Vitamin D and Genetic Ancestry are Associated with Apoptosis Rates in Benign and Malignant Prostatic Epithelium

Abstract

Purpose:

Vitamin D metabolites may be protective against prostate cancer (PCa). We conducted a cross-sectional analysis to evaluate associations between in vivo vitamin D status, genetic ancestry, and degree of apoptosis using prostatic epithelial TUNEL staining.

Experimental Design:

We performed indirect TUNEL staining on contralateral benign and dominant tumor epithelium using punch biopsies from radical prostatectomy tumor blocks. Participants had clinically localized PCa. Levels of prostatic and serum 25(OH)D and of serum 1,25 hydroxyvitamin D were assessed immediately before radical prostatectomy. Ancestry informative markers were used to estimate the percentage of genetic West African, Native American, and European ancestry.

Results:

Indirect TUNEL staining was performed on 121 of 182 newly diagnosed men, age 40–79, that were enrolled between 2013–2018. Serum 25(OH)D correlated positively with both tumor (ρ = 0.17, p = 0.03), and benign (ρ = 0.16, p = 0.04) prostatic epithelial TUNEL staining. Similarly, prostatic 25(OH)D correlated positively with both tumor (ρ = 0.31, p < 0.001) and benign (ρ = 0.20, p = 0.03) epithelial TUNEL staining. Only Native American ancestry was positively correlated with tumor (ρ = 0.22, p = 0.05) and benign (ρ = 0.27, p= 0.02) TUNEL staining. In multivariate regression models, increasing quartiles of prostatic 25(OH)D (β = 0.25, p = 0.04) and Native American ancestry (β = 0.327, p = 0.004) were independently associated with tumor TUNEL staining.

Conclusions:

Physiologic serum and prostatic 25(OH)D levels and Native American ancestry are positively associated with the degree of apoptosis in tumor and benign prostatic epithelium in clinically localized PCa. Vitamin D may have secondary chemoprevention benefits in preventing PCa progression in localized disease.

Introduction

Serum 25 hydroxyvitamin D [25(OH)D] deficiency has been associated with aggressive prostate cancer (PCa)^1–5 and there is evidence that 25(OH)D and its active metabolite, 1,25 dihydroxyvitamin D (1,25(OH)₂D), may be protective against PCa^6–11. Epidemiologic evidence from the randomized and placebo-controlled VITAL trial suggests vitamin D3 can reduce advanced and fatal cancer as a chemoprevention agent (p =0.04)¹². Within this study, there was a non-significant reduction (OR 0.88, p =0.07) in prostate cancer incidence in patients randomized to vitamin D3 of whom African Americans (AA) (HR 0.77, p =0.21) had the largest hazard ratio¹². Vitamin D deficiency may also contribute to the development of health disparities, as West African (WA) ancestry is associated with both risks of serum 25(OH)D deficiency^13–15 and aggressive PCa^16,17. One retrospective study demonstrated Hispanics with higher WA ancestry (i.e., Dominicans, Puerto Ricans, and Cubans) have higher PCa mortality rates relative to non-Hispanic Whites^18,19. Similarly, since Native Americans and Hispanics have the lowest PCa incidence and mortality rates, Native American ancestry may be protective from the development of aggressive PCa^20,21.

Cell line data and mouse models show vitamin D works in vitro to induce or increase apoptosis in prostate epithelium^22,23. In prostate cancer cell lines expressing the vitamin D receptor, vitamin D3 activates downstream effector protease, caspase-3, and upstream caspase-9 protease initiating the “intrinsic” pathway for apoptosis²⁴. There is also evidence that induction of apoptosis may be an important contributor to the efficacy of 1,25(OH)₂D, both as monotherapy and in combination with other agents^25–29. Clinical trials suggest vitamin D may be responsible for a decreased rate of PSA rise³⁵ or a radiologic partial response in tumor burden³⁰ in some PCa patients treated with adjuvant 1,25 dihydroxyvitamin D. Despite these effects, there are no prior studies that investigated the in vivo relationship between serum or prostatic levels of vitamin D metabolites and apoptosis in the prostatic epithelium. Further investigation of vitamin D could explain mortality disparities in AA PCa patients due to increased risk of progression from apoptosis inhibition. West African ancestry may confound potential associations between vitamin D and prostatic epithelial apoptosis. We aimed to assess the independent associations between in vivo serum and prostatic vitamin D with an established apoptosis marker in clinically localized PCa patients.

Materials and Methods

Study population

This is a single-center cross-sectional study of 121 men, aged 40–79 years who were diagnosed with clinically localized PCa and elected for radical prostatectomy as initial definitive therapy. Following protocol approval by the Jesse Brown Veterans Affairs (VA) Institutional Review Board, men were consecutively enrolled between November 2013 to 2019 at a Chicago VA Medical Center urology clinic. Of the 269 men with PSA < 100 ng/mL who were initially approached for consent after their outpatient urology visit, we identified 229 eligible candidates. Exclusion criteria included patients with clinical metastatic disease, those opting for active surveillance or other treatment options, history of end-stage renal or liver failure, parathyroid disorders, and prior treatment with androgen deprivation therapy or pelvic radiotherapy (for any pelvic malignancy). Forty-seven men refused enrollment, 182 men provided consent and 169 ultimately had their radical prostatectomy. Indirect Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was completed on the first 121 participants.

Clinical data

Clinical and demographic data were retrieved from patients’ electronic health records and patient-administered questionnaires at the time of enrollment or on postoperative day 1. Relevant clinical covariates obtained included age, first-degree family history of PCa, alcohol and tobacco use, marital status, and self-reported ethnicity and race. Ethnicity and race were classified as either non-Hispanic White, non-Hispanic Black, or Other (i.e., Native American, or Hispanic). Body mass index was calculated using standing height (m) and weight (kg) measurements from the date of enrollment. Total vitamin D intake was obtained from the validated “Nutrition Quest Vitamin D and Calcium” screener questionnaire³¹.

Specimen handling and pathological data

Whole blood for DNA and serum and fresh prostatic secretions were collected from each patient included in the analysis. Each prostatectomy specimen was gently wiped with saline-soaked gauze. Prostatic secretions were extracted by milking both seminal vesicles and the prostate from their most distal aspects to the urethral meatus. The expressed prostatic fluid was centrifuged to remove any cellular components and the supernatant was stored at −20°C. Whole blood was centrifuged for serum separation and samples were similarly stored at −20°C on the day of specimen collection. Serum and prostatic fluid were sent to Heartland Assays (Ames, Iowa) for measurement of 25(OH)D levels. Due to low fluid volumes, 1,25 dihydroxyvitamin D and 24,25 dihydroxyvitamin D levels in the prostate were not analyzed. Heartland Assays also provided serum levels of total 1,25(OH)₂D.

All prostatectomy specimens were graded and staged according to the 2014 Gleason grading system³² and guidelines³³ by two experienced pathologists (AKB, VM). A representative block containing the largest diameter of the highest Gleason grade group tumor was selected for indirect TUNEL staining. The corresponding tumor block was requested from the VA Pathology Department archives. The tumor blocks were then recut to produce 5-μm-thick slices.

Genetic ancestry estimation

DNA prepared from buffy coat drawn at the time of prostatectomy was used for ancestry genotyping. Variations in the distribution of single‐nucleotide polymorphisms (SNPs) can be used to differentiate human populations³⁴ and panels of ancestry informative markers (AIMs) have been developed to distinguish populations and to estimate population admixture. SNP genotyping was performed using the Sequenom MassARRAY genotyping platform with iPLEXchemistry according to manufacturer’s recommendations³⁵. We used a panel of 105 unlinked AIMs described by Kosoy³⁶ to estimate the proportion of West African (WA), European, and Native American (NA) genetic ancestry for each participant. This panel of AIMs consists of 105 unlinked SNPs and was previously reported in Grizzle et al¹⁷. Since the distribution of Native American ancestry was similar between Black and non-Black participants, we were able to analyze all participants together.

Indirect TUNEL staining and Quantification

Tissue sections were deparaffinized through graded changes of xylene and ethanol using a preset protocol on Autostainer XL (Leica Biosystems). TUNEL staining was carried out with ApopTag Plus Peroxidase In Situ Apoptosis Kit (Millipore-Sigma, #S7101) according to the manufacturer’s instructions. In brief, sections were pretreated with Proteinase K for 15 min at room temperature, and after washes, the endogenous peroxidase activity was quenched by 3% hydrogen peroxide for 5 minutes. Sections were briefly incubated in equilibration buffer, washed, covered with working strength solution of the TdT enzyme and incubated in a humid chamber at 37°C for 1 hour.

The reaction was stopped by agitating the samples for 15 seconds and then incubating them for 10 minutes in working strength Stop buffer at room temperature. After washes, anti-Digoxigenin conjugate was applied to the sections and incubated in a humid chamber for 30 minutes at room temperature. The signal was developed by completely covering the sections with peroxidase substrate and staining for 3 minutes while monitoring the progress under the microscope. Each slide was stained individually to keep staining time consistent. Sections were washed in water and counterstained with hematoxylin, dehydrated on Autostainer XL and mounted with Surgipath Micromount media (Leica Biosystems, Buffalo Grove, IL).

Slides were counted for 1000 epithelial nuclei by two independent trained reviewers (US, PG) by dividing the slides into quadrants and having both reviewers select quadrants at random for all the slides reviewed on a given day. To randomize the results, the nuclei were counted by starting the count with tumor glands near the center of the quadrant and working clockwise from this gland until 1000 nuclei were counted. When the reviewers’ percentages were discordant, these were recounted by both reviewers. The manual TUNEL counts were averaged across both readers’ new counts. The manual counts were performed on all slides and then reviewed for accuracy by a research pathologist (VM); VM would perform an additional count in cases of ongoing discordance and counts were averaged between the 3 manual counts.

Statistical analysis

For comparisons of clinical variables, levels of vitamin D metabolites and genetic ancestry across vitamin D statuses (deficient, insufficient, and normal) and racial group, we used non-parametric medians tests for continuous variables and Chi-square tests for categorical variables. Chi-square trend tests were used to compare distributions of non-binary categorical variables.

To assess for changes in TUNEL staining by tumor aggressiveness within clinically localized specimen, we used two analysis of variance (ANOVA) models with benign and tumor TUNEL staining as the dependent variables and used Gleason grade group and Cancer of the Prostate Risk Assessment-postsurgical (CAPRA-S) scores as the independent variables.

Spearman’s bivariate correlation was used to analyze the relationship between prostatic and serum 25 hydroxyvitamin, serum 1,25 dihydroxyvitamin D, and both tumor and benign TUNEL staining percent. Other putative predictors were also included in the correlation matrix (age, PSA, Gleason grade group, TNM pathologic staging, genetic ancestry, and race). The percentage of cells with positive indirect TUNEL staining was coded as a continuous measure of tumor apoptosis (0–100% labeled). The season was modeled as a binary variable with high ultraviolet radiation exposure (May 1–October 31) and low exposure months (November 1–April 30)¹⁵.

Best-fit linear regressions using -2 log-likelihood scores were constructed using the iterative model building to determine the associations between tumor and benign TUNEL staining percentage (100 x # of positively staining nuclei/1000 epithelial nuclei) and vitamin D status (i.e., serum 1,25(OH)₂D and serum and prostatic 25(OH)D). Serum 25(OH)D status was defined as deficiency (<20.0 ng/ml), insufficiency (20.0 – 29.9 ng/ml) and sufficient (≥30.0 ng/ml)³⁷. The linear models for tumor staining were adjusted for Gleason grade group³², age in 10-year increments, log₂(PSA), and the percent of benign epithelial cell TUNEL staining in case-matched samples. Batch number and the benign TUNEL staining variables were ultimately removed from the tumor TUNEL model to improve fit. The benign model was similarly constructed.

For the linear models, all genetic ancestry and the three vitamin D metabolites were modeled as continuous variables and quartiles after assessing the relationships graphically. Gleason grade group (1–5) and percent of benign epithelial cell TUNEL staining were modeled as continuous variables. Six participants’ prostatic 25(OH)D levels had to be imputed based on other Vitamin D metabolites. All statistical analyses were performed using the statistical package IBM SPSS version 27 (SPSS Inc., Chicago, IL, USA). The alpha = 0.05 and was two-sided for all analyses which were selected a priori.

Results

Patient demographics and clinical characteristics by serum 25(OH)D status are shown in Table 1. Overall, the participants included 90 non-Hispanic Blacks, 29 non-Hispanic Whites,1 Hispanic South American and 1 Native American. At baseline, the deficient, insufficient, and normal vitamin D status groups were well matched with no significant differences in age (p = 0.06), PSA (p = 0.37), BMI (p = 0.52) or prostate volume (p = 0.96), while varying significantly in both serum 1,25(OH)₂D (p <0.01) and prostatic 25(OH)D (p = 0.02) levels. Vitamin D status groups were similar regarding prevalence of 1^st degree relatives with PCa (p = 0.35), season of blood draw (p = 0.47), >pT2cN0M0 tumor staging (p = 0.66) and Gleason grade group ≥2 (p = 0.26). Men with the least WA ancestry (median WA ancestry, IQR: 0.46 – 78.7) had normal vitamin D levels compared to the deficient and insufficient groups with higher percentage of WA ancestry (p = 0.001). Similarly, Black men were over-represented among men with vitamin D deficiency (79% vs. 15%) (median EA ancestry, IQR: 6.75 – 37.5) and insufficiency (68% vs. 29%) (median EA ancestry, IQR: 8.2 – 69.6) as opposed to White men (p = 0.002).

Table 1A:

Demographic and clinical characteristics by serum 25(OH)D status

Continuous Variables	Serum 25(OH)D < 20.0 ng/ml N = 47	Serum 25(OH)D 20.0–29.9 ng/ml N = 44	Serum 25(OH)D ≥30.0 ng/ml N = 29	P valuea
Age (years)	63.0 (56.0–68.0)	66.0 (60.5–69.0)	66.0 (62.0–69.0)	0.06
Prostate Specific Antigen (ng/ml)	7.86 (5.41–13.1)	7.80 (4.8–15.1)	7.56 (5.13–13.7)	0.37
CAPRA-S	4 (2.0–5.0)	4 (2.0–5.0)	3 (1.75–5.0)	0.68
Body Mass Index (kg/m2)	29.5 (26.4–31.4)	27.8 (24.3–30.9)	28.7 (24.3–31.7)	0.52
West African Ancestry (%)	78.7 (60.0–85.7)	68.0 (28.7–78.4)	4.2 (0.46–78.7)	0.001
European Ancestry (%)	15.5 (6.7–37.5)	29.3 (8.2–69.6)	91.6 (20.0–97.0)	0.002
Native American Ancestry (%)	2.7 (1.5–7.8)	3.1 (1.5–6.7)	1.8 (0.7–6.8)	0.85
Prostate volumed (cm3)	28.7 (22.5–41.5)	31.9 (22.3–42.3)	29.3 (21.9–41.0)	0.96
PSA Density (ng/ml/cm3)	26.0 (16.4–41.4)	24.4 (12.6–41.2)	25.3 (17.4–47.5)	0.33
Tumor TUNEL (%)	0.8 (0.5–1.1)	0.8 (0.5–1.2)	0.8 (0.5–1.2)	0.85
Benign TUNEL (%)	0.3 (0.2–0.4)	0.3 (0.2–0.4)	0.3 (0.2–0.4)	0.96
Total Vitamin D Intake (IU/day)	61.7 (19.1–123.0)	53.5 (23.0–83.0)	66.0 (33.5–178.5)	0.15
Serum 1,25(OH)2D (pg/ml)	40.0 (34.5–55.2)	50.3 (35.9–80.0)	57.8 (43.0–79.5)	0.047
Prostatic 25(OH) D (pg/ml)	0.31 (0.02–0.73)	0.55 (0.31–1.74)	1.40 (0.19–2.59)	0.02
Categorical Variables	Serum 25(OH)D < 20.0 ng/ml N = 47	Serum 25(OH)D 20.0–29.9 ng/ml N = 44	Serum 25(OH)D ≥30.0 ng/ml N = 29	P valueb
1st degree prostate cancer family history	13 (28.3%)	8 (17.8%)	9 (31.0%)	0.35
Black Race	38 (80.9%)	37(82.2%)	15(51.7%)	0.006
Blood draw during high-UV period (May-October)	20 (58.8%)	11 (45.8%)	12 (63.2%)	0.47
Gleason Grade group 2–5	45 (95.7%)	38 (86.4%)	27 (93.1%)	0.26
Non-organ confined	21 (44.7%)	21 (46.7%)	16 (55.2%)	0.66

When comparing the same factors across Black (n=91) and non-Black race (n=30), both groups were similarly well-matched. As expected, the degree of West African and European ancestry was statistically significantly different. There was a trend towards the non-Black men having more aggressive prostate cancer (see Table 1B). Only Black men had radical prostatectomies for Gleason grade group 1 disease, but note, recruitment occurred before active surveillance in Black men was considered safe.

Table 1B:

Demographic and clinical characteristics by racial group

Continuous Variables	Non-Black N = 30 Median (IQR)	Black N = 91 Median (IQR)	P valuea
Age (years)	65.0 (60.0–69.0)	66.0 (59.0–69.0)	0.87
Prostate Specific Antigen (ng/ml)	6.3 (4.5–10.2)	7.9 (5.3–15.5)	0.56
Prostate volumed (cm3)	30.0 (22.0–46.0)	29.6 (22.8–41.0)	0.99
PSA Density (ng/ml/cm3)	0.23 (0.10–0.39)	0.28 (0.16–0.47)	0.56
Body Mass Index (kg/m2)	29.6 (26.9–32.6)	28.1 (24.4–31.2)	0.27
CAPRA-S	4.0 (2.0–7.0)	3.0 (2.0–5.0)	0.26
West African Ancestry (%)	1.4% (0.4–4.3%)	76.3% (66.2–85.8%)	<0.001
European Ancestry (%)	94.7% (90.2–98.8%)	6.8% (2.5–17.0%)	<0.001
Native American ancestry (%)	1.8% (0.8–5.8%)	2.9% (1.53–7.67%)	0.44
Benign TUNEL (%)	0.19% (0.26%)	0.30% (0.16%)	0.15
Tumor TUNEL (%)	0.40% (0.20–0.98%)	0.70% (0.40–1.10%)	0.63
Serum 25(OH)D (ng/ml)	29.6 (19.0–36.2)	21.2 (13.1–27.1)	0.052
Serum 1,25(OH)2D (pg/ml)	0.049 (0.031–0.065)	0.049 (0.037–0.071)	0.88
Prostatic 25(OH)D (pg/ml)	1.10 (0.10–1.68)	0.25 (0.03–1.69)	0.44
Categorical variable	Non-Black N = 30 n (%)	Black N = 91 n (%)	P valueb
1st degree prostate cancer family history	4 (13.3%)	26 (28.9%)	0.14
CAPRA-S ≥6	10 (33.3%)	16 (17.6%)	0.08
Non-organ confined	15 (50.0%)	28 (30.8%)	0.08
Gleason Grade group			0.10c
1	0 (0%)	10 (11.0%)
2	15 (50.0%)	50 (54.9%)
3	6 (20.0%)	17 (18.7%)
4	2 (6.7%)	1 (1.1%)
5	7 (23.3%)	13 (14.3%)

^a.Non-parametric Medians tests were performed for continuous variable comparisons across the vitamin D status and racial groups;

^b.Chi-square tests were used for comparisons of categorical variables;

^c.a Chi-square trend test evaluated the distribution of Gleason grade groups;

^d:transrectal ultrasound was used to estimate prostate volume; 1,25(OH)₂D: 1,25 dihydroxyvitamin D (calcitriol); 25(OH)D: 25 hydroxyvitamin D (calcidiol); CAPRA-S: The Cancer of the Prostate Risk Assessment-postsurgical is a composite measure of risk of biochemical recurrence after radical prostatectomy; high-UV period: 6 month period from May-October in Chicago, Illinois with sufficient ultraviolet radiation exposure for adequate cutaneous vitamin D synthesis; PCa: Prostate cancer; non-organ confined refers to pathologic stage greater than pT2cN_0/xM_o/x; TUNEL: Terminal deoxynucleotidyl transferase dUTP nick end labeling.

Using analysis of variance models, we found no statistical associations between Gleason grade group nor CAPRA-S score (modeled both as continuous and ordinal variables) with either benign or tumor TUNEL staining (See Table 2)³⁸. On post-hoc analysis, tumors with CAPRA-S scores from 0–2 (0.57%) have statistically lower tumor TUNEL staining than tumors with CAPRA-S scores from 3–11 (0.80%) on a bivariate analysis, p= 0.006. We also saw no associations with prostatic 25(OH)D concentrations by Gleason grade group or by CAPRA-S score (all p>0.10).

Table 2:

TUNEL staining stratified by Gleason Grade group and CAPRA-S score

Gleason Grade Group	Frequency N = 121	Percent Tumor TUNEL Staining	Percent Benign TUNEL Staining	Mean Prostatic 25(OHD), pg/ml
1	10	0.61%	0.23%	1.26
2	65	0.68%	0.29%	1.25
3	23	0.79%	0.28%	1.81
4–5	23	0.77%	0.26%	0.78
CAPRA-S Group	Frequency N = 121	Percent Tumor TUNEL Staining	Percent Benign TUNEL Staining	Mean Prostatic 25(OHD), pg/ml
0–2	44	0.57%	0.26%	1.08
3–5	51	0.79%	0.29%	1.55
6–11	26	0.80%	0.28%	1.05

Note: ANOVA models did not reach statistical significance for associations between TUNEL staining with Gleason grade group nor with CAPRA-S score; *on post-hoc analysis, CAPRA 0–2 (0.57%) has statistically lower tumor TUNEL staining than CAPRA 3–11 (0.80%) on a bivariate analysis, p= 0.006; CAPRA-S: The Cancer of the Prostate Risk Assessment-postsurgical score; TUNEL: Terminal deoxynucleotidyl transferase dUTP nick end labeling.

Serum 25(OH)D correlated positively with both tumor (ρ = 0.17, p = 0.03), and benign (ρ = 0.16, p = 0.04) prostatic epithelial TUNEL staining as shown in Table 3. Similarly, prostatic 25(OH)D correlated positively with both tumor (ρ = 0.31, p < 0.001) and benign (ρ = 0.20, p = 0.03) prostatic epithelial TUNEL staining. Serum 1,25(OH)₂D was also correlated positively with benign prostatic tissue TUNEL staining (ρ = 0.17, p = 0.03) and serum 25(OH)D (ρ = 0.25, p = 0.002).

Table 3:

Correlations between the degree of TUNEL staining, Vitamin D metabolites & genetic ancestry

	Tumor TUNEL Staining	Benign TUNEL Staining	Serum 25(OH)D	Prostatic 25(OH)D	Serum 1,25(OH)2D	West African ancestry	Native American ancestry	European ancestry	Black Race
Tumor TUNEL Staining %	1.00 -
Benign TUNEL Staining %	0.34, p< 0.001	1.00 -
Serum 25(OH)D, ng/ml	0.17, p= 0.03	0.16, p= 0.04	1.00 -
Prostatic 25(OH)D, ng/ml	0.31, p< 0.001	0.20, p= 0.03	0.26 p= 0.003	1.00 -
Serum 1,25(OH)2Dpg/ml	−0.05, p= 0.53	0.17, p= 0.03	0.25 p= 0.002	0.11 p= 0.22	1.00 -
West African ancestry	−0.06 p= 0.56	0.12 p= 0.30	-0.32 p= 0.002	−0.20 p= 0.07	0.04 p= 0.71	1.00 -
Native American ancestry	0.22, p= 0.05	0.27, p= 0.02	−0.19, p= 0.07	−0.17, p= 0.13	−0.09 p= 0.39	0.10, p= 0.31	1.00 -
European ancestry	0.01, p= 0.93	−0.17, p= 0.12	0.33, p= 0.01	0.19, p= 0.08	−0.02, p= 0.83	-0.96, p< 0.001	-0.28, p= 0.005	1.00 -
Black Race (Black vs. non-Black)	0.06 p= 0.57	0.15 p= 0.29	-0.21 p= 0.01	0.03 p= 0.74	0.01 p= 0.52	0.71 p< 0.001	0.25 0.01	-0.71 p< 0.001	1.00 -

The correlations are based on Spearman’s correlations rho (ρ); 25(OH)D = 25 hydroxyvitamin D; Serum 1,25(OH)₂D = 1,25 dihydroxyvitamin D; TUNEL: Terminal deoxynucleotidyl transferase dUTP nick end labeling; Black race is coded as non-Hispanic Black versus non-Hispanic White + Other.

In Table 3, West African ancestry was significantly negatively correlated with serum 25(OH)D (ρ = −0.32, p = 0.002) and had a trend for negative correlation with prostatic 25(OH)D (ρ = −0.20, p = 0.07). Native American ancestry was positively correlated with Black race (ρ = 0.25, p = 0.01), tumor TUNEL staining (ρ = 0.22, p = 0.05) and benign TUNEL staining (ρ = 0.27, p = 0.02). The negative correlation between serum 25(OH)D (ρ = −0.19, p = 0.07) and Native American ancestry approached significance, as did correlations between prostatic vitamin D levels and both European and WA ancestry. Serum 25(OH)D levels are significantly negatively correlated with Black race status and WA genetic ancestry.

Genetic European ancestry was also positively correlated with Serum 25(OH)D (ρ = 0.33, p = 0.01). There was a similar trend for prostatic 25(OH)D (ρ = 0.19, p = 0.08) levels. The proportions of European, WA (ρ = −0.96, p = 0.001) and NA (ρ = −0.28, p = 0.005) ancestries sum to 1 and are strongly inversely correlated.

The percentage of NA ancestry was also positively correlated with benign TUNEL staining proportions (ρ = 0.27, p = 0.02). NA ancestry correlation with tumor TUNEL was explored graphically in Figure 1 with boxplots. A stepwise increase was apparent between percent NA ancestry quartiles with tumor TUNEL staining proportions but was not statistically significant (p = 0.33) in an ANOVA model.

Figure 1: Benign and Tumor TUNEL Staining by Quartile of Native American ancestry.

The quartiles represent quartiles of Native American ancestry across the entire cohort and are not race-stratified. An Analysis of Variance model was done to test the univariate association of Native American quartiles with tumor and benign TUNEL staining. Both p values were greater than 0.05. TUNEL: Terminal deoxynucleotidyl transferase dUTP nick end labeling.

In Figure 2 are representative images of prominent regions of prostatic epithelium TUNEL assay staining. The frequency of brown TUNEL stained nuclei differs between two PCa patients with insufficient (2A) versus sufficient (2B) serum 25(OH)D levels. The patient with sufficient vitamin D levels (2B) had visual evidence of higher degrees of TUNEL staining relative to the participant with insufficient levels (2A). The benign epithelium had 6 TUNEL-stained nuclei per 1000 nuclei in the vitamin D sufficient participant vs. 3 TUNEL-stained nuclei in the participant with insufficient levels. The tumor epithelium had 12 TUNEL-stained vs. 7 TUNEL stained nuclei per 1000 nuclei in the sufficient and insufficient patient, respectively.

Figure 2: Representative TUNEL assay staining.

Figure 2A (left) demonstrates a 71-year-old White male with clinical stage T1c prostate cancer and a preoperative PSA of 3.8 ng/ml with a total serum 25 hydroxyvitamin D level of 20.7 ng/ml (insufficient) Final pathology demonstrated Gleason grade group 3 pT2cN0Mx adenocarcinoma of the prostate.

Figure 2B (right) demonstrates a 57-year-old Black man with clinical T1c prostate cancer and a preoperative PSA of 22.0 ng/ml with a total serum 25 hydroxyvitamin D level of 55.4 ng/ml (normal). Final pathology demonstrated Gleason grade group 5 pT3aN0Mx adenocarcinoma of the prostate. The participant’s tumor with vitamin D insufficiency had 58.3% of the TUNEL stained nuclei compared to the tumor from the participant with sufficiency. TUNEL: Terminal deoxynucleotidyl transferase dUTP nick end labeling.

In multivariate linear regressions for predicting tumor prostatic epithelial TUNEL staining proportion, quartiles of prostatic 25(OH)D (p = 0.035) and quartiles of NA ancestry (p = 0.009) were statistically significant (Table 4). Neither serum 25(OH)D nor 1,25(OH)₂D metabolites were associated with tumor TUNEL staining (both p >0.10). Notably, the batch number was not associated with the degree of tumor or benign TUNEL staining and was not included in the final models. Race-stratified analyses revealed similar effect estimates for NA ancestry and prostatic 25(OH)D but did not reach statistical significance and were not included.

Table 4:

Multivariate Linear Regressions for Percentage of Benign TUNEL & Tumor TUNEL Staining

Dependent = Tumor TUNEL staining %	β estimate (p-value) R2 = 0.178	β estimate (p-value) R2 = 0.152	β estimate (p-value) R2 = 0.128
Independent variables	Prostatic 25(OH)Da quartiles 0.254, p =0.035	Serum 25(OH)Da quartiles 0.181, p =0.125	Serum 1,25(OH)2Da quartiles 0.049, p = 0.651
	Pathologic stage 0.170, p = 0.175	Pathologic stage 0.164, p = 0.163	Pathologic stage 0.187, p = 0.115
	Age −0.059, p = 0.624	Age −0.077, p = 0.504	Age −.027, p = 0.811
	PSA, ng/ml −0.045, p = 0.719	PSA ng/ml −0.019, p = 0.874	PSA ng/ml −0.067, p = 0.565
	4th Quartile Native American Ancestry 0.311, p = 0.009	4th Quartile Native American Ancestry 0.327, p = 0.004	4th Quartile Native American Ancestry 0.303, p = 0.008
Dependent = Benign TUNEL staining %	β estimate (p-value) R2 = 0.274	β estimate (p-value) R2 = 0.236	β estimate (p-value) R2 = 0.235
Independent variables	Prostatic 25(OH)Da quartiles 0.236, p = 0.037	Serum 25(OH)Da quartiles 0.079, p = 0.478	Serum 1,25(OH)2Da quartiles 0.065, p = 0.527
	Pathologic stage 0.028, p = 0.811	Pathologic stage 0.060, p = 0.590	Pathologic stage 0.081 p = 0.463
	Age −0.320, p = 0.006	Age −0.300, p = 0.007	Age −0.275, p = 0.010
	PSA, ng/ml 0.366, p = 0.003	PSA ng/ml 0.335, p = 0.004	PSA ng/ml 0.317, p = 0.005
	4th Quartile Native American Ancestry 0.159, p = 0.147	4th Quartile Native American Ancestry 0.152, p = 0.146	4th Quartile Native American Ancestry 0.139, p= 0.183

Quartile 1 is the referent group for the prostate 25(OH)D variable. The pathologic stage is a 5-level ordinal variable (≤T2cN0/xM0/x, T3aN0/xM0/x, T3bN0M0/x, T3aN1M0/x, T3bN1M0/x). Age was coded in decades from age 40–49, 50–59, 60–69, 70–79 years old as a 4-level ordinal variable. Log base 2-transformed serum PSA was adjusted for 5-alpha reductase inhibitor use, and the Native American ancestry term is a binary term comparing quartile 4 to quartile 1–3 (referent). 25(OH)D = 25 hydroxyvitamin D.

In multivariate linear regressions for predicting benign prostatic epithelial TUNEL staining proportion, age, serum total PSA, and quartiles of prostatic 25(OH)D (p = 0.037) were significant predictors of benign TUNEL staining (Table 4). Relative to the tumor regressions, NA ancestry quartiles had reduced effect estimates in the benign models and were not statistically significant (all p>0.10). Neither serum 25(OH)D nor 1,25(OH)₂D metabolites were associated with benign TUNEL staining (both p >0.40).

Discussion

In the largest cross-sectional epidemiologic study to date on physiologic levels of serum and prostatic vitamin D metabolites and markers of prostate epithelial apoptosis, we found a positive correlation between indirect TUNEL labeling in prostatic tumor and benign epithelium and prostatic 25(OH)D levels. Furthermore, in multivariate regressions, prostatic 25(OH)D served as a significant predictor of tumor and benign epithelial TUNEL staining. Increasing prostatic 25(OH)D also had similar effect estimates for both benign and tumor TUNEL staining. We also note that serum PSA, which is known to correlate with prostate volume, was positively associated with benign TUNEL staining percentage.

These observations point towards a pro-apoptotic role for vitamin D in prostatic epithelial cells. Indeed, preclinical evidence has shown that vitamin D analogs can induce apoptosis in multiple PCa cell lines^26,39. 1,25 dihydroxyvitamin D has also been shown to promote apoptosis in cell lines of gastric⁴⁰ and breast cancer⁴⁷. Furthermore, active supplementation of vitamin D may affect apoptosis, as observed in one randomized trial in which supplementation with 800 IU/day of vitamin D₃ demonstrated evidence of increased apoptosis in colonic mucosa⁴¹.

A second interesting finding in this study was that Native American ancestry was strongly positively correlated with benign and tumor TUNEL staining in bivariate analyses. We also found a significant independent association between NA ancestry and tumor epithelial TUNEL staining. The 4^th quartile of NA ancestry (all p <0.01) was a significant independent predictor for tumor TUNEL staining (Table 3) across all three models. For tumor staining, the biggest change was observed between the 3^rd and 4^th quartiles (See Figure 1). We also observed a similar pattern in White and in Black men on stratified analysis, but sample size limitations may have led to statistically insignificant results (data not shown).

The association between prostatic 25(OH)D and TUNEL staining in vivo is novel. However, associations with vitamin D analogs with apoptosis have been reported in mouse models and cell lines^42–44. Our in vivo data could provide a rationale for an intervention for men with newly diagnosed PCa to increase their serum 25(OH)D levels which are positively correlated with prostatic 25(OH)D levels⁴³. This strategy could be especially beneficial for Blacks, who had 78% lower median prostatic 25(OH)D levels than non-Blacks in our cohort, tend to be chronically deficient in 25(OH)D,[Reference 15] and have higher rates of PCa and tumor progression relative to their White counterparts⁴⁵. While hypercalcemia as an adverse effect is often associated with 1,25 dihydroxyvitamin D supplementation ⁵³, it is less problematic with supplementation of 25(OH)D.

Guo et al did an age-matched comparison of Black and White PCas that demonstrated higher apoptosis rates (11.6% vs. 4.2%, P < 0. 001) in prostate tumors from Black men (n=79)⁴⁶. This is despite higher rates of vitamin D deficiency and similar tumor mutational burden^38,47. In our cohort, the mean tumor TUNEL staining in Blacks was 2.9% vs. 2.1% in Whites (p=0.23) and for Blacks, benign TUNEL staining was 1.9% vs. 1.1% (p=0.08) in Whites. While not statistically significant, this is in line with the Guo et al analysis. In addition, the Black men in this cohort had a significantly higher median (see Table 1B) and mean Native American ancestry (mean = 5.6%, SD 6.1% vs. mean = 3.0%, SD 3.3%, p= 0.01) in White participants. Together, these observations suggest that there are higher degrees of prostatic apoptosis in Black men and in men with higher degrees of Native American genetic ancestry. This finding is novel and warrants further exploration.

In another comparison of Black and Whites PCas, Khan et. al found that the extracellular vesicles from Black prostate tumors were enriched for inhibitors of apoptosis-like XIAP, cIAP-2, and Survivin⁴⁸. This could result in an excess of intracellular pro-apoptotic signaling and higher activation of apoptosis in Black PCa patients. This type of comparative analysis has not been done in men of NA ancestry or by the degree of genetic ancestry, but should be investigated.

This is the first time that NA ancestry has been associated with prostatic TUNEL staining. Epidemiologically, men with high NA ancestry, (e.g., some Hispanic⁴⁹ and Native Americans) have lower PCa incidence and mortality rates. This suggests that there may be protective genetic factors that are responsible for this association. Causation cannot be inferred given our cross-sectional study design.

Strengths of this study include adequate representation of clinical covariates important for PCa, such as race, genetic ancestry, age, family history, PSA, season, cancer grade, and pathologic stage in our prospectively identified patient population. Two pathologists with extensive experience with PCa research and diagnosis reviewed pathologic results, which helped reduce bias when determining Gleason grade group and the proportion of nuclei with indirect TUNEL-positive staining. Limitations of this study include its relatively small sample size which may limit the power of this associational study. Therefore, larger studies are needed to clarify and confirm the data presented in this paper. In addition, we were only able to measure prostatic 25(OH)D and not 1,25 dihydroxyvitamin D or 24,25 dihydroxyvitamin D given limited prostatic fluid after robotic prostatectomy from the insufflation. Although our metric of apoptosis, indirect TUNEL staining, is a well-established immunohistochemistry marker⁵⁰, it is not as robust as a functional analysis of apoptosis where we could distinguish between necrosis and apoptosis and between caspase-dependent and -independent apoptosis. Being an observational study, it is impossible to assert that low levels of vitamin D metabolites or genetic Native American ancestry are directly responsible for the observed differences in TUNEL staining⁵¹. Lastly, participants undergoing radical prostatectomy at large VA medical centers may not provide data generalizable to the rest of the population. Interventional studies are needed to longitudinally assess vitamin D metabolite status and the mechanisms through which it impacts prostatic epithelial cell apoptosis.

Conclusions

Our current findings suggest that physiologic levels of prostatic 25(OH)D levels and genetic Native American ancestry are associated with prostate benign and tumor epithelium apoptosis rates in men with predominantly clinically localized PCa. Our data provide evidence that higher serum 25(OH)D levels may be pro-apoptotic and warrants further exploration as a chemoprevention agent in non-metastatic PCa.

Data Availability Statement

A de-identified version of the analytic database would be made available as an SPSS or an Excel database for interested parties. Researchers should have a credible research idea and data sharing will be subject to Northwestern University’s Data Use agreement.