recommended by the NCCN and the EAU does not comprehensively capture recurrences. A surveillance algorithm based on the interaction between recurrence risk and competing health factors individualizes recommendations and may improve capture of recurrences and resource allocation. r 2015 Elsevier Inc. All rights reserved.
Keywords: Bladder cancer; Surveillance; Cystectomy; Guidelines; Recurrence
1. Introduction Currently, the National Comprehensive Cancer Network (NCCN) and the European Association of Urology (EAU) This publication was made possible by CTSA Grant: UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH), USA. The contents of the article are solely the responsibility of the authors and do not necessarily represent the official view of the NIH. * Corresponding author. Tel.:þ1-507-266-0191; fax:þ1-507-284-4951. E-mail address: [email protected] (I. Frank). http://dx.doi.org/10.1016/j.urolonc.2015.04.017 1078-1439/r 2015 Elsevier Inc. All rights reserved.
provide the most recognized oncologic surveillance guidelines following radical cystectomy (RC). However, these recommendations are not well supported by existing data and are disparate regarding total duration of follow-up. Specifically, the NCCN recommends that patients be followed up according to the same protocol, regardless of tumor stage, for 2 years [1]. Extension of surveillance beyond this time is then left up to clinical discretion. The EAU uses a stage-based approach to surveillance, recommending that patients with advanced-category disease be followed up more frequently. However, despite this
339.e2
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
risk-stratified approach, total follow-up for all stage groups is for 5 years [2]. The effect of stopping surveillance at 2 or 5 years has not been assessed, thereby making it unclear if this time frame is sufficient to comprehensively capture recurrences. Additionally, leaving the option of extending follow-up to the provider potentiates the existent heterogeneity in surveillance care [3]. The stopping points for surveillance are based on the appreciation that most of the recurrences following RC occur within 2 to 5 years. Such determinations are based on calculations of cumulative incidence of recurrences over time [4–10]. However, this approach does not account for how a patient's disease recurrence risk following RC changes with time and how competing comorbid conditions may influence this risk. Incorporating these features into guideline strategies may allow for development of more meaningful surveillance stopping points. Additionally, as health care transitions to a value-based system, and quality assessment and reimbursement become connected with adherence to standardized guidelines, improvement on surveillance strategies becomes imperative. Based on the earlier description, our objective was to first evaluate how many patients with recurrence following RC would have been detected if surveillance were strictly performed for the total time recommended by the NCCN or the EAU. Second, we present a novel approach to surveillance that dynamically models the relationship between a patient's risk of non–bladder cancer death and the risk of recurrence and propose individualized follow-up recommendations following RC.
2. Methods
outcomes annually through patient or treating physician correspondence. 2.2. Classification of disease recurrence and non–bladder cancer–related death Disease recurrence was defined as metastasis on imaging or biopsy at least 30 days after RC. Recurrences were classified according to location as urethral, upper urinary tract, abdomen/pelvis, thorax, and other (i.e., bone, central nervous system, and skin). For each patient, the first recurrence in time was counted as an event, and all subsequent recurrences were censored to prevent double counting. Recurrences occurring simultaneously in multiple locations were counted as individual events; however, it cannot be inferred if such events were independent. Non– bladder cancer death was defined as mortality at least 30 days after RC attributed to any cause except that related to the primary bladder cancer diagnosis, RC surgery, or postRC recurrence. 2.3. Evaluation of current guidelines We evaluated the NCCN and the EAU surveillance guidelines by determining the total number of recurrences that would have been captured if patients after RC were strictly followed up for the time prescribed: the EAU—5 years and the NCCN—2 years. The option of extending surveillance based on clinical discretion, as stated in the NCCN guidelines, was not examined owing to subjectivity. The total number of recurrences captured by these guidelines was also assessed according to pathologic stage (pT0Nx-0, pTa/CIS/1Nx-0, pT2Nx-0, pT3/4Nx-0, and pTanyNþ) and recurrence location.
2.1. Patient sample 2.4. Statistical analysis After obtaining institutional review board approval, we reviewed our cystectomy registry to identify 2,401 patients who underwent RC for pTanyNanyM0 urothelial carcinoma between 1980 and 2007. A total of 604 patients were excluded owing to nonurothelial histology (n = 251), fewer than 90 days of follow-up (n = 77), development of metastatic disease within 30 days of surgery (n = 69), receipt of neoadjuvant/adjuvant radiation therapy (n = 204), and unidentifiable site of recurrence (n = 3). Patients who had received perioperative chemotherapy (n = 228) were included. Given the retrospective nature of our study, postoperative oncologic surveillance regimens were not standardized. However, follow-up after RC generally consisted of a history/physical examination, urine cytology, and imaging of the chest/abdomen/pelvis every 3 months for the first 2 years, every 6 months for the next 2 years, and annually thereafter. Bone scan and brain imaging were performed when clinically indicated. For patients who were followed up outside our institution, registry coordinators captured overall health and bladder cancer–specific
The risks of both post-RC recurrence and non–bladder cancer death were estimated using Weibull models. Weibull models are an alternative parametric approach to Cox proportional hazard regressions [11]. Uniquely, Weibull distributions allow for simultaneous estimation of the hazard ratio and the relative increase or decrease in the risk of the event with time. This type of modeling provides a more reliable and precise method by which to predict how a particular risk of an event may mature over time [11]. For estimation of the risk of post-RC recurrence over time, patients were stratified according to both pathologic stage and relapse location, as described earlier. The risk of non– bladder cancer–related death over time was estimated after stratifying patients according to age: 60 years and younger, 61 to 70 years, 71 to 80 years, and older than 80 years. The Weibull models were then graphed, allowing us to identify the point in time where the risk of non–bladder cancer death exceeded the risk of recurrence. Statistical analysis was done using SAS version 9.2.
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8 Table 1 Clinicopathologic features of patients following radical cystectomy (n ¼ 1,797). Feature
n (%)
Age, y r60 61–70 71–80 4 80
348 650 624 175
(19.4) (36.2) (34.7) (9.7)
Sex Male Female
1,474 (82.0) 323 (18.0)
ECOG performance status 0 1 2 3 4
1,464 253 69 10 1
(81.5) (14.1) (3.8) (0.6) (0.1)
268 662 268 365 234
(14.9) (36.8) (14.9) (20.3) (13.0)
Pathologic category pT0Nx-0 pTa/CIS/1Nx-0 pT2Nx-0 pT3/4Nx-0 pTanyNþ Positive surgical margin
23 (1.5)
Perioperative chemotherapy
228 (12.7)
3. Results Table 1 summarizes patient clinicopathologic features. The median age for the cohort was 68.8 years (interquartile range [IQR]: 62.2–75.2), whereas the median number of lymph nodes dissected at RC was 11 (IQR: 7–18). Median postoperative follow-up was 10.6 years (IQR: 6.8–15.2), during which time 714 patients developed disease recurrence at a median of 1.1 years (IQR: 0.5–2.5) following surgery. Non–bladder cancer deaths occurred in 634 patients at a median of 6.9 years (IQR: 3.0–12.3) after surgery. Of those patients who died of non–bladder cancer causes, 57 (16.4%) were 60 years and younger, 222 (34.2%) were 61 to 70 years, 279 (44.7%) were 71 to 80 years, and 76 (43.4%) were older than 80 years.
339.e3
Of the 714 patients in whom disease recurred, 98 (10.5%) had recurrence in the urethra, 87 (9.3%) in the upper urinary tract, 448 (47.9%) in the abdomen/pelvis, 122 (13.1%) in the thorax, and 179 (19.2%) at other sites. Simultaneous recurrence in 2 or more locations was found in 195 (27.1%) patients. Location-specific recurrence patterns, stratified by pathologic stage from RC, are shown in Table 2. We then evaluated the number of recurrences captured by the NCCN and the EAU surveillance guidelines after RC (Table 3). Here, we found that a surveillance period of 2 years, as outlined by the NCCN, would capture 68.7% of all recurrences. Alternatively, monitoring patients for a period of 5 years, as recommended by the EAU, would allow detection of 89.9% of recurrences. Both the guidelines were most limited in their ability to detect recurrences among patients with lower-stage disease: pT0Nx-0 (NCCN, 60.7%; EAU, 80.4%) and pTa/CIS/1Nx-0 (NCCN, 47.9%; EAU, 79.9%) and among patients who developed a secondary upper urinary tract tumor (NCCN, 36.8%; EAU, 77.0%). The graphical interaction between Weibull models for the risk of recurrence stratified by stage and relapse location and the risk of non–bladder cancer death stratified by age group is shown in the Fig. Age-, stage, and relapse location–specific surveillance durations were then estimated from the Weibull models where the risk of non–bladder cancer death exceeded the risk of recurrence (Table 4). For example, in patients with pT0Nx-0 disease across all age groups, the risk of recurrence in the urethra, upper urinary tract, and thorax was never found to be greater than the risk of non–bladder cancer death following RC. For recurrences to the abdomen/pelvis among patients with pT0Nx-0 disease, the risk of non–bladder cancer–related death began to exceed their risk of recurrence at 2 years for those 60 years and younger, at 1 year for those aged 61 to 70 years, and at 6 months for those aged 71 to 80 years. In contrast, the risk of abdominal/pelvic recurrence for patients 60 years and younger, with either pT3/4Nx-0 or pTanyNþ disease, remained greater than their risk of non–bladder cancer death for more than 20 years. Based on these time points, when risk of non–bladder cancer death exceeds the risk of recurrence, we provide in Table 5 post-RC individualized surveillance duration options according to age, stage, and relapse location.
Table 2 Location-specific recurrence patterns stratified by pathologic stage. Stage group
pT0Nx-0 (n ¼ 51) pTa/CIS/1Nx-0 (n ¼ 219) pT2Nx-0 (n ¼ 106) pT3/4Nx-0 (n ¼ 186) pTanyNþ (n ¼ 152)
Number of recurrences by location (%) Urethra
Upper urinary tract
Abdomen/pelvis
Thorax
Other
7 60 12 12 7
10 54 7 10 6
27 90 72 139 120
8 23 19 40 32
9 34 28 64 44
(11.5) (22.9) (8.7) (4.5) (3.3)
(16.4) (20.7) (5.1) (3.8) (2.9)
(44.3) (34.5) (52.2) (52.4) (57.4)
(13.1) (8.8) (13.8) (15.1) (15.3)
(14.7) (13.0) (20.3) (24.2) (21.1)
144 (80.4) 174 (97.2) 87 (71.3) 113 (92.6) 314 (70.1) 408 (91.1) 32 (36.8) 67 (77.0) 76 (77.5) 91 (92.8) 122 (80.3) 148 (97.4) 158 (84.9) 181 (97.3) 75 (70.7) 97 (91.5) 105 (47.9) 175 (79.9)
pT3/4Nx-0 (n ¼ 186)
491 (68.7) 642 (89.9) NCCN (2 y) EAU (5 y)
pTa/CIS/1Nx-0 pT2Nx-0 (n ¼ 219) (n ¼ 106)
Number of recurrences captured (%) Total (n ¼ 714) Guideline
31 (60.7) 41 (80.4)
Urethral (n ¼ 98) pT0Nx-0 (n ¼ 51)
pTanyNþ (n ¼ 152)
By recurrence location By stage group
Upper tract (n ¼ 87)
Abdomen/Pelvis (n ¼ 448)
Thorax (n ¼ 122)
Other (n ¼ 179)
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
Table 3 Number of recurrences captured by the NCCN- and the EAU-prescribed surveillance periods.
339.e4
4. Discussion To our knowledge, this is the first study to both evaluate the accuracy of the NCCN and the EAU surveillance guidelines in capturing recurrences following cystectomy and provide a novel approach to surveillance by modeling competing risks. Overall, we found that the protocols prescribed by the NCCN and the EAU fail to comprehensively capture recurrences. The lowest number of recurrences captured by the NCCN and the EAU occurred among patients with either pT0Nx-0 or pTa/ CIS/1Nx-0 disease or recurrences to the upper urinary tract. Based on these shortcomings, we used a unique methodology to better predict a patient's bladder cancer course, as it naturally interacts with other competing health factors. Specifically, by identifying when the competing risk of non–bladder cancer death exceeds the risk of tumor recurrence, this approach allowed us to develop individualized surveillance recommendations and eliminate the oversimplified stopping points used by the current guidelines. The oncologic surveillance protocols from the NCCN and the EAU, as well as those from independent investigations [4,5,7,10], stop surveillance between 2 and 5 years. This time frame originates from the appreciation that most recurrences present within 5 years [4,5,7,10,12], and it is derived by calculating the cumulative incidence of recurrence. This oversimplified approach fails to incorporate known risk factors and patterns of recurrence [8,10, 13–15] and lacks the dynamic ability to capture how a patient's chance of recurrence changes over time. As exemplified in prostate cancer, a patient's recurrence-free interval correlates with his or her likelihood of recurrence in the future [16]. In bladder cancer, the concept of how a patient's risk profile changes with time, as it relates to cancer-specific and overall survival, has been demonstrated using an analysis termed conditional survival [17]. Weibull modeling has the capacity to predict how recurrence risk matures over time and interacts with competing health factors, thereby allowing guidelines to be better formulated to an individual's natural course of disease. Personalizing guidelines may provide a better balance between the derived benefit from surveillance and medical resource allocation. Currently, oncologic surveillance testing has been criticized as being costly from a health care standpoint [18–20]. However, at the same time, patients' value follow-up of their cancer. As exemplified in breast cancer surveillance, patients preferred to be seen as frequently as deemed necessary, even if recurrences were not identified [21]. Tailoring surveillance recommendations to the individual may allow a better balance to be achieved between follow-up that is desired and that which is medically necessary. One way to achieve more individualized recommendations is by incorporating a patient's competing risks for recurrence into surveillance decision making. The simplified approach used by the current
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
A)
B)
C)
D)
339.e5
E)
Fig. Weibull models illustrating the time point at which risk of non–bladder cancer death exceeds risk of recurrence. Decreasing hazard rates of risk of recurrence over time stratified by both stage and relapse location; (A) urethra, (B) upper urinary tract, (C) abdomen/pelvis, (D) thorax, and (E) other. Increasing hazard rates of risk of non–bladder cancer death over time stratified by age groups. Age-, stage, and relapse site–specific time points where risk of non–bladder cancer death exceeds risk of recurrence.
339.e6
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
Table 4 Age-, stage, and relapse location–specific time points when risk of non–bladder cancer death exceeds risk of recurrence after radical cystectomy in years. Stage group
Relapse location
Age (y) r60 (n ¼ 348)
61-70 (n ¼ 650)
71-80 (n ¼ 624)
480 (n ¼ 175)
pT0Nx-0 (n ¼ 268)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
– – 2 – 0.5
– – 1 – –
– – 0.5 – –
– – – – –
pTa/CIS/1Nx-0 (n ¼ 662)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
2 1.5 3 – 1
1 0.5 1 – 0.5
0.5 – 0.5 –
– – – –
pT2Nx-0 (n ¼ 268)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
1 – 9 1.5 2
0.5 – 3 1 1.5
– – 2 0.5 1
– – 1 – 0.5
pT3/4Nx-0 (n ¼ 365)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
1 1 420 4 6
0.5 – 8 1.5 2.5
– – 3.5 1 1.5
– – 2 0.5 1
pTanyNþ (n ¼ 234)
Urethra Upper urinary tract Abdomen/pelvis Thorax Other
1 – 420 6 8
0.5 – 16 2 3
– – 6 1 1.5
– – 3.5 0.5 1
– ¼ Dash mark indicates that risk of non–bladder cancer death exceeded the risk of recurrence starting at 30 days following radical cystectomy, suggesting surveillance may not be necessary.
guidelines is unable to capture the interaction that occurs between a patient's natural course of disease with bladder cancer and his or her competing health risks. Although the influence of competing risks on decision making is yet to be used in bladder cancer research, this concept has been explored in surgical management decisions for renal cell carcinoma [22–24]. For example, Sun et al. [22] showed that in patients who had a high competing risk of mortality compared with that of renal
cell carcinoma, such as in the elderly and in those with low-risk disease, the benefit from surgical intervention became muted. Using our novel statistical approach, we found a similar result in which both elderly and low-risk patients had a higher competing risk of non–bladder cancer death when compared with their level of recurrence risk. However, in contrast, for patients with pT3/4Nx-0 and pTanyNþ disease, their level of recurrence risk in particular locations remained higher than that for their
Table 5 Duration of age- and stage-specific oncologic surveillance testing in years needed before risk of non–bladder cancer death exceeds risk of recurrence.a,b Stage group
pT0Nx-0 pTa/CIS/1Nx-0 pT2Nx-0 pT3/4Nx-0 pTanyNþ
Abdominal/pelvic imaging
Chest imaging
Urethral washing/cytology
r60
61–70
71–80
480
r60
61–70
71–80
480
r60
61–70
71–80
480
2 3 9 420 420
1 1 3 8 16
0.5 0.5 2 3.5 6
– – 1 2 3.5
– – 1.5 4 6
– – 1 1.5 2
– – 0.5 1 1
– – – 0.5 0.5
– 2 1 1 1
– 1 0.5 0.5 0.5
– 0.5 – – –
– – – – –
– ¼ Dash mark indicates that risk of non–bladder cancer death exceeded the risk of recurrence starting at 30 days following radical cystectomy, suggesting surveillance may not be necessary. a Patient symptoms and clinical suspicion should direct non–routinely scheduled testing and testing beyond the recommended stopping points. b Routine physical examination and pertinent laboratory testing are recommended throughout the duration of surveillance testing.
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
competing risk of non–bladder cancer death for more than 20 years. In these situations, where risk of recurrence remains higher than competing factors, the medical utility of continuing oncologic surveillance becomes more substantiated. Weibull modeling of both recurrence risk and risk of non–bladder cancer death enables patients and clinicians to individualize surveillance regimens that optimize capturing recurrences, minimize unnecessary testing, and direct medical resources to those patients who are remaining at risk. In developing these individualized oncologic surveillance durations following RC, it was not our intention to provide recommendations regarding the frequency or the type of surveillance testing used. As independent series [4,25] have shown a survival benefit to detecting asymptomatic recurrences with standard surveillance testing such as urine cytology and abdominal/chest imaging, we do not recommend deviating from these tests when following the proposed surveillance durations (Table 5). Furthermore, it was also not our intention to refute the benefit of continued follow-up pertaining to urinary reservoir function or nononcologic factors. After understanding the limitations of using the NCCN and the EAU guidelines, we believe it was first important to demonstrate how dynamic modeling of a patient's recurrence risk alongside his or her competing risks of mortality produces vastly different surveillance durations when compared with current practice. Future studies using alternative methods are being planned to provide recommendations for both the frequency and the type of testing when following these individualized surveillance durations. We recognize that by using a retrospective study design, our findings may be subject to inherent biases. For example, our oncologic follow-up was not standardized. However, the protocol used was consistent among all patients, and only 3% of patients were lost to follow-up. Additionally, both imaging modalities and practice patterns varied over the course of the study period. We also acknowledge that the poor survival seen in patients who develop recurrence following RC [26–29] has called into question the value of using oncologic surveillance. However, as mentioned previously, a survival benefit has been demonstrated with the detection of asymptomatic recurrences [4,25] as well as from elements of post-RC surveillance [14]. Although these findings lack prospective validation, they are promising and substantiate surveillance. We realize that the surveillance strategy presented could incorporate other prognostic factors to stratify risk of recurrence or non–bladder cancer death. However, with the addition of more variables, the model would become more complex and cumbersome to navigate. Thus, when proposing a new strategy, we felt it was important to provide a balance in both specificity and simplicity. A balanced algorithm may also be important if health care transitions into a more value-based system in which the surveillance of bladder cancer is being primarily provided
339.e7
by the general practitioner. Nevertheless, despite the degree of specificity, clinical judgment remains critical in defining if follow-up needs to be shortened or extended for the individual. 5. Conclusions The NCCN and the EAU bladder cancer surveillance guidelines do not comprehensively capture disease recurrence following RC. Both the guidelines underperform in patients with no evidence of disease or non–muscleinvasive disease at time of cystectomy and in those with secondary upper urinary tract recurrences. Using a novel statistical approach that dynamically models when a patient's risk of non–bladder cancer death exceeds his or her own risk of recurrence, we provide individualized recommendations for oncologic surveillance following RC. We believe this approach is an improvement over the current guidelines, as it eliminates oversimplified stopping points and improves the balance between the benefit derived from surveillance and medical resource allocation. References [1] Clark PE, NCCN Clinical Practice Guidelins in Oncology: Bladder Cancer, version.1.2013;2014. [2] Stenzl A, Cowan NC, De Santis M, Jakse G, Kuczyk MA, Merseburger AS, et al. The updated EAU guidelines on muscle-invasive and metastatic bladder cancer. Eur Urol 2009;55:815–25. [3] Dalbagni G, Bochner BH, Cronin A, Herr HW, Donat SM. A plea for a uniform surveillance schedule after radical cystectomy. J Urol 2011; 185:2091–6. [4] Giannarini G, Kessler TM, Thoeny HC, Nguyen DP, Meissner C, Studer UE. Do patients benefit from routine follow-up to detect recurrences after radical cystectomy and ileal orthotopic bladder substitution? Eur Urol 2010;58:486–94. [5] Kuroda M, Meguro N, Maeda O, Saiki S, Kinouchi T, Usami M, et al. Stage specific follow-up strategy after cystectomy for carcinoma of the bladder. Int J Urol 2002;9:129–33. [6] Montie JE. Follow-up after cystectomy for carcinoma of the bladder. Urol Clin North Am 1994;21:639–43. [7] Slaton JW, Swanson DA, Grossman HB, Dinney CP. A stage specific approach to tumor surveillance after radical cystectomy for transitional cell carcinoma of the bladder. J Urol 1999;162:710–4. [8] Soukup V, Babjuk M, Bellmunt J, Dalbagni G, Giannarini G, Hakenberg OW, et al. Follow-up after surgical treatment of bladder cancer: a critical analysis of the literature. Eur Urol 2012;62:290–302. [9] Vrooman OP, Witjes JA. Follow-up of patients after curative bladder cancer treatment: guidelines vs. practice. Curr Opin Urol 2010;20: 437–42. [10] Yafi FA, Aprikian AG, Fradet Y, Chin JL, Izawa J, Rendon R, et al. Surveillance guidelines based on recurrence patterns after radical cystectomy for bladder cancer: the Canadian Bladder Cancer Network experience. BJU Int 2012;110:1317–23. [11] Carroll KJ. On the use and utility of the Weibull model in the analysis of survival data. Control Clin Trials 2003;24:682–701. [12] Witjes JA, European Association of Urology: Guidelines on muscleinvasive and metastatic bladder cancer; 2013;77–9. [13] Huguet J. Follow-up after radical cystectomy based on patterns of tumour recurrence and its risk factors. Acta Urologica Espanol 2013;37:376–82.
339.e8
S.B. Stewart-Merrill et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 339.e1–339.e8
[14] Strope SA, Chang SH, Chen L, Sandhu G, Piccirillo JF, Schootman M. Survival impact of followup care after radical cystectomy for bladder cancer. J Urol 2013;190:1698–703. [15] Umbreit EC, Crispen PL, Shimko MS, Farmer SA, Blute ML, Frank I. Multifactorial site-specific recurrence model after radical cystectomy for urothelial carcinoma. Cancer 2010;116:3399–407. [16] Tollefson MK, Blute ML, Rangel LJ, Karnes RJ, Frank I. Lifelong yearly prostate specific antigen surveillance is not necessary for low risk prostate cancer treated with radical prostatectomy. J Urol 2010;184:925–9. [17] Ploussard G, Shariat SF, Dragomir A, Kluth LA, Xylinas E, MassonLecomte A, et al. Conditional survival after radical cystectomy for bladder cancer: evidence for a patient changing risk profile over time. Eur Urol 2013;66:361–70. [18] Kaplan AL, Litwin MS, Chamie K. The future of bladder cancer care in the USA. Nat Rev Urol 2013;66:59–62. [19] Riley GF, Potosky AL, Lubitz JD, Kessler LG. Medicare payments from diagnosis to death for elderly cancer patients by stage at diagnosis. Med Care 1995;33:828–41. [20] Svatek RS, Hollenbeck BK, Holmang S, Lee R, Kim SP, Stenzl A, et al. The economics of bladder cancer: costs and considerations of caring for this disease. Eur Urol 2014;66:253–62. [21] The GIVIO Investigators. Impact of follow-up testing on survival and health-related quality of life in breast cancer patients. A multicenter randomized controlled trial. J Am Med Assoc 1994;271:1587–92. [22] Sun M, Becker A, Tian Z, Roghmann F, Abdollah F, Larouche A, et al. Management of localized kidney cancer: calculating cancerspecific mortality and competing risks of death for surgery and nonsurgical management. Eur Urol 2014;65:235–41.
[23] Chen DY, Uzzo RG, Viterbo R. Thinking beyond surgery in the management of renal cell carcinoma: the risk to die from renal cell carcinoma and competing risks of death. World J Urol. 2014;32:607–13. [24] Kutikov A, Egleston BL, Wong YN, Uzzo RG. Evaluating overall survival and competing risks of death in patients with localized renal cell carcinoma using a comprehensive nomogram. J Clin Oncol 2010;28:311–7. [25] Boorjian SA, Tollefson MK, Cheville JC, Costello BA, Thapa P, Frank I. Detection of asymptomatic recurrence during routine oncological followup after radical cystectomy is associated with improved patient survival. J Urol 2011;186:1796–802. [26] Hassan JM, Cookson MS, Smith JA Jr., Chang SS. Patterns of initial transitional cell recurrence in patients after cystectomy. J Urol 2006;175:2054–7. [27] International Bladder Cancer Nomogram Consortium, Bochner BH, Kattan MW, Vora KC. Postoperative nomogram predicting risk of recurrence after radical cystectomy for bladder cancer. J Clin Oncol 2006;24:3967–72. [28] Shariat SF, Karakiewicz PI, Palapattu GS, Lotan Y, Rogers CG, Amiel GE, et al. Outcomes of radical cystectomy for transitional cell carcinoma of the bladder: a contemporary series from the Bladder Cancer Research Consortium. J Urol 2006;176:2414–22: [discussion 22]. [29] Stein JP, Lieskovsky G, Cote R, Groshen S, Feng AC, Boyd S, et al. Radical cystectomy in the treatment of invasive bladder cancer: long-term results in 1,054 patients. J Clin Oncol 2001;19: 666–75.