Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
Original article
Bladder cancer risk: Use of the PLCO and NLST to identify a suitable screening cohort Laura-Maria Krabbe, M.D.a,b, Robert S. Svatek, M.D.c, Shahrokh F. Shariat, M.D.a,d, Edward Messing, M.D.e, Yair Lotan, M.D.a,* a
Department of Urology, University of Texas Southwestern Medical Center, Dallas, TX Department of Urology, University of Muenster Medical Center, Muenster, Germany c Department of Urology, University of Texas Health Science Center, San Antonio TX d Department of Urology, Medical University of Vienna, Vienna, Austria e Department of Urology, University of Rochester Medical Center, Rochester, NY
b
Received 28 April 2014; received in revised form 14 June 2014; accepted 16 June 2014
Abstract Purpose: Bladder cancer (BC) screening is not accepted in part owing to low overall incidence. We used the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) and National Lung Cancer Screening Trial (NLST) to identify optimal high-risk populations most likely to benefit from screening. Materials and methods: Data were extracted from PLCO and NLST to stratify risk of BC by overall population, sex, race, age at inclusion, and smoking status. Incidence rates between groups were compared using chi-square test. Results: BC was identified in 1,430/154,898 patients in PLCO and 439/53,173 patients in NLST. BCs were grade III/IV in 36.8% and 41.3%. Incidence rates were significantly higher in men than in women (PLCO: 1.4 vs. 0.31/1,000 person-years and NLST: 1.84 vs. 0.6/1,000 person-years, both P o 0.0001). In proportional hazards models, male sex, higher age, and duration and intensity of smoking were associated with higher risk of BC (all P o 0.0001). In men older than 70 years with smoking exposure of 30 pack-years (PY) and more, incidence rates were as high as 11.92 (PLCO) and 5.23 (NLST) (per 1,000 person-years). In current high-intensity smokers (Z50 PY), the sex disparity in incidence persists in both trials (0.78 vs. 2.99 per 1,000 person-years in PLCO and 1.12 vs. 2.65 per 1,000 person-years in NLST). Conclusions: Men older than 60 years with a smoking history of 430 PY had incidence rates of more than 2/1,000 person-years, which could serve as an excellent population for screening trials. Sex differences in the incidence of BC cannot be readily explained by the differences in exposure to tobacco, as sex disparity persisted regardless of smoking intensity. r 2014 Elsevier Inc. All rights reserved.
Keywords: Bladder cancer; Screening; Risk factors; Sex; Risk stratification
1. Introduction Bladder cancer (BC) is the sixth most common cancer overall in North America with an estimated 72,570 new cases and 15,210 deaths in 2013 in the United States [1]. Men are about 3 times more likely to develop BC in their lifetime than women are, which has been attributed to different historical smoking patterns or occupational exposures [1–3]. Earlier reports suggested that the population attributable risk (PAR) Corresponding author. Tel.: þ1-214-648-0389; fax: þ1-214-648-8786. E-mail address: [email protected] (Y. Lotan). *
http://dx.doi.org/10.1016/j.urolonc.2014.06.009 1078-1439/r 2014 Elsevier Inc. All rights reserved.
owing to smoking was 50% to 65% in men and 20% to 30% in women, although more recent studies showed more balanced PARs of approximately 50% for both sexes owing to increased rates of smoking in women [4]. PARs explain what proportion of BC is attributable to a risk factor such as smoking; however, limited data have been available to compare the incidence rates of BC in men and women with similarly high exposures to tobacco and to evaluate if the sex disparity remains. Approximately 25% of all patients newly diagnosed with BC have muscle-invasive or metastatic disease, which leads to high morbidity and mortality as well as high cost for treatment [1]. The goal of screening is to achieve a shift to
65.e20
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
lower stages at diagnosis thereby improving morbidity and survival, especially because as many as 80% of muscleinvasive cancers are already muscle invasive at first diagnosis. Screening for BC has been studied by several investigators; however, partly owing to the low overall incidence of BC (37.5 and 9.3 per 100,000 in men and women), screening is currently not recommended in routine practice [1,5–10]. There are multiple factors contributing to low BC detection rates in prior screening studies, including healthy volunteer bias, but the main reason was the inclusion of patients at too low risk for BC. Identifying a cohort with potentially higher disease incidence is critical for instituting an effective screening program. Both the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) and the National Lung Cancer Screening Trial (NLST) performed longitudinal prospective evaluations of large populations and tracked secondary malignancies. Using these data sets, we evaluated the effect of age, sex, ethnicity, and smoking intensity on likelihood of developing BC [11,12]. This information should provide an excellent resource to identify superior populations for BC screening efforts. 2. Material and methods The study designs of PLCO and NLST have been described previously [11,12]. Both trials did not screen for BC but tracked diagnoses of other malignancies during the trial period. Briefly, in PLCO, 154,900 participants from the general population aged 55 through 74 years were enrolled between 1993 and 2001 [11]. Participants had to have a negative history of prostate, lung, colorectal, and ovarian cancer to be eligible, and they were randomized to screening vs. no screening for these malignancies and were followed through the end of 2008. No restrictions to minimum or maximum smoking exposure applied in PLCO. Follow-up was conducted annually via mailed questionnaires, and reported cancers were confirmed by retrieving results from medical record systems and further entered as abstracted information out of the cancer registry system. In NLST, 53,456 subjects considered to be at high risk for lung cancer and aged 55 through 74 years were enrolled between 2002 and 2004 [12]. Participants had to have a negative history of lung cancer, a smoking history of at least 30 pack-years (PY), and former smokers had to have stopped smoking no longer than 15 years before enrollment. Study subjects were randomized to lung cancer screening via computer tomography of the chest vs. conventional x-ray imaging of the chest. Follow-up was conducted annually via mailed questionnaires, and reported cancers were confirmed by retrieving results from medical record systems and further entered as abstracted information out of the cancer registry system. 3. Statistical analysis Differences in person-time incidence rates between men and women stratified by race, age, and smoking exposure
were computed with the chi-square test for both trials separately. Furthermore, proportional hazards models were computed separately to identify high-risk groups with incidence rates above the threshold needed for screening possibly to be considered beneficial and cost-efficient. These thresholds were extrapolated from colon cancer rates that have led to colon cancer screening recommendations [13,14]. To subsequently refine the selection of high-risk patients, difference in incidence rates in men with a smoking history of 430 PY and an exit age of r80 years were computed stratified by race, age, and smoking exposure. All statistical analyses were conducted with SAS version 9.2.
4. Results From the PLCO, 78,215 women and 76,683 men were included in our analyses with 270 and 1,160 BCs, respectively. Overall incidence of BC was 0.31/1,000 personyears for women and 1.4/1,000 person-years for men (P o 0.0001) (Table 1). Incidence rates were highest in white study subjects, subjects Z70 years of age at study entry, and current smokers with a smoking history of Z50 PY. When incidence rates were compared between men and women stratified by race, age, or smoking exposure, men always had a significantly higher incidence rate for BC than women (Table 1). In NLST, 21,922 women and 31,251 men were included in our analyses with 83 and 356 BCs, respectively. Incidence of BC was significantly higher in men compared with women in this study set as well (1.84 vs. 0.6 per 1,000 person-years, respectively) (Table 2). As in PLCO, incidence rates were highest in white study subjects, subjects Z70 years of age at study entry, and current smokers with a smoking history of Z50 PY. When stratified by race, age, or smoking exposure, men always had a significantly higher incidence rate for BC than women (Table 2). Cancer characteristics of the detected cancers are displayed in Table 3. Overall, 1,430/154,898 and 436/53,173 BC were detected in PLCO and NLST, and proportion of grade III/IV cancers were 36.8% vs. 41.3%, respectively, without substantial sex differences. In NLST, a higher proportion of grade IV cancers were observed (14.7% vs. 5.4% in PLCO). In proportional hazards models, study subjects with higher age, male sex, white race (in PLCO only), and higher smoking exposure were more likely to develop BC (Tables 4 and 5). Compared with a reference age of o60 years, study participants with an entry age of 65 to 69 years harbored a hazard ratio (HR) of 2.1 (95% CI: 1.8–2.5, P o 0.0001) in PLCO and 2.7 (95% CI: 2.1–3.6, P o 0.001) in NLST, whereas subjects Z70 years of age rendered an HR of 2.7 (95% CI: 2.4–3.2, P o 0.0001) and 3.9 (95% CI: 3.0–5.3, P o 0.001) in PLCO and NLST, respectively (Tables 4 and 5).
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
65.e21
Table 1 Descriptive statistics for all men and women included in PLCO regarding incidence of bladder cancer Women All
Cases of BC Cases of BC Person(n) (%) years
All 78,215 270 Race White 67,401 249 Black 4,338 7 Other 6,476 14 Age at randomization, y o60 26,927 65 60–64 23,529 80 65–69 17,169 82 Z70 10,591 43 Smoking (up to 50 PY) 30–50 PY, Z15 y 1,773 13 stopped 30–50 PY, o15 y 3,741 19 stopped 30–50 PY, current 2,707 28 smoker Smoking (Z50 PY) Z50 PY, Z15 y 667 5 stopped Z50 PY, o15 y 2,804 22 stopped Z50 PY, current 2,396 19 smoker a
P valueb
Men Incidence ratea
All
Cases of BC Cases of BC Person(n) (%) years
Incidence ratea
0.3
862,404
0.31
76,683 1,160
1.5
827,313
1.4
o0.0001
0.4 0.2 0.2
750,493 43,858 68,053
0.33 0.16 0.21
65,179 1,074 3,370 22 8,134 64
1.6 0.7 0.8
711,846 32,801 82,666
1.51 0.67 0.77
o0.0001 0.0003 o0.0001
0.2 0.3 0.5 0.4
294,902 266,164 189,121 112,217
0.22 0.30 0.43 0.38
24,759 24,024 17,763 10,137
233 350 347 230
0.9 1.5 2.0 2.3
268,619 266,754 190,513 101,427
0.87 1.31 1.82 2.27
o0.0001 o0.0001 o0.0001 o0.0001
0.7
19,375
0.67
4,749
96
2.0
51,445
1.87
0.0003
0.5
40,501
0.47
4,262
91
2.1
45,484
2.0
1.0
28,659
0.98
2,639
47
1.8
27,049
1.74
0.0145
0.7
7,154
0.70
2,873
57
2.0
30,163
1.89
0.0263
0.8
29,668
0.74
6,232
173
2.8
63,587
2.72
o0.0001
0.8
24,272
0.78
4,500
129
2.9
43,210
2.99
o0.0001
o0.0001
Number of cases per 1,000 person-years. P value compares incidence rate of women vs. men.
b
Women were significantly less likely to develop BC with an HR of 0.27 (95% CI: 0.24–0.31, P o 0.0001) and 0.35 (95% CI: 0.28–0.46, P o 0.0001) in PLCO and NLST, respectively (Tables 4 and 5). White subjects had a higher probability to develop BC when compared with nonwhite study subjects in PLCO but not in NLST (Tables 4 and 5). As smoking exposure increased in both trials, risk for BC also increased rendering highest risk in current smokers with 450 PY (Tables 4 and 5). Compared with never smokers, current smokers with 450 PY of smoking history had an HR of 4.3 (95% CI: 3.6–5.3, P o 0.0001) in PLCO, whereas compared with smokers o50 PY who stopped o15 years ago, current smokers with 450 PY had an HR of 2.0 (95% CI: 1.5–2.6, P o 0.0001) (Tables 4 and 5). To further characterize a possible screening population, incidence rates for men with a minimum smoking exposure of 30 PY as well as a screening exit age of 80 years were computed and are displayed in Tables 6 and 7. In PLCO, men with white race, older than 60 years and a smoking exposure of a minimum of 30 PY showed incidence rates of 2.0 per 1,000 person-years. In NLST, men of all races older than 65 years and a smoking exposure of 450 PY showed incidence rates of approximately 2.0 per 1,000 person-years. Incidences rates were as high as
5.23 (NLST) to 11.92 per 1,000 person-years (PLCO) in men 470 years and 2.65 (NLST) to 3.19 per 1,000 person-years (PLCO) in current smokers with 450 PY (Tables 6 and 7). If one were to only screen men older than 60 years with more than 30 PY smoking history, then some cancers would be missed. Not screening women would avoid screening 50.5% and 42.6% of PLCO and NLST populations while missing 19% and 18.9% of cancers. If we do not screen men younger than 60 years, then 20% and 22.2% of cancers would be missed, but 32% and 41.8% of men would avoid screening in PLCO and NLST, respectively. In PLCO, 318 participants reported a previous history of BC, of which, 13 had a cancer recurrence during the study course (5/98 women and 8/220 men). This represented less than 1% of all BCs in PLCO. In NLST, 240 participants reported a history of BC, of which, 11 had a BC recurrence during the study course (3/39 women and 8/201 men). This represented 2.5% of all BCs in NLST.
5. Discussion To develop rational screening policies, populations at risk need to be identified with sufficient incidence of
65.e22
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
Table 2 Descriptive statistics for all men and women included in the NLST regarding incidence of bladder cancer Women All All 21,922 Race White 19,989 Black 1,076 Other 857 Age at randomization, y o60 9,801 60–64 6,684 65–69 3,743 Z70 1,694 Smoking (up to 50 PY) o50 PY, o15 y 6,536 stopped o50 PY, current 6,796 smoker Smoking (Z50 PY) Z50 PY, o15 y 4,397 stopped Z50 PY, current 4,193 smoker a
P valueb
Men Cases of BC Cases of BC Person(n) (%) years
Incidence ratea
All
Cases of BC Cases of BC Person(n) (%) years
Incidence ratea
83
0.4
137,780
0.60
31,251 356
1.1
192,959
1.84
o0.0001
79 3 1
0.4 0.3 0.1
126,195 6,490 5,094
0.63 0.46 0.20
28,303 322 1,295 14 1,653 20
1.1 1.1 1.2
175,536 7,510 9,913
1.83 1.86 2.02
o0.0001 0.0176 0.0047
26 31 18 8
0.3 0.5 0.5 0.5
61,928 41,971 23,511 10,369
0.42 0.74 0.77 0.77
13,064 79 9,685 88 5,774 106 2,728 83
0.6 0.9 1.8 3.0
81,561 60,339 35,188 15,870
0.97 1.46 3.01 5.23
o0.0001 0.0009 o0.0001 o0.0001
16
0.2
41,691
0.38
7,511
65
0.9
47,362
1.37
o0.0001
21
0.3
42,477
0.49
7,003
55
0.8
43,131
1.28
o0.0001
17
0.4
27,822
0.61
9,067 114
1.3
56,470
2.02
o0.0001
29
0.7
25,789
1.12
7,670 122
1.6
45,995
2.65
o0.0001
Number of cases per 1,000 person-years. P value compares incidence rate of women vs. men.
b
disease. This has led to age cutoffs for screening of malignancies such as colon, prostate, and breast cancer. Currently BC screening is not used, but it represents the fourth most common cancer in men [1]. Moreover, although mortality for prostate and colon cancers is decreasing partly owing to screening, that of BC is unchanged over decades. PLCO and NLST provide valuable longitudinal information on incidence of BC stratified by known risk factors for BC such as age, sex, ethnicity, and smoking intensity [3]. Analyzing these studies, we identified high-risk populations of BC that could be used for BC screening. The exact minimal incidence of cancer necessary to justify screening is uncertain as it depends on survival benefit of screening. However, we were able to identify populations with at least 2 cases per 1,000 person-years, which is similar to yields
for colorectal cancer per 1,000 screened, which was 1.1 to 2.5 in women and 2.4 to 5.6 in men [13]. BC incidence rates were highest in men older than 70 years with 50 PY smoking history. There are considerable potential benefits to early diagnosis of BC. Currently, 25% of the newly diagnosed patients with BC already have muscle-invasive or metastatic disease. Patients who are diagnosed when their disease is muscle invasive have a significantly worse survival compared with non–muscle-invasive BC with average group estimated 5-year survival of 50% vs. greater than 90% [15]. Furthermore, diagnosis at the time of muscle invasion limits treatment options to highly morbid approaches such as radical cystectomy and chemotherapy. Identifying disease early with screening can thus have tremendous benefits on both survival and quality of life.
Table 3 Cancer characteristics stratified by grade in PLCO and NLST PLCO n (%)
NLST n (%)
All All cancers Well differentiated (grade 1) Moderately differentiated (grade 2) Poorly differentiated (grade 3) Undifferentiated (grade 4) Not determined/stated/applicable
1,430 358 325 449 77 221
(100.0) (25.0) (22.7) (31.4) (5.4) (15.5)
Women
Men
270 83 65 64 14 44
1,160 275 260 385 63 177
(100.0) (30.7) (24.1) (23.7) (5.2) (16.3)
All (100.0) (23.7) (22.4) (33.2) (5.4) (15.3)
436 110 91 116 64 55
(100.0) (25.2) (20.9) (26.6) (14.7) (12.6)
Women
Men
82 20 15 24 16 7
354 90 76 92 48 48
(100.0) (24.4) (18.3) (29.3) (19.5) (8.5)
(100.0) (25.4) (21.4) (26.0) (13.6) (13.6)
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25 Table 4 Proportional hazards model of bladder cancer incidence in PLCO Parameter Age, y o60 60–64 65–69 Z70 Sex Men Women Race White Black Other Smoking Never smoker 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped o15 y ago 450 PY, current smoker
Hazard ratio 95% CI
P value
Reference 1.467 2.106 2.692
– 1.265–1.701 1.814–2.446 2.277–3.182
– o0.0001 o0.0001 o0.0001
Reference 0.271
– – 0.236–0.310 o0.0001
Reference 0.438 0.569
– – 0.303–0.634 o0.0001 0.447–0.723 o0.0001
Reference 2.904 3.693 3.722 4.343
– 2.335–3.612 2.861–4.767 3.104–4.463 3.555–5.305
– o0.0001 o0.0001 o0.0001 o0.0001
There have been several attempts to demonstrate that BC screening results in earlier detection of BC before muscle invasion and metastasis [5–9]. These studies found that screening for BC in the general population rendered very low incidence rates of BC and concluded that screening for BC would be ineffective in the general population. Lotan et al. [8] attempted to screen a population at high risk (smoking Z10 y or occupational exposure Z15 y) in 1,811 subjects but only 3 cancers were found (0.16%). Future screening studies will benefit from limiting populations to patients older than 60 years with more than 30 to 40 PY smoking history and focusing on men primarily, as in screening trials it is of utmost importance to demonstrate a benefit from screening (decreased mortality of disease) so the population screened Table 5 Proportional hazards model of bladder cancer incidence in NLST Parameter Age, y o60 60–64 65–69 Z70 Sex Men Women Race White Black Other Smoking 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped o15 y ago 450 PY, current smoker
Hazard ratio
95% CI
P value
Reference 1.549 2.734 3.944
– 1.190–2.016 2.102–3.557 2.962–5.253
– 0.0011 o0.0001 o0.0001
Reference 0.351
– 0.284–0.460
– o0.0001
Reference 1.017 1.018
– 0.616–1.681 0.656–1.579
–
Reference 1.179 1.333 1.995
– 0.860–1.617 1.008–1.763 1.521–2.618
–
0.9466 0.9380
0.3067 0.0438 o0.0001
65.e23
must be at high enough risk to justify the inherent risks of screening interventions. There have been 2 studies that have used BC incidence rates in PLCO to develop models for predicting BC and evaluate screening trial feasibility [16,17]. Vickers et al. [16] used decision curve analysis and a risk score implementing multiple risk factors for BC (age, smoking history, sex, and family history) to determine likelihood of high-grade or invasive disease and identify which population should be included in a BC screening trial. Mir et al. [17] developed nomograms for the detection of any BC or high-grade BC with c-indices of 0.746 and 0.759, respectively. In contrast to these studies, we provide the actual rates of cancer stratified by age, sex, and smoking history (Table 1). We identify intensity of smoking based on duration and stratified by current and former smokers. This allows identification of cancer rates by individualized risk groupings. We also stratify the grade of cancer by sex (Table 3). Our study also provides HRs by different risk factors (Tables 4 and 5), and we stratify the risk of BC in the highest risk group (men with over 30 PY) by age and degree of smoking (Tables 6 and 7). These models reinforce the concept that screening the general population is not feasible owing to low incidence and that a high-risk population will be necessary to reduce size of any prospective screening study and provide a potential survival benefit. In fact, a Markov model, used to estimate cumulative cancer-related costs and efficacy of screening vs. no screening of a high-risk population for BC, found that screening in a population with a BC incidence of 41.6% could improve overall survival as well as result in cost savings [18]. Several groups within the PLCO and NLST achieved a period prevalence over the extrapolated threshold (Tables 1 and 2). Another interesting observation in this study is the persistent sex disparity between men and women despite similar intensity of smoking. In the National Bladder Cancer Study from 1978, when adjusted for the effect of age, men had a BC incidence of 27.5/100,000 person-years vs. 7.0/100,000 person-years in women, but at that time smoking was much more common in men than in women [19]. As currently there is closer parity in smoking habits between men and women, one might expect similar incidence in BC as smoking is the greatest common risk factor to develop BC [4]. However, in both the PLCO and NLST, men had significantly higher rates of BC than women despite similar intensities of smoking. Although smoking intensity does not narrow the sex gap, heavy smoking does increase the risk of BC in both men and women [4,19]. There has been evidence in animal models that hormonal levels of estrogen, testosterone, and their receptors' activities influence development of BC in the presence of carcinogens, which might explain sex disparity and has important implications for our understanding and potential treatments of BC [20–23].
65.e24
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
Table 6 Incidence rates of bladder cancer for men in PLCO (430 PY and exit age r80 y)
All Race White Black Other Age at randomization, y o60 60–64 65–69 Z70 Smoking 30–50 PY, stopped Z15 y ago 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped Z15 y ago 450 PY, stopped o15 y ago 450 PY, current smoker a
Incidence ratea
P value
All
Cases of BC (n)
Cases of BC (%)
Person-years
22,305
562
2.5
225,259
2.49
–
20,074 997 1,234
538 9 15
2.7 0.9 1.2
203,986 8,901 12,372
2.64 1.01 1.21
–
7,817 8,031 5,251 1,206
148 175 167 72
1.9 2.2 3.2 6.0
82,424 85,964 50,831 6,040
1.80 2.04 3.29 11.92
–
3,805 3,919 2,579 2,240 5,574 4,188
85 86 47 50 168 126
2.2 2.2 1.8 2.2 3.0 3.0
39,969 41,268 26,314 22,586 55,684 39,438
2.13 2.08 1.79 2.21 3.01 3.19
0.0030 0.0023
0.2607 o0.0001 o0.0001 – 0.8945 0.3413 0.8218 0.0083 0.0035
Incidence rates per 1,000 person-years.
Limitations of the study are a lack of information regarding occupational exposure to carcinogens leading to BC. Furthermore, both studies might suffer from healthy volunteer bias and might therefore overestimate the threshold of age and smoking exposure, which is potentially ideal to investigate screening for BC, and therefore exclude people who might benefit from screening. 6. Conclusion Screening for BC in men older than 60 years with a smoking history of 430 PY generates incidence rates of more than 2/1,000 person-years, which could serve as the highest risk population for screening trials. Sex differences in BC incidence cannot be readily explained
by disparity in exposure to tobacco, as in PLCO and NLST, the sex disparity persisted even in high-intensity smokers.
Acknowledgments The authors thank the National Cancer Institute for access to NCI's data collected by the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. The authors thank the National Cancer Institute for access to NCI's data collected by the National Lung Screening Trial. The statements contained herein are solely those of the authors and do not represent or imply concurrence or endorsement by NCI.
Table 7 Incidence rates of bladder cancer for men in NLST (exit age r80 y)
All Race White Black Other Age at randomization, y o60 60–64 65–69 Z70 Smoking 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped o15 y ago 450 PY, current smoker a
All
Cases of BC (n)
Cases of BC (%)
Person-years
Incidence ratea
P value
31,251
356
1.1
192,959
1.84
–
28,303 1,295 1,653
322 14 20
1.1 1.1 1.2
175,536 7,510 9,913
1.83 1.86 2.02
– 0.9529 0.6795
13,064 9,685 5,774 2,728
79 88 106 83
0.6 0.9 1.8 3.0
81,561 60,339 35,188 15,870
0.97 1.46 3.01 5.23
– 0.0078 o0.0001 o0.0001
7,511 7,003 9,067 7,670
65 55 114 122
0.9 0.8 1.3 1.6
47,362 43,131 56,470 45,995
1.37 1.28 2.02 2.65
– 0.6883 0.0125 o0.0001
Incidence rates per 1,000 person-years.
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
References [1] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin 2013;63:11–30. [2] Brown T, Slack R, Rushton L. Occupational cancer in Britain. Urinary tract cancers: bladder and kidney. Br J Cancer 2012;107(Suppl. 1): S76–S84. [3] Burger M, Catto JW, Dalbagni G, Grossman HB, Herr H, Karakiewicz P, et al. Epidemiology and risk factors of urothelial bladder cancer. Eur Urol 2013;63:234–41. [4] Freedman ND, Silverman DT, Hollenbeck AR, Schatzkin A, Abnet CC. Association between smoking and risk of bladder cancer among men and women. J Am Med Assoc 2011;306:737–45. [5] Messing EM, Young TB, Hunt VB, Wehbie JM, Rust P. Urinary tract cancers found by homescreening with hematuria dipsticks in healthy men over 50 years of age. Cancer 1989;64:2361–7. [6] Britton JP, Dowell AC, Whelan P, Harris CM. A community study of bladder cancer screening by the detection of occult urinary bleeding. J Urol 1992;148:788–90. [7] Bangma CH, Loeb S, Busstra M, Zhu X, Bouazzaoui SE, Refos J, et al. Outcomes of a bladder cancer screening program using home hematuria testing and molecular markers. Eur Urol 2013;64(1):41–7. doi: 10.1016/j.eururo.2013.02.036. [8] Lotan Y, Elias K, Svatek RS, Bagrodia A, Nuss G, Moran B, et al. Bladder cancer screening in a high risk asymptomatic population using a point of care urine based protein tumor marker. J Urol 2009;182:52–7:[discussion 8]. [9] Hedelin H, Jonsson K, Salomonsson K, Boman H. Screening for bladder tumours in men aged 60–70 years with a bladder tumour marker (UBC) and dipstick-detected haematuria using both white-light and fluorescence cystoscopy. Scand J Urol Nephrol 2006;40:26–30. [10] Larre S, Catto JW, Cookson MS, Messing EM, Shariat SF, Soloway MS, et al. Screening for bladder cancer: rationale, limitations, whom to target, and perspectives. Eur Urol 2013;63(6):1049–58. doi: 10.1016/j.eururo.2012.12.062. [11] Prorok PC, Andriole GL, Bresalier RS, Buys SS, Chia D, Crawford ED, et al. Design of the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 2000;21:273S–309S. [12] Aberle DR, Berg CD, Black WC, Church TR, Fagerstrom RM, Galen B, et al. The National Lung Screening Trial: overview and study design. Radiology 2011;258:243–53.
65.e25
[13] Weissfeld JL, Schoen RE, Pinsky PF, Bresalier RS, Church T, Yurgalevitch S, et al. Flexible sigmoidoscopy in the PLCO cancer screening trial: results from the baseline screening examination of a randomized trial. J Natl Cancer Inst 2005;97:989–97. [14] Whitlock EP, Lin J, Liles E, Beil T, Fu R, O'Connor E, et al, editors. Screening for Colorectal Cancer: An Updated Systematic Review. Rockville, MD: Agency for Healthcare Research and Quality (US); 2008 [Report No. 08-05-05124-EF-1]. [15] Lughezzani G, Burger M, Margulis V, Matin SF, Novara G, Roupret M, et al. Prognostic factors in upper urinary tract urothelial carcinomas: a comprehensive review of the current literature. Eur Urol 2012;62:100–14. [16] Vickers AJ, Bennette C, Kibel AS, Black A, Izmirlian G, Stephenson AJ, et al. Who should be included in a clinical trial of screening for bladder cancer? a decision analysis of data from the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial. Cancer 2013;119:143–9. [17] Mir MC, Stephenson AJ, Grubb RL 3rd, Black A, Kibel AS, Izmirlian G. Predicting risk of bladder cancer using clinical and demographic information from prostate, lung, colorectal, and ovarian cancer (PLCO) screening trial participants. Cancer Epidemiol Biomarkers Prev 2013;22(12):2241–9. doi: 10.1158/1055-9965.EPI-130632. [18] Lotan Y, Svatek RS, Sagalowsky AI. Should we screen for bladder cancer in a high-risk population? a cost per life-year saved analysis. Cancer 2006;107:982–90. [19] Hartge P, Harvey EB, Linehan WM, Silverman DT, Sullivan JW, Hoover RN, et al. Unexplained excess risk of bladder cancer in men. J Natl Cancer Inst 1990;82:1636–40. [20] Zhang Y. Understanding the gender disparity in bladder cancer risk: the impact of sex hormones and liver on bladder susceptibility to carcinogens. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 2013;31:287–304. [21] Miyamoto H, Yang Z, Chen YT, Ishiguro H, Uemura H, Kubota Y, et al. Promotion of bladder cancer development and progression by androgen receptor signals. J Natl Cancer Inst 2007;99:558–68. [22] Johnson AM, O'Connell MJ, Miyamoto H, Huang J, Yao JL, Messing EM, et al. Androgenic dependence of exophytic tumor growth in a transgenic mouse model of bladder cancer: a role for thrombospondin1. BMC Urol 2008;8:7. [23] Johnson AM, O'Connell MJ, Messing EM, Reeder JE. Decreased bladder cancer growth in parous mice. Urology 2008;72:470–3.
Original article
Bladder cancer risk: Use of the PLCO and NLST to identify a suitable screening cohort Laura-Maria Krabbe, M.D.a,b, Robert S. Svatek, M.D.c, Shahrokh F. Shariat, M.D.a,d, Edward Messing, M.D.e, Yair Lotan, M.D.a,* a
Department of Urology, University of Texas Southwestern Medical Center, Dallas, TX Department of Urology, University of Muenster Medical Center, Muenster, Germany c Department of Urology, University of Texas Health Science Center, San Antonio TX d Department of Urology, Medical University of Vienna, Vienna, Austria e Department of Urology, University of Rochester Medical Center, Rochester, NY
b
Received 28 April 2014; received in revised form 14 June 2014; accepted 16 June 2014
Abstract Purpose: Bladder cancer (BC) screening is not accepted in part owing to low overall incidence. We used the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) and National Lung Cancer Screening Trial (NLST) to identify optimal high-risk populations most likely to benefit from screening. Materials and methods: Data were extracted from PLCO and NLST to stratify risk of BC by overall population, sex, race, age at inclusion, and smoking status. Incidence rates between groups were compared using chi-square test. Results: BC was identified in 1,430/154,898 patients in PLCO and 439/53,173 patients in NLST. BCs were grade III/IV in 36.8% and 41.3%. Incidence rates were significantly higher in men than in women (PLCO: 1.4 vs. 0.31/1,000 person-years and NLST: 1.84 vs. 0.6/1,000 person-years, both P o 0.0001). In proportional hazards models, male sex, higher age, and duration and intensity of smoking were associated with higher risk of BC (all P o 0.0001). In men older than 70 years with smoking exposure of 30 pack-years (PY) and more, incidence rates were as high as 11.92 (PLCO) and 5.23 (NLST) (per 1,000 person-years). In current high-intensity smokers (Z50 PY), the sex disparity in incidence persists in both trials (0.78 vs. 2.99 per 1,000 person-years in PLCO and 1.12 vs. 2.65 per 1,000 person-years in NLST). Conclusions: Men older than 60 years with a smoking history of 430 PY had incidence rates of more than 2/1,000 person-years, which could serve as an excellent population for screening trials. Sex differences in the incidence of BC cannot be readily explained by the differences in exposure to tobacco, as sex disparity persisted regardless of smoking intensity. r 2014 Elsevier Inc. All rights reserved.
Keywords: Bladder cancer; Screening; Risk factors; Sex; Risk stratification
1. Introduction Bladder cancer (BC) is the sixth most common cancer overall in North America with an estimated 72,570 new cases and 15,210 deaths in 2013 in the United States [1]. Men are about 3 times more likely to develop BC in their lifetime than women are, which has been attributed to different historical smoking patterns or occupational exposures [1–3]. Earlier reports suggested that the population attributable risk (PAR) Corresponding author. Tel.: þ1-214-648-0389; fax: þ1-214-648-8786. E-mail address: [email protected] (Y. Lotan). *
http://dx.doi.org/10.1016/j.urolonc.2014.06.009 1078-1439/r 2014 Elsevier Inc. All rights reserved.
owing to smoking was 50% to 65% in men and 20% to 30% in women, although more recent studies showed more balanced PARs of approximately 50% for both sexes owing to increased rates of smoking in women [4]. PARs explain what proportion of BC is attributable to a risk factor such as smoking; however, limited data have been available to compare the incidence rates of BC in men and women with similarly high exposures to tobacco and to evaluate if the sex disparity remains. Approximately 25% of all patients newly diagnosed with BC have muscle-invasive or metastatic disease, which leads to high morbidity and mortality as well as high cost for treatment [1]. The goal of screening is to achieve a shift to
65.e20
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
lower stages at diagnosis thereby improving morbidity and survival, especially because as many as 80% of muscleinvasive cancers are already muscle invasive at first diagnosis. Screening for BC has been studied by several investigators; however, partly owing to the low overall incidence of BC (37.5 and 9.3 per 100,000 in men and women), screening is currently not recommended in routine practice [1,5–10]. There are multiple factors contributing to low BC detection rates in prior screening studies, including healthy volunteer bias, but the main reason was the inclusion of patients at too low risk for BC. Identifying a cohort with potentially higher disease incidence is critical for instituting an effective screening program. Both the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) and the National Lung Cancer Screening Trial (NLST) performed longitudinal prospective evaluations of large populations and tracked secondary malignancies. Using these data sets, we evaluated the effect of age, sex, ethnicity, and smoking intensity on likelihood of developing BC [11,12]. This information should provide an excellent resource to identify superior populations for BC screening efforts. 2. Material and methods The study designs of PLCO and NLST have been described previously [11,12]. Both trials did not screen for BC but tracked diagnoses of other malignancies during the trial period. Briefly, in PLCO, 154,900 participants from the general population aged 55 through 74 years were enrolled between 1993 and 2001 [11]. Participants had to have a negative history of prostate, lung, colorectal, and ovarian cancer to be eligible, and they were randomized to screening vs. no screening for these malignancies and were followed through the end of 2008. No restrictions to minimum or maximum smoking exposure applied in PLCO. Follow-up was conducted annually via mailed questionnaires, and reported cancers were confirmed by retrieving results from medical record systems and further entered as abstracted information out of the cancer registry system. In NLST, 53,456 subjects considered to be at high risk for lung cancer and aged 55 through 74 years were enrolled between 2002 and 2004 [12]. Participants had to have a negative history of lung cancer, a smoking history of at least 30 pack-years (PY), and former smokers had to have stopped smoking no longer than 15 years before enrollment. Study subjects were randomized to lung cancer screening via computer tomography of the chest vs. conventional x-ray imaging of the chest. Follow-up was conducted annually via mailed questionnaires, and reported cancers were confirmed by retrieving results from medical record systems and further entered as abstracted information out of the cancer registry system. 3. Statistical analysis Differences in person-time incidence rates between men and women stratified by race, age, and smoking exposure
were computed with the chi-square test for both trials separately. Furthermore, proportional hazards models were computed separately to identify high-risk groups with incidence rates above the threshold needed for screening possibly to be considered beneficial and cost-efficient. These thresholds were extrapolated from colon cancer rates that have led to colon cancer screening recommendations [13,14]. To subsequently refine the selection of high-risk patients, difference in incidence rates in men with a smoking history of 430 PY and an exit age of r80 years were computed stratified by race, age, and smoking exposure. All statistical analyses were conducted with SAS version 9.2.
4. Results From the PLCO, 78,215 women and 76,683 men were included in our analyses with 270 and 1,160 BCs, respectively. Overall incidence of BC was 0.31/1,000 personyears for women and 1.4/1,000 person-years for men (P o 0.0001) (Table 1). Incidence rates were highest in white study subjects, subjects Z70 years of age at study entry, and current smokers with a smoking history of Z50 PY. When incidence rates were compared between men and women stratified by race, age, or smoking exposure, men always had a significantly higher incidence rate for BC than women (Table 1). In NLST, 21,922 women and 31,251 men were included in our analyses with 83 and 356 BCs, respectively. Incidence of BC was significantly higher in men compared with women in this study set as well (1.84 vs. 0.6 per 1,000 person-years, respectively) (Table 2). As in PLCO, incidence rates were highest in white study subjects, subjects Z70 years of age at study entry, and current smokers with a smoking history of Z50 PY. When stratified by race, age, or smoking exposure, men always had a significantly higher incidence rate for BC than women (Table 2). Cancer characteristics of the detected cancers are displayed in Table 3. Overall, 1,430/154,898 and 436/53,173 BC were detected in PLCO and NLST, and proportion of grade III/IV cancers were 36.8% vs. 41.3%, respectively, without substantial sex differences. In NLST, a higher proportion of grade IV cancers were observed (14.7% vs. 5.4% in PLCO). In proportional hazards models, study subjects with higher age, male sex, white race (in PLCO only), and higher smoking exposure were more likely to develop BC (Tables 4 and 5). Compared with a reference age of o60 years, study participants with an entry age of 65 to 69 years harbored a hazard ratio (HR) of 2.1 (95% CI: 1.8–2.5, P o 0.0001) in PLCO and 2.7 (95% CI: 2.1–3.6, P o 0.001) in NLST, whereas subjects Z70 years of age rendered an HR of 2.7 (95% CI: 2.4–3.2, P o 0.0001) and 3.9 (95% CI: 3.0–5.3, P o 0.001) in PLCO and NLST, respectively (Tables 4 and 5).
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
65.e21
Table 1 Descriptive statistics for all men and women included in PLCO regarding incidence of bladder cancer Women All
Cases of BC Cases of BC Person(n) (%) years
All 78,215 270 Race White 67,401 249 Black 4,338 7 Other 6,476 14 Age at randomization, y o60 26,927 65 60–64 23,529 80 65–69 17,169 82 Z70 10,591 43 Smoking (up to 50 PY) 30–50 PY, Z15 y 1,773 13 stopped 30–50 PY, o15 y 3,741 19 stopped 30–50 PY, current 2,707 28 smoker Smoking (Z50 PY) Z50 PY, Z15 y 667 5 stopped Z50 PY, o15 y 2,804 22 stopped Z50 PY, current 2,396 19 smoker a
P valueb
Men Incidence ratea
All
Cases of BC Cases of BC Person(n) (%) years
Incidence ratea
0.3
862,404
0.31
76,683 1,160
1.5
827,313
1.4
o0.0001
0.4 0.2 0.2
750,493 43,858 68,053
0.33 0.16 0.21
65,179 1,074 3,370 22 8,134 64
1.6 0.7 0.8
711,846 32,801 82,666
1.51 0.67 0.77
o0.0001 0.0003 o0.0001
0.2 0.3 0.5 0.4
294,902 266,164 189,121 112,217
0.22 0.30 0.43 0.38
24,759 24,024 17,763 10,137
233 350 347 230
0.9 1.5 2.0 2.3
268,619 266,754 190,513 101,427
0.87 1.31 1.82 2.27
o0.0001 o0.0001 o0.0001 o0.0001
0.7
19,375
0.67
4,749
96
2.0
51,445
1.87
0.0003
0.5
40,501
0.47
4,262
91
2.1
45,484
2.0
1.0
28,659
0.98
2,639
47
1.8
27,049
1.74
0.0145
0.7
7,154
0.70
2,873
57
2.0
30,163
1.89
0.0263
0.8
29,668
0.74
6,232
173
2.8
63,587
2.72
o0.0001
0.8
24,272
0.78
4,500
129
2.9
43,210
2.99
o0.0001
o0.0001
Number of cases per 1,000 person-years. P value compares incidence rate of women vs. men.
b
Women were significantly less likely to develop BC with an HR of 0.27 (95% CI: 0.24–0.31, P o 0.0001) and 0.35 (95% CI: 0.28–0.46, P o 0.0001) in PLCO and NLST, respectively (Tables 4 and 5). White subjects had a higher probability to develop BC when compared with nonwhite study subjects in PLCO but not in NLST (Tables 4 and 5). As smoking exposure increased in both trials, risk for BC also increased rendering highest risk in current smokers with 450 PY (Tables 4 and 5). Compared with never smokers, current smokers with 450 PY of smoking history had an HR of 4.3 (95% CI: 3.6–5.3, P o 0.0001) in PLCO, whereas compared with smokers o50 PY who stopped o15 years ago, current smokers with 450 PY had an HR of 2.0 (95% CI: 1.5–2.6, P o 0.0001) (Tables 4 and 5). To further characterize a possible screening population, incidence rates for men with a minimum smoking exposure of 30 PY as well as a screening exit age of 80 years were computed and are displayed in Tables 6 and 7. In PLCO, men with white race, older than 60 years and a smoking exposure of a minimum of 30 PY showed incidence rates of 2.0 per 1,000 person-years. In NLST, men of all races older than 65 years and a smoking exposure of 450 PY showed incidence rates of approximately 2.0 per 1,000 person-years. Incidences rates were as high as
5.23 (NLST) to 11.92 per 1,000 person-years (PLCO) in men 470 years and 2.65 (NLST) to 3.19 per 1,000 person-years (PLCO) in current smokers with 450 PY (Tables 6 and 7). If one were to only screen men older than 60 years with more than 30 PY smoking history, then some cancers would be missed. Not screening women would avoid screening 50.5% and 42.6% of PLCO and NLST populations while missing 19% and 18.9% of cancers. If we do not screen men younger than 60 years, then 20% and 22.2% of cancers would be missed, but 32% and 41.8% of men would avoid screening in PLCO and NLST, respectively. In PLCO, 318 participants reported a previous history of BC, of which, 13 had a cancer recurrence during the study course (5/98 women and 8/220 men). This represented less than 1% of all BCs in PLCO. In NLST, 240 participants reported a history of BC, of which, 11 had a BC recurrence during the study course (3/39 women and 8/201 men). This represented 2.5% of all BCs in NLST.
5. Discussion To develop rational screening policies, populations at risk need to be identified with sufficient incidence of
65.e22
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
Table 2 Descriptive statistics for all men and women included in the NLST regarding incidence of bladder cancer Women All All 21,922 Race White 19,989 Black 1,076 Other 857 Age at randomization, y o60 9,801 60–64 6,684 65–69 3,743 Z70 1,694 Smoking (up to 50 PY) o50 PY, o15 y 6,536 stopped o50 PY, current 6,796 smoker Smoking (Z50 PY) Z50 PY, o15 y 4,397 stopped Z50 PY, current 4,193 smoker a
P valueb
Men Cases of BC Cases of BC Person(n) (%) years
Incidence ratea
All
Cases of BC Cases of BC Person(n) (%) years
Incidence ratea
83
0.4
137,780
0.60
31,251 356
1.1
192,959
1.84
o0.0001
79 3 1
0.4 0.3 0.1
126,195 6,490 5,094
0.63 0.46 0.20
28,303 322 1,295 14 1,653 20
1.1 1.1 1.2
175,536 7,510 9,913
1.83 1.86 2.02
o0.0001 0.0176 0.0047
26 31 18 8
0.3 0.5 0.5 0.5
61,928 41,971 23,511 10,369
0.42 0.74 0.77 0.77
13,064 79 9,685 88 5,774 106 2,728 83
0.6 0.9 1.8 3.0
81,561 60,339 35,188 15,870
0.97 1.46 3.01 5.23
o0.0001 0.0009 o0.0001 o0.0001
16
0.2
41,691
0.38
7,511
65
0.9
47,362
1.37
o0.0001
21
0.3
42,477
0.49
7,003
55
0.8
43,131
1.28
o0.0001
17
0.4
27,822
0.61
9,067 114
1.3
56,470
2.02
o0.0001
29
0.7
25,789
1.12
7,670 122
1.6
45,995
2.65
o0.0001
Number of cases per 1,000 person-years. P value compares incidence rate of women vs. men.
b
disease. This has led to age cutoffs for screening of malignancies such as colon, prostate, and breast cancer. Currently BC screening is not used, but it represents the fourth most common cancer in men [1]. Moreover, although mortality for prostate and colon cancers is decreasing partly owing to screening, that of BC is unchanged over decades. PLCO and NLST provide valuable longitudinal information on incidence of BC stratified by known risk factors for BC such as age, sex, ethnicity, and smoking intensity [3]. Analyzing these studies, we identified high-risk populations of BC that could be used for BC screening. The exact minimal incidence of cancer necessary to justify screening is uncertain as it depends on survival benefit of screening. However, we were able to identify populations with at least 2 cases per 1,000 person-years, which is similar to yields
for colorectal cancer per 1,000 screened, which was 1.1 to 2.5 in women and 2.4 to 5.6 in men [13]. BC incidence rates were highest in men older than 70 years with 50 PY smoking history. There are considerable potential benefits to early diagnosis of BC. Currently, 25% of the newly diagnosed patients with BC already have muscle-invasive or metastatic disease. Patients who are diagnosed when their disease is muscle invasive have a significantly worse survival compared with non–muscle-invasive BC with average group estimated 5-year survival of 50% vs. greater than 90% [15]. Furthermore, diagnosis at the time of muscle invasion limits treatment options to highly morbid approaches such as radical cystectomy and chemotherapy. Identifying disease early with screening can thus have tremendous benefits on both survival and quality of life.
Table 3 Cancer characteristics stratified by grade in PLCO and NLST PLCO n (%)
NLST n (%)
All All cancers Well differentiated (grade 1) Moderately differentiated (grade 2) Poorly differentiated (grade 3) Undifferentiated (grade 4) Not determined/stated/applicable
1,430 358 325 449 77 221
(100.0) (25.0) (22.7) (31.4) (5.4) (15.5)
Women
Men
270 83 65 64 14 44
1,160 275 260 385 63 177
(100.0) (30.7) (24.1) (23.7) (5.2) (16.3)
All (100.0) (23.7) (22.4) (33.2) (5.4) (15.3)
436 110 91 116 64 55
(100.0) (25.2) (20.9) (26.6) (14.7) (12.6)
Women
Men
82 20 15 24 16 7
354 90 76 92 48 48
(100.0) (24.4) (18.3) (29.3) (19.5) (8.5)
(100.0) (25.4) (21.4) (26.0) (13.6) (13.6)
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25 Table 4 Proportional hazards model of bladder cancer incidence in PLCO Parameter Age, y o60 60–64 65–69 Z70 Sex Men Women Race White Black Other Smoking Never smoker 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped o15 y ago 450 PY, current smoker
Hazard ratio 95% CI
P value
Reference 1.467 2.106 2.692
– 1.265–1.701 1.814–2.446 2.277–3.182
– o0.0001 o0.0001 o0.0001
Reference 0.271
– – 0.236–0.310 o0.0001
Reference 0.438 0.569
– – 0.303–0.634 o0.0001 0.447–0.723 o0.0001
Reference 2.904 3.693 3.722 4.343
– 2.335–3.612 2.861–4.767 3.104–4.463 3.555–5.305
– o0.0001 o0.0001 o0.0001 o0.0001
There have been several attempts to demonstrate that BC screening results in earlier detection of BC before muscle invasion and metastasis [5–9]. These studies found that screening for BC in the general population rendered very low incidence rates of BC and concluded that screening for BC would be ineffective in the general population. Lotan et al. [8] attempted to screen a population at high risk (smoking Z10 y or occupational exposure Z15 y) in 1,811 subjects but only 3 cancers were found (0.16%). Future screening studies will benefit from limiting populations to patients older than 60 years with more than 30 to 40 PY smoking history and focusing on men primarily, as in screening trials it is of utmost importance to demonstrate a benefit from screening (decreased mortality of disease) so the population screened Table 5 Proportional hazards model of bladder cancer incidence in NLST Parameter Age, y o60 60–64 65–69 Z70 Sex Men Women Race White Black Other Smoking 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped o15 y ago 450 PY, current smoker
Hazard ratio
95% CI
P value
Reference 1.549 2.734 3.944
– 1.190–2.016 2.102–3.557 2.962–5.253
– 0.0011 o0.0001 o0.0001
Reference 0.351
– 0.284–0.460
– o0.0001
Reference 1.017 1.018
– 0.616–1.681 0.656–1.579
–
Reference 1.179 1.333 1.995
– 0.860–1.617 1.008–1.763 1.521–2.618
–
0.9466 0.9380
0.3067 0.0438 o0.0001
65.e23
must be at high enough risk to justify the inherent risks of screening interventions. There have been 2 studies that have used BC incidence rates in PLCO to develop models for predicting BC and evaluate screening trial feasibility [16,17]. Vickers et al. [16] used decision curve analysis and a risk score implementing multiple risk factors for BC (age, smoking history, sex, and family history) to determine likelihood of high-grade or invasive disease and identify which population should be included in a BC screening trial. Mir et al. [17] developed nomograms for the detection of any BC or high-grade BC with c-indices of 0.746 and 0.759, respectively. In contrast to these studies, we provide the actual rates of cancer stratified by age, sex, and smoking history (Table 1). We identify intensity of smoking based on duration and stratified by current and former smokers. This allows identification of cancer rates by individualized risk groupings. We also stratify the grade of cancer by sex (Table 3). Our study also provides HRs by different risk factors (Tables 4 and 5), and we stratify the risk of BC in the highest risk group (men with over 30 PY) by age and degree of smoking (Tables 6 and 7). These models reinforce the concept that screening the general population is not feasible owing to low incidence and that a high-risk population will be necessary to reduce size of any prospective screening study and provide a potential survival benefit. In fact, a Markov model, used to estimate cumulative cancer-related costs and efficacy of screening vs. no screening of a high-risk population for BC, found that screening in a population with a BC incidence of 41.6% could improve overall survival as well as result in cost savings [18]. Several groups within the PLCO and NLST achieved a period prevalence over the extrapolated threshold (Tables 1 and 2). Another interesting observation in this study is the persistent sex disparity between men and women despite similar intensity of smoking. In the National Bladder Cancer Study from 1978, when adjusted for the effect of age, men had a BC incidence of 27.5/100,000 person-years vs. 7.0/100,000 person-years in women, but at that time smoking was much more common in men than in women [19]. As currently there is closer parity in smoking habits between men and women, one might expect similar incidence in BC as smoking is the greatest common risk factor to develop BC [4]. However, in both the PLCO and NLST, men had significantly higher rates of BC than women despite similar intensities of smoking. Although smoking intensity does not narrow the sex gap, heavy smoking does increase the risk of BC in both men and women [4,19]. There has been evidence in animal models that hormonal levels of estrogen, testosterone, and their receptors' activities influence development of BC in the presence of carcinogens, which might explain sex disparity and has important implications for our understanding and potential treatments of BC [20–23].
65.e24
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
Table 6 Incidence rates of bladder cancer for men in PLCO (430 PY and exit age r80 y)
All Race White Black Other Age at randomization, y o60 60–64 65–69 Z70 Smoking 30–50 PY, stopped Z15 y ago 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped Z15 y ago 450 PY, stopped o15 y ago 450 PY, current smoker a
Incidence ratea
P value
All
Cases of BC (n)
Cases of BC (%)
Person-years
22,305
562
2.5
225,259
2.49
–
20,074 997 1,234
538 9 15
2.7 0.9 1.2
203,986 8,901 12,372
2.64 1.01 1.21
–
7,817 8,031 5,251 1,206
148 175 167 72
1.9 2.2 3.2 6.0
82,424 85,964 50,831 6,040
1.80 2.04 3.29 11.92
–
3,805 3,919 2,579 2,240 5,574 4,188
85 86 47 50 168 126
2.2 2.2 1.8 2.2 3.0 3.0
39,969 41,268 26,314 22,586 55,684 39,438
2.13 2.08 1.79 2.21 3.01 3.19
0.0030 0.0023
0.2607 o0.0001 o0.0001 – 0.8945 0.3413 0.8218 0.0083 0.0035
Incidence rates per 1,000 person-years.
Limitations of the study are a lack of information regarding occupational exposure to carcinogens leading to BC. Furthermore, both studies might suffer from healthy volunteer bias and might therefore overestimate the threshold of age and smoking exposure, which is potentially ideal to investigate screening for BC, and therefore exclude people who might benefit from screening. 6. Conclusion Screening for BC in men older than 60 years with a smoking history of 430 PY generates incidence rates of more than 2/1,000 person-years, which could serve as the highest risk population for screening trials. Sex differences in BC incidence cannot be readily explained
by disparity in exposure to tobacco, as in PLCO and NLST, the sex disparity persisted even in high-intensity smokers.
Acknowledgments The authors thank the National Cancer Institute for access to NCI's data collected by the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. The authors thank the National Cancer Institute for access to NCI's data collected by the National Lung Screening Trial. The statements contained herein are solely those of the authors and do not represent or imply concurrence or endorsement by NCI.
Table 7 Incidence rates of bladder cancer for men in NLST (exit age r80 y)
All Race White Black Other Age at randomization, y o60 60–64 65–69 Z70 Smoking 30–50 PY, stopped o15 y ago 30–50 PY, current smoker 450 PY, stopped o15 y ago 450 PY, current smoker a
All
Cases of BC (n)
Cases of BC (%)
Person-years
Incidence ratea
P value
31,251
356
1.1
192,959
1.84
–
28,303 1,295 1,653
322 14 20
1.1 1.1 1.2
175,536 7,510 9,913
1.83 1.86 2.02
– 0.9529 0.6795
13,064 9,685 5,774 2,728
79 88 106 83
0.6 0.9 1.8 3.0
81,561 60,339 35,188 15,870
0.97 1.46 3.01 5.23
– 0.0078 o0.0001 o0.0001
7,511 7,003 9,067 7,670
65 55 114 122
0.9 0.8 1.3 1.6
47,362 43,131 56,470 45,995
1.37 1.28 2.02 2.65
– 0.6883 0.0125 o0.0001
Incidence rates per 1,000 person-years.
L.-M. Krabbe et al. / Urologic Oncology: Seminars and Original Investigations 33 (2015) 65.e19–65.e25
References [1] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin 2013;63:11–30. [2] Brown T, Slack R, Rushton L. Occupational cancer in Britain. Urinary tract cancers: bladder and kidney. Br J Cancer 2012;107(Suppl. 1): S76–S84. [3] Burger M, Catto JW, Dalbagni G, Grossman HB, Herr H, Karakiewicz P, et al. Epidemiology and risk factors of urothelial bladder cancer. Eur Urol 2013;63:234–41. [4] Freedman ND, Silverman DT, Hollenbeck AR, Schatzkin A, Abnet CC. Association between smoking and risk of bladder cancer among men and women. J Am Med Assoc 2011;306:737–45. [5] Messing EM, Young TB, Hunt VB, Wehbie JM, Rust P. Urinary tract cancers found by homescreening with hematuria dipsticks in healthy men over 50 years of age. Cancer 1989;64:2361–7. [6] Britton JP, Dowell AC, Whelan P, Harris CM. A community study of bladder cancer screening by the detection of occult urinary bleeding. J Urol 1992;148:788–90. [7] Bangma CH, Loeb S, Busstra M, Zhu X, Bouazzaoui SE, Refos J, et al. Outcomes of a bladder cancer screening program using home hematuria testing and molecular markers. Eur Urol 2013;64(1):41–7. doi: 10.1016/j.eururo.2013.02.036. [8] Lotan Y, Elias K, Svatek RS, Bagrodia A, Nuss G, Moran B, et al. Bladder cancer screening in a high risk asymptomatic population using a point of care urine based protein tumor marker. J Urol 2009;182:52–7:[discussion 8]. [9] Hedelin H, Jonsson K, Salomonsson K, Boman H. Screening for bladder tumours in men aged 60–70 years with a bladder tumour marker (UBC) and dipstick-detected haematuria using both white-light and fluorescence cystoscopy. Scand J Urol Nephrol 2006;40:26–30. [10] Larre S, Catto JW, Cookson MS, Messing EM, Shariat SF, Soloway MS, et al. Screening for bladder cancer: rationale, limitations, whom to target, and perspectives. Eur Urol 2013;63(6):1049–58. doi: 10.1016/j.eururo.2012.12.062. [11] Prorok PC, Andriole GL, Bresalier RS, Buys SS, Chia D, Crawford ED, et al. Design of the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 2000;21:273S–309S. [12] Aberle DR, Berg CD, Black WC, Church TR, Fagerstrom RM, Galen B, et al. The National Lung Screening Trial: overview and study design. Radiology 2011;258:243–53.
65.e25
[13] Weissfeld JL, Schoen RE, Pinsky PF, Bresalier RS, Church T, Yurgalevitch S, et al. Flexible sigmoidoscopy in the PLCO cancer screening trial: results from the baseline screening examination of a randomized trial. J Natl Cancer Inst 2005;97:989–97. [14] Whitlock EP, Lin J, Liles E, Beil T, Fu R, O'Connor E, et al, editors. Screening for Colorectal Cancer: An Updated Systematic Review. Rockville, MD: Agency for Healthcare Research and Quality (US); 2008 [Report No. 08-05-05124-EF-1]. [15] Lughezzani G, Burger M, Margulis V, Matin SF, Novara G, Roupret M, et al. Prognostic factors in upper urinary tract urothelial carcinomas: a comprehensive review of the current literature. Eur Urol 2012;62:100–14. [16] Vickers AJ, Bennette C, Kibel AS, Black A, Izmirlian G, Stephenson AJ, et al. Who should be included in a clinical trial of screening for bladder cancer? a decision analysis of data from the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial. Cancer 2013;119:143–9. [17] Mir MC, Stephenson AJ, Grubb RL 3rd, Black A, Kibel AS, Izmirlian G. Predicting risk of bladder cancer using clinical and demographic information from prostate, lung, colorectal, and ovarian cancer (PLCO) screening trial participants. Cancer Epidemiol Biomarkers Prev 2013;22(12):2241–9. doi: 10.1158/1055-9965.EPI-130632. [18] Lotan Y, Svatek RS, Sagalowsky AI. Should we screen for bladder cancer in a high-risk population? a cost per life-year saved analysis. Cancer 2006;107:982–90. [19] Hartge P, Harvey EB, Linehan WM, Silverman DT, Sullivan JW, Hoover RN, et al. Unexplained excess risk of bladder cancer in men. J Natl Cancer Inst 1990;82:1636–40. [20] Zhang Y. Understanding the gender disparity in bladder cancer risk: the impact of sex hormones and liver on bladder susceptibility to carcinogens. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 2013;31:287–304. [21] Miyamoto H, Yang Z, Chen YT, Ishiguro H, Uemura H, Kubota Y, et al. Promotion of bladder cancer development and progression by androgen receptor signals. J Natl Cancer Inst 2007;99:558–68. [22] Johnson AM, O'Connell MJ, Miyamoto H, Huang J, Yao JL, Messing EM, et al. Androgenic dependence of exophytic tumor growth in a transgenic mouse model of bladder cancer: a role for thrombospondin1. BMC Urol 2008;8:7. [23] Johnson AM, O'Connell MJ, Messing EM, Reeder JE. Decreased bladder cancer growth in parous mice. Urology 2008;72:470–3.
