Lipoproteins, sex hormones and inflammatory markers in association with prostate cancer

Abstract

Objective. To evaluate lipoprotein profile and sex hormones in patients with prostate cancer (PCa) and benign prostatic hyperplasia (BPH) and their possible associations with some inflammatory markers linked to PCa.

Methods. A total of 150 men (50–65 years), matched by age and body mass index (BMI), included in this study and divided into three groups according to total prostate specific antigen (PSA), digital rectal examination and prostate biopsy: 50 PCa, 50 BPH and 50 controls. Total cholesterol (Chol), HDL-chol, LDL-chol, triglycerides (TG), total testosterone (T), free T (FT), bioavailable T (BioT), estradiol and SHBG were measured. The free androgen index (FAI) and TG/HDL-chol were calculated. In 25 PCa and 25 controls, C-reactive protein (hs-CRP), adiponectin and insulin were determined.

Results. Patients with PCa showed higher TG/HDL-chol and diminished HDL-chol than Controls and BPH. PSA correlated inversely with HDL-chol and directly with TG/HDL-chol. FAI, FT, BioT and estradiol levels were higher, and SHBG and adiponectin were lower in PCa than in Controls. No differences were found in androgens between BPH and PCa.

Conclusion. Our most novel findings are that the patients with PCa presented lower total Chol and HDL-chol and higher TG/HDL-chol than BPH and Controls. Patients with PCa showed higher androgens and lower adiponectin than Controls.

Keywords:

• Prostate cancer
• sex hormones
• lipoprotein profile
• inflammation
• adiponectin

Introduction

At present, prostate cancer (PCa) is the third leading cause of death from cancer in men [1]. Epidemiologic data linking PCa risk and overweight/obesity are controversial. Currently, different studies show that Western lifestyle, reduced physical activity, high fat intake are related to PCa risk [2–5]; however, others do not reach the same conclusion [6,7]. A relationship between the degree in obesity – evaluated through body mass index (BMI) – and the severity and progression of PCa has been described [8,9]. Contradictory results have also been found concerning the relationship of benign prostatic hyperplasia (BPH), whose etiology is still not completely understood, with total caloric or fat intake, BMI and diabetes [10,11].

Insulin resistance (IR)/hyperinsulinemic states, usually related with overweight/obesity, are also associated with high PCa risk [12,13], although different factors could be responsible for this correlation. Recently, a relationship between hypertriglyceridemia and remnant lipoproteins with PCa has been reported [14], suggesting that hypertriglyceridemia, characteristic of IR, could be one of the factors associated with this neoplasia. However, the association of lipoprotein profile with PCa has not been thoroughly studied.

As regards to sex hormones, although a recent study has found that they were not linked to PCa risk [15] – considering the hormone dependent nature of these tumors – the ratio of androgens to estrogens might be important in prostatic growth, not only in PCa but in BPH as well. Moreover, a recent study in an experimental animal model has shown that both androgens and estrogens together play a significant role in the induction of inflammation and prostatic cancer [16]. Low testosterone concentrations are associated with higher malignancy and less differentiation of prostatic tumors [17]. However, it has also been suggested that the loss of testosterone could be expected to exert a suppressive effect on PCa development [18].

Chronic inflammation has also been found to correlate with the onset or progression of cancer; however, opposite results specifically related with PCa have been found [19,20]. Adipose tissue produces different adipocytokines, like adiponectin, which has been evaluated in different stages of PCa, though contradictory results have also been published [21,22]. Whether the increase of adipocytokines and other inflammatory markers are related only to obesity or to PCa as well needs to be investigated.

Taking into account the previous considerations, the aims of this study were to evaluate possible associations between the lipoprotein profile and sex hormones in patients with PCa and BPH and to investigate the role of adiponectin and high sensitive C-reactive protein (hs-CRP) as markers of the inflammatory process linked to PCa risk.

Methods

A total number of 3000 men (50–65 years old) completed a prostatic evaluation at the Urology Division, Hospital de Clínicas ‘José de San Martín’, between 2005 and 2007. Digital rectal examination (DRE) and total prostate specific antigen (PSA) test were performed on all of them, and according to the results, 150 men were later recruited for this study. The study groups were as follows: 50 patients with PCa, with PSA ≲ 2.50 ng/ml and <10.00 ng/ml with prostatic adenocarcinoma diagnosed by biopsy, (all the patients presented an organ confined cancer and 80% of them showed a Gleason score ≤ 6); 50 BPH patients, with PSA ≲ 2.50 ng/ml and <10.00 ng/ml with prostatic adenoma diagnosed by biopsy; and a Control group, 50 men without any prostatic disease, with PSA < 2.50 ng/ml and normal DRE. Patients with PCa, BPH and Controls were matched by age and BMI. Patients with other neoplasias, previous prostatic disease, diabetes, thyroid and renal disorders, hormonal therapy or any other drug modifying lipid metabolism were excluded before the groups were conformed. All subjects gave their informed consent, and the original protocol was approved by the Ethics Committee of Facultad de Farmacia y Bioquimica-Universidad de Buenos Aires.

To calculate the BMI, weight and height were obtained for each patient. Blood samples were obtained by venipuncture after 12 h fasting, before performing the DRE. Separated sera were aliquoted and stored at –70°C until processing.

Analytical methods

PSA was determined by Microparticle Enzyme Immunoassay (MEIA, Abbott Laboratories, Chicago, USA) with a coefficient of variation (CV) intra- and interassay of 3.20 and 4.31%, respectively, (lower (L) and upper (U) limits of detection 0.02–50.00 ng/ml).

Total cholesterol (chol) and triglycerides (TG) were measured in a Hitachi 917 autoanalyser by enzymatic colorimetric methods (Roche Diagnostics GmbH, Mannheim, Germany), CV intra- and interassay of 1.1 and 2.2% for Chol and 1.3 and 2.4% for TG, respectively. High-density lipoprotein (HDL)-chol was determined by selective precipitation, and low-density lipoprotein (LDL)-chol was determined by the Friedwald equation. TG/HDL-chol index was calculated as a surrogate marker of IR [23]. Total testosterone (T) and estradiol were measured through radioimmunoassay (RIA) (Coat-A-Count Total Testosterone and Cout-A-Count Estradiol Diagnostic Products Corporation, Los Angeles, CA); L and U limits of detection were 0.7–55 nmol/l for T and 73–13200 pmol/l for estradiol; CV intraassay <7% and CV interassay <12% for T and <7% and <8.1% for estradiol. Sex hormone-binding globulin (SHBG) was performed through a non-competitive chemoluminiscent method (Immulite autoanalyzer, Diagnostic Products Corporation, Los Angeles, CA), L and U limits of detection were 0.2–180 nmol/l, CV intraassay <8% and a CV interassay <13.5%. Free testosterone (FT) and bioavailable testosterone (BioT) were calculated from T and SHBG, according to the law of mass action. In addition, the free androgen index (FAI) was calculated as T/SHBG × 100.

In a subgroup of 25 patients with PCa and 25 Controls, in whom enough serum samples were available; adiponectin, hs-CRP and insulin were also determined. Adiponectin was measured through RIA (LINCO-Research, St Louis, MO), L and U limits of detection were 1–200 ng/ml and the samples are diluted 1/500; CV intraassay 3.9% and interassay 8.5%, hs-CRP by immunoturbidimetry (Tina Quant, Roche Diagnostics Corporation, Indianapolis, IN) CV < 3% and insulin by Micro Particle Enzyme Immunoassay (MEIA, Abbott Laboratories, Chicago, IL), CV intraassay 2.9% and interassay 3.40%.

Statistical analysis

Results are expressed as mean ± standard deviation (SD) or median (range), according to the data distribution. Differences were considered significant at a p-value <0.05. Mean differences among groups were performed by an analysis of variance (one way ANOVA) or using Tukey test as post-hoc analysis. When the analysis was performed between two groups, t-test or Mann–Whitney test were used, according to the data distribution. In addition, correlations between variables were calculated using Spearman test. Multiple regression analysis was performed to assess the correlation of SHBG, androgenic hormones and TG/HDL-chol with PSA and hs-CRP levels using step-up regression to build the model. Simple linear regressions were carried out first on each of the selected variables, and the result accounting for the larger variation was chosen and kept as the first variable. Statistical analysis was performed using SPSS 11.5 software (SPSS, Chicago, IL).

Results

Table I shows that patients with PCa presented, as expected, higher values of PSA than BPH and Controls. Total Chol was significantly lower in patients with PCa and BPH than in Controls (p = 0.05 and 0.012, respectively). HDL-chol was lower in patients with PCa than in BPH and Controls (p = 0.005 and p = 0.001, respectively). Regarding LDL-chol, lower levels were observed in patients with BPH and PCa as compared with Controls; however, significant differences were observed only between BPH and Controls (p = 0.022).

Table I.  PSA, age, BMI and lipoprotein profile in PCa, BPH and Controls.

	PCa (n = 50)	BPH (n = 50)	Controls (n = 50)
PSA (ng/ml)	5.08 ± 0.38*	2.44 ± 0.36^†	0.22 ± 0.36
Age (years)	60 ± 0.6	60 ± 0.6	60 ± 0.6
BMI (kg/m^2)	28 ± 0.8	27 ± 0.7	27 ± 0.8
Total-chol (mmol/l)	4.73 ± 1.14^‡	4.60 ± 1.46^§	5.54 ± 2.05
TG (mmol/l)	1.49 ± 0.67	1.29 ± 0.33	1.50 ± 0.76
HDL-chol (mmol/l)	0.88 ± 0.31^¶,**	1.14 ± 0.42	1.20 ± 0.36
LDL-chol (mmol/l)	3.15 ± 1.09	2.91 ± 1.38^††	3.77 ± 2.08
TG/HDL-chol	1.88 ± 1.22^‡‡	1.31 ± 0.66	1.31 ± 0.83

Results are expressed as mean ± SD.

*p = 0.0001 versus BPH patients and Controls.

^†p = 0.02 versus Controls.

^‡p = 0.05 versus Controls.

^§p = 0.012 versus Controls.

^¶p = 0.005 versus BPH.

**p = 0.001 versus Controls.

^††p = 0.022 versus Controls.

^‡‡p = 0.008 versus BPH patients and Controls.

TG/HDL-chol ratio was higher in patients with PCa than in BPH and Controls (p = 0.008). In addition, PSA correlated negatively with HDL-chol (r = − 0.216, p = 0.013) and LDL-chol (r = − 0.164, p = 0.05) and positively with TG/HDL-chol (r = 0.188, p = 0.032).

With regards to androgenic profile (Table II), there were no differences in T among groups. However, when compared with Controls, patients with PCa and BPH presented higher values of FT (p = 0.002 and p = 0.015, respectively), BioT (p = 0.002 and p = 0.015, respectively) and FAI (p = 0.0001 and p = 0.003, respectively). Regarding SHBG levels, patients with PCa showed lower values than BPH and Controls (p = 0.016), while estradiol levels were higher in patients with PCa than in Controls (p = 0.016). As can be seen in Table III, positive correlations were found between PSA and T (r = 0.165, p = 0.045), BioT (r = 0.318, p = 0.0001), FT (r = 0.315, p = 0.0001), FAI (r = 0.389, p = 0.0001) and estradiol (r = 0.345, p = 0.003). In addition, PSA correlated negatively with SHBG (r = − 0.217, p = 0.01).

Table II.  Androgenic hormones, FAI, SHBG and estradiol in PCa, BPH and Controls.

	PCa (n = 50)	BPH (n = 50)	Controls (n = 50)
T (nmol/l)	16.64 ± 5.89	16.64 ± 5.20	14.56 ± 5.89
FT (pmol/l)	307.5 ± 108.9*	293.3 ± 84.2^†	238.5 ± 98.8
BioT (nmol/l)	7.11 ± 2.53*	6.76 ± 1.94^†	5.48 ± 2.25
FAI	41.6 ± 13.2^‡	39.2 ± 10.7^§	31.2 ± 11.8
SHBG (nmol/l)	39.9 ± 13.5^¶	43.7 ± 14.3	48.4 ± 17.6
Estradiol (pmol/l)	200.1 ± 64.6^¶	190.9 ± 59.1	156.4 ± 55.8

Results are expressed as mean ± SD.

*p = 0.002 versus Controls.

^†p = 0.015 versus Controls.

^‡p = 0.0001 versus Controls.

^§p = 0.003 versus Controls.

^¶p = 0.016 versus Controls.

Table III.  Correlations between PSA and androgenic hormones, estradiol, SHBG and FAI.

	r	p
T	0.165	0.045
FT	0.315	0.0001
BioT	0.318	0.0001
Estradiol	0.345	0.003
SHBG	−0.217	0.01
FAI	0.389	0.0001

After adjustment by SHBG and androgenic hormones in a multivariate linear regression model, the relationship between TG/HDL-chol and PSA remained significant (β = 5.63, p = 0.019).

Finally, in a subgroup of 25 patients with PCa and 25 Controls, adiponectin, insulin and hs-CRP levels were also investigated. As can be seen in Table IV, patients with PCa presented lower values of adiponectin than Controls (p = 0.049); adiponectin also showed an inverse correlation with PSA, which did not reach statistical significance (r = − 0.274, p = 0.08). There were no significant differences between groups, neither in insulin levels, nor in hs-CRP levels; however, a positive correlation was observed between PSA and hs-CRP (r = 0.267, p = 0.031). Moreover, 23% of patients with PCa presented hs-CRP values higher than 3.0 mg/l, while only 4% of Controls showed those values. After adjustment by TG/HDL-chol, the relationship between hs-CRP and PSA remained significant (β = 0.26, p = 0.05). When SHBG was included in the model, hs-CRP and PSA showed a tendency to correlate (β = 3.76, p = 0.06).

Table IV.  Adiponectin, insulin and hs-CRP in patients with PCa and Controls.

	PCa (n = 25)	Controls (n = 25)
Adiponectin (μg/ml)*	10.3 (3.7–28.5)^†	20.5 (4.6–48.5)
Insulin (μUI/ml)	10.9 ± 7.1	10.5 ± 7.2
hs-CRP (mg/l)*	1.2 (0.1–14.9)	0.89 (0.1–3.7)

*Adiponectin and hs-CPR results are expressed as median (range). Insulin are expressed as mean ± SD.

^†p = 0.049.

Discussion

In this study, we present results of lipoprotein profile, sex hormones, adiponectin, insulin and hs-CRP in patients with PCa. The most novel findings of our research are that patients with PCa presented lower total Chol and HDL-chol levels and higher TG/HDL-chol index than BPH and Controls. Furthermore, patients with PCa showed higher sex hormones like BioT, FT, FAI and estradiol and lower levels of SHBG than Controls. In addition, adiponectin levels were lower in patients with PCa than in Controls.

Given that epidemiological evidence that links PCa and obesity is still inconclusive and controversial [4,11,24,25] and considering the strong effect of overweight/obesity on metabolic status, we matched subjects according to BMI. Mean BMI was in the overweight range (between 25 and 29 kg/m²) in the three groups studied.

Data linking serum lipid/lipoproteins levels and PCa development are limited. We consider that the total Chol decrease in patients with PCa in relation to Controls, observed in this study, is due to a HDL-chol reduction. Meanwhile, some authors found no differences in lipoprotein profile among patients with PCa, BPH and Controls [26], in recent studies in patients with PCa [27] and in men who died of PCa [28], a decrease in HDL-chol levels was found, although a decline in total Chol levels was not observed. Moreover, an increase in LDL-chol was also reported; however, we found that this lipoprotein showed a tendency to be lower. As regards TG plasma concentration, in contrast to data obtained by Wuermli et al. [29], we found no differences among groups, as well as no correlation between TG and PSA levels. However, our results evidence a suggestive increase in the ratio TG/HDL-chol, which is nowadays considered an attractive surrogate marker of IR [23].

Regarding sex hormones, in this study we found that patients with PCa showed higher levels of FT, BioT, FAI and lower SHBG than Controls. Given that total T did not show significant differences between these two groups, the differences seen with FT and BioT must be due to the differences in levels of SHBG. A recently published collaborative analysis on worldwide data about endogenous hormones and PCa risk showed that serum concentrations of sex hormones were not associated with the risk of this type of cancer [15]. However, in this collaboration study the assessment of hormone levels was done pre-diagnostically, while in our study hormonal measurements were performed at the time of diagnosis. Nevertheless, studies are still controversial; in particular, Travis et al. [30] find a positive correlation between FT and PCa risk in young men, while others do not find significant associations with androgens [31].

Obesity may contribute to hormonal changes through the decrease of SHBG levels and concomitant increase of bioavailable androgens and also by enhancing peripheral aromatisation of androgens to estrogens. There is some evidence that estrogens, either alone or together with androgens, can induce aberrant prostatic growth and subsequent neoplasic transformation [32]. In our study, the relationship found between sex hormones and PSA was significant, suggesting that the effect of sex hormones on PCa risk could be partly independent of BMI, considering that there were no differences in this parameter between groups.

Given that sex steroids are known to modulate serum lipoproteins, we evaluated the effect of those confounding variables, as independent factors, on the relationship between PSA and TG/HDL-chol. The correlation observed between these parameters continued being significant after the multiple regression analysis.

We did not observe higher levels of hs-CRP in PCa patients compared to Controls; however, almost a quarter of the subjects presented hs-CRP values over 3.0 mg/l, the cut-off value indicating high relative risk category [33], against 4% in the Control group. Conflicting results have been published about circulating levels of inflammatory markers and PCa. Some authors found that inflammatory markers are more strongly associated with cancer death risk than with cancer incidence [19,34]; while others [35] found a strong association between hs-CRP and PSA. Hs-CRP was found to be associated with the risk of developing lung [36] or colon cancer [20] but not PCa. In our study, the association observed between hs-CRP and PSA persisted significantly even after adjustment made by TG/HDL-chol, suggesting that this relationship could be independent of changes in lipoproteins.

We observed lower adiponectin levels in patients with PCa than in Controls. Regarding this adipocytokine, there are also inconsistent results. Some authors found lower levels of adiponectin in PCa patients in comparison to controls; moreover, they described that patients with advanced cancer presented the lowest levels of this adipocytokine [21,26,37]. On the other hand, Housa et al. [22] have recently reported that serum adiponectin levels are higher in locally advanced PCa than in organ-confined PCa and may thus serve as an auxiliary marker providing further improvement for discrimination between different tumor stages.

Given the high prevalence of BPH in adult men, we were interested to include this group of patients in our study. In patients with BPH, TG and HDL-chol showed similar results with those observed by Parsons et al. [11]; however, in our study, lower levels of LDL-chol were observed in comparison to Controls. Androgens, which are essential in growth and development of the prostatic gland, could have an important role in this disease [38]; there is also some evidence showing that metabolic alterations could be implicated in the pathogenesis of BPH [39]. Considering hormone profile, we observed an increase in free androgens in patients with BPH as compared to Controls.

Our study has some limitations, such as the cross-sectional design, the lack of glucose levels data and the evaluation of body fat distribution, the latter being more significant than BMI alone in assessing obesity. Consequently, we were not able to evaluate IR state, which would have been very important, given that we found that other parameters characteristics of IR state were altered. Using strict inclusion criteria, this study included 150 patients; we consider that for future studies a larger sample should be used.

In conclusion, our study provides new confirmatory data about PCa and its strong associations with metabolic, hormonal and inflammatory profile. Nevertheless, further studies are necessary to confirm the potential mechanisms involved.

Acknowledgements

This work was supported by grants from University of Buenos Aires (B401) and Roemmers Fundation (2006–2008).