The association of weight change in young adulthood and smoking status with risk of prostate cancer recurrence

Abstract

The decades before prostate cancer diagnosis represent an etiologically relevant time period for prostate cancer carcinogenesis. However, the association of weight gain in young adulthood with subsequent biochemical recurrence among men with prostate cancer is not well studied, particularly among smokers. We conducted a prospective cohort study of 1082 men with prostate cancer and treated with either radical prostatectomy or radiation between 2003–2010. The association of weight at age 20, weight at age 50, and weight change from age 20 to 50 with biochemical recurrence was assessed using Cox Proportional Hazards with adjustment for confounders. Stratum specific hazard ratio (HR) estimates by smoking status were evaluated. In the overall cohort, weight at age 20 (HR per 30 kg: 1.56, 95% Confidence Interval (CI): 1.02, 2.38, p-trend: 0.039), weight at age 50 (HR per 30 kg: 1.80, 95% CI: 1.32, 2.47, p-trend: <0.001), and weight change from age 20 to 50 (HR per 30 kg: 1.84, 95% CI: 1.24, 2.74, p-trend: 0.003) were associated with biochemical recurrence. In stratified analyses, weight change from age 20 to 50 was significantly associated with biochemical recurrence only in former smokers (HR per 30 kg: 3.87, 95% CI: 1.88, 8.00, p-trend: <0.001) and ever smokers (HR per 30 kg: 2.38, 95% CI: 1.27, 4.45, p-trend: 0.007). No significant association was observed between weight gain in young adulthood and biochemical recurrence in never smokers. This study adds further evidence that weight gain during early adult years conveys long-term risk for adverse cancer outcomes.

INTRODUCTION

Prostate cancer is the most common cancer among men in the United States, and there will be an estimated 161,360 new cases of prostate cancer in 2017.¹ Within 10 years of diagnosis, 20–30% of men treated with radical prostatectomy and 30–50% of men treated with radiation will experience prostate cancer recurrence.² Previous studies have suggested that obesity, weight, and smoking are all predictors of prostate cancer recurrence.^3–18 However, existing studies have not examined the association of long-term, weight gain from young adulthood to middle-age with prostate cancer recurrence. Here, we examined the association of weight in young adulthood and weight change from age 20 to age 50, by smoking status, with risk of biochemical recurrence in men treated for prostate cancer at the Washington University School of Medicine in St. Louis. The decades before diagnosis are an etiologically relevant, but understudied, time period in prostate cancer carcinogenesis with pre-cancerous lesions observable in some men by early adulthood.¹⁹ We are one of the first studies to examine weight change in this critical period from young adulthood to middle age.

Moreover, few studies have specifically examined the impact of smoking on the potential association between weight gain and biochemical recurrence. Among the three studies that have investigated the role of smoking status on this association one was limited to examining never smokers, another was limited to measures of obesity at time of treatment and consisted only of men outside the United States thereby limiting generalizability, while the third study only examined outcomes other than recurrence.^{15, 20, 21} In contrast to our study, these studies have not examined the impact of long-term (30 years) weight change in a etiological relevant time period several years prior to diagnosis, with risk of biochemical recurrence, specifically among smokers.^{15, 20, 21}

MATERIALS and METHODS

Study Population and Data Collection

The study population consisted of men from the Washington University Prostate Cancer Prospective Cohort (PCPC). The PCPC is a cohort of men with biopsy-diagnosed prostate cancer that received treatment at the Washington University School of Medicine in St. Louis between 2003 and 2010. Men were recruited at the time of prostate cancer diagnosis, but prior to treatment. Medical records were abstracted for clinical characteristics of disease, treatment, and follow-up visits. Study participants completed a survey upon enrollment which included questions on demographics, smoking status, and general health information. Medical charts were reviewed bi-annually. Men receiving follow-up care outside the Washington University School of Medicine system were contacted by phone and mailings, and medical records were obtained from the current medical provider. Follow-up was 98% complete on December 31, 2012, the end date for follow-up. Informed consent was obtained at time of study enrollment and the study was approved by the Institutional Review Board at the Washington University School of Medicine.

1314 participants enrolled in the PCPC cohort and either partially or fully completed the survey. Participants were eligible for inclusion in our study if they received definitive (i.e. curative) treatment with radical prostatectomy or radiation, self-identified as either Black or White, and were diagnosed with prostate cancer after 50 years of age. Men with missing covariate information were excluded. The covariates most likely to be missing were Gleason Sum (n=111) and Prostate Specific Antigen (PSA) (n=100). The final analytic cohort consisted of 1082 men.

Exposure, Modifier, Covariates, and Outcome

Our primary exposures were weight in young adulthood and weight change from young adulthood to middle age. Weight at age 20 and weight at age 50 were self-reported on the baseline survey and evaluated as continuous variables. Weight change from age 20 to age 50 was evaluated both as a continuous variable and an ordinal variable. For the continuous variable, we report results per 30 kg of weight change, as this is the approximate weight gain from the 5^th to 95^th percentile of weight change. The ordinal variable was coded using the following categories: Weight loss: >10 kg, no change: (≥ −10 kg to <10 kg), and weight gain: (≥10 kg to <20 kg), (≥ 20 kg to <30 kg), or (≥ 30 kg). Weight at diagnosis was determined using medical records and coded as a continuous variable. Smoking status, evaluated primarily as an effect modifier, was based on self-report, and participants were categorized as never, former, or ever smokers (current + former smokers).

Participants self-reported race, and were categorized as “Black” or “White”. Education was based on self-report, and participants were characterized as not having a college degree or being a college graduate or above. Clinical factors including age at diagnosis, Gleason Sum, stage, and PSA at diagnosis were determined using medical records. Age at diagnosis was coded as a continuous variable. Gleason Sum was defined as ≤7 vs. >7, stage was defined as T1 vs.T2/T3, and PSA at diagnosis was defined as <4 vs. ≥4. Primary treatment type was characterized as either surgery (radical prostatectomy) or radiation.

Our primary outcome of interest was biochemical recurrence. Among participants treated with radical prostatectomy, biochemical recurrence was defined as a PSA of 0.2 ng/mL or higher for two consecutive assays without treatment.²² Among participants treated with radiation, biochemical recurrence was defined as rise of 2 ng/mL or more above the nadir achieved after radiation therapy.²³ Secondary treatment for rise in PSA, even less than 0.2 ng/mL, was classified as a biochemical recurrence for both patients treated with radical prostatectomy or radiation. However, receipt of adjuvant therapies without a rise in PSA were not coded as a biochemical recurrence.

Statistical Analysis

Cox Proportional Hazards Models were used to assess the association between weight at age 20, weight at age 50, weight change from age 20 to 50, and weight at diagnosis with biochemical recurrence. Each measure of weight was evaluated in a separate model. In adjusted analyses, models were adjusted for age at diagnosis, race, and education. Models for weight change were additionally adjusted for weight at age 20. In a sensitivity analyses, we additionally adjusted for clinical factors including stage, Gleason sum, and PSA. We chose not to adjust for these factors in our primary models, as clinical factors are mediators on the pathway from exposure to outcome, and it is generally recommended that mediators not be part of the adjustment set.²⁴ Several previous studies have shown that obesity is associated with a higher Gleason sum or more aggressive prostate cancer at diagnosis, and as such it is likely that clinical factors such as Gleason sum are potential mediators. ^{3, 8, 25–27} Effect modification by smoking was examined by stratifying by smoking status.

The proportional hazard assumption for each categorical variable was assessed by examining Kaplan-Meier and log (-log (Survival Probability)) plots. If the plots indicated a potential violation of the proportional hazards assumption, we examined whether the interaction between time to biochemical recurrence and the covariate of interest was statistically significant using an a priori alpha equal to 0.05. If the interaction was significant, the interaction term between time to biochemical recurrence and the covariate was retained in the final model. The proportional hazards assumption for continuous variables was assessed using Schoenfeld Residuals.

Patients that died during follow-up for reasons not related to prostate cancer were censored on date of death. By contrast patients that had a death attributable to prostate cancer, but did not meet our definition of recurrence, were counted as having a recurrence on date of death.

All analyses were conducted using SAS 9.4 (Cary, NC).

RESULTS

Characteristics of Study Population

Characteristics of the analytic cohort (n=1082) can be seen in Table 1. Most participants were white. Over half of the men in our cohort (56.9%) gained more than 10 kg from age 20 to age 50. Approximately half of the cohort were never-smokers, and only 10% of the cohort were current smokers. College graduates or above comprised half the cohort. The majority of the cohort had a Gleason Sum ≤ 7, were diagnosed at Stage T1, and had a PSA of >4 at diagnosis. Surgery was, by far, the predominant treatment type. The mean weight at age 50 (surgery: 88.2 kg; radiation: 87.2 kg) and mean weight at diagnosis (surgery: 92.3 kg; radiation: 93.9 kg) were similar across treatment type.

Table 1.

Participant characteristics of the Washington University Prostate Cancer Prospective Cohort, by recurrence status

	All Participants (n=1082) n(%)	Participants with Recurrence (n=131) n (%)	Participants without Recurrence (n=951) n (%)
Age at diagnosis (Mean, SD)	62.0 (6.3)	61.7 (6.0)	62.0 (6.3)
Race
Black	66 (6.1)	6 (4.6)	60 (6.3)
White	1016 (93.9)	125 (95.4)	891 (93.7)
Weight (kg) at diagnosis (Mean, SD)	92.4 (15.8)	95.7 (15.3)	92.0 (15.8)
Weight (kg) at age 20 (Mean, SD)	74.8 (11.8)	76.7 (14.2)	74.5 (11.4)
Weight (kg) at age 50 (Mean, SD)	88.1 (14.6)	92.4 (17.5)	87.5 (14.0)
Weight change from age 20 to age 50 (kg) (Mean, SD)	+13.4 (11.6)	+15.6 (16.0)	+13.1 (10.7)
Weight change from age 20 to age 50 (kg) (Mean, SD), categorical
Weight loss >10 kg	15 (1.4)	2 (1.5)	13 (1.4)
No change (≥ −10, <10 kg)	451 (41.7)	50 (38.2)	401 (42.2)
Weight gain (≥ 10, <20 kg)	356 (32.9)	36 (27.5)	320 (33.7)
Weight gain (≥ 20, <30 kg)	189 (17.5)	29 (22.1)	160 (16.7)
Weight gain (≥ 30 kg)	71 (6.6)	14 (10.7)	57 (6.0)
Height (cm) (Mean, SD)	178.8 (7.1)	180.2 (6.5)	178.6 (7.1)
Smoking Status
Never	521 (48.2)	60 (45.8)	461 (48.5)
Former	453 (41.9)	53 (40.5)	400 (42.1)
Current	108 (10.0)	18 (13.7)	90 (9.5)
Education
No college degree	525 (48.5)	72 (55.0)	453(47.6)
College graduate or above	557 (51.5)	59 (45.0)	498 (52.4)
Gleason Sum
≤7	673 (62.2)	28 (21.4)	645 (67.8)
>7	409 (37.8)	103 (78.6)	306 (32.2)
Stage
T1	852 (78.7)	83 (63.4)	769 (80.9)
T2/T3	230 (21.3)	48 (36.6)	182 (19.1)
PSA at diagnosis
≤4	260 (24.0)	19 (14.5)	241 (25.3)
>4	822 (76.0)	112 (85.5)	710 (74.7)
Treatment
Surgery	994 (91.9)	123 (93.9)	871 (91.6)
Radiation	88 (8.1)	8 (6.1)	80 (8.4)

Twelve percent of participants experienced biochemical recurrence. Age at diagnosis was similar in both men with biochemical recurrence and without biochemical recurrence. Race, education, and smoking were similarly distributed in both men that experienced biochemical recurrence and men that did not. A weight change of greater or equal to than 20 kg, from age 20 to 50, was significantly more prevalent in men with biochemical recurrence (32.8%) than men without recurrence (22.8%) (p=0.01). Men that experienced biochemical recurrence were more likely to have a Gleason Sum >7, be diagnosed at Stage II/III, and have a PSA ≥ 4 at diagnosis. Surgery was the predominant treatment type in both men with and without biochemical recurrence.

Biochemical Recurrence

The mean biochemical recurrence-free survival time was 5.95 years using the Kaplan-Meir estimator of the biochemical recurrence-free survival probability. Smoking status was found to violate the proportional hazards assumption, and as such we included an interaction term for smoking status and time to biochemical recurrence in models that included the smoking status variable (p-interaction former smoking and time to recurrence: 0.012; p-interaction current smoking and time to recurrence: 0.529). No other study variable was found to violate the proportional hazards assumption.

Thirty-eight patients (3.5% of cohort) died for reasons not related to prostate cancer and were censored on date of death. Only one patient had a death attributable to prostate cancer, and this patient was counted as having a recurrence on date of death.

Weight at diagnosis, weight at age 20, and weight at age 50 were significantly associated with biochemical recurrence in both models adjusted for age at diagnosis and in fully-adjusted models adjusted for age, race, education, smoking status, and the interactions between smoking status and time to recurrence (Table 2). Weight change from age 20 to 50 was significantly associated with biochemical recurrence in models with additional adjustment for weight at age 20 (HR per 30 kg weight change 1.84, 95% CI: 1.24, 2.74, p-trend 0.003). We observed a clear dose-response relationship between weight change from age 20 to age 50 and biochemical recurrence in models where weight change was examined as an ordinal variable (p-trend: 0.013). In sensitivity analysis, with additional adjustment for clinical characteristics (i.e. Gleason sum, stage, and PSA at diagnosis) our results for weight change from age 20 to 50 were consistent with our primary model (HR per 30 kg: 1.51, 95% CI: 1.03, 2.22, p-trend: 0.033).

Table 2.

Association of weight in early adult years, weight gain and prostate cancer recurrence, 131 recurrences, 4829.74 person-years, Washington University Prostate Cancer Prospective Cohort

	# of recurr-ences	Person-years	Age- adjusted Model 3 HR (95% CI)	p trend Age-adjusted models	Adjusted Model 1 4 HR (95% CI)	Adjusted Model 2 4 HR (95% CI)	Adjusted Model 3 4 HR (95% CI)	Adjusted Model 4 5 HR (95% CI)	Adjusted Model 5 5 HR (95% CI)	p trend Adjusted models
Overall	131	4829.74
Weight at Diagnosis (kg)1			1.52 (1.12, 2.04)	0.006	1.53 (1.13, 2.06)					0.006
Weight at age 20 (kg) 1			1.55 (1.02, 2.37)	0.043		1.56 (1.02, 2.38)				0.039
Weight at age 50 (kg)1			1.82 (1.33, 2.49)	<0.001			1.80 (1.32, 2.47)			<0.001
Weight change age 20 to 50 (kg)1			1.71 (1.14, 2.57)	0.001				1.84 (1.24, 2.74)		0.003
Weight change age 20 to 50 (kg)2				0.025						0.013
Weight loss (>10 kg)	2	67.76	0.82 (0.69, 0.98)						0.80 (0.67, 0.95)
No change (≥ −10, <10 kg)	50	2064.79	1.00						1.00
Weight gain (≥ 10, <20 kg)	36	1613.32	1.22 (1.02, 1.46)						1.25 (1.05, 1.50)
Weight gain (≥ 20, <30 kg)	29	803.65	1.50 (1.05, 2.13)						1.57 (1.10, 2.24)
Weight gain (≥ 30 kg)	14	280.22	1.83 (1.08, 3.10)						1.96 (1.15, 3.35)

¹HR per 30 kg weight change. (The approximate interval from the 5^th to the 95^th percentile of weight change was 30 kg).

²Weight change coded as an ordinal variable using the categories listed

³Model adjusted for diagnosis age

⁴Model adjusted for diagnosis age, race, education, smoking status, and interaction terms for smoking status and time to recurrence as smoking status violated the proportional hazards assumption.

⁵Model adjusted for diagnosis age, race, education, smoking status, weight at age 20, and interaction terms for smoking status and time to recurrence as smoking status violated the proportional hazards assumption.

Interaction between smoking status and time to recurrence (age-adjusted): p for interaction former smokers 0.012; p for interaction current smokers 0.529.

To examine effect modification by smoking, we stratified by smoking status (never smokers, former smokers, and ever smokers) (Table 3). Since only 10% of our cohort were current smokers, we did not have adequate statistical power to examine current smokers alone. Weight at age 50 was significantly associated with biochemical recurrence in never smokers, former smokers, and ever smokers in adjusted models. After controlling for weight at age 20, weight change from age 20 to age 50 was significantly associated with biochemical recurrence only among former smokers (HR per 30 kg: 3.87, 95% CI: 1.88, 8.00 p-trend: <0.001) and ever smokers (HR per 30 kg: 2.38, 95% CI: 1.27, 4.45, p-trend: 0.007). By contrast, no signification association was observed in never smokers (HR per 30 kg: 1.50, 95% CI: 0.88, 2.57, p-trend; 0.136).

Table 3.

Effect of weight on prostate cancer recurrence by smoking status, Washington University Prostate Cancer Prospective Cohort

	Number of Recurrences	Person-years	Adjusted Model 1 2 HR (95% CI)	Adjusted Model 2 2 HR (95% CI)	Adjusted Model 3 2 HR (95% CI)	Adjusted Model 4 3 HR (95% CI)	p trend
Never Smokers (n=521)
Overall	60	2334.94
Weight at Diagnosis (kg)1			1.59 (1.06, 2.38)				0.026
Weight at age 20 (kg)1				2.25 (1.30, 3.92)			0.004
Weight at age 50 (kg)1					1.86 (1.24, 2.80)		0.003
Weight change age 20 to 50 (kg)1						1.50 (0.88, 2.57)	0.136
Former Smokers (n=453)
Overall	53	2037.65
Weight at Diagnosis (kg)1			1.76 (1.07, 2.95)				0.025
Weight at age 20 (kg)1				1.35 (0.65, 2.82)			0.418
Weight at age 50 (kg)1					2.58 (1.46, 4.55)		0.001
Weight change age 20 to 50 (kg)1						3.87 (1.88, 8.00)	< 0.001
Ever Smokers (Former + Current Smokers) (n=561)
Overall	71	2494.80
Weight at Diagnosis (kg)1			1.44 (0.92, 2.25)				0.112
Weight at age 20 (kg)1				1.05 (0.57, 1.95)			0.869
Weight at age 50 (kg)1					1.74 (1.06, 2.85)		0.029
Weight change age 20 to 50 (kg)1						2.38 (1.27, 4.45)	0.007

¹HR per 30 kg weight change

²Model adjusted for diagnosis age, race, and education

³Model adjusted for diagnosis age, race, education, and weight at age 20

In a sensitivity analysis, with additional adjustment for clinical variables (Gleason sum, stage, and PSA at prostate cancer diagnosis), results were consistent with our primary models. Weight change from age 20 to 50 was significantly associated with biochemical recurrence in former smokers (HR per 30 kg: 3.85, 95% CI: 1.86, 7.97, p-trend: <0.001) and ever smokers (HR per 30 kg: 2.34, 95% CI: 1.25, 4.39, p-trend: 0.008), but not in never smokers (HR per 30 kg: 1.10, 95% CI: 0.65, 1.85, p-trend: 0.725).

DISCUSSION

In this study of 1082 men diagnosed with prostate cancer, weight in young adulthood (age 20) and middle age (age 50), and weight change from young adulthood to middle age (age 20 to 50) were significantly associated with risk of subsequent biochemical recurrence. In analyses stratified by smoking status, the association between weight change and biochemical recurrence was limited to former and ever smokers. Weight change in never smokers was not significantly associated with increased risk for biochemical recurrence.

A key feature of this study is the focus on weight gain from young adulthood to middle age. We examined weight gain over a 30-year period to age 50, on average, a decade before prostate cancer diagnosis. This period represents a potentially critical period for future prostate cancer carcinogenesis, and by time of diagnosis or treatment key carcinogenic events may have already be initiated. By the time men are aged 19–29, atrophic lesions in the prostate (potential prostate cancer precursor lesions) may be present in up to half the men.^{19, 28} Moreover, by the time men reach their thirties, some will already have evidence of high-grade prostatic intraepithelial neoplasia (HGPIN) and histological prostate cancer.¹⁹

Several previous studies have reported a positive association between obesity at time of diagnosis or time of treatment and risk for biochemical recurrence.^{3–8, 10} However, the few existing studies that have examined the association of pre-diagnostic weight gain and biochemical recurrence have reported inconsistent results.^{21, 29–31} In accordance with our findings, among a cohort of men treated with radical prostatectomy, Strom et al. found that obesity at age 40 and weight change of greater than 1.5 kg/year from age 25 to prostate cancer diagnosis were significantly associated with biochemical recurrence in unadjusted models.³⁰ Moreover, both Joshu et al. and Whitley et al. reported that weight gain in the years immediately prior to radical prostatectomy was positively associated with biochemical recurrence.^{29, 31} In contrast to our study, in each of these studies the endpoint for weight change was either prostate cancer diagnosis or radical prostatectomy.^{21, 29–31} Our endpoint for weight change - age 50 - was, on average, 10 years before prostate cancer diagnosis for the men in our cohort. The 30 year period between our weight measures (age 20 and age 50) encompasses an etiologically relevant time period. As mentioned previously, pre-neoplastic and neoplastic alterations prostate alterations can be observed in men as young as 19–29 years of age.¹⁹

In addition to weight and obesity, several previous studies have observed that smoking is positively associated with biochemical recurrence.^11–18 However, few existing studies have examined effect modification by smoking status on the association between obesity, weight, and biochemical recurrence. A study among a cohort of Korean men found that smoking was significantly associated with biochemical recurrence only among those with a Body Mass Index (BMI) ≥ 25 kg/m² at time of treatment.¹⁵ However, this study was different from ours as the BMI measure is not prior to diagnosis, and the authors analyzed the smoking-biochemical recurrence association among obese men instead of the obesity-biochemical recurrence association among smokers.¹⁵ Moreover, a study among Korean men is not likely generalizable to men from the United States. Indeed, the mean BMI among men in this study is 24 kg/m² while the mean BMI at time of diagnosis for the men in our study was 29 kg/m^2.15 In a large consortium study of white men, Yuan et al. found that the association between pre-diagnostic BMI and prostate cancer death was significantly modified by smoking status.²⁰ In line with this finding, we observed that weight change from age 20 to age 50 was significantly associated with biochemical recurrence only among former smokers or ever smokers. However, biochemical recurrence and mortality are not interchangeable outcomes, and many men with biochemical recurrence do not die from prostate cancer.³² Finally, consistent with our findings the Health Professional’s Follow-up study, reported no association between weight change from age 21 to prostate cancer diagnosis among non-smokers.²¹ However, this study did not report the association between weight change and biochemical recurrence among smokers.²¹ Moreover, the study differs from our study as the endpoint for the weight change was prostate cancer diagnosis whereas we examined weight change, on average, a decade prior to prostate cancer diagnosis.²¹ To our knowledge, this is one of the first studies to examine the potential association of long-term weight gain, several years prior to diagnosis, with risk of biochemical recurrence, specifically among smokers. Our results suggest that smoking is an important effect modifier on the association between weight change in early life and biochemical recurrence.

Potential biological mechanisms can explain the impact of both weight gain and smoking. Briefly, an increased weight or obesity can contribute to hyperinsulinemia and lower testosterone levels; both of which have been associated with aggressive disease at diagnosis and recurrence.^{9, 33–38} Specifically, men with hyperinsulinemia are more likely to have high levels of insulin-like growth factor, a known prostate cancer carcinogen.³⁹ Smoking could induce prostate cancer tumor growth through multiple pathways including hormonal changes (increase in androgen levels), tobacco-related carcinogens (both cadmium and N-nitroso compounds have been associated with prostate tumorigenesis), genetic mutations, tumor angiogenesis, and inflammation (cytokine release).^{40, 41} Moreover, increased weight, obesity, and smoking can all hamper prostate cancer detection. Obesity is associated with benign prostatic hyperplasia (BPH) and less accurate digital rectal exams (DRE), and both smoking and obesity are associated with lower PSA levels.^{34, 41, 42} These factors can all contribute to a delayed diagnosis, more aggressive disease at diagnosis, and thus an increased potential for recurrence.^{34, 41, 42} It can be hypothesized that these factors are compounded in men with dual exposure (i.e. smoking and weight gain) resulting in smokers that gained weight being at greatest risk for biochemical recurrence as was observed in our study.

Our study was limited by the use of self-reported height and weight both at age 20 and age 50. However, middle-aged men have been shown to recall height and weight accurately 27–37 years later.⁴³ Both weight gain and BMI have been shown to be minimally underestimated during this recall period (on average, weight gain is underestimated by 3 kg and BMI is underestimated by 1 kg/m²).⁴³ In addition, we did not have adequate statistical power to examine current smoking alone. However, we do present results for ever smokers, a classification that includes current smokers. Moreover, former smokers were likely smokers in young adulthood or middle age, the same time period in which we are evaluating weight gain. As such it is likely former smokers were smokers in an etiologically relevant time period. In addition, to date only 12% of cohort had experienced biochemical recurrence. The proportion of men with biochemical recurrence will likely increase with increased follow-up time. However, despite this relatively short follow-up time, we were able to observe significant associations with biochemical recurrence for weight at age 20, weight at age 50, and weight gain from age 20 to 50. Finally, we do not have information on PSA screening prior to diagnosis. It is possible that men who received regular PSA screening were healthier (i.e. less likely to smoke or gain weight) than men that were not screened.

Major strengths of this study include a prospective design and the availability of weight measures in young adulthood and middle age, smoking status, and post-diagnostic, follow-up clinical data. This study fills a crucial gap in the current literature on weight and prostate cancer recurrence, by examining weight gain in a critical, etiologically relevant, time period a decade before diagnosis ¹⁹. Moreover, this is one of the first studies to examine effect modification by smoking status on the weight change- biochemical recurrence association. Our results suggest that men who smoke and gain weight from young adulthood to middle age are at increased risk for prostate cancer recurrence. Smokers that have history of weight gain may represent a high-risk group that could benefit from prostate cancer screening.

In conclusion, weight at both age 20 and age 50 and weight gain from age 20 to 50 were each associated with biochemical recurrence in this prospective study. In stratified analysis, weight gain from age 20 to 50 was significantly associated with biochemical recurrence only among former or ever smokers. These data suggest that weight gain during early adult years may be associated with a long term risk for adverse cancer outcomes among prostate cancer patients who received definitive treatment even after additionally controlling for markers of disease aggressiveness at diagnosis. Future studies among diverse populations are needed to further examine exposures in early adulthood and adverse cancer outcomes.

Novelty and Impact.

Obesity in the years preceding diagnosis has previously been shown to increase the risk of adverse prostate cancer outcomes. However, here the authors found that weight gain in young adulthood, a decade prior to diagnosis, is associated with biochemical recurrence among men with prostate cancer. This association was modified by smoking status. These results support weight control initiatives in young adulthood to reduce the risk of adverse cancer outcomes later in life.

Abbreviations Used

BMI

Body Mass Index

BPH

Benign Prostatic Hyperplasia

CI

Confidence Interval

DRE

Digital Rectal Exam

HR

Hazard Ratio

PCPC

Prostate Cancer Prospective Cohort

PSA

Prostate Specific Antigen

SD

Standard Deviation