Association of serum carotenoids and retinoids with intraprostatic inflammation in men without prostate cancer or clinical indication for biopsy in the placebo arm of the Prostate Cancer Prevention Trial

Abstract

Non-supplemental carotenoids and retinol may potentiate antioxidant and anti-inflammatory mechanisms. Chronic intraprostatic inflammation is linked to prostate carcinogenesis. We investigated the association of circulating carotenoids and retinol with intraprostatic inflammation in benign tissue. We included 235 men from the Prostate Cancer Prevention Trial placebo arm who had a negative end-of-study biopsy, most (92.8%) done without clinical indication. α-carotene, β-carotene, β-cryptoxanthin, lycopene, and retinol were assessed by high-performance liquid chromatography using pooled year 1 and 4 serum. Presence and extent of intraprostatic inflammation in benign tissue was assessed in 3 (of 6–10) biopsy cores. Logistic (any core with inflammation vs none) and polytomous logistic (some or all cores with inflammation vs none) regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI) of intraprostatic inflammation by concentration tertile adjusting for age, race, prostate cancer family history, and serum cholesterol. None of the carotenoids or retinol was associated with intraprostatic inflammation, except β-cryptoxanthin, which appeared to be positively associated with any core with inflammation [vs none, T2: OR (95% CI)=2.67 (1.19,5.99); T3: 1.80 (0.84,3.82), p-trend=0.12]. These findings suggest that common circulating carotenoids and retinol are not useful dietary intervention targets for preventing prostate cancer via modulating intraprostatic inflammation.

Introduction

There has been a long-standing research interest in carotenoids, especially lycopene, and retinol as potential protective factors against prostate cancer ^1–18. Carotenoids are singlet oxygen and peroxyl radical scavengers that can inhibit inflammatory mediators and activate antioxidants by influencing intracellular signaling ^{19; 20}. In addition to their antioxidant activity ²¹, retinoids modulate the immune system ^{22; 23}. Given the biological effects of carotenoids and retinol, the hypothesis that higher intake primarily from high-dose dietary supplements or serum concentrations of carotenoids and retinol would reduce prostate cancer risk has been repeatedly investigated ⁸. However, it is now thought that dose may be relevant, and beneficial effects of carotenoids, including β-carotene (which can be converted to retinol), are observed primarily in doses obtained from a standard diet, but that very high doses from dietary supplements, especially in a non-oxidative environment, may be harmful ²⁰. Randomized, placebo controlled trials have found that supplemental β-carotene doses may increase risk of cancer, in particular lung cancer, in men with more oxidative potential/stress, for example in smokers ^{24; 25}. Taking together trials and epidemiologic observational studies, regardless of dose and environmental context, findings on the association of blood carotenoid and retinol levels, including lycopene, with risk of prostate cancer are mixed ^{1–9; 11; 13–17; 26; 27}.

Increasing evidence suggests that intraprostatic inflammation contributes to the development and progression of prostate cancer ^28–30. Whether carotenoids and retinol influence intraprostatic inflammation has not been addressed. Few epidemiologic studies have been able to investigate intraprostatic inflammation in the etiology of prostate cancer in men without a clinical indication for biopsy ^{29; 30}. Restricting to men without an indication for biopsy is critical to avoiding bias, because intraprostatic inflammation can cause elevated PSA, which when used for prostate cancer screening, increases the likelihood of prostate biopsy and thus, occult prostate cancer detection. The only two such studies to our knowledge, one in the Prostate Cancer Prevention Trial (PCPT) and one linking PCPT and the Selenium and Vitamin E Cancer Prevention Trial (SELECT), restricted to men without clinical indication for biopsy, support the intraprostatic inflammation-prostate cancer hypothesis ^{29; 30}.

We examined the association of serum concentrations of carotenoids and retinol with intraprostatic inflammation in benign tissue in men without an indication for biopsy in the PCPT. We hypothesized that circulating concentrations of α-carotene, β-carotene, β-cryptoxanthin, lycopene, and their sum are inversely associated with intraprostatic inflammation. Given that serum retinol concentrations are homeostatically regulated ³¹, that retinol has either been not associated or positively associated with other cancers in epidemiologic studies ^{1; 3–5; 7; 9; 11–17}, that circulating retinol concentrations were positively associated with prostate cancer risk in a worldwide-pooled collaborative study ⁸, and that in the PCPT the positive association between circulating retinol and high-grade prostate cancer was stronger in vitamin A/β-carotene supplement users ¹², we hypothesized that serum retinol would not be associated with intraprostatic inflammation.

Methods

This study was approved by the Institutional Review Board at The Johns Hopkins Bloomberg School of Public Health (IRB00011055).

Study Population:

We used data from a subset of men randomized to the placebo arm of the PCPT who were not diagnosed with prostate cancer on the end-of-study biopsy (controls). The PCPT was a randomized, placebo-controlled trial that examined the potential of finasteride to reduce risk of prostate cancer over seven years ³². The PCPT randomized 18,880 men from 1994 to 1997 over 55 years old with prostate-specific antigen (PSA) ≤3 ng/mL, a normal digital-rectal examination (DRE), and a score <20 on the American Urological Association Symptom Index. Men were screened for prostate cancer at each annual visit. During the trial, if PSA >4 ng/mL or the DRE was suspicious for prostate cancer (i.e., abnormal) a biopsy was recommended. At the end of the trial and regardless of serum PSA concentration or DRE result, men who were not previously diagnosed with prostate cancer were asked to undergo a biopsy. This study includes 235 controls from the placebo arm who were part of a nested case-control study and had available measures of serum carotenoids, retinol, and intraprostatic inflammation ³³. In that nested case-control study, non-white controls were oversampled; and cases and controls were frequency matched on age at randomization, family history of prostate cancer, and treatment arm ³³. Of the 235 men, 92.8% did not have a clinical indication for biopsy (e.g., elevated PSA or suspicious DRE) and was performed per trial protocol.

Serum Carotenoids, Retinol, and Cholesterol Measurement:

Serum concentrations (μg/mL) of α-carotene, β-carotene, β-cryptoxanthin, lycopene, and retinol were measured by high performance liquid chromatography in serum pooled from years 1 and 4 as previously described ^{10; 12}. Coefficients of variation for pooled quality control samples were 15% for β-cryptoxanthin and 13% for all other carotenoids, and 8% for retinol ^{10; 12}. We summed α-carotene, β-carotene, β-cryptoxanthin, and lycopene concentrations. We categorized concentrations in tertiles based on the distribution in the 235 controls. Serum concentration of total cholesterol was measured on the Roche Cobas Mira Plus Chemistry Analyzer ¹².

Intraprostatic Inflammation Assessment:

Inflammation was assessed in benign tissue in prostate biopsy cores by a pathologist trained to review for inflammation ³⁰. Six to ten biopsy cores were collected per man. Of these one or more were embedded in each tissue block. We selected tissue blocks such that we had three cores per man for review. One hematoxylin and eosin stained section per tissue block was digitally imaged and visually reviewed by a single pathologist ³⁰. Few men exhibited acute inflammation (e.g., polymorphonuclear cells), thus most of the inflammation that was present was chronic (e.g., lymphocytes, macrophages). The presence of inflammation was expressed as at least one biopsy core with inflammation (any biopsy core with inflammation) and the extent of inflammation as none, some, or all biopsy cores with inflammation, as previously described ¹².

Covariates:

At trial entry, men had a clinic visit and completed a questionnaire, which included age, race, smoking history, history of diabetes, and other lifestyle and medical factors. Weight and height were measured by trained staff at study entry, and weight was measured annually. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Participants were asked how physically active (i.e., walking, swimming, dancing, recreational sports) they were in the past four weeks and were classified as sedentary, light, moderate, or active. Serum cholesterol concentration was measured at baseline. Participants were asked to report on frequency of use of multivitamins and single supplements in the past 12 months. We classified men as regular users of a vitamin A supplement if they reported use at least 3 times per week. We classified men as daily beta-carotene supplement users if their dose of beta-carotene from any type of supplement was >0 μg/day.

Statistical analyses:

We natural logarithm transformed serum carotenoid and retinol concentrations and calculated least square geometric mean concentrations and 95% confidence intervals (CI) by the presence or extent of inflammation using linear regression adjusting for age, race (White, non-White), family history of prostate cancer, and serum cholesterol (as may impact serum concentration of carotenoids in circulation ³⁴). Results were comparable after further adjustment for BMI (tertiles: <25.80, 25.80-<28.66, ≥28.66 kg/m²), height, diabetes, smoking status; thus only age-race-family history adjusted results are presented. To test for linear trend in geometric mean concentrations across increasing extent of inflammation, we entered into a linear regression model extent of inflammation as an ordinal variable with the natural logarithm carotenoid or retinol concentration as the dependent variable, and tested the coefficient for the ordinal variable using the Type III Sums of Squares F test statistic. Logistic regression (presence) or nominal polytomous logistic regression (extent) was used to estimate the association of tertiles of concentrations of the carotenoids or retinol with inflammation adjusting for age, race, family history of prostate cancer, and serum cholesterol.

Results

Table 1 shows baseline characteristics of the men by extent of biopsy cores inflamed. Mean age at baseline and biopsy slightly increased across extent of cores inflamed. The prevalence of a high school education or less was higher in men with any cores inflamed. Mean baseline and biopsy PSA levels increased with increasing extent of cores inflamed. The prevalence of a history of diabetes decreased across the increasing extent of cores inflamed. None of the lifestyle factors differed notably across extent of cores with inflammation, except for a lower prevalence of moderate/active physical activity in men with any cores inflamed. Of the 222 individuals who had information on vitamin A supplement use, 8.6% were users (3+ times/week); and of the 107 individuals who had information on daily use of any type of supplement containing β-carotene, 48.8% were users.

Table 1.

Characteristics by Extent of Prostate Biopsy Cores Inflamed in 235 Men without Prostate Cancer and Irrespective of Indication for Biopsy^*, Placebo Arm, PCPT

Cores Inflamed	None	Any	Some	All
N	53	182	122	60
Age, baseline, mean	62.7	64.6	64.4	65.1
Age, biopsy, mean	69.7	71.6	71.4	72.1
Non-White, %	13.2	17.0	17.2	16.7
Smoking status, %
Current	5.7	5.5	4.9	6.7
Former	66.0	66.4	59.8	61.7
Never	28.3	34.1	35.3	31.7
Education, %
High School or below	11.3	21.4	21.3	21.7
College, some or graduate	60.4	40.1	41.0	38.3
Post-graduate	28.3	38.5	37.7	40.0
Physical Activity, %
Sedentary/light	49.1	58.2	59.8	55.0
Moderate/active	50.9	41.7	40.2	45.0
Family history of PC, %	17.0	17.0	15.6	20.0
History of diabetes, %	13.2	7.7	9.0	5.0
BMI, mean (kg/m2), mean	27.8	27.3	27.2	27.5
Alcohol (g/day), mean	9.4	7.4	7.1	8.0
PSA, baseline (ng/ml), mean	0.89	1.26	1.20	1.36
PSA, biopsy (ng/ml), mean	1.27	1.80	1.55	2.29

^*92.8% did not have a clinical indication for biopsy

Table 2 shows geometric mean serum concentrations by extent of cores inflamed. Mean carotenoid and retinol concentrations did not differ across extent of cores inflamed.

Table 2.

Geometric Mean Concentrations^* (nmol/mL) of Carotenoids and Retinol by Extent of Prostate Biopsy Cores Inflamed in 235 Men without Prostate Cancer and Irrespective of Indication for Biopsy^**, Placebo Arm, PCPT

	None	95%CI	Any	95%CI	Some	95%CI	All	95%CI	p-trend
β-cryptoxanthin	0.15	(0.13,0.097)	0.16	(0.14,0.17)	0.16	(0.15,0.18)	0.15	(0.132,0.17)	0.91
Lycopene	0.63	(0.56,0.17)	0.64	(0.61,0.69)	0.64	(0.59,0.69)	0.66	(0.59,0.74)	0.54
α-carotene	0.088	(0.074,0.11)	0.085	(0.077,0.093)	0.088	(0.079,0.098)	0.078	(0.066,0.092)	0.46
β-carotene	0.46	(0.38,0.55)	0.43	(0.39,0.47)	0.44	(0.39,0.50)	0.41	(0.34,0.48)	0.35
Sum of these carotenoids	1.42	(1.26,1.59)	1.41	(1.31,1.50)	1.41	(1.31,1.53)	1.38	(1.24,1.55)	0.78
Retinol	2.42	(2.27,2.58)	2.34	(2.27,2.43)	2.36	(2.26,2.45)	2.32	(2.19,2.46)	0.18

^*Adjusted for baseline age, race, family history of prostate cancer, serum cholesterol; linear models for P-trend across none, some, all cores inflamed adjusted for baseline age, race, family history of prostate cancer, serum cholesterol, tested at α=0.05 with no cores inflamed as the reference.

^**92.8% did not have a clinical indication for biopsy

Table 3 gives the odds ratios (OR) and 95% CI for the association of serum carotenoids and retinol with the presence and extent of inflammation. None of the carotenoids or retinol was statistically significantly associated with presence or extent of inflammation, with the exception of β-cryptoxanthin. With respect to presence of inflammation, compared to tertile 1 (lowest concentration) of β-cryptoxanthin, tertile 2 was positively associated with any biopsy core inflamed [OR (95%CI), 2.67 (1.19,5.99)] as was tertile 3 (highest concentration), albeit not statistically significantly [1.80 (0.84,3.82)]. With respect to extent of inflammation, tertiles 2 and 3 of β-cryptoxanthin were similarly positively associated with having some [OR (95%CI), T2: 2.98 (1.27,6.97), T3: 2.03 (0.91,4.53)], and all [T2: 2.16 (0.83,5.63), T3: 1.40 (0.55, 3.53)] cores inflamed, albeit not consistently statistically significant.

Table 3.

Association^* of Carotenoids and Retinol with Extent of Prostate Biopsy Cores Inflamed in 235 Men without Prostate Cancer and Irrespective of Indication for Biopsy^**, Placebo Arm, PCPT

		Extent of Biopsy Cores Inflamed
		None	Any		Some		All
	Tertile***	N	N	OR (95%CI)	N	OR (95%CI)	N	OR (95%CI)
β-cryptoxanthin	1	25	53	1.00	33	1.00	20	1.00
	2	12	67	2.67 (1.19,5.99)	46	2.98 (1.27,6.97)	21	2.16 (0.83,5.63)
	3	16	62	1.80 (0.84,3.82)	43	2.03 (0.91,4.53)	19	1.40 (0.55,3.53)
Lycopene	1	16	62	1.00	42	1.00	20	1.00
	2	22	57	0.56 (0.26,1.21)	38	0.57 (0.26,1.28)	19	0.54 (0.21,1.38)
	3	15	63	1.38 (0.59,3.25)	42	1.33 (0.55,3.23)	21	1.51(0.56,4.08)
α-carotene	1	19	59	1.00	37	1.00	22	1.00
	2	16	63	1.30 (0.60,2.79)	43	1.39 (0.63,3.12)	20	1.13 (0.45,2.83)
	3	18	60	1.02 (0.49,2.22)	42	1.16 (0.52,2.58)	18	0.82 (0.33,2.07)
β-carotene	1	16	62	1.00	41	1.00	21	1.00
	2	19	60	0.73 (0.33,1.58)	41	0.76 (0.34,1.72)	19	0.65 (0.26,1.66)
	3	18	60	0.81 (0.37,1.79)	40	0.83 (0.36,1.89)	20	0.79 (0.31,2.03)
Sum of these carotenoids	1	22	56	1.00	38	1.00	18	1.00
	2	13	66	1.35 (0.63,2.90)	43	1.34 (0.60,2.99)	23	1.38 (0.55,2.99)
	3	18	60	1.33 (0.61,2.90)	41	1.32 (0.58,3.00)	19	1.35 (0.53,3.49)
Retinol	1	17	61	1.00	42	1.00	19	1.00
	2	15	64	1.31 (0.59,2.89)	41	1.21 (0.53,2.78)	23	1.53 (0.60,3.92)
	3	21	57	0.78 (0.35,1.65)	39	0.75 (0.33,1.70)	18	0.77 (0.30,1.99)

^*Models adjusted for baseline age, cholesterol, race, family history prostate cancer.

^**92.8% did not have a clinical indication for biopsy

^***Tertile ranges (nmol/mL): β-cryptoxanthin: T1: 0.0139 to <0.122, T2: 0.122 to <0.203, T3: 0.203 to 0.728; lycopene: T1: 0.0700 to <0.548, T2:0.548–0.782, T3:0.786 to 1.90; α-carotene: T1:0.0169 to <0.0655, T2: 0.0655 to <0.111, T3: 0.111 to 0.894; β-carotene: T1: 0.0437 to <0.326, T2: 0.326 to <0.567, T3: 0.567 to 6.63; sum of these carotenoids: T1: 0.145 to <1.18, T2: 1.18 to <1.67, T3: 1.67–9.31; retinol: T1: 1.36 to <2.15, T2: 2.15 to <2.609, T3: 2.609 to 5.15.

We sought possible non-causal explanations for the positive association between β-cryptoxanthin and the presence and extent of inflammation. The number of biopsy cores or slides assessed did not differ among tertiles of β-cryptoxanthin or by the presence of inflammation and the results were unchanged when adjusting for number of biopsy cores or slides assessed. The results were also similar after adjusting for baseline PSA, PSA velocity, and after excluding the 17 men whose biopsy was performed for clinical indication [any vs. none, T2: 2.47 (1.02,5.98), T3: 2.43 (1.05,5.60)]. The results were similar after restricting to men without select oxidative states [any vs. none OR (95%CI): not current smokers (N=222) – T2: 1.79 (0.82,3.90), T3: 2.81 (1.20,6.57); not diabetic (N=214) – T2: 2.47 (1.04,5.85), T3: 1.76 (0.79,3.93)].

Discussion

In this subset of controls from the placebo arm of the PCPT, we found no association of serum concentrations of lycopene, α-carotene, β-carotene, and retinol with intraprostatic inflammation. Higher serum β-cryptoxanthin level was positively associated with both the presence and extent of intraprostatic inflammation for tertile 2 but not for tertile 3. These patterns for β-cryptoxanthin were also observed when restricting to controls not enriched for oxidative states such as cigarette smoking or diabetes. We conducted this study in a setting in which the opportunity to investigate the link of carotenoids and retinol with intraprostatic inflammation was less likely to be biased by clinical indication for biopsy. Nevertheless, the findings did not support our hypothesis.

We had hypothesized that carotenoids would be associated with reduced inflammation in the prostate, and if so, intraprostatic inflammation could mediate any associations between circulating carotenoids and prostate cancer risk. Our hypothesis was, in part, based on the fact that some carotenoids or their metabolic products are anti-oxidants and are anti-inflammatory in laboratory settings ²⁰, and on prior findings in the PCPT that intraprostatic inflammation was associated with prostate cancer ^{29; 30}. With respect to the link between carotenoids and prostate cancer, a large pooled analysis of data from 15 cohorts reported an inverse association between serum lycopene and aggressive but not total prostate cancer, but no associations were observed for the other carotenoids ⁸. In that same pooled analysis, serum retinol was weakly positively associated with prostate cancer risk ⁸. Previous prospective studies in the placebo arm of the PCPT found that retinol and α-carotene concentrations were positively associated ¹², whereas β-carotene, β-cryptoxanthin ¹², and serum lycopene ¹⁰ were not associated with incidence of total and high-grade prostate cancer; these PCPT data were included in the pooled analysis ⁸. Our findings do not support that circulating carotenoids or retinol are inversely associated with intraprostatic inflammation or that intraprostatic inflammation mediates any observed associations of carotenoids and retinol with prostate cancer risk.

Unexpectedly, we found a suggestion that β-cryptoxanthin was positively associated with intraprostatic inflammation. We considered possible non-causal explanations, including by removing men with pro-oxidant states, but the suggestive positive association remained. β-cryptoxanthin, which can be converted into vitamin A, and in the US its major food sources are citrus fruits and juices, is one of the most absorbable carotenoids ³⁵. Citrus fruit intake is not associated or weakly inversely associated with prostate cancer risk ^{36; 37}. Normal ranges for carotenoid concentrations in the blood have not been established, because carotenoids are not essential nutrients, and it is unknown what a high or low level of any carotenoid, including β-cryptoxanthin, indicates with respect to health status.

A major strength of our study is that prostate tissue inflammation was measured in men most of whom did not have a clinical indication for a prostate biopsy; results were comparable after excluding the 7% of men with a conventional clinical indication for biopsy. Since inflammation can cause serum PSA to rise, and a high PSA can lead to a biopsy recommendation, selecting men without a clinical indication for biopsy minimizes the bias of over-selecting tissue biopsies in men who are likely to be inflamed. Another strength is that carotenoid and retinol concentrations were measured in samples that were pooled over two time points, thus reducing intra-participant variability. We expect that our overall findings may be generalizable to other population of men, since circulating carotenoid, including beta-cryptoxanthin, and retinol concentrations did not notably differ from those reported in other large cohorts^{27; 38} aside from higher carotenoid concentrations in our study when compared with a cohort of current and former smokers⁴. Limitations include that we were not able to assess whether circulating carotenoids reflect prostate tissue levels, and that the circulating levels (pooled serum from Years 1 and 4) and tissue inflammation (Year 7) were not measured concurrently. Further, we visually assessed inflammation on H&E slides but did not measured immune cell function or cellular phenotypes. Thus, we cannot rule out other influences of carotenoids on, say, the balance of effector versus suppressor cells in the prostate stroma.

In summary, serum concentrations of α-carotene, β-carotene, lycopene, and their sum, and retinol were not associated with inflammation in prostate tissue. We cannot rule out that the suggestive positive association between β-cryptoxanthin and intraprostatic inflammation that we observed is due to chance or bias, particularly since we did not observe any dose-response effect or a linear pattern of association. These findings suggest that circulating carotenoids and retinol are not useful dietary intervention targets for preventing prostate cancer via modulating intraprostatic inflammation.