Hypertension, obesity and prostate cancer biochemical recurrence after radical prostatectomy

Abstract

Background

The metabolic syndrome (MetS) comprises a constellation of risk factors associated with an increased risk for cardiovascular disease. Components of MetS have emerged as putative risk factors for prostate carcinoma. In this study, we examine the association between three features of the MetS (obesity, hypertension and diabetes) and the risk of biochemical recurrence (BCR) after radical prostatectomy (RP).

Methods

We examined data from 1428 men in the University of Michigan Prostate Cancer Data Bank who elected to have RP as their primary treatment. We calculated body mass index from patients' weight and height measured at the time of prostate cancer diagnosis. We used the University of Michigan's Electronic Medical Record Search Engine to identify subjects with hypertension and/or diabetes before their prostate cancer diagnosis.

Results

Of 1428 men who underwent RP, 107 (8%) subsequently developed BCR with a median length of follow-up post-surgery of 3.6 years. Obesity and hypertension were each associated with an increased risk of BCR (adjusted hazard ratio (aHR) = 1.37; 95% CI 0.92–2.09 and aHR = 1.51, 95% CI 1.01–2.26), whereas no association was observed between diabetes and BCR (aHR = 0.73; 95% CI 0.40–1.33).

Conclusions

Obesity and hypertension were each associated with an increased risk for BCR of prostate cancer after RP, independent of age at diagnosis and tumor pathological features. Given the increasing rates of obesity, hypertension and prostate cancer, a better understanding of the relationship between these entities is of significant public health importance. Elucidation of the involved pathogenic mechanisms will be needed to establish causality.

Introduction

The metabolic syndrome (MetS) comprises a constellation of risk factors associated with an increased risk for cardiovascular disease. This cluster of metabolic derangements is also referred to as ‘insulin resistance syndrome’, emphasizing the critical pathophysiological role of subnormal sensitivity to the beneficial metabolic and vascular effects of insulin.¹ The syndrome is common, with an age-adjusted prevalence of 34.2% based on findings from the combined National Nutrition and Health Examination Survey, 1999–2006.² However, a variety of differing definitions for MetS influence prevalence estimates.³ The definition of MetS developed by the National Cholesterol Education Program (NCEP) is commonly used⁴ and includes any three of the following five features: (1) abdominal obesity (waist circumference > 102 cm in men and > 88 cm in women); (2) blood pressure ≥ 130/85 mm Hg; (3) high-density lipoprotein cholesterol < 40 mg dl⁻¹ in men and < 50 mg dl⁻¹ in women; (4) triglycerides > 150 mg dl⁻¹; and (5) fasting glucose > 110 mg dl⁻¹. The World Health Organization (WHO) has expanded the obesity criterion to include patients with a body mass index (BMI) > 30 kg m⁻². The frequency of MetS occurrence was recently found to be similar using the NCEP and WHO definitions.⁵

The MetS as well as a number of its individual components have emerged as putative risk factors for prostate carcinoma.^6–9 Results from the Flint Men's Health Study indicate that both abdominal obesity and hypertension were independently related to prostate cancer in African–American men.⁸ A longitudinal case–control study in 2007 showed that men in the highest quartile of systolic blood pressure (>150 mm Hg) had a modest increase in risk of prostate cancer.¹⁰ More recently, hypertension was shown to contribute to poorer prognosis in surgically treated prostate cancer patients.¹¹ Obesity has been consistently associated with an increased risk for a variety of malignancies, most notably breast and colon cancer.¹² A growing body of evidence suggests that obesity also confers an increased risk for aggressive or high-grade prostate cancer and is associated with poorer quality of life, post-treatment outcomes and higher mortality.^13–15

Given the increase in prevalence of certain components of the MetS (obesity, hypertension and diabetes) in the United States over the past decade,² further elucidation of their relationships with prostate cancer risk and progression of disease is clearly warranted. We hypothesize that these components are each associated with an increased risk for disease relapse after definitive treatment. In this study, we examine the association between obesity, hypertension and diabetes and risk of biochemical recurrence (BCR) after radical prostatectomy (RP), an area where data are currently limited.

Materials and Methods

Study participants

The University of Michigan Prostate Cancer Data Bank is a repository of tissue, blood, urine, DNA and clinicopathological information used for prostate cancer research. The 1428 patients eligible for this particular study had histologically confirmed adenocarcinoma of the prostate and elected to have RP as their primary treatment between 1994 and 2007. Informed consent was obtained from all patients before inclusion in the database and the Institutional Review Board at the University of Michigan approved all protocols. All patients were treated in the University of Michigan Health System. Available data on patients used in this investigation included pathologic stage (TNM), pathologic Gleason grade, postoperative surgical margin status (negative/focal/extensive) and the use of neoadjuvant androgen deprivation therapy before RP. Subjects who received preoperative radiation therapy were excluded from our analysis. PSA levels were checked postoperatively at 6 weeks, and then every 6 months thereafter with a median length of follow-up post-surgery of 43 months (3.6 years). PSA concentrations are generally expected to become undetectable within 4 weeks after RP.¹⁶ Subjects who did not reach a PSA nadir of < 0.2 ng ml⁻¹ by 6 weeks post-RP (suggestive of residual disease) were excluded. BCR was defined as two consecutive rising postoperative PSA values > 0.2 ng ml⁻¹.¹⁷

Assessment of obesity, hypertension and diabetes

BMI was used as a surrogate measure for waist circumference to identify subjects who were obese. We calculated BMI (in kg m⁻²) for each patient from weight and height measured and recorded at the time of prostate cancer diagnosis. We used the WHO-defined cut-point to classify patients as either non-obese (BMI < 30 kg m⁻²) or obese (BMI ≥ 30 kg m⁻²).⁵ We used the Electronic Medical Record Search Engine (EMERSE) to identify subjects with hypertension and/or diabetes (diagnosed at any point before prostate cancer diagnosis). EMERSE was developed at the University of Michigan for the purposes of securely searching the hospital's electronic medical record system (CareWeb) for research and data abstraction.¹⁸ EMERSE is able to search the medical record in its entirety, including the problem summary list, patient notes, and pathology and radiology reports. It also has the ability to look for potential spelling errors in documents, allowing detection of misspelled words. An unlimited number of search terms can be queried. A complete list of the search terms used to determine hypertension and diabetes is available in supplementary material (See Supplementary Appendix). The most frequently observed terms to determine hypertension were: (1) hypertension; (2) HTN; (3) hypertensive; (4) high BP; (5) high blood pressure; and the most frequently observed terms to determine diabetes were: (1) diabetes; (2) DM; (3) NIDDM; (4) type-2 diabetes; (5) diabetic. A list of medications (both generic and brand name) was included in the search, allowing us to also identify subjects as hypertensive or diabetic based on the medications they were taking (See Supplementary Appendix).

Statistical analysis

All statistical analyses were performed using Statistical Analysis Software (SAS Institute, v. 9.1, Cary, NC, USA). Crude associations between patient (age, race, hypertension, BMI, diabetes) and tumor (TNM stage, Gleason grade, surgical margin status) characteristics and prostate cancer recurrence were tested using either a χ² test for categorical variables or the Wilcoxon rank-sum test for continuous variables. Kaplan–Meier plots were used in preliminary analyses to verify the proportional hazards assumption inherent to the Cox model and all variables were determined to be consistent with this assumption. Cox proportional hazards regression was used to estimate adjusted hazard ratios (aHRs) and 95% CIs for disease recurrence associated with each MetS component controlling for potential confounders. Subjects were followed from time of diagnosis until date of recurrence, date of death or loss to follow-up or end of study (22 September 2008). Final models simultaneously adjusted for age at diagnosis, tumor stage, surgical margin status, Gleason sum, and history of diabetes, hypertension, and obesity (BMI ≥ 30 kg m⁻²). Inclusion of neoadjuvant androgen deprivation therapy before RP in the models did not appreciably change the HRs and was therefore not included in the analysis. Models were also constructed to estimate the risk of recurrence with various combinations of the three components, however, the only group of patients large enough to produce a robust measure of risk were those patients who were both hypertensive and obese.

Results

The baseline characteristics of the entire cohort are shown in Table 1, with the associations between these features and BCR presented in Table 2. Of 1428 men diagnosed with prostate cancer who underwent RP as their primary treatment, 107 (8%) recurred based upon our definition. The average age at diagnosis among patients in the cohort was 59 years (± 7.3 years) with no difference in age at diagnosis between patients with and without evidence of disease recurrence. Eighty-one percent of the total patient sample was Caucasian, 8% African American and the remaining 11% of patients were of either other or unknown race, with no difference in the distribution by race between patients with and without recurrence. Patients with evidence of disease recurrence were more likely to have an advanced tumor stage at time of diagnosis, more aggressive surgical Gleason sum and have extensive positive surgical margins (Table 2).

Table 1. Baseline frequencies or means of patient age, tumor characteristics and metabolic syndrome features according to biochemical recurrence status.

Characteristic	N (%) or mean (s.d.)
Total	1428
Age (years)	59.1 (7.3)
Surgical margins
Negative	1213 (84.9)
Focal	176 (12.3)
Extensive	39 (2.7)
T stage
T2A	329 (23.3)
T2B	872 (61.8)
T3A	159 (11.3)
T3B	44 (3.1)
T4	8 (0.6)
Gleason sum
≤6	403 (28.2)
7	929 (65.1)
≥8	96 (6.7)
Obesity
No	961 (67.3)
Yes	467 (32.7)
Diabetes
No	1245 (87.2)
Yes	183 (12.8)
Hypertension
No	735 (51.5)
Yes	693 (48.5)

Table 2. Adjusted^a hazard ratios (95% CIs) for biochemical recurrence associated with patient age, tumor characteristics and metabolic syndrome features.

Characteristic	Recurrence N (%) or mean (s.d.)	No recurrence N (%) or mean (s.d.)	Hazard ratio (95% CI)
Total	107 (7.5%)	1321 (92.5%)
Age (years)	59 (7.4)	59.1 (7.3)	0.99 (0.96, 1.02)
Surgical margins
Negative	76 (71.0)	1137 (86.1)	1.0
Focal	18 (16.8)	158 (12.0)	1.02 (0.60, 1.75)
Extensive	13 (12.2)	26 (2.0)	2.19 (1.14, 4.22)
T stage
T2A	12 (11.2)	317 (24.3)	1.0
T2B	49 (45.8)	823 (63.1)	1.31 (0.69, 2.49)
T3A	28 (26.2)	131 (10.0)	2.75 (1.35, 5.60)
T3B	15 (14.0)	29 (2.2)	4.74 (2.05, 11.0)
T4	3 (2.8)	4 (0.4)	7.60 (1.93, 29.9)
Gleason sum
≤6	9 (8.4)	394 (29.8)	1.0
7	70 (65.4)	859 (65.0)	2.73 (1.34, 5.57)
≥8	28 (26.2)	68 (5.2)	7.36 (3.17, 17.1)
Obesity
No	65 (60.8)	896 (67.8)	1.0
Yes	42 (39.3)	425 (32.2)	1.37 (0.92, 2.06)
Diabetes
No	94 (87.9)	1151 (87.1)	1.0
Yes	13 (12.2)	170 (12.9)	0.73 (0.40, 1.33)
Hypertension
No	43 (40.2)	692 (52.4)	1.0
Yes	64 (59.8)	629 (47.6)	1.51 (1.01, 2.26)

Abbreviation: CI, confidence interval.

^aHazard ratios adjusted for age, surgical margin status, tumor stage, Gleason sum and metabolic syndrome features.

A total of 467 patients (33%) were characterized as obese (BMI ≥ 30 kg m⁻²). After adjustment for age at diagnosis, tumor stage, surgical margin status, Gleason sum, and the remaining two MetS features (hypertension, diabetes), obesity was associated with a modest increase in risk of BCR (aHR = 1.37; 95% CI 0.92–2.06). We identified 693 men in the dataset with hypertension (49%) and these patients had an approximate 50% greater risk of recurrence compared with patients classified as normotensive (aHR = 1.51, 95% CI 1.01–2.26). A history of diabetes was documented in 183 patients (13%), and although there was an inverse association between diabetes and BCR, the results were far from achieving statistical significance (aHR = 0.73; 95% CI 0.40–1.33). Furthermore, subjects who were both obese and hypertensive had a more than twofold increase in risk of recurrence, (aHR = 2.07; 95% CI 1.23–3.53) (Table 3). Unfortunately, the number of subjects in the cohort with all three of the measured components was too small to analyze with BCR risk.

Table 3. Risk of biochemical recurrence in men with hypertension and obesity.

Features of metabolic syndrome	Recurrence N (%)	No recurrence N (%)	Adjusted HRa(95% CI)
Neither HTN or obese	30 (28.0)	522 (39.5)	1.0
Exhibits either HTN or obesity	48 (44.9)	544 (41.2)	1.40 (0.87, 2.23)
Exhibits both HTN and obesity	29 (27.1)	255 (19.3)	2.07 (1.23, 3.53)

Abbreviations: CI, confidence interval; HR, hazard ratio; HTN, hypertension.

^aHazard ratios adjusted for age, surgical margin status, T stage, Gleason sum and diabetes status.

Discussion

Prostate cancer is the most common malignancy diagnosed among men in the United States and a major cause of cancer-related morbidity and mortality.¹⁹ Incidence increased dramatically due to the introduction of widespread prostate cancer screening in the late 1980s and early 1990s.²⁰ Because of such screening efforts, the malignancy is often still confined to the prostate gland itself at the time of diagnosis. RP has been shown to reduce mortality, risk of metastasis and local tumor progression in patients with organ-confined disease.²¹ Nonetheless, a significant proportion of patients develop recurrent disease post-surgery.^22,23

Although age, race, and family history are well-established risk factors for prostate cancer, both obesity and the MetS have emerged as additional putative risk factors.²⁴ We observed that obesity was associated with a modest increased risk of BCR after RP. This finding is consistent with previously reported post-RP data by Freedland et al.²⁵ In the past decade, a consistent association has been observed between higher BMI and prostate cancer aggressiveness, progression and mortality.^26–29 Our data add to this growing body of evidence. Our observation may be explained at least in part by the fact that surgery is more technically challenging in the obese or morbidly obese patient, or due to delayed detection of depressed PSA concentrations among obese men.³⁰ However, our increased recurrence rate was independent of surgical margin status, arguing in favor of a biological mechanism by which obesity confers risk.

Our results also indicate that subjects with hypertension were more likely to experience recurrence. There is currently a limited amount of data in the literature supporting a link between high blood pressure and prostate cancer. Earlier observational studies showed no statistically significant association between hypertension and prostate cancer incidence.^31,32 However, the use of antihypertensive medication was noted to be inversely associated with incident prostate cancer.³² Hammarsten and Högstedt³³ were the first to report an association between hypertension and advanced tumor stage in a case-series of 299 men with prostate cancer. More recently, a community-based case–control study of African Americans reported a 2.4-fold increased risk in prostate cancer incidence among subjects with hypertension.⁸ Our findings are consistent with those of a recent study by Post et al.,¹¹ which reported that hypertension increased the risk for BCR by more than twofold in the post-RP setting. Our findings do not suggest effect modification between hypertension and obesity on the risk of BCR, in that the risk among patients who were both hypertensive and obese was similar to the additive risks associated with each individual feature.

We did not observe diabetes to be associated with an increased risk for prostate cancer recurrence in this population. In fact, multiple prospective and case–control studies have reported diabetes to be associated with a reduced risk for prostate cancer, despite its direct ties to obesity and insulin resistance.³⁴ Although this may at first seem counterintuitive, it is important to consider the complex nature of diabetes. Insulin concentrations vary considerably throughout the disease's natural history. Although diabetic men may initially be hyperinsulinemic (a consequence of insulin resistance), they ultimately become insulin deficient on account of irreversibly damaged pancreatic β-cells. Given the indolent nature of prostate cancer, it is certainly possible that β-cell function has been lost or greatly reduced by the time of clinical presentation.³⁵ This is challenging to measure epidemiologically and further investigation is necessary.

The pathogenic mechanisms potentially linking obesity and other components of the MetS to prostate carcinogenesis are poorly understood. The central derangement of MetS—a defect in insulin-stimulated glucose uptake with secondary hyperinsulinemia—may in fact be present in patients with prostate cancer.³² Hyperin-sulinemic men have been shown to have higher circulating levels of insulin-like growth factor 1 (IGF-1), a putative prostate cancer mitogen.³⁶ IGF-1 has a role in cellular proliferation and apoptosis reduction; thus, increased bioavailability of IGF-1 may be relevant to prostate cancer incidence and aggressiveness, though data demonstrating causality is lacking.³⁷

Another mechanism by which prostate cancer and features of MetS may be linked is chronic inflammation. There is a substantial amount of epidemiological data correlating prostatic inflammation with prostate malignancy.³⁸ For example, prostatitis and sexually transmitted infections have been associated with increased prostate cancer risk, whereas exposure to antioxidants and anti-inflammatory agents is associated with decreased risk. We also understand that the MetS is itself a pro-inflammatory state. The enhanced risk for coronary artery disease, which MetS is known to confer, is mediated by vascular inflammation and subsequent endothelial dysfunction.³⁹ Hypertension itself has been linked to chronic inflammation and oxidative stress, while excess reactive oxygen species are thought to promote prostate cancer cell growth.⁴⁰ Furthermore, adipose tissue is now understood to actively secrete inflammatory cytokines (TNF-alpha, IL-6, CRP) and hormones (leptin, adiponectin) that have been shown to enhance tumor growth. Leptin in particular is thought to influence prostate cancer cell differentiation and progression.⁴¹ Chronic, sub-clinical inflammation has in turn been linked to the pathogenesis of insulin resistance.⁴² Thus, obesity and MetS likely affect prostate carcinogenesis via a complex interaction between altered androgen metabolism, insulin resistance, and chronic inflammation. Further study is needed to clarify these pathways.

We acknowledge that our study has certain limitations. First, we note that comorbidities may not be accurately reported in the electronic medical record. For example, we were unable to account for subjects who fail to report a history of hypertension or diabetes to their medical providers. In this regard, we were reliant on self-reported information, introducing the possibility of misclassification bias. The prevalence of both diabetes (13%) and hypertension (49%) reported in this study are lower than age-specific estimates from the general population of men,⁴³ suggesting that underreporting may be an issue. However, the likelihood of differential underreporting of medical history before the time of diagnosis between patients who recurred and did not is unlikely, so that any bias introduced by underreport of these conditions would bias findings toward the null. Furthermore, findings published by Klabunde et al.⁴⁴ suggest that men with recent prostate cancer diagnoses are generally capable of providing reliable and accurate comorbidity information. Additionally, we feel that by including a comprehensive list of medications in our EMERSE search, risk for misclassification was minimized. Another potential limitation is the racial homogeneity of our cohort, with 81% of subjects being Caucasian and only 8% African American, and therefore these findings may not be generalizable to the African–American patient. This may be significant, as African–American men have roughly a 1.6-fold greater risk of being diagnosed with prostate cancer (compared with Caucasian men), as well as a 2.4-fold greater chance of dying from the disease.⁸ Furthermore, hypertension and diabetes are more prevalent in the African–American community.⁴⁵ However, Post et al.¹¹ recently demonstrated that despite the racial differences in prevalence of individual MetS components, the observed associations with BCR did not differ appreciably by race. We also acknowledge that our study examines only three out of the five components of MetS. Dyslipidemia and hypertriglyceridemia were not addressed in this study, because we did not have consistent and reliable knowledge of the lipid status of our subjects. Last, we used BMI to define abdominal obesity, rather than waist circumference (which is traditionally used in the MetS definition), because we did not have access to waist measurements. However, BMI > 30 kg m⁻² is an accepted criterion for MetS according to the WHO definition.⁵

In conclusion, we provide evidence for the association between hypertension and obesity and risk of BCR after RP for prostate cancer using a large sample of patients from a tertiary care center. Further investigation should be performed to confirm these findings, as hypertension and obesity are potentially modifiable risk factors for post-surgical recurrence in men with prostate cancer.