Cancer Causes Control DOI 10.1007/s10552-014-0393-3
ORIGINAL PAPER
Body mass index and weight change in men with prostate cancer: progression and mortality Stephanie E. Bonn • Fredrik Wiklund • Arvid Sjo¨lander Robert Szulkin • Pa¨r Stattin • Erik Holmberg • Henrik Gro¨nberg • Katarina Ba¨lter
•
Received: 18 October 2013 / Accepted: 28 April 2014 Ó Springer International Publishing Switzerland 2014
Abstract Purpose Body mass index (BMI) is a modifiable lifestyle factor that has been associated with an increased risk of fatal prostate cancer and biochemical recurrence. The main purpose of the present study was to investigate the association between the exposure BMI at the time of a prostate cancer diagnosis and weight change after diagnosis, and the outcomes of prostate cancer progression and mortality in a large cohort study. Methods Data from 4,376 men diagnosed with clinically localized prostate cancer between 1997 and 2002 were analyzed. BMI and weight change were self-reported in 2007. Hazard ratios (HRs) with 95 % confidence intervals (CIs) were estimated in complete-case analysis (n = 3,214) using Cox proportional hazards models. Results Progression was experienced among 639 (14.6 %) of the study participants, and in total, 450
(10.3 %) deaths of any cause and 134 (3.1 %) prostate cancer-specific deaths were recorded during follow-up. Obese men had a 47 % increased rate of overall mortality compared to normal weight men (HR 1.47, 95 % CI 1.03–2.10). No statistically significant associations were found for BMI and prostate cancer progression or prostate cancer-specific mortality. A weight loss [5 % after diagnosis almost doubled the rate of overall mortality compared to maintaining a stable weight (HR 1.94, 95 % CI 1.41–2.66), while a weight gain [5 % was associated with an almost doubled increased rate of prostate cancer-specific mortality (HR 1.93, 95 % CI 1.18–3.16). Conclusions Being obese was associated with an increased rate of overall mortality, and gaining weight after a prostate cancer diagnosis was associated with an increased rate of prostate cancer-specific mortality. Keywords Prostatic neoplasms Disease progression Mortality Body mass index Weight change Epidemiology
Introduction S. E. Bonn (&) F. Wiklund A. Sjo¨lander R. Szulkin H. Gro¨nberg K. Ba¨lter Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Nobels va¨g 12A, 171 77 Stockholm, Sweden e-mail: [email protected] P. Stattin Department of Surgical and Perioperative Sciences, Urology and Andrology, Umea˚ University, Umea˚, Sweden E. Holmberg Department of Oncology, Institute of Clinical Sciences, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden
In Sweden, one in eight men younger than 75 years of age is diagnosed with prostate cancer, and while the 5-year survival is approximately 90 %, the risk of a relapse among men undergoing curative treatment is almost 30 % [1]. The increasing incidence of prostate cancer, together with the concomitant decrease in mortality rates, has led to increasing numbers of men living with a previous prostate cancer diagnosis. Greater utilization of prostate-specific antigen (PSA) testing has led to the detection of not only clinically relevant tumors, but also of slow growing tumors that
123
Cancer Causes Control
otherwise would have escaped diagnosis [2]. Since many men are diagnosed at an early stage, a large proportion have indolent cancer that may never progress to a clinical tumor [3] and active surveillance instead of invasive treatment is an increasingly common option among men with low-risk tumors [4]. Thus, it is important to find complementary ways to influence cancer progression. Body mass index (BMI) is a modifiable lifestyle factor with the potential to reduce or delay progression of prostate cancer in both treated and untreated men [5]. BMI is an important proxy for body fat and is used in both clinical and research settings where overweight and obesity are commonly defined as BMI being C25 or C30 kg/m2, respectively [6]. With the prevalence of overweight and obesity increasing worldwide [7], and the co-occurring increase in prostate cancer incidence, it is highly relevant to clarify the possible association between overweight and obesity and prostate cancer. Results from the previous studies have shown no association between BMI and the overall risk of prostate cancer, while men with a high BMI were more likely to be diagnosed with aggressive or fatal prostate cancer [8–10]. This suggests that BMI may be of greater importance for prostate cancer progression than for the risk of developing the disease [11]. Previous studies have also found a high BMI to be associated with an increased risk of biochemical recurrence [12–18] and an increased risk of prostate cancer-specific mortality [18–23]. Weight change is another lifestyle factor with potential to affect both prostate cancer risk and survival. Results are similar to those for BMI, and while previous studies of weight gain during adulthood have shown no association with the incidence of prostate cancer, an increased risk of aggressive prostate cancer and prostate cancer mortality was found [23, 24]. The outcome after a prostate cancer diagnosis might not only be affected by BMI at the time of prostate cancer diagnosis but also by a potential weight gain thereafter. Two previous studies have found an increased risk of biochemical recurrence among men gaining weight in the years before a radical prostatectomy [25, 26]. However, to our knowledge, there are no previous studies investigating the association between weight gain after a prostate cancer diagnosis and survival. We aimed to study BMI at the time of prostate cancer diagnosis and weight change after diagnosis in relation to prostate cancer progression and mortality in a large cohort of Swedish men with localized prostate cancer. Results from this study may guide future studies regarding whether BMI, weight maintenance, or weight change after a prostate cancer diagnosis can influence the survival of prostate cancer patients.
123
Methods Study design All incident cancers in Sweden are reported to the Swedish National Cancer Register (NCR) [27]. For prostate cancer specifically, the National Prostate Cancer Registry (NPCR) of Sweden includes all patients with a prostate cancer diagnosis [28] and holds additional information of serum PSA levels; tumor, node, metastasis (TNM) stage; tumor differentiation at the time of diagnosis; and primary treatment within 6 months of diagnosis. Study participants in the present study (PROCAP, PROgression in CAncer of the Prostate) were derived from a retrospective nationwide cohort study of patients with localized prostate cancer, the NPCR of Sweden Follow-up Study, which has previously been described in detail [4]. In brief, patients were eligible for inclusion in the NPCR of Sweden Follow-up Study if they were registered with a localized prostate cancer in the NPCR between 1 January 1997 (1 January 1998 in one region) and 31 December 2002; B70 years of age at diagnosis; diagnostic serum PSA \ 20 ng/ml; local tumor stage T1–T2; and no signs of lymph node metastasis (NX or N0) or bone metastasis (MX or M0). Out of the 8,304 patients fulfilling the criteria, 7,960 (96 %) accepted inclusion to the study. PROCAP is an extension of the NPCR of Sweden Follow-up Study and has previously been described elsewhere [29]. In short, all patients in the NPCR of Sweden Follow-up Study who were still alive in 2007 (n = 7,074) were eligible for inclusion in PROCAP and invited to respond to a questionnaire assessing lifestyle factors and to donate a blood sample for genetic analysis. In total, 5,779 (82 %) patients answered the questionnaire and were included. However, patients with unknown primary treatment or missing progression data (n = 880) were excluded from the analysis as the definition of disease progress is dependent on primary treatment. Further, patients primarily treated with hormones (n = 131), patients with unknown date of last record in medical journals or a missing date of termination of deferred treatment (n = 29), and patients with completely missing questionnaire data or incomplete data on self-reported BMI (n = 363) were excluded from the analysis. In total, 4,376 patients were included in further analysis. All patients included in PROCAP gave their written informed consent for participation. The study has been approved by the research ethics committee at Karolinska Institutet. End points With a median time period of 4 years after diagnosis, research nurses in each of six health care regions in
Cancer Causes Control
Sweden extracted information at one occasion for each patient on the date of last follow-up, reason for and date of termination of surveillance, subsequent PSA testing, and signs of local progress and distant metastasis from medical records. Among patients treated with curative intent (radiation therapy or radical prostatectomy), either having biochemical recurrence, local progress, or distant metastasis, was considered disease progress. However, biochemical recurrence in the present study was defined differently depending on primary treatment and has previously been described in detail [29]. Briefly, biochemical recurrence was defined as a doubling in PSA above the post-treatment nadir value and exceeding at least 1 ng/ml for patients who underwent radiation therapy. For patients treated with radical prostatectomy, biochemical recurrence was defined as two consecutive tests with PSA levels [0.2 ng/ml; the date for this event was set to the first of the two occasions. For operated patients with only one registered PSA test value, a value [0.5 ng/ml was considered biochemical recurrence. Further, the event of termination of deferred treatment with biochemical progression as a reason for termination defined disease progression in patients on surveillance. The time to event was defined as the time from the prostate cancer diagnosis to the earliest observed progressive event for each treatment-specific definition of progressive events. Patients without disease progression were censored at the last date of follow-up recorded in the medical journal, whereas patients on surveillance who ended their deferred treatment without any signs of disease progression were censored at the recorded date of termination. In total, 639 men (14.6 %) experienced a progress event during the follow-up, with 214 (13.2 %), 355 (15.6 %), and 70 (14.5 %) progression events recorded among men with a BMI \ 25, 25–30, and [30 kg/m2, respectively. The median time to disease progression was 2.3 years, and the median time to censoring was 3.9 years. Cause of death, prostate cancer specific or other cause, and date of event were obtained from the Swedish Causeof-Death registry using national identification numbers. Time to event was defined as time from prostate cancer diagnosis to the date of death reported in the registry. All patients were left-truncated by study design at the date of inclusion to PROCAP, since only patients alive at this date could be included in the analysis. Patients still alive on 8 June 2012 were censored at this date. The median survival times for overall mortality, prostate cancer-specific mortality, or censoring were 10.1, 9.9, and 11.6 years, respectively. Lifestyle factors Lifestyle factors were assessed in a questionnaire sent out to patients between January 2007 and June 2008. Patients
chose to respond to the questionnaire either in paper format (50 %) or via the Web. Data from the Web-based questionnaires were directly saved in digital format while paperbased questionnaires were scanned into digital format after being checked for completeness by study personnel. Patients self-reported their current height, weight, and weight change since diagnosis in the questionnaire. Weight change was reported as no change in weight or as a weight increase or decrease from the time of prostate cancer diagnosis. In the case of weight increase or decrease, how many kilograms that was gained or lost was reported. BMI (kg/m2) at diagnosis was estimated based on self-reported current height and weight, and weight change since diagnosis. Normal weight, overweight, and obesity were defined according to the National Institute of Health as normal weight if BMI \ 25 kg/m2, overweight if BMI 25 \ 30 kg/m2, and obese if BMI C 30 kg/m2 [6]. In what follows, we use ‘‘BMI’’ as shorthand for ‘‘estimated BMI at diagnosis.’’ Based on reported current weight and weight change and the distribution of data, patients were categorized into three groups: no change or a change B5 %, an increase [5 % or a decrease [5 % since diagnosis. Additional lifestyle factors considered as potential confounding factors were also assessed in the self-reported questionnaire. Tobacco users were defined as ‘‘current,’’ ‘‘former,’’ or ‘‘never’’ smokers based on reported smoking habits since the time of prostate cancer diagnosis. Physical activity at the age of 50 and after diagnosis was defined as total metabolic equivalent time (MET-h) based on questions previously validated by Norman et al. [30]. Total energy intake was assessed by a food frequency questionnaire similar to that previously validated by Messerer et al. [31] asking specifically about food habits since the time of prostate cancer diagnosis. Further, variables of overall stress during the last year, education level, current occupation, and family history of prostate cancer were assessed. Statistical methods Distributions and means of demographic and clinical variables were studied across BMI categories. Statistically significant associations were tested for using one-way ANOVA for continuous variables and chi-square test for categorical variables. Progression-free, overall, and prostate cancer-specific survivals were analyzed using the Kaplan–Meier method. Time to event for the different BMI and weight change categories was compared using log-rank test. Life tables were used to assess the 5-year survival. Cox proportional hazards models were used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95 % confidence intervals (95 % CIs) in the analysis of progression and mortality [32]. All three models included only complete cases, i.e., participants with
123
Cancer Causes Control
information on all covariates included in the most adjusted model. Time since prostate cancer diagnosis was used as the underlying timescale. BMI and weight change were included as categorical exposures in the Cox proportional hazards models. To examine whether men who had reported a change in weight due to illness were influencing the results, we carried out additional sensitivity analysis with 18-months lag time. Model-free (i.e., not estimated using a Cox proportional hazards model) HRs and CIs were assessed using the statistical software R. To assess whether measured covariates should be considered as potential confounding factors and adjusted for in the Cox proportional hazards models, we tested whether the covariates were statistically associated with both the exposure (BMI and weight change) and the outcome (progression and mortality). The association between covariates and the exposure was assessed using linear regression models, treating BMI as continuous. The association between the covariates and the outcome was assessed using Cox proportional hazards models. Covariates tested were age at diagnosis; MET-h at age 50; education level; smoking habits after diagnosis; family history of prostate cancer; PSA level at diagnosis; T-, N-, and M-stages; tumor grade and Gleason score at diagnosis; and primary treatment. In mortality analysis, a potential confounding effect of MET-h after diagnosis and stress level after diagnosis was also evaluated. In the analysis of BMI, confounding by weight change was tested for. Correspondingly, in the analysis of weight change, confounding of BMI was tested for. Based on this confounder selection, multivariable-adjusted models for progression, overall mortality, and prostate cancer-specific mortality included age at diagnosis (5-year categories), mode of primary treatment (curative intent, radical prostatectomy, or radiation therapy), and Gleason score (\6, 6, [6). We also present results from fully adjusted models that were further adjusted for PSA at diagnosis, clinical stage (T1 or T2), N-stage (N0, N1, or NX), M-stages (M0 or MX), smoking habits (current, former, never), and MET-h at age 50. Interaction between the two exposures and all confounding factors were tested for using likelihood ratio tests. The Cox proportional hazards assumption was tested using Schoenfelds residuals, and no statistically significant deviation from the proportional hazards assumption was detected. The level of significance was set to p \ 0.05. All analyses were performed using STATA 12.1 (STATA Corporation, College Station, TX).
Results In total, 4,376 men were included in the analysis. The median time from prostate cancer diagnosis to inclusion in
123
the study was 7.25 years with an interquartile range of 2.15 years. Among the included men, 37 % were normal weight (BMI \ 25 kg/m2), 52 % of the men were overweight (BMI 25 \ 30 kg/m2), and 11 % were obese (BMI C 30 kg/m2) at diagnosis. Among those classified as normal weight, eight men had a BMI \ 18.5 kg/m2. The mean BMI at diagnosis was 26.2 ± 3.2 kg/m2. In total, 1,040 (23.8 %) men had reported a weight change greater than 5 % after diagnosis; 649 (14.8 %) had reported an increase and 391(8.9 %) a decrease. The mean (±SD) age at diagnosis was 63.0 ± 5.1, and overweight and obese men were statistically significantly younger at diagnosis compared to normal weight men. There were also statistically significant differences between the BMI categories with regard to reported MET-h at age 50 and after diagnosis with normal weight, overweight, and obese men having reported 23.5 ± 7.6, 24.7 ± 8.3, and 24.4 ± 8.5 MET-h, respectively, at age 50 and corresponding MET-h of 24.0 ± 7.0, 24.0 ± 7.0, and 22.8 ± 7.3 after diagnosis. There were also statistically significant differences between the BMI categories with regard to weight change, education, smoking, and M-stage at diagnosis. Men in the higher BMI categories had reported weight change, both increase and decrease, to a larger extent than normal weight men. Higher BMI also seemed to be associated with a lower level of education. Normal weight men were more likely to be never or current smokers compared to overweight and obese men that reported to be former smokers to a greater extent. Regarding M-stage at diagnosis, the number of men classified as MX was increased in the two higher BMI categories compared to among normal weight men. The mean PSA value at diagnosis was 8.3 ± 4.2 ng/ ml; there was no statistically significant difference in PSA between the BMI categories. Characteristics of participants are shown in Table 1. Progression was experienced among 639 (14.6 %) of the study participants. In total, 450 (10.3 %) deaths of any cause and 134 (3.1 %) prostate cancer-specific deaths were recorded. The prostate cancer progression, overall mortality, and prostate cancer-specific mortality rates in the cohort were 40.2, 24.0, and 7.1 per 1,000 person-years, respectively. Table 2 displays model-free mortality rate ratios comparing BMI categories for overall and prostate cancer-specific mortality, and includes the number of events, person-years, and corresponding mortality rates stratified on weight change. Kaplan–Meier curves with log-rank analysis showed a statistically significant difference in progression (Fig. 1), overall mortality, and prostate cancer-specific mortality (Fig. 2). Although no clear pattern between the BMI categories is seen for progression, the survival among obese men appears different from that over normal and overweight men for overall and prostate cancer-specific
Cancer Causes Control Table 1 Characteristics of study participants included in the analyses in the PROCAP study divided by body mass index (BMI) at diagnosis BMI (kg/m2) at diagnosis All (n = 4,376) n (%)
pc
\25 (n = 1,617) n (%)
25 \ 30 (n = 2,276) n (%)
[30 (n = 483) n (%)
Age at diagnosis (years) \55
0.035 320 (7.3)
105 (6.5)
174 (7.6)
41 (8.5)
55 \ 60 60 \ 65
877 (20.0) 1,385 (31.7)
296 (18.3) 514 (31.8)
471 (20.7) 720 (31.6)
110 (22.8) 151 (31.3)
65 \ 70
1,560 (35.7)
600 (37.1)
794 (34.9)
166 (34.4)
234 (5.4)
102 (6.3)
117 (5.1)
15 (3.1)
3,336 (76.2)
1,334 (82.5)
1,714 (75.3)
288 (59.6)
[5 % Increase
649 (14.8)
199 (12.3)
369 (16.2)
81 (16.8)
[5 % Decrease
391 (8.9)
84 (5.2)
193 (8.5)
114 (23.6)
C70 Weight change since diagnosisa No change or B5 % change
0.000
Tumor stage
0.678
T1a
209 (4.8)
72 (4.5)
111 (4.9)
T1b
99 (2.3)
42 (2.6)
43 (1.9)
26 (5.4) 14 (2.9)
T1c
2,234 (51.1)
822 (50.8)
1,168 (51.3)
244 (50.5)
T2
1,834 (41.9)
681 (42.1)
954 (41.9)
199 (41.2)
Gleason scoreb
0.062
\6
1,098 (25.1)
386 (23.9)
600 (26.4)
112 (23.2)
6 [6
1,794 (41.0) 811 (18.5)
683 (42.2) 275 (17.0)
916 (40.2) 429 (18.8)
195 (40.4) 107 (22.2)
1,456 (33.3)
543 (33.6)
757 (33.3)
156 (32.3)
N-classification
0.575
N0 N1
25 (0.6)
NX
2,854 (65.2)
Missing data
41 (0.9)
10 (0.6) 1,045 (64.6) 19 (1.2)
10 (0.4) 1,490 (65.5) 19 (0.8)
5 (1.0) 319 (66.0) 3 (0.6)
M-classification
0.012
M0
2,213 (50.6)
852 (52.7)
1,143 (50.2)
218 (45.1)
MX
2,133 (48.7)
754 (46.6)
1,117 (49.1)
261 (54.2)
Missing data
30 (0.7)
11 (0.7)
16 (0.7)
3 (0.6)
409 (25.3)
551 (24.2)
111 (23.0)
Primary treatment
0.068
Surveillance
1,071 (24.5)
Radical prostatectomy
2,390 (54.6)
884 (54.7)
1,259 (55.3)
247 (51.1)
915 (20.9)
324 (20.0)
466 (20.5)
125 (25.9)
1,851 (42.3) 2,141 (48.9)
766 (47.4) 684 (42.3)
929 (40.8) 1,159 (50.9)
156 (32.3) 298 (61.7)
344 (7.9)
154 (9.5)
165 (7.2)
25 (5.2)
40 (0.9)
13 (0.8)
23 (1.0)
4 (0.8)
565 (34.9)
972 (42.7)
240 (49.7)
Radiation therapy Smoking
a
0.000
Never smoker Past smoker Current smoker Missing data a
Education level B9 years
[9 B 12 years
1,563 (35.7)
579 (35.8)
829 (36.4)
155 (32.1)
[12 years
1,008 (23.0)
462 (28.6)
462 (20.3)
84 (17.4)
11 (0.7)
14 (0.6)
4 (0.8)
Missing data a
0.000 1,777 (40.6)
28 (0.6)
Self-reported in PROCAP
b
Missing data from n = 673
c
p values from chi-square test
123
Cancer Causes Control Table 2 Model-freea hazard ratios (HRs) of overall and prostate cancer-specific mortality by body mass index (BMI) at diagnosis and selfreported weight change since diagnosis BMI \ 25 kg/m2
BMI 25 \ 30 kg/m2
BMI C 30 kg/m2
No. of events
Personyears
Events/1,000 person-years
No. of events
Personyears
Events/1,000 person-years
No. of events
Personyears
Events/1,000 person-years
121
5,762.1
21.0
157
7,357.0
21.3
34
1,214.2
28.0
Mortality Overall No weight change Weight increase [5 %
23
851.6
27.0
33
1,594.5
20.7
10
344.8
29.0
Weight decrease [5 %
16
342.3
46.7
35
796.4
43.9
21
454.7
46.2
160
6,956.0
23.0
225
9,747.9
23.1
65
2,013.7
32.3
Total
HR (95 % CI)
HR (95 % CI)
HR (95 % CI)
1.00
1.00 (0.82–1.23)
1.40 (1.05–1.87)
Prostate cancer specific No weight change
30
5,762.1
5.2
47
7,357.0
6.4
12
1,214.2
9.9
Weight increase [5 %
12
851.6
14.1
10
1,594.5
6.3
7
344.8
20.3
Weight decrease [5 % Total
2
342.3
5.8
10
796.4
12.6
4
454.7
8.8
44
6,956.0
6.3
67
9,747.9
6.9
23
2,013.7
11.4
HR (95 % CI) 1.00 a
HR (95 % CI) 1.09 (0.74–1.59)
HR (95 % CI) 1.81 (1.10–2.97)
Not estimated from a Cox proportional hazards model
mortality. Five years after diagnosis, 83.7 % of the normal weight men were progression-free compared to 80.4 and 80.9 % among the overweight and obese men, respectively. The 5-year overall survival after inclusion to PROCAP was 89.5, 89.3, and 85.8 % for normal weight, overweight, and obese men, respectively, while the corresponding prostate cancer-specific survival was 97.0, 96.7, and 94.7 %. For men in different weight change categories, Kaplan– Meier curves with log-rank analysis showed a statistically significant difference in both overall and prostate cancerspecific mortality (Fig. 3). The 5-year overall survival after inclusion to PROCAP was 90.0, 89.0, and 80.0 % for men with stable weight, men who gained weight, and men who lost weight, respectively. The corresponding prostate cancer-specific survival was 97.0, 95.2, and 95.6 %. Model-based unadjusted and adjusted HRs comparing BMI categories and BMI as continuous variable for progression, overall, and prostate cancer-specific mortality from Cox proportional hazard regression models included only complete cases (n = 3,214) and are shown in Table 3. There were no statistically significant differences in progression rates between the BMI categories, although point estimates indicated higher rates among overweight and obese men compared to normal weight. When analyzing BMI as a continuous exposure, there was a 3 % increased rate of progression per one unit increase in BMI in multivariate-adjusted models, HR 1.03 (95 % CI 1.00–1.06). While the point estimates were similar per BMI unit for overall and prostate cancer-specific mortality, the CIs
123
Fig. 1 Kaplan–Meier survival curve for progression
showed only borderline significance. While obese men had no increased rate of progression, they had an increased rate of overall mortality, HR 1.47 (95 % CI 1.03–2.10). No statistically significant association was found between BMI categories and prostate cancer-specific mortality. Men who experienced a weight reduction [5 % after diagnosis had an almost doubled rate of overall mortality compared to men with a stable weight, HR 1.94 (95 % CI 1.41–2.66) (Table 4). Weight reduction was not statistically significantly associated with prostate cancer-specific mortality, HR 1.56 (95 % CI 0.80–3.05). When men who had died within 18 months after inclusion to the study were excluded in the sensitivity analysis, the association for
Cancer Causes Control
Fig. 2 Kaplan–Meier survival curves for mortality and BMI categories. On the x-axis, time from inclusion in PROCAP to death or censoring is shown, origin (time = 0) is the date of diagnosis in lefttruncated Cox proportional hazard regressions
Fig. 3 Kaplan–Meier survival curves for mortality and weight change categories. On the x-axis, time from inclusion in PROCAP to death or censoring is shown, origin (time = 0) is the date of diagnosis in left-truncated Cox proportional hazard regressions
overall mortality remained statistically significant while the HR point estimate was slightly attenuated, HR 1.67 (95 % CI 1.14–2.46). For prostate cancer-specific mortality, the point estimate was attenuated and statistically nonsignificant, HR 1.07 (95 % CI 0.45–2.55). While a weight increase [5 % was not associated with overall mortality, HR 1.14 (95 % CI 0.82–1.58), men who increased in weight after diagnosis had a statistically significantly increased rate of prostate cancer-specific mortality, HR 1.93 (95 % CI 1.18–3.16). In the sensitivity analysis, the results for overall mortality remained statistically nonsignificant, HR 1.29 (95 % CI 0.91–1.83), while the results for prostate cancer-specific mortality remained statistically significant, HR 1.97 (95 % CI 1.16–3.35).
overall mortality compared to normal weight men. We also found that a weight loss[5 % after diagnosis increased the overall mortality rate compared to maintaining a stable weight, whereas weight gain was associated with an increased rate of prostate cancer-specific mortality. Established risk factors of prostate cancer include age, ethnicity, and heredity [2], while lifestyle factors, including BMI, are still debated. A recent meta-analysis by Discacciati et al. [33] showed that high BMI was associated with a decreased risk of localized prostate cancer but an increased risk of advanced prostate cancer, indicating a dual effect of BMI on prostate cancer risk. More recently, attention has shifted from studies of risk to studies regarding prostate cancer progression and mortality. A majority of previous studies, but not all [34], have found increasing BMI to be associated with an increased risk of biochemical recurrence [12–18]. High BMI has also been associated with a higher risk of prostate cancer-specific mortality in several studies [18–23]. A review and meta-analysis by Cao and Ma [18] showed increased risks of both biochemical recurrence and
Discussion In this large cohort study of men with localized prostate cancer, we found that obese men had an increased rate of
123
Cancer Causes Control Table 3 Model-based age-adjusted and multivariable-adjusted hazard ratios (HRs) from complete-case analysis (n = 3,214) according to body mass index (BMI) at the time of prostate cancer diagnosis obtained through Cox proportional hazards models No. of events
BMI (kg/m2) categories
n
\25 (ref) HR
25 \ 30 HR (95 % CI)
C30 HR (95 % CI)
HR (95 % CI)
Adjusteda
1.00
1.15 (0.94–1.41)
1.07 (0.78–1.46)
1.02 (1.00–1.06)
Adjustedb
1.00
1.15 (0.94–1.40)
1.06 (0.77–1.46)
1.02 (1.00–1.05)
1.00
1.13 (0.92–1.38)
1.11 (0.81–1.53)
1.03 (1.00–1.06)
Adjusteda
1.00
1.03 (0.80–1.31)
1.42 (1.00–1.31)
1.02 (0.99–1.06)
Adjustedb
1.00
1.03 (0.81–1.32)
1.37 (0.97–1.95)
1.02 (0.98–1.05)
Adjustedc
1.00
1.06 (0.83–1.36)
1.47 (1.03–2.10)
1.02 (0.99–1.06)
Adjusteda
1.00
0.98 (0.63–1.51)
1.30 (0.68–2.45)
1.03 (0.97–1.10)
Adjustedb
1.00
0.97 (0.63–1.50)
1.16 (0.61–2.21)
1.02 (0.96–1.08)
Adjustedc
1.00
0.96 (0.62–1.50)
1.29 (0.67–2.48)
1.02 (0.96–1.09)
Measure
Progression
Adjusted
455
c
Overall mortality
311
Prostate cancer-specific mortality
a
Per unit of BMI (kg/m2)
96
Adjusted for age at diagnosis (5-year categories)
b
Adjusted for age at diagnosis, primary treatment (curative intent, radical prostatectomy, or radiation therapy), and Gleason score at diagnosis (\6, 6, or [6)
c
Adjusted for age at diagnosis; primary treatment; Gleason score at diagnosis; PSA level; T-, N-, and M-stages at diagnosis; smoking habits (current, former, never); and total MET-h at age 50
Table 4 Model-based age-adjusted and multivariable-adjusted hazard ratios (HRs) from complete-case analysis (n = 3,214) according to weight change categories at the time of prostate cancer diagnosis obtained through Cox proportional hazards models No. of events
No change (ref) HR
Category of weight change
Adjusteda Adjustedb Adjustedc
Increase [5 %
Decrease [5 %
HR (95 % CI)
HR (95 % CI)
1.00
1.19 (0.87–1.64)
2.11 (1.56–1.64)
1.00
1.17 (0.85–1.61)
2.11 (1.54–2.88)
1.00
1.14 (0.82–1.58)
1.94 (1.41–2.66)
Adjusteda
1.00
2.05 (1.27–3.31)
1.51 (0.79–2.87)
Adjustedb
1.00
1.92 (1.18–3.12)
1.63 (0.84–3.14)
Adjustedc
1.00
1.93 (1.18–3.16)
1.56 (0.80–3.05)
n Overall mortality
Prostate cancerspecific mortality
a
311
96
Adjusted for age at diagnosis (5-year categories)
b
Adjusted for age at diagnosis, primary treatment (curative intent, radical prostatectomy, or radiation therapy), Gleason score at diagnosis (\6, 6, or [6), and BMI category at the time of diagnosis
c
Adjusted for age at diagnosis; primary treatment; Gleason score at diagnosis; BMI category at the time of diagnosis; PSA level; T-, N-, and M-stages at diagnosis; smoking habits (current, former, never); and total MET-h at age 50
123
prostate cancer-specific mortality with estimates of a 15–21 % increased risk per 5 kg/m2 increase in BMI. In contrast to many previous findings, our results showed no association between BMI and rate of progression during a median follow-up of 4 years or prostate cancer-specific mortality. However, we found a positive association between high BMI and overall mortality, in contrast to a previous study where no association between BMI and overall or non-prostate cancer-specific mortality was seen [20]. High weight gain during adulthood has been associated with an increased risk of aggressive and fatal prostate cancer [23, 24]. Larger weight gain from the age of 25 years has also been associated with an increased risk of biochemical recurrence after prostate cancer diagnosis [15]. Few studies have investigated the association between weight change after diagnosis and survival, but Joshu et al. [25] showed that weight gain 5 years before and 1 year after a prostate cancer diagnosis was associated with an almost doubled risk of prostate cancer recurrence and Whitley et al. [26] found that a weight gain C2.5 kg in the year preceding a radical prostatectomy increased the risk of biochemical recurrence by 65 % compared to men who gained \2.5 kg in weight. In line with the previous findings, we showed an increased rate of prostate cancer-specific mortality among men who experienced weight gain [5 % after diagnosis. The fact that the increased rate
Cancer Causes Control
remained even after excluding men who died within the first 18 months after inclusion to the study strengthens our finding since the weight increase is less likely to have occurred due to other illness. The increased rates seen among men who decreased in weight were, on the contrary, slightly attenuated in the analysis of overall mortality and disappeared in the analysis of prostate cancer-specific mortality, indicating that the association seen for weight decrease might in fact be due to weight loss because of illness. Nevertheless, our results are based on self-reported weight change and no information on whether or not the change was intentional was available. Intentional weight change may have different effects than unintended weight changes, which may be due to something else, e.g., other diseases, health issues, or malnutrition. The effect of intentional weight change after a prostate cancer diagnosis needs to be studied further. One frequently proposed explanation to the observed association between BMI and prostate cancer mortality is a delayed diagnosis among obese men, leading to a potentially more advanced disease at diagnosis and thereby poorer survival. Obesity-related plasma hemodilution, generating lower PSA values, may be one explanation for a delayed diagnosis and poorer outcome among obese patients [35]. However, high BMI has been associated with a higher risk of prostate cancer-specific mortality both before and during the PSA screening era [21], indicating that hemodilution of PSA values does not fully explain the association. Nor do we believe that a more advanced disease among overweight and obese subjects in our study fully explains the associations found between high BMI and progression and mortality since all men in our cohort were diagnosed with localized prostate cancer and there was no statistically significant difference in PSA values, tumor stage, or grade between the different BMI categories. Several possible mechanisms linking adiposity and prostate cancer have been proposed. The main hypothesis involves hormonal and metabolic changes including pathways for insulin and insulin-like growth factors (IGFs), sex hormones, and adipokine signaling [36]. High BMI is associated with hyperinsulinemia that, through reduction of IGF-binding proteins, leads to increased serum levels of IGF-I, which has been shown to promote tumor development [37]. Obesity is also associated with decreased androgen levels and lower serum testosterone levels in men, the latter having been linked to an increased risk of aggressive prostate tumors [38]. Further, altered levels of adipokines with increased levels of leptin and decreased levels of adiponectin are a result of obesity and have been associated with tumor development [36]. Hormonal and metabolic pathways that are altered by the level of adiposity might also be affected by weight
change once a tumor has developed and could be one explanation for the increased rate of prostate cancer-specific mortality we found among men who increased in weight from the time of diagnosis. While BMI at the time of diagnosis is a fixed factor for each patient once diagnosed, weight maintenance or change after a diagnosis has the potential to complement treatment as it can be influenced by the patient himself. This study has several strengths including a populationbased design with a large sample size. An additional strength is the complete follow-up regarding disease progression during on average the first 4 years of follow-up and survival. The short follow-up times for progression limit our ability to study the long-term effects of BMI on progression. However, among patients with localized prostate cancer, disease progression is an early event and we believe that the limited follow-up time would only have minor effects on our observed results. The condition of being alive up to 5–10 years after diagnosis in order to be eligible for inclusion in the PROCAP study is a major limitation in our study. However, it can be argued that any bias induced by this left truncation is likely to act in opposite direction as a true causal effect, thus resulting in conservative estimates. Although the response rate was high, we were only able to study progression and mortality among patients surviving long enough to participate in PROCAP. Nonetheless, the survival rate among men with prostate cancer is almost 70 % during the first 10 years after diagnosis in Sweden [1]. The inability to include the most severe cases, patients that die early after a prostate cancer diagnosis may partly explain why we, in contrast to other studies, found no association between BMI and prostate cancer progression and mortality. The retrospective self-report of height, weight, and change in weight is another limitation. However, previous validation studies have shown a strong correlation between self-reported and measured weight, although the long-term recall is influenced by current weight [39]. Also, it seems plausible that the outcome in our study (progression/mortality) had no major effect on self-reported measures, and thus any misclassification error is likely to be non-differential (i.e., bias toward a null effect). Even though BMI is often used as a proxy for body fat, it is not without limitations. It does not factor in body composition (fat mass vs. fat free mass) and fat distributions (superficial vs. visceral fat). Subjects with high muscle mass and low fat mass may be incorrectly classified as overweight or obese [40]. Limitations in BMI as an exposure may be a reason for differences between study outcomes. A previous study investigating anthropometric differences among obese men showed that the fat distribution among men experiencing biochemical failure was different from that of men who did not, despite similar weights and BMI; the authors
123
Cancer Causes Control
concluded that the differences in fat distribution might account for conflicting results regarding BMI and prostate cancer [41]. In conclusion, we found an increased rate of overall mortality among obese men and men who lost a large amount of weight. Men who increased in weight following diagnosis had an increased rate of prostate cancer-specific mortality. Although not conclusive, our findings indicate a positive association between high BMI and prostate cancer progression and mortality. Future studies investigating the role of adiposity and weight maintenance or change in relation to prostate cancer progression and mortality are needed to clarify the association. Acknowledgments We thank all of the participants in the PROCAP study. We also thank Carin Cavalli-Bjo¨rkman and Ami Ro¨nnberg for their work during data collection and Michael Broms for his work with the databases. We also acknowledge the NPCR steering group and all the research nurses who extracted data for the follow-up study. The present study was supported by the Swedish Cancer Society, Grant No. CAN 2011/868, and the Swedish research Council for Health, Working life and Welfare, Grant No. 2011-0650. Conflict of interest
The authors declare no conflict of interests.
References 1. Bergman O, Hont G, Johansson E (ed) (2013) The National Board of Health and Welfare, Stockholm, Sweden. http://www. socialstyrelsen.se/Lists/Artikelkatalog/Attachments/19108/20136-5.pdf. Accessed 18 Oct 2013 (In Swedish: Cancer i siffror 2013) 2. Jemal A, Bray F, Center MM et al (2011) Global cancer statistics. CA Cancer J Clin 61(2):69–90 3. Meyerhardt JA, Ma J, Courneya KS (2010) Energetics in colorectal and prostate cancer. J Clin Oncol 28(26):4066–4073 4. Stattin P, Holmberg E, Bratt O et al (2008) Surveillance and deferred treatment for localized prostate cancer. Population based study in the National Prostate Cancer Register of Sweden. J Urol 180(6):2423–2429 (discussion 2429–2430) 5. Ornish D, Weidner G, Fair WR et al (2005) Intensive lifestyle changes may affect the progression of prostate cancer. J Urol 174(3):1065–1070 6. Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults (1998) Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: executive summary. Am J Clin Nutr 68(4):899–917 7. World Health Organization (2012) World Health Statistics 2012. http://www.who.int/gho/publications/world_health_statistics/EN_ WHS2012_Full.pdf. Accessed 18 Oct 2013 8. Giovannucci E, Liu Y, Platz EA, Stampfer MJ, Willett WC (2007) Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer 121(7):1571–1578 9. Rodriguez C, Freedland SJ, Deka A et al (2007) Body mass index, weight change, and risk of prostate cancer in the Cancer Prevention Study II Nutrition Cohort. Cancer Epidemiol Biomarker Prev 16(1):63–69 10. Stocks T, Hergens MP, Englund A, Ye W, Stattin P (2010) Blood pressure, body size and prostate cancer risk in the Swedish Construction Workers cohort. Int J Cancer 127(7):1660–1668
123
11. Mistry T, Digby JE, Desai KM, Randeva HS (2007) Obesity and prostate cancer: a role for adipokines. Eur Urol 52(1):46–53 12. Efstathiou JA, Chen MH, Renshaw AA, Loffredo MJ, D’Amico AV (2007) Influence of body mass index on prostate-specific antigen failure after androgen suppression and radiation therapy for localized prostate cancer. Cancer 109(8):1493–1498 13. Palma D, Pickles T, Tyldesley S (2007) Obesity as a predictor of biochemical recurrence and survival after radiation therapy for prostate cancer. BJU Int 100(2):315–319 14. Strom SS, Kamat AM, Gruschkus SK et al (2006) Influence of obesity on biochemical and clinical failure after external-beam radiotherapy for localized prostate cancer. Cancer 107(3):631–639 15. Strom SS, Wang XM, Pettaway CA et al (2005) Obesity, weight gain, and risk of biochemical failure among prostate cancer patients following prostatectomy. Clin Cancer Res 11(19):6889–6894 16. Bassett WW, Cooperberg MR, Sadetsky N et al (2005) Impact of obesity on prostate cancer recurrence after radical prostatectomy: data from CaPSURE. Urology 66(5):1060–1065 17. Ly D, Reddy CA, Klein EA, Ciezki JP (2010) Association of body mass index with prostate cancer biochemical failure. J Urol 183(6):2193–2199 18. Cao Y, Ma J (2011) Body-mass index, prostate cancer-specific mortality and biochemical recurrence: a systematic review and meta-analysis. Cancer Prev Res 4(4):486–501 19. Gong Z, Agalliu I, Lin DW, Stanford JL, Kristal AR (2007) Obesity is associated with increased risks of prostate cancer metastasis and death after initial cancer diagnosis in middle-aged men. Cancer 109(6):1192–1202 20. Efstathiou JA, Bae K, Shipley WU et al (2007) Obesity and mortality in men with locally advanced prostate cancer: analysis of RTOG 85–31. Cancer 110(12):2691–2699 21. Ma J, Li H, Giovannucci E et al (2008) Prediagnostic body-mass index, plasma C-peptide concentration, and prostate cancer-specific mortality in men with prostate cancer: a long-term survival analysis. Lancet Oncol 9(11):1039–1047 22. Haggstrom C, Stocks T, Ulmert D et al (2012) Prospective study on metabolic factors and risk of prostate cancer. Cancer 118(24):6199–6206 23. Bassett JK, Severi G, Baglietto L et al (2012) Weight change and prostate cancer incidence and mortality. Int J Cancer 131(7):1711–1719 24. Wright ME, Chang SC, Schatzkin A et al (2007) Prospective study of adiposity and weight change in relation to prostate cancer incidence and mortality. Cancer 109(4):675–684 25. Joshu CE, Mondul AM, Menke A et al (2011) Weight gain is associated with an increased risk of prostate cancer recurrence after prostatectomy in the PSA era. Cancer Prev Res (Phila) 4(4):544–551 26. Whitley BM, Moreira DM, Thomas JA et al (2011) Preoperative weight change and risk of adverse outcome following radical prostatectomy: results from the Shared Equal Access Regional Cancer Hospital database. Prostate Cancer Prostatic Dis 14(4):361–366 27. Barlow L, Westergren K, Holmberg L, Talback M (2009) The completeness of the Swedish Cancer Register: a sample survey for year 1998. Acta Oncol 48(1):27–33 28. Van Hemelrijck M, Wigertz A, Sandin F et al (2012) Cohort profile: the National Prostate Cancer Register of Sweden and Prostate Cancer data Base Sweden 2.0. Int J Epidemiol 42(4):956–967 29. Szulkin R, Holmberg E, Stattin P et al (2012) Prostate cancer risk variants are not associated with disease progression. Prostate 72(1):30–39 30. Norman A, Bellocco R, Bergstrom A, Wolk A (2001) Validity and reproducibility of self-reported total physical activity—
Cancer Causes Control
31.
32. 33.
34.
35.
differences by relative weight. Int J Obes Relat Metab Disord 25(5):682–688 Messerer M, Johansson SE, Wolk A (2004) The validity of questionnaire-based micronutrient intake estimates is increased by including dietary supplement use in Swedish men. J Nutr 134(7):1800–1805 Cox DR (1972) Regression models and life-tables. J R Stat Soc B 34(2):187 Discacciati A, Orsini N, Wolk A (2012) Body mass index and incidence of localized and advanced prostate cancer—a dose– response meta-analysis of prospective studies. Ann Oncol 23(7):1665–1671 Tomaszewski JJ, Chen YF, Bertolet M et al (2013) Obesity is not associated with aggressive pathologic features or biochemical recurrence after radical prostatectomy. Urology 81(5):992–996 Banez LL, Hamilton RJ, Partin AW et al (2007) Obesity-related plasma hemodilution and PSA concentration among men with prostate cancer. JAMA 298(19):2275–2280
36. Roberts DL, Dive C, Renehan AG (2010) Biological mechanisms linking obesity and cancer risk: new perspectives. Annu Rev Med 61:301–316 37. Renehan AG, Frystyk J, Flyvbjerg A (2006) Obesity and cancer risk: the role of the insulin–IGF axis. Trends Endocr Metab 17(8):328–336 38. Schnoeller T, Jentzmik F, Rinnab L et al (2013) Circulating free testosterone is an independent predictor of advanced disease in patients with clinically localized prostate cancer. World J Urol 31(2):253–259 39. Kovalchik S (2009) Validity of adult lifetime self-reported body weight. Public Health Nutr 12(8):1072–1077 40. Prentice AM, Jebb SA (2001) Beyond body mass index. Obes Rev 2(3):141–147 41. Mucksavage P, Mitchell C, Kutikov A et al (2012) Anthropometric differences in obese men with biochemical failure after radical retropubic prostatectomy. Urol Oncol 30(5):590–595
123
ORIGINAL PAPER
Body mass index and weight change in men with prostate cancer: progression and mortality Stephanie E. Bonn • Fredrik Wiklund • Arvid Sjo¨lander Robert Szulkin • Pa¨r Stattin • Erik Holmberg • Henrik Gro¨nberg • Katarina Ba¨lter
•
Received: 18 October 2013 / Accepted: 28 April 2014 Ó Springer International Publishing Switzerland 2014
Abstract Purpose Body mass index (BMI) is a modifiable lifestyle factor that has been associated with an increased risk of fatal prostate cancer and biochemical recurrence. The main purpose of the present study was to investigate the association between the exposure BMI at the time of a prostate cancer diagnosis and weight change after diagnosis, and the outcomes of prostate cancer progression and mortality in a large cohort study. Methods Data from 4,376 men diagnosed with clinically localized prostate cancer between 1997 and 2002 were analyzed. BMI and weight change were self-reported in 2007. Hazard ratios (HRs) with 95 % confidence intervals (CIs) were estimated in complete-case analysis (n = 3,214) using Cox proportional hazards models. Results Progression was experienced among 639 (14.6 %) of the study participants, and in total, 450
(10.3 %) deaths of any cause and 134 (3.1 %) prostate cancer-specific deaths were recorded during follow-up. Obese men had a 47 % increased rate of overall mortality compared to normal weight men (HR 1.47, 95 % CI 1.03–2.10). No statistically significant associations were found for BMI and prostate cancer progression or prostate cancer-specific mortality. A weight loss [5 % after diagnosis almost doubled the rate of overall mortality compared to maintaining a stable weight (HR 1.94, 95 % CI 1.41–2.66), while a weight gain [5 % was associated with an almost doubled increased rate of prostate cancer-specific mortality (HR 1.93, 95 % CI 1.18–3.16). Conclusions Being obese was associated with an increased rate of overall mortality, and gaining weight after a prostate cancer diagnosis was associated with an increased rate of prostate cancer-specific mortality. Keywords Prostatic neoplasms Disease progression Mortality Body mass index Weight change Epidemiology
Introduction S. E. Bonn (&) F. Wiklund A. Sjo¨lander R. Szulkin H. Gro¨nberg K. Ba¨lter Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Nobels va¨g 12A, 171 77 Stockholm, Sweden e-mail: [email protected] P. Stattin Department of Surgical and Perioperative Sciences, Urology and Andrology, Umea˚ University, Umea˚, Sweden E. Holmberg Department of Oncology, Institute of Clinical Sciences, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden
In Sweden, one in eight men younger than 75 years of age is diagnosed with prostate cancer, and while the 5-year survival is approximately 90 %, the risk of a relapse among men undergoing curative treatment is almost 30 % [1]. The increasing incidence of prostate cancer, together with the concomitant decrease in mortality rates, has led to increasing numbers of men living with a previous prostate cancer diagnosis. Greater utilization of prostate-specific antigen (PSA) testing has led to the detection of not only clinically relevant tumors, but also of slow growing tumors that
123
Cancer Causes Control
otherwise would have escaped diagnosis [2]. Since many men are diagnosed at an early stage, a large proportion have indolent cancer that may never progress to a clinical tumor [3] and active surveillance instead of invasive treatment is an increasingly common option among men with low-risk tumors [4]. Thus, it is important to find complementary ways to influence cancer progression. Body mass index (BMI) is a modifiable lifestyle factor with the potential to reduce or delay progression of prostate cancer in both treated and untreated men [5]. BMI is an important proxy for body fat and is used in both clinical and research settings where overweight and obesity are commonly defined as BMI being C25 or C30 kg/m2, respectively [6]. With the prevalence of overweight and obesity increasing worldwide [7], and the co-occurring increase in prostate cancer incidence, it is highly relevant to clarify the possible association between overweight and obesity and prostate cancer. Results from the previous studies have shown no association between BMI and the overall risk of prostate cancer, while men with a high BMI were more likely to be diagnosed with aggressive or fatal prostate cancer [8–10]. This suggests that BMI may be of greater importance for prostate cancer progression than for the risk of developing the disease [11]. Previous studies have also found a high BMI to be associated with an increased risk of biochemical recurrence [12–18] and an increased risk of prostate cancer-specific mortality [18–23]. Weight change is another lifestyle factor with potential to affect both prostate cancer risk and survival. Results are similar to those for BMI, and while previous studies of weight gain during adulthood have shown no association with the incidence of prostate cancer, an increased risk of aggressive prostate cancer and prostate cancer mortality was found [23, 24]. The outcome after a prostate cancer diagnosis might not only be affected by BMI at the time of prostate cancer diagnosis but also by a potential weight gain thereafter. Two previous studies have found an increased risk of biochemical recurrence among men gaining weight in the years before a radical prostatectomy [25, 26]. However, to our knowledge, there are no previous studies investigating the association between weight gain after a prostate cancer diagnosis and survival. We aimed to study BMI at the time of prostate cancer diagnosis and weight change after diagnosis in relation to prostate cancer progression and mortality in a large cohort of Swedish men with localized prostate cancer. Results from this study may guide future studies regarding whether BMI, weight maintenance, or weight change after a prostate cancer diagnosis can influence the survival of prostate cancer patients.
123
Methods Study design All incident cancers in Sweden are reported to the Swedish National Cancer Register (NCR) [27]. For prostate cancer specifically, the National Prostate Cancer Registry (NPCR) of Sweden includes all patients with a prostate cancer diagnosis [28] and holds additional information of serum PSA levels; tumor, node, metastasis (TNM) stage; tumor differentiation at the time of diagnosis; and primary treatment within 6 months of diagnosis. Study participants in the present study (PROCAP, PROgression in CAncer of the Prostate) were derived from a retrospective nationwide cohort study of patients with localized prostate cancer, the NPCR of Sweden Follow-up Study, which has previously been described in detail [4]. In brief, patients were eligible for inclusion in the NPCR of Sweden Follow-up Study if they were registered with a localized prostate cancer in the NPCR between 1 January 1997 (1 January 1998 in one region) and 31 December 2002; B70 years of age at diagnosis; diagnostic serum PSA \ 20 ng/ml; local tumor stage T1–T2; and no signs of lymph node metastasis (NX or N0) or bone metastasis (MX or M0). Out of the 8,304 patients fulfilling the criteria, 7,960 (96 %) accepted inclusion to the study. PROCAP is an extension of the NPCR of Sweden Follow-up Study and has previously been described elsewhere [29]. In short, all patients in the NPCR of Sweden Follow-up Study who were still alive in 2007 (n = 7,074) were eligible for inclusion in PROCAP and invited to respond to a questionnaire assessing lifestyle factors and to donate a blood sample for genetic analysis. In total, 5,779 (82 %) patients answered the questionnaire and were included. However, patients with unknown primary treatment or missing progression data (n = 880) were excluded from the analysis as the definition of disease progress is dependent on primary treatment. Further, patients primarily treated with hormones (n = 131), patients with unknown date of last record in medical journals or a missing date of termination of deferred treatment (n = 29), and patients with completely missing questionnaire data or incomplete data on self-reported BMI (n = 363) were excluded from the analysis. In total, 4,376 patients were included in further analysis. All patients included in PROCAP gave their written informed consent for participation. The study has been approved by the research ethics committee at Karolinska Institutet. End points With a median time period of 4 years after diagnosis, research nurses in each of six health care regions in
Cancer Causes Control
Sweden extracted information at one occasion for each patient on the date of last follow-up, reason for and date of termination of surveillance, subsequent PSA testing, and signs of local progress and distant metastasis from medical records. Among patients treated with curative intent (radiation therapy or radical prostatectomy), either having biochemical recurrence, local progress, or distant metastasis, was considered disease progress. However, biochemical recurrence in the present study was defined differently depending on primary treatment and has previously been described in detail [29]. Briefly, biochemical recurrence was defined as a doubling in PSA above the post-treatment nadir value and exceeding at least 1 ng/ml for patients who underwent radiation therapy. For patients treated with radical prostatectomy, biochemical recurrence was defined as two consecutive tests with PSA levels [0.2 ng/ml; the date for this event was set to the first of the two occasions. For operated patients with only one registered PSA test value, a value [0.5 ng/ml was considered biochemical recurrence. Further, the event of termination of deferred treatment with biochemical progression as a reason for termination defined disease progression in patients on surveillance. The time to event was defined as the time from the prostate cancer diagnosis to the earliest observed progressive event for each treatment-specific definition of progressive events. Patients without disease progression were censored at the last date of follow-up recorded in the medical journal, whereas patients on surveillance who ended their deferred treatment without any signs of disease progression were censored at the recorded date of termination. In total, 639 men (14.6 %) experienced a progress event during the follow-up, with 214 (13.2 %), 355 (15.6 %), and 70 (14.5 %) progression events recorded among men with a BMI \ 25, 25–30, and [30 kg/m2, respectively. The median time to disease progression was 2.3 years, and the median time to censoring was 3.9 years. Cause of death, prostate cancer specific or other cause, and date of event were obtained from the Swedish Causeof-Death registry using national identification numbers. Time to event was defined as time from prostate cancer diagnosis to the date of death reported in the registry. All patients were left-truncated by study design at the date of inclusion to PROCAP, since only patients alive at this date could be included in the analysis. Patients still alive on 8 June 2012 were censored at this date. The median survival times for overall mortality, prostate cancer-specific mortality, or censoring were 10.1, 9.9, and 11.6 years, respectively. Lifestyle factors Lifestyle factors were assessed in a questionnaire sent out to patients between January 2007 and June 2008. Patients
chose to respond to the questionnaire either in paper format (50 %) or via the Web. Data from the Web-based questionnaires were directly saved in digital format while paperbased questionnaires were scanned into digital format after being checked for completeness by study personnel. Patients self-reported their current height, weight, and weight change since diagnosis in the questionnaire. Weight change was reported as no change in weight or as a weight increase or decrease from the time of prostate cancer diagnosis. In the case of weight increase or decrease, how many kilograms that was gained or lost was reported. BMI (kg/m2) at diagnosis was estimated based on self-reported current height and weight, and weight change since diagnosis. Normal weight, overweight, and obesity were defined according to the National Institute of Health as normal weight if BMI \ 25 kg/m2, overweight if BMI 25 \ 30 kg/m2, and obese if BMI C 30 kg/m2 [6]. In what follows, we use ‘‘BMI’’ as shorthand for ‘‘estimated BMI at diagnosis.’’ Based on reported current weight and weight change and the distribution of data, patients were categorized into three groups: no change or a change B5 %, an increase [5 % or a decrease [5 % since diagnosis. Additional lifestyle factors considered as potential confounding factors were also assessed in the self-reported questionnaire. Tobacco users were defined as ‘‘current,’’ ‘‘former,’’ or ‘‘never’’ smokers based on reported smoking habits since the time of prostate cancer diagnosis. Physical activity at the age of 50 and after diagnosis was defined as total metabolic equivalent time (MET-h) based on questions previously validated by Norman et al. [30]. Total energy intake was assessed by a food frequency questionnaire similar to that previously validated by Messerer et al. [31] asking specifically about food habits since the time of prostate cancer diagnosis. Further, variables of overall stress during the last year, education level, current occupation, and family history of prostate cancer were assessed. Statistical methods Distributions and means of demographic and clinical variables were studied across BMI categories. Statistically significant associations were tested for using one-way ANOVA for continuous variables and chi-square test for categorical variables. Progression-free, overall, and prostate cancer-specific survivals were analyzed using the Kaplan–Meier method. Time to event for the different BMI and weight change categories was compared using log-rank test. Life tables were used to assess the 5-year survival. Cox proportional hazards models were used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95 % confidence intervals (95 % CIs) in the analysis of progression and mortality [32]. All three models included only complete cases, i.e., participants with
123
Cancer Causes Control
information on all covariates included in the most adjusted model. Time since prostate cancer diagnosis was used as the underlying timescale. BMI and weight change were included as categorical exposures in the Cox proportional hazards models. To examine whether men who had reported a change in weight due to illness were influencing the results, we carried out additional sensitivity analysis with 18-months lag time. Model-free (i.e., not estimated using a Cox proportional hazards model) HRs and CIs were assessed using the statistical software R. To assess whether measured covariates should be considered as potential confounding factors and adjusted for in the Cox proportional hazards models, we tested whether the covariates were statistically associated with both the exposure (BMI and weight change) and the outcome (progression and mortality). The association between covariates and the exposure was assessed using linear regression models, treating BMI as continuous. The association between the covariates and the outcome was assessed using Cox proportional hazards models. Covariates tested were age at diagnosis; MET-h at age 50; education level; smoking habits after diagnosis; family history of prostate cancer; PSA level at diagnosis; T-, N-, and M-stages; tumor grade and Gleason score at diagnosis; and primary treatment. In mortality analysis, a potential confounding effect of MET-h after diagnosis and stress level after diagnosis was also evaluated. In the analysis of BMI, confounding by weight change was tested for. Correspondingly, in the analysis of weight change, confounding of BMI was tested for. Based on this confounder selection, multivariable-adjusted models for progression, overall mortality, and prostate cancer-specific mortality included age at diagnosis (5-year categories), mode of primary treatment (curative intent, radical prostatectomy, or radiation therapy), and Gleason score (\6, 6, [6). We also present results from fully adjusted models that were further adjusted for PSA at diagnosis, clinical stage (T1 or T2), N-stage (N0, N1, or NX), M-stages (M0 or MX), smoking habits (current, former, never), and MET-h at age 50. Interaction between the two exposures and all confounding factors were tested for using likelihood ratio tests. The Cox proportional hazards assumption was tested using Schoenfelds residuals, and no statistically significant deviation from the proportional hazards assumption was detected. The level of significance was set to p \ 0.05. All analyses were performed using STATA 12.1 (STATA Corporation, College Station, TX).
Results In total, 4,376 men were included in the analysis. The median time from prostate cancer diagnosis to inclusion in
123
the study was 7.25 years with an interquartile range of 2.15 years. Among the included men, 37 % were normal weight (BMI \ 25 kg/m2), 52 % of the men were overweight (BMI 25 \ 30 kg/m2), and 11 % were obese (BMI C 30 kg/m2) at diagnosis. Among those classified as normal weight, eight men had a BMI \ 18.5 kg/m2. The mean BMI at diagnosis was 26.2 ± 3.2 kg/m2. In total, 1,040 (23.8 %) men had reported a weight change greater than 5 % after diagnosis; 649 (14.8 %) had reported an increase and 391(8.9 %) a decrease. The mean (±SD) age at diagnosis was 63.0 ± 5.1, and overweight and obese men were statistically significantly younger at diagnosis compared to normal weight men. There were also statistically significant differences between the BMI categories with regard to reported MET-h at age 50 and after diagnosis with normal weight, overweight, and obese men having reported 23.5 ± 7.6, 24.7 ± 8.3, and 24.4 ± 8.5 MET-h, respectively, at age 50 and corresponding MET-h of 24.0 ± 7.0, 24.0 ± 7.0, and 22.8 ± 7.3 after diagnosis. There were also statistically significant differences between the BMI categories with regard to weight change, education, smoking, and M-stage at diagnosis. Men in the higher BMI categories had reported weight change, both increase and decrease, to a larger extent than normal weight men. Higher BMI also seemed to be associated with a lower level of education. Normal weight men were more likely to be never or current smokers compared to overweight and obese men that reported to be former smokers to a greater extent. Regarding M-stage at diagnosis, the number of men classified as MX was increased in the two higher BMI categories compared to among normal weight men. The mean PSA value at diagnosis was 8.3 ± 4.2 ng/ ml; there was no statistically significant difference in PSA between the BMI categories. Characteristics of participants are shown in Table 1. Progression was experienced among 639 (14.6 %) of the study participants. In total, 450 (10.3 %) deaths of any cause and 134 (3.1 %) prostate cancer-specific deaths were recorded. The prostate cancer progression, overall mortality, and prostate cancer-specific mortality rates in the cohort were 40.2, 24.0, and 7.1 per 1,000 person-years, respectively. Table 2 displays model-free mortality rate ratios comparing BMI categories for overall and prostate cancer-specific mortality, and includes the number of events, person-years, and corresponding mortality rates stratified on weight change. Kaplan–Meier curves with log-rank analysis showed a statistically significant difference in progression (Fig. 1), overall mortality, and prostate cancer-specific mortality (Fig. 2). Although no clear pattern between the BMI categories is seen for progression, the survival among obese men appears different from that over normal and overweight men for overall and prostate cancer-specific
Cancer Causes Control Table 1 Characteristics of study participants included in the analyses in the PROCAP study divided by body mass index (BMI) at diagnosis BMI (kg/m2) at diagnosis All (n = 4,376) n (%)
pc
\25 (n = 1,617) n (%)
25 \ 30 (n = 2,276) n (%)
[30 (n = 483) n (%)
Age at diagnosis (years) \55
0.035 320 (7.3)
105 (6.5)
174 (7.6)
41 (8.5)
55 \ 60 60 \ 65
877 (20.0) 1,385 (31.7)
296 (18.3) 514 (31.8)
471 (20.7) 720 (31.6)
110 (22.8) 151 (31.3)
65 \ 70
1,560 (35.7)
600 (37.1)
794 (34.9)
166 (34.4)
234 (5.4)
102 (6.3)
117 (5.1)
15 (3.1)
3,336 (76.2)
1,334 (82.5)
1,714 (75.3)
288 (59.6)
[5 % Increase
649 (14.8)
199 (12.3)
369 (16.2)
81 (16.8)
[5 % Decrease
391 (8.9)
84 (5.2)
193 (8.5)
114 (23.6)
C70 Weight change since diagnosisa No change or B5 % change
0.000
Tumor stage
0.678
T1a
209 (4.8)
72 (4.5)
111 (4.9)
T1b
99 (2.3)
42 (2.6)
43 (1.9)
26 (5.4) 14 (2.9)
T1c
2,234 (51.1)
822 (50.8)
1,168 (51.3)
244 (50.5)
T2
1,834 (41.9)
681 (42.1)
954 (41.9)
199 (41.2)
Gleason scoreb
0.062
\6
1,098 (25.1)
386 (23.9)
600 (26.4)
112 (23.2)
6 [6
1,794 (41.0) 811 (18.5)
683 (42.2) 275 (17.0)
916 (40.2) 429 (18.8)
195 (40.4) 107 (22.2)
1,456 (33.3)
543 (33.6)
757 (33.3)
156 (32.3)
N-classification
0.575
N0 N1
25 (0.6)
NX
2,854 (65.2)
Missing data
41 (0.9)
10 (0.6) 1,045 (64.6) 19 (1.2)
10 (0.4) 1,490 (65.5) 19 (0.8)
5 (1.0) 319 (66.0) 3 (0.6)
M-classification
0.012
M0
2,213 (50.6)
852 (52.7)
1,143 (50.2)
218 (45.1)
MX
2,133 (48.7)
754 (46.6)
1,117 (49.1)
261 (54.2)
Missing data
30 (0.7)
11 (0.7)
16 (0.7)
3 (0.6)
409 (25.3)
551 (24.2)
111 (23.0)
Primary treatment
0.068
Surveillance
1,071 (24.5)
Radical prostatectomy
2,390 (54.6)
884 (54.7)
1,259 (55.3)
247 (51.1)
915 (20.9)
324 (20.0)
466 (20.5)
125 (25.9)
1,851 (42.3) 2,141 (48.9)
766 (47.4) 684 (42.3)
929 (40.8) 1,159 (50.9)
156 (32.3) 298 (61.7)
344 (7.9)
154 (9.5)
165 (7.2)
25 (5.2)
40 (0.9)
13 (0.8)
23 (1.0)
4 (0.8)
565 (34.9)
972 (42.7)
240 (49.7)
Radiation therapy Smoking
a
0.000
Never smoker Past smoker Current smoker Missing data a
Education level B9 years
[9 B 12 years
1,563 (35.7)
579 (35.8)
829 (36.4)
155 (32.1)
[12 years
1,008 (23.0)
462 (28.6)
462 (20.3)
84 (17.4)
11 (0.7)
14 (0.6)
4 (0.8)
Missing data a
0.000 1,777 (40.6)
28 (0.6)
Self-reported in PROCAP
b
Missing data from n = 673
c
p values from chi-square test
123
Cancer Causes Control Table 2 Model-freea hazard ratios (HRs) of overall and prostate cancer-specific mortality by body mass index (BMI) at diagnosis and selfreported weight change since diagnosis BMI \ 25 kg/m2
BMI 25 \ 30 kg/m2
BMI C 30 kg/m2
No. of events
Personyears
Events/1,000 person-years
No. of events
Personyears
Events/1,000 person-years
No. of events
Personyears
Events/1,000 person-years
121
5,762.1
21.0
157
7,357.0
21.3
34
1,214.2
28.0
Mortality Overall No weight change Weight increase [5 %
23
851.6
27.0
33
1,594.5
20.7
10
344.8
29.0
Weight decrease [5 %
16
342.3
46.7
35
796.4
43.9
21
454.7
46.2
160
6,956.0
23.0
225
9,747.9
23.1
65
2,013.7
32.3
Total
HR (95 % CI)
HR (95 % CI)
HR (95 % CI)
1.00
1.00 (0.82–1.23)
1.40 (1.05–1.87)
Prostate cancer specific No weight change
30
5,762.1
5.2
47
7,357.0
6.4
12
1,214.2
9.9
Weight increase [5 %
12
851.6
14.1
10
1,594.5
6.3
7
344.8
20.3
Weight decrease [5 % Total
2
342.3
5.8
10
796.4
12.6
4
454.7
8.8
44
6,956.0
6.3
67
9,747.9
6.9
23
2,013.7
11.4
HR (95 % CI) 1.00 a
HR (95 % CI) 1.09 (0.74–1.59)
HR (95 % CI) 1.81 (1.10–2.97)
Not estimated from a Cox proportional hazards model
mortality. Five years after diagnosis, 83.7 % of the normal weight men were progression-free compared to 80.4 and 80.9 % among the overweight and obese men, respectively. The 5-year overall survival after inclusion to PROCAP was 89.5, 89.3, and 85.8 % for normal weight, overweight, and obese men, respectively, while the corresponding prostate cancer-specific survival was 97.0, 96.7, and 94.7 %. For men in different weight change categories, Kaplan– Meier curves with log-rank analysis showed a statistically significant difference in both overall and prostate cancerspecific mortality (Fig. 3). The 5-year overall survival after inclusion to PROCAP was 90.0, 89.0, and 80.0 % for men with stable weight, men who gained weight, and men who lost weight, respectively. The corresponding prostate cancer-specific survival was 97.0, 95.2, and 95.6 %. Model-based unadjusted and adjusted HRs comparing BMI categories and BMI as continuous variable for progression, overall, and prostate cancer-specific mortality from Cox proportional hazard regression models included only complete cases (n = 3,214) and are shown in Table 3. There were no statistically significant differences in progression rates between the BMI categories, although point estimates indicated higher rates among overweight and obese men compared to normal weight. When analyzing BMI as a continuous exposure, there was a 3 % increased rate of progression per one unit increase in BMI in multivariate-adjusted models, HR 1.03 (95 % CI 1.00–1.06). While the point estimates were similar per BMI unit for overall and prostate cancer-specific mortality, the CIs
123
Fig. 1 Kaplan–Meier survival curve for progression
showed only borderline significance. While obese men had no increased rate of progression, they had an increased rate of overall mortality, HR 1.47 (95 % CI 1.03–2.10). No statistically significant association was found between BMI categories and prostate cancer-specific mortality. Men who experienced a weight reduction [5 % after diagnosis had an almost doubled rate of overall mortality compared to men with a stable weight, HR 1.94 (95 % CI 1.41–2.66) (Table 4). Weight reduction was not statistically significantly associated with prostate cancer-specific mortality, HR 1.56 (95 % CI 0.80–3.05). When men who had died within 18 months after inclusion to the study were excluded in the sensitivity analysis, the association for
Cancer Causes Control
Fig. 2 Kaplan–Meier survival curves for mortality and BMI categories. On the x-axis, time from inclusion in PROCAP to death or censoring is shown, origin (time = 0) is the date of diagnosis in lefttruncated Cox proportional hazard regressions
Fig. 3 Kaplan–Meier survival curves for mortality and weight change categories. On the x-axis, time from inclusion in PROCAP to death or censoring is shown, origin (time = 0) is the date of diagnosis in left-truncated Cox proportional hazard regressions
overall mortality remained statistically significant while the HR point estimate was slightly attenuated, HR 1.67 (95 % CI 1.14–2.46). For prostate cancer-specific mortality, the point estimate was attenuated and statistically nonsignificant, HR 1.07 (95 % CI 0.45–2.55). While a weight increase [5 % was not associated with overall mortality, HR 1.14 (95 % CI 0.82–1.58), men who increased in weight after diagnosis had a statistically significantly increased rate of prostate cancer-specific mortality, HR 1.93 (95 % CI 1.18–3.16). In the sensitivity analysis, the results for overall mortality remained statistically nonsignificant, HR 1.29 (95 % CI 0.91–1.83), while the results for prostate cancer-specific mortality remained statistically significant, HR 1.97 (95 % CI 1.16–3.35).
overall mortality compared to normal weight men. We also found that a weight loss[5 % after diagnosis increased the overall mortality rate compared to maintaining a stable weight, whereas weight gain was associated with an increased rate of prostate cancer-specific mortality. Established risk factors of prostate cancer include age, ethnicity, and heredity [2], while lifestyle factors, including BMI, are still debated. A recent meta-analysis by Discacciati et al. [33] showed that high BMI was associated with a decreased risk of localized prostate cancer but an increased risk of advanced prostate cancer, indicating a dual effect of BMI on prostate cancer risk. More recently, attention has shifted from studies of risk to studies regarding prostate cancer progression and mortality. A majority of previous studies, but not all [34], have found increasing BMI to be associated with an increased risk of biochemical recurrence [12–18]. High BMI has also been associated with a higher risk of prostate cancer-specific mortality in several studies [18–23]. A review and meta-analysis by Cao and Ma [18] showed increased risks of both biochemical recurrence and
Discussion In this large cohort study of men with localized prostate cancer, we found that obese men had an increased rate of
123
Cancer Causes Control Table 3 Model-based age-adjusted and multivariable-adjusted hazard ratios (HRs) from complete-case analysis (n = 3,214) according to body mass index (BMI) at the time of prostate cancer diagnosis obtained through Cox proportional hazards models No. of events
BMI (kg/m2) categories
n
\25 (ref) HR
25 \ 30 HR (95 % CI)
C30 HR (95 % CI)
HR (95 % CI)
Adjusteda
1.00
1.15 (0.94–1.41)
1.07 (0.78–1.46)
1.02 (1.00–1.06)
Adjustedb
1.00
1.15 (0.94–1.40)
1.06 (0.77–1.46)
1.02 (1.00–1.05)
1.00
1.13 (0.92–1.38)
1.11 (0.81–1.53)
1.03 (1.00–1.06)
Adjusteda
1.00
1.03 (0.80–1.31)
1.42 (1.00–1.31)
1.02 (0.99–1.06)
Adjustedb
1.00
1.03 (0.81–1.32)
1.37 (0.97–1.95)
1.02 (0.98–1.05)
Adjustedc
1.00
1.06 (0.83–1.36)
1.47 (1.03–2.10)
1.02 (0.99–1.06)
Adjusteda
1.00
0.98 (0.63–1.51)
1.30 (0.68–2.45)
1.03 (0.97–1.10)
Adjustedb
1.00
0.97 (0.63–1.50)
1.16 (0.61–2.21)
1.02 (0.96–1.08)
Adjustedc
1.00
0.96 (0.62–1.50)
1.29 (0.67–2.48)
1.02 (0.96–1.09)
Measure
Progression
Adjusted
455
c
Overall mortality
311
Prostate cancer-specific mortality
a
Per unit of BMI (kg/m2)
96
Adjusted for age at diagnosis (5-year categories)
b
Adjusted for age at diagnosis, primary treatment (curative intent, radical prostatectomy, or radiation therapy), and Gleason score at diagnosis (\6, 6, or [6)
c
Adjusted for age at diagnosis; primary treatment; Gleason score at diagnosis; PSA level; T-, N-, and M-stages at diagnosis; smoking habits (current, former, never); and total MET-h at age 50
Table 4 Model-based age-adjusted and multivariable-adjusted hazard ratios (HRs) from complete-case analysis (n = 3,214) according to weight change categories at the time of prostate cancer diagnosis obtained through Cox proportional hazards models No. of events
No change (ref) HR
Category of weight change
Adjusteda Adjustedb Adjustedc
Increase [5 %
Decrease [5 %
HR (95 % CI)
HR (95 % CI)
1.00
1.19 (0.87–1.64)
2.11 (1.56–1.64)
1.00
1.17 (0.85–1.61)
2.11 (1.54–2.88)
1.00
1.14 (0.82–1.58)
1.94 (1.41–2.66)
Adjusteda
1.00
2.05 (1.27–3.31)
1.51 (0.79–2.87)
Adjustedb
1.00
1.92 (1.18–3.12)
1.63 (0.84–3.14)
Adjustedc
1.00
1.93 (1.18–3.16)
1.56 (0.80–3.05)
n Overall mortality
Prostate cancerspecific mortality
a
311
96
Adjusted for age at diagnosis (5-year categories)
b
Adjusted for age at diagnosis, primary treatment (curative intent, radical prostatectomy, or radiation therapy), Gleason score at diagnosis (\6, 6, or [6), and BMI category at the time of diagnosis
c
Adjusted for age at diagnosis; primary treatment; Gleason score at diagnosis; BMI category at the time of diagnosis; PSA level; T-, N-, and M-stages at diagnosis; smoking habits (current, former, never); and total MET-h at age 50
123
prostate cancer-specific mortality with estimates of a 15–21 % increased risk per 5 kg/m2 increase in BMI. In contrast to many previous findings, our results showed no association between BMI and rate of progression during a median follow-up of 4 years or prostate cancer-specific mortality. However, we found a positive association between high BMI and overall mortality, in contrast to a previous study where no association between BMI and overall or non-prostate cancer-specific mortality was seen [20]. High weight gain during adulthood has been associated with an increased risk of aggressive and fatal prostate cancer [23, 24]. Larger weight gain from the age of 25 years has also been associated with an increased risk of biochemical recurrence after prostate cancer diagnosis [15]. Few studies have investigated the association between weight change after diagnosis and survival, but Joshu et al. [25] showed that weight gain 5 years before and 1 year after a prostate cancer diagnosis was associated with an almost doubled risk of prostate cancer recurrence and Whitley et al. [26] found that a weight gain C2.5 kg in the year preceding a radical prostatectomy increased the risk of biochemical recurrence by 65 % compared to men who gained \2.5 kg in weight. In line with the previous findings, we showed an increased rate of prostate cancer-specific mortality among men who experienced weight gain [5 % after diagnosis. The fact that the increased rate
Cancer Causes Control
remained even after excluding men who died within the first 18 months after inclusion to the study strengthens our finding since the weight increase is less likely to have occurred due to other illness. The increased rates seen among men who decreased in weight were, on the contrary, slightly attenuated in the analysis of overall mortality and disappeared in the analysis of prostate cancer-specific mortality, indicating that the association seen for weight decrease might in fact be due to weight loss because of illness. Nevertheless, our results are based on self-reported weight change and no information on whether or not the change was intentional was available. Intentional weight change may have different effects than unintended weight changes, which may be due to something else, e.g., other diseases, health issues, or malnutrition. The effect of intentional weight change after a prostate cancer diagnosis needs to be studied further. One frequently proposed explanation to the observed association between BMI and prostate cancer mortality is a delayed diagnosis among obese men, leading to a potentially more advanced disease at diagnosis and thereby poorer survival. Obesity-related plasma hemodilution, generating lower PSA values, may be one explanation for a delayed diagnosis and poorer outcome among obese patients [35]. However, high BMI has been associated with a higher risk of prostate cancer-specific mortality both before and during the PSA screening era [21], indicating that hemodilution of PSA values does not fully explain the association. Nor do we believe that a more advanced disease among overweight and obese subjects in our study fully explains the associations found between high BMI and progression and mortality since all men in our cohort were diagnosed with localized prostate cancer and there was no statistically significant difference in PSA values, tumor stage, or grade between the different BMI categories. Several possible mechanisms linking adiposity and prostate cancer have been proposed. The main hypothesis involves hormonal and metabolic changes including pathways for insulin and insulin-like growth factors (IGFs), sex hormones, and adipokine signaling [36]. High BMI is associated with hyperinsulinemia that, through reduction of IGF-binding proteins, leads to increased serum levels of IGF-I, which has been shown to promote tumor development [37]. Obesity is also associated with decreased androgen levels and lower serum testosterone levels in men, the latter having been linked to an increased risk of aggressive prostate tumors [38]. Further, altered levels of adipokines with increased levels of leptin and decreased levels of adiponectin are a result of obesity and have been associated with tumor development [36]. Hormonal and metabolic pathways that are altered by the level of adiposity might also be affected by weight
change once a tumor has developed and could be one explanation for the increased rate of prostate cancer-specific mortality we found among men who increased in weight from the time of diagnosis. While BMI at the time of diagnosis is a fixed factor for each patient once diagnosed, weight maintenance or change after a diagnosis has the potential to complement treatment as it can be influenced by the patient himself. This study has several strengths including a populationbased design with a large sample size. An additional strength is the complete follow-up regarding disease progression during on average the first 4 years of follow-up and survival. The short follow-up times for progression limit our ability to study the long-term effects of BMI on progression. However, among patients with localized prostate cancer, disease progression is an early event and we believe that the limited follow-up time would only have minor effects on our observed results. The condition of being alive up to 5–10 years after diagnosis in order to be eligible for inclusion in the PROCAP study is a major limitation in our study. However, it can be argued that any bias induced by this left truncation is likely to act in opposite direction as a true causal effect, thus resulting in conservative estimates. Although the response rate was high, we were only able to study progression and mortality among patients surviving long enough to participate in PROCAP. Nonetheless, the survival rate among men with prostate cancer is almost 70 % during the first 10 years after diagnosis in Sweden [1]. The inability to include the most severe cases, patients that die early after a prostate cancer diagnosis may partly explain why we, in contrast to other studies, found no association between BMI and prostate cancer progression and mortality. The retrospective self-report of height, weight, and change in weight is another limitation. However, previous validation studies have shown a strong correlation between self-reported and measured weight, although the long-term recall is influenced by current weight [39]. Also, it seems plausible that the outcome in our study (progression/mortality) had no major effect on self-reported measures, and thus any misclassification error is likely to be non-differential (i.e., bias toward a null effect). Even though BMI is often used as a proxy for body fat, it is not without limitations. It does not factor in body composition (fat mass vs. fat free mass) and fat distributions (superficial vs. visceral fat). Subjects with high muscle mass and low fat mass may be incorrectly classified as overweight or obese [40]. Limitations in BMI as an exposure may be a reason for differences between study outcomes. A previous study investigating anthropometric differences among obese men showed that the fat distribution among men experiencing biochemical failure was different from that of men who did not, despite similar weights and BMI; the authors
123
Cancer Causes Control
concluded that the differences in fat distribution might account for conflicting results regarding BMI and prostate cancer [41]. In conclusion, we found an increased rate of overall mortality among obese men and men who lost a large amount of weight. Men who increased in weight following diagnosis had an increased rate of prostate cancer-specific mortality. Although not conclusive, our findings indicate a positive association between high BMI and prostate cancer progression and mortality. Future studies investigating the role of adiposity and weight maintenance or change in relation to prostate cancer progression and mortality are needed to clarify the association. Acknowledgments We thank all of the participants in the PROCAP study. We also thank Carin Cavalli-Bjo¨rkman and Ami Ro¨nnberg for their work during data collection and Michael Broms for his work with the databases. We also acknowledge the NPCR steering group and all the research nurses who extracted data for the follow-up study. The present study was supported by the Swedish Cancer Society, Grant No. CAN 2011/868, and the Swedish research Council for Health, Working life and Welfare, Grant No. 2011-0650. Conflict of interest
The authors declare no conflict of interests.
References 1. Bergman O, Hont G, Johansson E (ed) (2013) The National Board of Health and Welfare, Stockholm, Sweden. http://www. socialstyrelsen.se/Lists/Artikelkatalog/Attachments/19108/20136-5.pdf. Accessed 18 Oct 2013 (In Swedish: Cancer i siffror 2013) 2. Jemal A, Bray F, Center MM et al (2011) Global cancer statistics. CA Cancer J Clin 61(2):69–90 3. Meyerhardt JA, Ma J, Courneya KS (2010) Energetics in colorectal and prostate cancer. J Clin Oncol 28(26):4066–4073 4. Stattin P, Holmberg E, Bratt O et al (2008) Surveillance and deferred treatment for localized prostate cancer. Population based study in the National Prostate Cancer Register of Sweden. J Urol 180(6):2423–2429 (discussion 2429–2430) 5. Ornish D, Weidner G, Fair WR et al (2005) Intensive lifestyle changes may affect the progression of prostate cancer. J Urol 174(3):1065–1070 6. Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults (1998) Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: executive summary. Am J Clin Nutr 68(4):899–917 7. World Health Organization (2012) World Health Statistics 2012. http://www.who.int/gho/publications/world_health_statistics/EN_ WHS2012_Full.pdf. Accessed 18 Oct 2013 8. Giovannucci E, Liu Y, Platz EA, Stampfer MJ, Willett WC (2007) Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer 121(7):1571–1578 9. Rodriguez C, Freedland SJ, Deka A et al (2007) Body mass index, weight change, and risk of prostate cancer in the Cancer Prevention Study II Nutrition Cohort. Cancer Epidemiol Biomarker Prev 16(1):63–69 10. Stocks T, Hergens MP, Englund A, Ye W, Stattin P (2010) Blood pressure, body size and prostate cancer risk in the Swedish Construction Workers cohort. Int J Cancer 127(7):1660–1668
123
11. Mistry T, Digby JE, Desai KM, Randeva HS (2007) Obesity and prostate cancer: a role for adipokines. Eur Urol 52(1):46–53 12. Efstathiou JA, Chen MH, Renshaw AA, Loffredo MJ, D’Amico AV (2007) Influence of body mass index on prostate-specific antigen failure after androgen suppression and radiation therapy for localized prostate cancer. Cancer 109(8):1493–1498 13. Palma D, Pickles T, Tyldesley S (2007) Obesity as a predictor of biochemical recurrence and survival after radiation therapy for prostate cancer. BJU Int 100(2):315–319 14. Strom SS, Kamat AM, Gruschkus SK et al (2006) Influence of obesity on biochemical and clinical failure after external-beam radiotherapy for localized prostate cancer. Cancer 107(3):631–639 15. Strom SS, Wang XM, Pettaway CA et al (2005) Obesity, weight gain, and risk of biochemical failure among prostate cancer patients following prostatectomy. Clin Cancer Res 11(19):6889–6894 16. Bassett WW, Cooperberg MR, Sadetsky N et al (2005) Impact of obesity on prostate cancer recurrence after radical prostatectomy: data from CaPSURE. Urology 66(5):1060–1065 17. Ly D, Reddy CA, Klein EA, Ciezki JP (2010) Association of body mass index with prostate cancer biochemical failure. J Urol 183(6):2193–2199 18. Cao Y, Ma J (2011) Body-mass index, prostate cancer-specific mortality and biochemical recurrence: a systematic review and meta-analysis. Cancer Prev Res 4(4):486–501 19. Gong Z, Agalliu I, Lin DW, Stanford JL, Kristal AR (2007) Obesity is associated with increased risks of prostate cancer metastasis and death after initial cancer diagnosis in middle-aged men. Cancer 109(6):1192–1202 20. Efstathiou JA, Bae K, Shipley WU et al (2007) Obesity and mortality in men with locally advanced prostate cancer: analysis of RTOG 85–31. Cancer 110(12):2691–2699 21. Ma J, Li H, Giovannucci E et al (2008) Prediagnostic body-mass index, plasma C-peptide concentration, and prostate cancer-specific mortality in men with prostate cancer: a long-term survival analysis. Lancet Oncol 9(11):1039–1047 22. Haggstrom C, Stocks T, Ulmert D et al (2012) Prospective study on metabolic factors and risk of prostate cancer. Cancer 118(24):6199–6206 23. Bassett JK, Severi G, Baglietto L et al (2012) Weight change and prostate cancer incidence and mortality. Int J Cancer 131(7):1711–1719 24. Wright ME, Chang SC, Schatzkin A et al (2007) Prospective study of adiposity and weight change in relation to prostate cancer incidence and mortality. Cancer 109(4):675–684 25. Joshu CE, Mondul AM, Menke A et al (2011) Weight gain is associated with an increased risk of prostate cancer recurrence after prostatectomy in the PSA era. Cancer Prev Res (Phila) 4(4):544–551 26. Whitley BM, Moreira DM, Thomas JA et al (2011) Preoperative weight change and risk of adverse outcome following radical prostatectomy: results from the Shared Equal Access Regional Cancer Hospital database. Prostate Cancer Prostatic Dis 14(4):361–366 27. Barlow L, Westergren K, Holmberg L, Talback M (2009) The completeness of the Swedish Cancer Register: a sample survey for year 1998. Acta Oncol 48(1):27–33 28. Van Hemelrijck M, Wigertz A, Sandin F et al (2012) Cohort profile: the National Prostate Cancer Register of Sweden and Prostate Cancer data Base Sweden 2.0. Int J Epidemiol 42(4):956–967 29. Szulkin R, Holmberg E, Stattin P et al (2012) Prostate cancer risk variants are not associated with disease progression. Prostate 72(1):30–39 30. Norman A, Bellocco R, Bergstrom A, Wolk A (2001) Validity and reproducibility of self-reported total physical activity—
Cancer Causes Control
31.
32. 33.
34.
35.
differences by relative weight. Int J Obes Relat Metab Disord 25(5):682–688 Messerer M, Johansson SE, Wolk A (2004) The validity of questionnaire-based micronutrient intake estimates is increased by including dietary supplement use in Swedish men. J Nutr 134(7):1800–1805 Cox DR (1972) Regression models and life-tables. J R Stat Soc B 34(2):187 Discacciati A, Orsini N, Wolk A (2012) Body mass index and incidence of localized and advanced prostate cancer—a dose– response meta-analysis of prospective studies. Ann Oncol 23(7):1665–1671 Tomaszewski JJ, Chen YF, Bertolet M et al (2013) Obesity is not associated with aggressive pathologic features or biochemical recurrence after radical prostatectomy. Urology 81(5):992–996 Banez LL, Hamilton RJ, Partin AW et al (2007) Obesity-related plasma hemodilution and PSA concentration among men with prostate cancer. JAMA 298(19):2275–2280
36. Roberts DL, Dive C, Renehan AG (2010) Biological mechanisms linking obesity and cancer risk: new perspectives. Annu Rev Med 61:301–316 37. Renehan AG, Frystyk J, Flyvbjerg A (2006) Obesity and cancer risk: the role of the insulin–IGF axis. Trends Endocr Metab 17(8):328–336 38. Schnoeller T, Jentzmik F, Rinnab L et al (2013) Circulating free testosterone is an independent predictor of advanced disease in patients with clinically localized prostate cancer. World J Urol 31(2):253–259 39. Kovalchik S (2009) Validity of adult lifetime self-reported body weight. Public Health Nutr 12(8):1072–1077 40. Prentice AM, Jebb SA (2001) Beyond body mass index. Obes Rev 2(3):141–147 41. Mucksavage P, Mitchell C, Kutikov A et al (2012) Anthropometric differences in obese men with biochemical failure after radical retropubic prostatectomy. Urol Oncol 30(5):590–595
123
