Coffee, PI3K signaling pathway, and prostate cancer: a prospective study in the Health Professionals Follow-up Study

Abstract

Background

Higher coffee intake has been associated with reduced risk of prostate cancer, particularly aggressive forms. The activation of the PI3K signaling pathway plays an important role in prostate carcinogenesis.

Objective

To evaluate associations between pre-diagnostic coffee intake and a PI3K activation score, the expression/presence of PI3K regulators, and downstream effectors in tumor tissue from men with prostate cancer in the Health Professionals Follow-up Study, a prospective cohort study conducted in the US.

Design

A case-only study design was applied. Coffee intake was assessed using validated food frequency questionnaires completed in 1986 and every four years thereafter until prostate cancer diagnosis.

Participants/setting

Study participants comprised 1,242 men diagnosed with prostate cancer from 1986 to 2009 and with tumor markers assessed from tissue microarrays constructed from tumor specimens.

Main outcome measures

The outcomes include the PI3K activation score; expression of insulin receptor and IGF1 receptor; angiogenesis markers; and presence of the tumor suppressor PTEN, chronic and acute inflammation, simple atrophy, and post-atrophic hyperplasia.

Statistical analyses performed

Multivariable linear or logistic regression was conducted to estimate associations between coffee intake and tumor marker expression/presence.

Results

Among coffee drinkers (86.6% of the population), median (25^th-75^th) coffee intake was 2 (1–3) cups/day. The associations between coffee consumption and the tumor markers of interest were generally weak with modest precision. When comparing men who drank >3 cups/day of coffee with nondrinkers, the absolute percent difference in the PI3K activation score and angiogenesis markers ranged from 0.6% to 3.6%. The odds ratios for PTEN loss, IGF1 receptor and insulin receptor expression, and presence of chronic and acute inflammation, simple atrophy, and post-atrophic hyperplasia also were not statistically significant, were imprecise, and ranged from 0.82 to 1.58.

Conclusions

Coffee intake was not observed to be associated with PI3K activation, related regulators, and several effectors in prostate tumor tissue. Studies exploring alternative pathways or earlier steps in carcinogenesis are needed to investigate the underlying mechanisms of the coffee and prostate cancer association.

INTRODUCTION

The activation of the phosphoinositide 3-kinase (PI3K) signaling pathway is a critical driving event in prostate cancer development and progression, by controlling cell proliferation, adhesion, migration, invasion, metabolism, inflammation, angiogenesis, and survival.^1–4 Insulin receptor (IR) and insulin-like growth factor I receptor (IGF1R) are key activators for PI3K signaling, whereas phosphatase and tensin homolog (PTEN), a tumor suppressor gene, is a key negative regulator.^5–7 Previous studies also show that TMPRSS2:ERG gene fusion may modify the associations between IGF1R and IR, as well as PTEN loss, with prostate cancer progression.^8,9

In the Health Professionals Follow-up Study (HPFS), we previously reported that coffee consumption of at least 6 cups a day was associated with a 60% lower risk (95% CI, 25%, 78%) of lethal prostate cancer (defined as prostate cancer-specific death or the presence of metastases at prostate cancer diagnosis or during follow-up) compared to non-drinkers.¹⁰ However, the mechanisms underlying this association are unknown. Experimental studies have reported that compounds in coffee including chlorogenic acid and coffee diterpenes downregulate the PI3K pathway in cancers.^11,12 In addition, emerging evidence suggests that coffee consumption is associated with lower concentrations of inflammatory biomarkers and reduced inflammation.¹³ Therefore, to explore potential mechanisms linking coffee consumption and prostate cancer outcomes, the associations between coffee intake with PI3K activation, PI3K regulators, angiogenic markers, inflammation, and proliferative inflammation atrophy were investigated in prostate tumors among men diagnosed with prostate cancer in the Health Professionals Follow-up Study (HPFS). We hypothesized that higher coffee consumption would be associated with lower PI3K activation, markers of lower angiogenic potential, and lower levels of inflammatory markers in prostate tumor tissue.

MATERIALS AND METHODS

Study Population

The HPFS was initiated in 1986 when 51,529 male health professionals living in the US aged 40–75 years completed a self-administered baseline questionnaire inquiring about demographic factors, lifestyle habits, medical history, and disease diagnoses including prostate cancer.¹⁴ Subsequently, biennial questionnaires have been completed by participants to collect updated information. Incident primary prostate cancers were identified by self-report, then confirmed by review of medical records and pathology reports. Clinical information including stage was extracted from the medical records. The study population in this report was a subset of men with prostate cancer diagnosed between 1986 and 2009 for whom archival formalin-fixed paraffin-embedded tumor tissue from prostatectomy or transurethral resection of the prostate was collected from the treating hospitals.^8,15–17 Each individual protein biomarker was assayed and quantified at the same time on all included cases. The study protocol was approved by the institutional review boards of the Harvard T.H. Chan School of Public Health and Brigham and Women’s Hospital, and those of participating registries as required. Completing and returning study questionnaires implied informed consent.

Coffee intake assessment

Usual diet over the past year was assessed by validated semi-quantitative food frequency questionnaires (FFQ) mailed every four years since 1986. The FFQs contained a standard portion size for 135–150 foods including regular and decaffeinated coffee and nine possible frequency-of-consumption responses for each food ranging from “never or less than once per month” to “six or more times per day”.¹⁴ Average daily nutrient intake for each food was calculated by multiplying the reported frequency of intake by the nutrient content of that food estimated primarily from U.S. Department of Agriculture databases;^18–22 nutrient values were then summed across all foods. Nutrient intakes were energy-adjusted using the residual method.²³ Total coffee intake was calculated as the sum of regular coffee and decaffeinated coffee intakes. Processed meat consumption, a confounding variable, was derived as the number of servings of all processed meat on the FFQ such as bacon, hotdogs, and sausage. A validation study in HPFS demonstrated a high correlation (r=0.93) of self-reported coffee intake from FFQ with two 7-day dietary records.¹⁴

Assessment of tumor tissue markers

Tissue microarrays

High-density tumor tissue microarrays (TMA) were constructed by sampling 0.6 mm paraffin embedded tissue cores (typically three cores per prostate tumor) and embedding the cores on a recipient array block. With the exception of the angiogenesis and inflammation measures, all biomarkers were assessed on TMAs. Scoring for each tumor marker was conducted by study pathologists or using automated image analysis blinded to disease outcome and clinical data.

Insulin Receptor and IGF1R

Immunohistochemical staining was conducted on each TMA using established procedures and validated antibodies. The slides were scanned with the BLISS system (Bacus Lab, North Lombard, IL). The percent of staining was reported for both receptors.^16,24

Angiogenesis

Protein expression of endothelial cell marker CD34 was ascertained on each TMA. The size and architecture of microvessels were quantified by semi-automated image analysis through Image ProPlus 4.5 software (Media Cybernetics, Rockville, MD).¹⁶ Vessel area was measured as the average area comprised by a vessel (μm²). Vessel irregularity was calculated as perimeter²/4 × π × area, with a value of 1.0 indicating a perfect circle and values greater than 1.0 indicating increasing irregularity. Higher vessel number and smaller irregular vessels are more angiogenic.

PTEN loss and TMPRSS2:ERG gene fusion

PTEN and ERG immunohistochemical staining were performed on TMAs using validated assays and antibodies.^8,25 A tissue core was considered to have PTEN loss if the intensity of cytoplasmic and nuclear staining was markedly decreased or entirely lost among any TMA cores compared with surrounding benign glands and/or stroma. Tumors were classified as ERG-positive (i.e., carrying the TMPRSS2:ERG gene fusion) if at least one TMA core stained positive for ERG and ERG-negative if all cores stained negative for ERG.⁸

PI3K activation score

As described previously,¹⁵ activation of the PI3K pathway was assessed on a single cell-level based on protein expression of the pS6 and stathmin proteins and the negative pathway regulator PTEN. Cell-level activation scores, with positive contributions of pS6 and stathmin and inverse contributions of PTEN expression, were corrected for batch effects, rank-normalized, and then summarized on a tumor level as the median of the cell-level scores across all tumor cells of a tumor (range, 0 to 100, with higher values indicating activation of the pathway).

Inflammation and proliferative inflammation atrophy

The full set of hematoxylin and eosin slides, not only the TMAs, were reviewed for histopathological evidence of inflammation using widely accepted approaches.²⁶ Acute inflammation was considered by the presence of neutrophils and chronic inflammation by the presence of mononuclear cells such as lymphocytes and macrophages. Two markers of proliferative inflammatory atrophy (PIA) which occur in association with inflammation were also measured. PIA is a discrete focus of proliferative glandular epithelium with the morphological appearance of simple atrophy (SA) or post-atrophic hyperplasia (PAH).^27–29

Covariate assessment

To control for confounding, several demographic, lifestyle, and dietary factors that are established or suspected risk factors for prostate cancer³⁰ and/or that varied by coffee consumption (Table 1) were evaluated. The selected covariates included age at diagnosis (continuous), smoking habits (never smokers, past smokers, quit > 10 years, past smokers quit ≤ 10 years, current smokers), physical activity (continuous, MET-hours/week), body mass index (BMI, continuous, kg/m²), height (continuous, m), alcohol intake (nondrinkers, >0-≤ 28, >28g/day), family history of prostate cancer in brother or father (yes, no), PSA testing history in the last two years (yes, no), total energy intake (continuous, kcal/day), processed meat intake (continuous, servings/day), tomato sauce intake (continuous, servings/day), and total calcium intake (from foods and supplements; continuous, mg/day). For physical activity, participants reported nearly every two years the amount of time spent each week participating in several moderate and vigorous activities and exercises including walking, jogging, running, and cycling. Each activity was assigned a MET score, the ratio of the metabolic rate divided by the resting metabolic rate, and MET-hours per week calculated for each activity and overall. For alcohol consumption, 28 g/d was selected as the cutpoint because the USDA Dietary Guidelines recommend men consume two alcoholic drinks or less in a day and one alcoholic drink equivalent contains 14 grams of pure alcohol.³¹ Height was assessed at study enrollment. Data on family history of prostate cancer, PSA testing history, and smoking habits from the latest questionnaire completed before prostate cancer diagnosis were used. The cumulative average across all questionnaires from baseline to the last questionnaire before prostate cancer diagnosis was used for the remaining covariates to account for long term effects. Additional analyses were conducted in which we further adjusted for the remaining covariates in the models in our prior report on coffee consumption and prostate cancer analyses:¹⁰ year of prostate cancer diagnosis (continuous), race (white, other), BMI at age 21 (<20, 20- <22.5, 22.5-<25, ≥25 kg/m²), history of diabetes (yes, no), multivitamin use (yes, no), alpha linolenic acid intake (quintiles, g/day), and supplemental vitamin E intake (quintiles, mg/day).

Table 1.

Characteristics^a of men in the Health Professionals Follow-up Study diagnosed with prostate cancer between 1986 and 2009 with tumor tissue marker measurements by total coffee intake

	Total coffee intake
	None (n=166)	>0–1 cup/d (n=293)	>1–3 cups/d (n=523)	>3 cups/d (n=260)
Coffee intake, median (Q25, Q75)
Total coffee intake, cups/day	-	0.5 (0.2, 0.9)	2.1 (1.6, 2.5)	4.2 (3.5, 4.8)
Regular coffee intake, cups/day	-	0.1 (0.0, 0.4)	1.3 (0.6, 2.2)	3.2 (2.0, 4.3)
Decaffeinated coffee intake, cups/day	-	0.1 (0.0, 0.5)	0.4 (0.0, 1.1)	0.7 (0.0, 2.3)
Intake of other dietary variables, mean (SD)
Total energy intake, kcal/day	1970 (551)	1881 (493)	1975 (520)	2049 (470)
Processed meat intake, servings/week	2.0 (2.3)	2.0 (2.6)	2.4 (2.8)	2.7 (2.3)
Tomato sauce intake, servings/week	0.8 (0.8)	0.9 (0.8)	0.9 (0.7)	0.9 (0.7)
Total calcium intake,b mg/day	1022 (426)	948 (395)	923 (350)	891 (292)
Alcohol drinkers, %	50.0	85.6	88.7	89.1
Alcohol intake in drinkers, g/day, mean (SD)	8.3 (10.1)	11.4 (11.7)	16.1 (15.7)	16.4 (13.4)
Alcohol intake >28 g/day, %	2.4	7.5	16.2	16.1
Racec
Asian, %	1.0	2.1	0.92	1.2
Black, %	2.8	2.1	0.65	<0.01
Other, %	1.2	1.2	0.96	0.3
White, %	85.0	90.7	92.2	95.6
Unknown, %	10.0	3.9	5.2	2.9
Smoking habits
Never smoker, %	65.0	49.7	40.7	23.9
Past smoker, quit >10 years ago, %	13.9	28.6	33.9	43.1
Past smoker, quit ≤10 years, %	4.4	8.6	9.6	14.0
Current smoker, %	0.5	3.0	5.6	9.8
Physical activity, MET-hours/week,d mean (SD)	28.3 (21.9)	30.4 (21.8)	29.9 (21.4)	28.0 (21.4)
Body mass index, kg/m2, mean (SD)	25.3 (2.7)	25.2 (3.4)	25.6 (2.8)	25.9 (2.8)
Height, m, mean (SD)	1.78 (0.06)	1.78 (0.08)	1.78 (0.07)	1.79 (0.05)
Prostate-specific antigen (PSA) screening history, %	57.4	59.7	61.6	58.5
Family history of prostate cancer, %	27.7	21.3	23.9	20.5
Age at prostate cancer diagnosis, years, mean (SD)	64.6 (5.4)	65.4 (6.1)	65.9 (6.5)	65.5 (6.1)
Prostate cancer characteristics
High-grade with 4+3 and above, %	44.7	44.7	45.8	45.2
Prostate cancer-specific death, %	8.8	8.9	11.0	11.6
Lethal prostate cancer,e %	10.9	12.2	14.0	14.2

^aAll values are standardized to the age distribution of the study population except age at prostate cancer diagnosis.

^bEnergy adjusted total (from foods and supplements) calcium intake.

^cParticipants self-reported their race.

^dMET: Metabolic equivalent of task

^eLethal prostate cancer was defined as metastases at diagnosis, metastases over follow-up, or prostate cancer-specific death.

Statistical analysis.

This study used a case-only design as we were interested in whether prostate tumor characteristics varied by pre-diagnostic coffee consumption. Participants with a history of cancer at baseline, missing baseline dietary data including coffee intake, and missing data for all tumor markers of interest were excluded from the analyses. Cumulative average coffee intake representing intake assessed between 1986 and the FFQ completed prior to prostate cancer diagnosis was analyzed. Total, regular, and decaffeinated coffee intakes were modeled continuously; total coffee consumption also was modeled as a categorical variable in which participants were separated into four categories based on consumption (nondrinkers, >0–1, >1–3, and >3 cups/day). The PI3K activation score and each angiogenic marker was log transformed to improve normality. Multivariable linear regression was used to estimate the relative difference as the percent difference and corresponding 95% confidence interval (CI) associated with coffee intake, with nondrinkers as the reference. For PTEN loss, inflammation, and proliferative inflammation atrophy markers, multivariable logistic regression was conducted to estimate the odds ratio (OR) and 95% confidence intervals (CI) for their associations with coffee intake. When investigating types of coffee, both regular and decaffeinated coffee intakes were simultaneously included in the model. To test for linear trend, a variable was generated by assigning the median coffee intake of each category to men within that category. Then, that variable was modeled as a continuous variable and the p-trend was calculated from the Wald test. Analyses restricted to men who had never smoked or had quit over ten years before prostate cancer diagnosis were conducted to eliminate potential residual confounding by smoking status. Stratified analyses were conducted by tumor grade and ERG status.

For all tests, two-sided p-values < 0.05 were considered statistically significant. All analyses were conducted using SAS software, version 9.4 of the SAS System for Unix (Cary, NC).³²

RESULTS

The current analysis included 1,242 men with prostate cancer for whom data on at least one of the tissue biomarkers of interest was available. Median (25^th-75^th) total coffee intake among drinkers (86.6% of the study population) was 2 (1–3) cups/day. Men who drank the most coffee had higher processed meat intake, had lower calcium intake, and were more likely to be ever smokers. They were also more likely to be alcohol drinkers, and, among drinkers, had higher alcohol intake (Table 1).

Results from age-adjusted models were similar to those from multivariable models; therefore, only the results from the multivariable models are discussed in the text. Weak and nonsignificant associations were observed between coffee consumption and the PI3K activation score and angiogenic markers (Table 2). When comparing men who drank >3 cups/day of coffee with nondrinkers, the PI3K activation score was 1.9% higher (95% CI: −6.6%, 11.2%). Among the angiogenesis markers, the strongest association was observed for the number of vessels which was 3.6% higher (95% CI: −9.0%, 18.0%) among men who drank >3 cups/day of coffee compared to nondrinkers. Compared with nondrinkers, men who drank >3 cups/day of coffee were more likely to be diagnosed with tumors with PTEN-loss (OR = 1.34, 95% CI: 0.71, 2.53), positive ERG expression (OR=1.27, 95% CI: 0.79, 2.04), high IGF1 receptor expression (OR=1.58, 95% CI: 0.59, 4.24), and high insulin receptor expression (OR=1.15, 95% CI: 0.51, 2.56), although none of the associations were statistically significant. The associations for the inflammation markers and proliferative inflammation atrophy markers were weak and nonsignificant (Table 3). Results were similar but less precise after further adjustment for the remaining variables in the models from our prior report on coffee consumption and prostate cancer¹⁰ (Supplementary Tables 4–5).

Table 2.

Differences (95% confidence intervals) in tumor marker expression by categories of total coffee intake among men with prostate cancer in the Health Professionals Follow-up Study

Tumor marker	Categories of total coffee intake
	Nondrinkers	>0–1 cup/d	>1–3 cups/d	>3 cups/d	p-trend	Per 1 cup/d increase
PI3Ka activation score, overall mean (SD) = 50.8 (15.0)
N	88	160	270	151		669
Model 1b	Ref	4.7% (−3.4%, 13.6%)	0.8% (−6.7%, 8.9%)	1.9% (−6.3%, 10.9%)	0.76	0.1% (−1.5%, 1.7%)
Model 2c	Ref	5.0% (−3.2%, 13.9%)	1.0% (−6.7%, 9.4%)	1.9% (−6.6%, 11.2%)	0.70	0.0% (−1.7%, 1.6%)
Angiogenesis markers
N	76	127	231	128		562
Vessel number, overall mean (SD) = 76.1 (37.9)
Model 1b	Ref	2.3% (−10.0%, 16.3%)	7.9% (−3.7%, 21.0%)	5.7% (−6.3%, 19.2%)	0.36	0.8% (−1.4%, 3.0%)
Model 2c	Ref	2.0% (−10.3%, 15.9%)	6.5% (−5.5%, 20.0%)	3.6% (−9.0%, 18.0%)	0.62	0.7% (−1.6%, 3.0%)
Vessel irregularity,d overall mean (SD) = 4.0 (1.1)
Model 1b	Ref	2.3% (−5.2%, 10.4%)	−0.9% (−7.7%, 6.3%)	−2.9% (−10.3%, 5.1%)	0.18	−0.2% (−1.6%, 1.2%)
Model 2c	Ref	4.0% (−3.3%, 11.9%)	2.0% (−4.7%, 9.3%)	0.6% (−6.9%, 8.7%)	0.59	0.0% (−1.3%, 1.4%)
Vessel diameter,e μm, overall mean (SD) = 25.2 (5.3)
Model 1b	Ref	−1.9% (−7.4%, 3.9%)	−3.1% (−8.2%, 2.3%)	−0.2% (−5.8%, 5.8%)	0.79	−0.3% (−1.3%, 0.7%)
Model 2c	Ref	−2.5% (−8.2%, 3.5%)	−3.8% (−9.2%, 2.0%)	−1.0% (−7.0%, 5.5%)	0.89	−0.3% (−1.3%, 0.8%)
Vessel area,e μm2, overall mean (SD) = 541 (279)
Model 1b	Ref	−2.1% (−14.0%, 11.5%)	−5.1% (−15.8%, 7.0%)	3.4% (−9.4%, 18.0%)	0.47	−0.1% (−2.4%, 2.2%)
Model 2c	Ref	−5.0% (−17.2%, 8.9%)	−8.5% (−19.7%, 4.3%)	−0.8% (−14.1%, 14.6%)	0.76	−0.3% (−2.7%, 2.1%)

^aPI3K: Phosphoinositide 3-kinase

^bModel 1 adjusted for age at prostate cancer diagnosis (continuous, years).

^cModel 2: Model 1 plus further adjustment for smoking habits (never smokers, past smokers, quit > 10 years, past smokers quit ≤ 10 years, current smokers), physical activity (continuous, MET-hours/week), body mass index (continuous, kg/m²), height (continuous, m), alcohol intake (nondrinkers, > 0–28, >28 g/day), family history of prostate cancer (yes, no), PSA testing history (yes, no), total energy intake (continuous, kcal/day), processed meat intake (continuous, servings/day), tomato sauce intake (continuous, servings/day), and total calcium intake (continuous, mg/day).

^dIrregular vessels are more angiogenic; a score of 1.0 indicates a perfect circle.

^eSmaller vessels are more angiogenic.

Table 3.

Odds ratios (95% confidence intervals) representing associations between total coffee intake and tumor marker expression among men with prostate cancer in the Health Professionals Follow-up Study

Tumor marker	Categories of total coffee intake
	Nondrinkers	>0–1 cup/d	>1–3 cups/d	>3 cups/d	p-trend	Per 1-cup/day increase
PTENa loss, overall 24.1 % any loss
Loss/Intact, N	21/84	54/138	70/251	45/127		190/600
Model 1b	Ref	1.56 (0.88, 2.77)	1.11 (0.64, 1.92)	1.41 (0.78, 2.54)	0.87	1.04 (0.94, 1.15)
Model 2c	Ref	1.43 (0.78, 2.60)	1.03 (0.58, 1.84)	1.34 (0.71, 2.53)	0.94	1.03 (0.92, 1.15)
ERGd expression (indicating TMPRSS2:ERGe fusion), overall 47.2% positive
Positive/Negative, N	62/75	116/121	187/215	101/110		466/521
Model 1b	Ref	1.21 (0.79, 1.84)	1.09 (0.74, 1.61)	1.15 (0.75, 1.78)	0.86	1.00 (0.92, 1.08)
Model 2c	Ref	1.26 (0.80, 1.96)	1.19 (0.78, 1.81)	1.27 (0.79, 2.04)	0.58	1.02 (0.93, 1.11)
IGF1f receptor, overall 81.5% high
High/Low,g N	43/11	87/16	132/36	78/14		340/77
Model 1b	Ref	1.58 (0.67, 3.73)	1.04 (0.48, 2.25)	1.53 (0.63, 3.68)	0.76	1.00 (0.86, 1.17)
Model 2c	Ref	1.79 (0.69, 4.62)	1.07 (0.46, 2.52)	1.58 (0.59, 4.24)	0.88	1.01 (0.85, 1.20)
Insulin receptor, overall 56.6% high
High/Low,g N	23/26	60/37	88/68	48/37		219/168
Model 1b	Ref	2.03 (1.00, 4.10)	1.58 (0.82, 3.03)	1.54 (0.75, 3.14)	0.91	0.96 (0.85, 1.09)
Model 2c	Ref	1.92 (0.89, 4.13)	1.30 (0.63, 2.65)	1.15 (0.51, 2.56)	0.39	0.89 (0.77, 1.03)
Inflammation
Chronic inflammation, overall 83.9% positive
Presence/Absence, N	106/20	188/38	336/58	154/34		784/150
Model 1b	Ref	0.96 (0.53, 1.74)	1.13 (0.65, 1.97)	0.88 (0.48, 1.61)	0.75	1.00 (0.90, 1.12)
Model 2c	Ref	1.16 (0.62, 2.18)	1.31 (0.73, 2.38)	0.99 (0.51, 1.91)	0.78	1.00 (0.89, 1.13)
Acute inflammation, overall 29.1% positive
Presence/Absence, N	34/92	75/150	111/275	49/138		269/655
Model 1b	Ref	1.37 (0.85, 2.22)	1.11 (0.70, 1.74)	0.97 (0.58, 1.62)	0.31	0.96 (0.88, 1.05)
Model 2c	Ref	1.53 (0.92, 2.56)	1.21 (0.75, 1.97)	1.03 (0.59, 1.79)	0.30	0.96 (0.87, 1.06)
Proliferative inflammation atrophy
Simple atrophy, overall 73.0% positive
Presence/Absence, N	83/43	165/60	293/95	134/52		675/250
Model 1b	Ref	1.47 (0.91, 2.36)	1.65 (1.07, 2.56)	1.37 (0.84, 2.24)	0.44	1.03 (0.94, 1.13)
Model 2c	Ref	1.58 (0.96, 2.61)	1.65 (1.03, 2.64)	1.32 (0.77, 2.26)	0.79	1.02 (0.92, 1.12)
Post-atrophic hyperplasia, overall 22.1% positive
Presence/Absence, N	31/95	44/181	91/297	38/148		204/721
Model 1b	Ref	0.73 (0.43, 1.23)	0.92 (0.57, 1.47)	0.77 (0.45, 1.33)	0.78	0.97 (0.88, 1.07)
Model 2c	Ref	0.85 (0.49, 1.48)	1.01 (0.60, 1.67)	0.82 (0.45, 1.48)	0.73	0.97 (0.87, 1.07)

^aPTEN: Phosphatase and TENsin homolog deleted on chromosome 10

^bModel 1: adjusted for age at prostate cancer diagnosis (continuous, years).

^cModel 2: Model 1+ smoking habits (never smokers, past smokers, quit > 10 years, past smokers quit ≤ 10 years, current smokers), physical activity (continuous, MET-hours/week), body mass index (continuous, kg/m²), height (continuous, m), alcohol intake (nondrinkers, >0–28, >28 g/day), family history of prostate cancer (yes, no), PSA testing history (yes, no), total energy intake (continuous, kcal/day), processed meat intake (continuous, servings/day), tomato sauce intake (continuous, servings/day), and total calcium intake (continuous, mg/day).

^dERG: ETS transcription factor ERG

^eTMPRSS2:ERG: transmembrane protease serine 2: ETS transcription factor ERG

^fIGF1: Insulin-like growth factor I.

^gMean tumor intensity was dichotomized as low (weak to none) and high (moderate to strong).

Similar results were observed for regular and decaffeinated coffee intake, among never smokers and past smokers who had quit over ten years prior to prostate cancer diagnosis, by grade, and by ERG expression (Supplementary Table 6).

DISCUSSION

In HPFS, an 18% lower risk of overall prostate cancer was previously reported among men who drank ≥6 cups/day of coffee compared with nondrinkers. The association was even stronger for lethal prostate cancer with a relative risk (RR) of 0.40 (95% CI, 0.22, 0.75) for the same comparison.¹⁰ An inverse association was also reported in a recent meta-analysis of 15 prospective cohort studies including the HPFS (RR of overall prostate cancer for the highest versus lowest coffee intake category = 0.91; 95% CI 0.84, 0.98).³³ In the present study, no statistically significant associations were found between total coffee intake and prostate tumor PI3K activation level, PI3K regulators, and PI3K downstream effectors, including the PI3K activation score and markers of insulin receptor and IGF1 receptor expression, angiogenesis, PTEN, inflammation, or proliferative inflammation atrophy. Similar results were found for regular and decaffeinated coffee, when the study population was limited to never smokers and past smokers who quit over 10 years before their prostate cancer diagnosis, and by tumor grade and ERG expression.

The mechanism through which coffee may reduce prostate cancer risk is unknown. Coffee has diverse biological effects, and compounds in coffee have been shown to influence the insulin/PI3K/AKT signaling pathways,³⁴ which play a significant role in the development and progression of prostate cancer. PI3K activation is elevated in a large proportion of prostate cancer patients.³⁵ Accordingly, inhibitors targeting the PI3K pathway are being studied as therapeutic agents against prostate cancer.^36,37 Phenolic compounds in coffee show inhibitory potential for adipogenesis-related inflammation, mitochondrial dysfunction, and insulin resistance through regulation of insulin/PI3K/AKT signaling pathways.³⁴ Cafestol, a coffee-specific diterpene, can induce apoptosis in cancer cells by inhibiting the PI3K/AKT signal pathway.¹¹ However, the concentration of diterpenes is very low in filtered coffee, which is the most popular type of coffee consumed in the US.³⁸ Further, the insulin receptor and IGF1 receptor both mediate the activation of the PI3K pathway, while PTEN negatively regulates PI3K activation.²⁴ Emerging evidence suggests that ERG can interact with the IGF1 receptor, insulin receptor, and PTEN, which could regulate PI3K activation and promote prostate cancer progression.^8,9 However, the results stratified by ERG expression did not suggest a difference in the current study, although results were imprecise. Once activated, PI3K initiates a wave of downstream events, which play key roles in the control of cell activities including inflammation and angiogenesis.^39,40 However, in the current study no significant associations were found between coffee consumption and PI3K activation level, angiogenesis, expression of insulin receptor, IGF1 receptor, and PTEN, inflammation, and proliferative inflammation atrophy. Other potential mechanisms by which coffee may reduce prostate cancer risk including through regulation of glucose and lipid metabolism, increased antioxidant activity, improved protection against DNA damage, and influencing steroid metabolism.⁴¹ Another potential reason could be that the study participants were restricted to prostate cancer patients. The potential beneficial effect of coffee intake in preventing tumor development might be more obvious in the prostate tissue of men without cancer.

The strengths of this study include repeated and validated measurement of diet including coffee intake over a long period of time, as well as central pathologic review of all morphologic features of the prostate tumor specimens. However, several limitations should be noted. First, statistical power was limited by the number of prostate cancer cases, particularly for lethal cases. Second, misclassification in coffee consumption is possible as participants estimated their intakes over the past year. However, a validation study in this cohort showed high reproducibility and validity for coffee consumption.¹⁴ Further, cumulative average coffee consumption that incorporated intake measures assessed every four years up to a participant’s prostate cancer diagnosis was analyzed, thereby reducing random within-person measurement error. Third, our findings apply to light and moderate coffee consumption as approximately 90% of the population consumed <4 cups/day of coffee. Fourth, the study population was comprised primarily of highly educated, white, nonsmoking men, which may limit the generalizability of the findings. However, the homogeneous study population would reduce potential confounding.

CONCLUSIONS

The results from this study suggest that coffee intake is not associated with activation of PI3K, expression of the insulin receptor and IGF1 receptor, PTEN loss, ERG expression, angiogenesis, inflammation, and proliferative inflammation atrophy, among men with prostate cancer. Future studies evaluating steps earlier in the process of carcinogenesis, rather than established tumors are needed. Moreover, investigations on different potential pathways and corresponding biomarkers are needed to explore the underlying mechanisms of the association between coffee intake and prostate cancer development.

RESEARCH SNAPSHOT.

Research Question:

Is coffee intake associated with PI3K activation, expression of PI3K regulators, and PI3K downstream effectors in prostate tumor tissue?

Key Findings:

In this study of 1,242 prostate cancer patients from the Health Professionals Follow-up Study (HPFS) with tumor tissue marker data, associations between total coffee intake and the PI3K activation score, expression/presence of insulin receptor, insulin-like growth factor I receptor, angiogenesis, the tumor suppressor PTEN, inflammation, and proliferative inflammation atrophy were generally weak and not statistically significant. Results were similar for regular and decaffeinated coffee intake, among non-smokers, and by tumor grade and ERG gene fusion status.