Summary
Prostate cancer cells rely heavily on cholesterol. In prospective cohort studies, we assessed the transcriptome of intratumoral cholesterol regulators. Cancers that ultimately became lethal were characterized by de novo cholesterol synthesis rather than transcellular uptake or esterification of cholesterol.
Abstract
Lethal prostate cancers have higher expression of squalene monooxygenase (SQLE), the second rate-limiting enzyme of cholesterol synthesis. Preclinical studies suggested that aberrant cholesterol regulators, receptors and transporters contribute to cholesterol accumulation uniformly. We assessed their association with features of aggressive cancers. In the prospective prostate cancer cohorts within the Health Professional Follow-up Study, the Physicians’ Health Study and the Swedish Watchful Waiting Study, tumor mRNA expression profiling was performed. Lethal disease was defined as mortality or metastases from prostate cancer (n = 266) in contrast to non-lethal disease without metastases after >8 years of follow-up (n = 476). Associations with Gleason grade were additionally assessed using The Cancer Genome Atlas primary prostate cancer dataset (n = 333). Higher Gleason grade was associated with lower LDLR expression, lower SOAT1 and higher SQLE expression. Besides high SQLE expression, cancers that became lethal despite primary treatment were characterized by low LDLR expression (odds ratio for highest versus lowest quintile, 0.37; 95% CI 0.18–0.76) and by low SOAT1 expression (odds ratio, 0.41; 95% CI 0.21–0.83). The association of LDLR expression and lethality was not present in tumors with high IDOL expression. ABCA1, PCSK9 or SCARB1 expressions were not associated with Gleason grade or lethal cancer. In summary, prostate cancers that progress to lethal disease rely on de novo cholesterol synthesis (via SQLE), rather than transcellular uptake (via LDLR) or cholesterol esterification (via SOAT1). These results may help design pharmacotherapy for high-risk patients.
Introduction
As an integral component of cell membranes and precursor of steroid hormones, cancer cells take up cholesterol from the bloodstream or synthesize it de novo (Figure 1). Prostate cancers also rely on cholesterol for androgen biosynthesis (1), which is critical for tumor progression.
Figure 1.
A synopsis of intratumoral cholesterol metabolism. Cholesterol regulators assessed in this study are noted in bold; broken arrows indicate regulation and continuous arrows indicate metabolic steps. Transcellular cholesterol uptake is mediated mainly by endocytosis of cholesterol-laden low-density lipoproteins (LDL) via the LDL receptor (LDLR) and partly by endocytosis of cholesterol esters from high-density lipoproteins (HDL) via the HDL receptor (SCARB1) (2,3). LDLR levels are under tight control of PCSK9 (3) and IDOL (4). The transporter ABCA1 mediates transcellular cholesterol efflux (5). Intracellularly, cholesterol can be produced de novo via the mevalonate pathway with its rate-controlling enzyme 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase (HMGCR), the target of statin drugs (3). The product of the mevalonate pathway, squalene, is then cyclicized and elaborated into cholesterol; this pathway is regulated by the activity of squalene monooxygenase (SQLE) (6). Excess intracellular cholesterol can be esterified for storage in lipid droplets by sterol O-acyltransferase 1 (gene name, SOAT1) (7). Although SOAT1 is often referred to as ‘ACAT1’ (acyl-coenzyme A: cholesterol acyltransferase 1) (7,8), it should not be confused with the enzyme acetyl-CoA acetyltransferase encoded by ACAT1. Sterol regulatory element binding transcription factor 2 (SREBF2) is the main transcription factor that regulates cholesterol homeostasis (3,9).
In vitro and animal studies of prostate cancer as well as assessments of small series of clinical specimens have reported differences in expression and activity of key elements of cholesterol homeostasis. Compared to normal prostate cells, prostate tumors had increased levels of intracellular cholesterol precursors, loss of ABCA1-mediated cholesterol efflux and upregulated LDLR, HMGCR, SCARB1 and SOAT1 expression (2,5,7,10–13). These findings might suggest that all elements of cholesterol homeostasis are deregulated in a rather uniform fashion such that all contribute to cellular cholesterol accumulation in parallel. In particular, higher LDLR expression has been reported in prostate and other cancers (11,14,15).
Translation of these findings into clinically actionable drug targets and into biomarkers requires understanding of their association with cancer progression and lethal disease in patients. Using a conservative discovery approach, we recently identified high mRNA expression of SQLE as strongly associated with higher Gleason grade and as a strong predictor of lethal prostate cancer, independent from standard clinicopathologic factors (16). In this study, we aimed at guiding cholesterol-directed cancer therapy by characterizing the individual associations of specific cholesterol regulators and prostate cancer progression. Studying three large cohorts of prostate cancer patients and The Cancer Genome Atlas (TCGA), we hypothesized that the cholesterol regulators ABCA1, LDLR, SCARB1, SOAT1 and SQLE are differentially associated with prostate cancer aggressiveness as defined by Gleason grade and lethal outcome.
Methods
Patient populations
A unique aspect of the selected patient populations is the inclusion of both treated and untreated prostate cancer patients. The first set of patients included men who were diagnosed with prostate cancer during follow-up of the Health Professionals Follow-up Study (HPFS) and the Physicians’ Health Study (PHS). The HPFS is a prospective cohort study of initially 51529 U.S. male health professionals aged 40–75 years (17). The PHS was initially a randomized, controlled trial of acetylsalicylic acid and micronutrients in primary prevention of cardiovascular disease and cancer among 29071 healthy U.S. male physicians initially aged ≥40 years (18,19). For both cohorts, data on prostate cancer was obtained via biennial questionnaires, systematic medical record review and in-depth ascertainment of death causes (98 and 99% complete in HPFS and PHS, respectively). We developed a tumor tissue repository among a subset of men with prostate cancer in these cohorts, primarily (92%) from those who underwent curative-intent primary prostatectomy. For this study, we included data from men for whom we had whole genome gene expression profiling data.
We also included prostate cancer patients from the population-sampled Swedish Watchful Waiting Study (SWWS) who were diagnosed with prostate cancer when undergoing transurethral resection of the prostate (TURP) for presumed benign prostatic hyperplasia (20). Patients were initially left untreated, but medical or surgical androgen-deprivation therapy was initiated upon symptomatic progression. Metastases and deaths were ascertained via medical records, autopsies and the national death registry. From this cohort, we included men with gene expression profiling data.
The TCGA dataset Prostate Adenocarcinoma (PRAD) in its published version (21) was obtained from cBioportal and the National Cancer Institute (22,23).
Tissue and biomarkers
The hematoxylin–eosin slides of tumor tissue in the HPFS, PHS and SWWS cohorts underwent standardized pathological re-review, as previously described (24), and tissue microarrays were constructed. mRNA expression profiling from high-density tumor areas was performed in PHS and HPFS using the GeneChip Human Gene 1.0 ST array (Affymetrix, Santa Clara, CA; for HPFS and PHS), resulting in 20 254 unique named genes (Gene Expression Omnibus accession number GSE62872). For the SWWS study, we used the DASL platform (Illumina, San Diego, CA), resulting in 6144 unique named genes (GSE8402), as previously described (20,25,26). mRNA expression profiling in TCGA was performed by others as described elsewhere (21).
Cholesterol regulators of interest were ABCA1, LDLR, SCARB1, SOAT1 and, for comparison, SQLE. Further, two regulators of LDLR expression, PCSK9 and IDOL (alternative name, MYLIP), were included. IDOL, PCSK9 and SOAT1 expressions were not measured with the DASL platform in SWWS. Both ACAT1 and HMGCR were not associated with lethal cancer in our previous study (16), and we anticipated HMGCR activity would be reflected by SQLE expression.
Statistical analysis
The three study populations HPFS, PHS and SWWS were sampled nested within the three prostate cancer cohorts using an extreme case–control design (27). Cases of lethal cancer were defined as patients dying from prostate cancer or developing metastases while controls with non-lethal cancer had prostate cancer but were alive without evidence of metastases at 8 years of follow-up.
To assess changes in cholesterol regulators with tumor dedifferentiation, mRNA expression (outcome) was compared across Gleason grade (exposure) using linear regression. Tests for trend across Gleason grades (5–6, 3 + 4, 4 + 3, 8, 9–10) used ordinal indices as a continuous predictor.
Associations of mRNA expression of the cholesterol regulators with progression to lethal cancer were assessed in univariable logistic regression models using quintiles of expression of the individual cholesterol regulators. Tests for trend used ordinal indices of quintiles as a continuous predictor. Follow up for cancer-specific mortality was not available for TCGA. Plots of mRNA expression and Gleason grade as well as lethal cancer were visually inspected for nonlinear associations.
To assess the independent prognostic value of regulators significantly associated with lethal cancer, a logistic regression model was adjusted for age at cancer diagnosis (continuous), year of diagnosis (quintiles), body mass index (<25, 25–30, >30 kg/m2), current tobacco use at diagnosis (binary), family history of prostate cancer in brother or father (binary), diabetes mellitus (binary), statin use at diagnosis (binary), Gleason grade (categorized as above), stage (T1/T2, T3, T4/N1/M1) and SQLE expression (continuous) given its strong previously identified association with lethal disease. A second model additionally included both SOAT1 and LDLR. Associations between SQLE and LDLR as well as SOAT1 and LDLR were assessed by inspecting scatterplots and by using Pearson correlation.
Multiplicative effect modification of the association between LDLR expression and lethal cancer according to PCSK9 and IDOL expressions was assessed in HPFS/PHS. Subgroups of interest were the lowest tertile of PCSK9 expression, as potentially caused by PCSK9 inhibitors (3), and the highest tertile of IDOL expression, as during cholesterol abundance (4). Analyses were repeated for SQLE and SOAT1.
All tests were two sided and P < 0.05 was considered significant. Reported P-values were not adjusted for multiple comparisons, but Bonferroni-adjusted significance levels within one analysis were provided.
The research protocol was approved by the institutional review board at the Harvard T.H. Chan School of Public Health and Partners Healthcare.
Results
Study population
This study included 404 prostate cancer patients (254 from HPFS, 150 from PHS) with initial curative-intent therapy (92% radical prostatectomy) as well as 388 patients from SWWS initially treated with watchful waiting (Table 1). In total, there were 266 lethal cancers and 476 non-lethal cancers. The TCGA data set, used for comparison, contained mRNA expression and Gleason score data for 333 tumors; baseline characteristics were previously reported (21).
Table 1.
Baseline characteristics of men with prostate cancer in the Health Professionals Follow-up Study (HPFS), the Physicians’ Health Study (PHS) and the Swedish Watchful Waiting Study (SWWS)
| HPFS and PHSa | SWWSb | |||
|---|---|---|---|---|
| Lethal cancer | Nonlethal | Lethal cancer | Nonlethal | |
| n | 113 | 291 | 153 | 185 |
| Age at diagnosis, median (range) | 68 (47–81) | 65 (49–80) | 74 (54–91) | 73 (51–89) |
| Year of diagnosis, median (range) | 1992 (1983–2005) | 1995 (1982–2003) | 1991 (1977–1998) | 1992 (1977–1998) |
| Diagnosis before 1993, n (%) | 59 (52) | 78 (27) | No PSA screening | |
| Stage, n (%) | ||||
| pT1a | 24 (16) | 64 (35) | ||
| pT1b | 129 (84) | 121 (65) | ||
| pT1/T2 | 41 (36) | 198 (68) | ||
| pT3 | 46 (41) | 86 (30) | ||
| pT4/N1/M1 | 26 (23) | 7 (2) | ||
| Gleason, n (%) | ||||
| 5–6 | 1 (1) | 56 (19) | 18 (12) | 87 (47) |
| 7 (3 + 4) | 13 (12) | 126 (43) | 28 (18) | 52 (28) |
| 7 (4 + 3) | 35 (31) | 67 (23) | 35 (23) | 33 (18) |
| 8 | 18 (16) | 25 (9) | 12 (8) | 8 (4) |
| 9–10 | 46 (41) | 17 (6) | 60 (39) | 5 (3) |
| PSA, n (%) | ||||
| <4 | 4 (6) | 29 (11) | ||
| 4–10 | 35 (50) | 163 (60) | ||
| ≥10 | 31 (44) | 79 (29) | ||
| Missing | 43 | 20 | ||
| Tissue, n (%) | ||||
| TURPc | 27 (24) | 8 (3) | 153 (100) | 185 (100) |
| Prostatectomy | 86 (76) | 283 (97) | ||
| BMI, mean (SD)d | 25.8 (3.3) | 25.1 (2.8) | 25.3 (3.2) | 26.0 (3.4) |
| Family history, n (%) | 28 (25) | 73 (25) | ||
| Smoker, n (%)e | 17 (15) | 20 (7) | ||
| Diabetes, n (%) | 5 (4) | 9 (3) | ||
| High cholesterol, n (%) | 36 (32) | 79 (27) | ||
| Statin use at diagnosis, n (%) | 10 (9) | 33 (11) | ||
aMostly treated with curative intent (92% radical prostatectomy).
bAll initially treated with watchful waiting.
cTransurethral resection of the prostate.
dBody mass index, with standard deviation. In SWWS, BMI only available for 268 patients (120 cases and 148 controls).
eTobacco use at diagnosis.
Gleason grade
Associations of cholesterol regulators with Gleason grade are presented in Table 2. With increasing dedifferentiation measured by Gleason grade, the expression of LDLR and SOAT1 decreased. As previously reported, tumors with higher Gleason grades had higher SQLE expression (16). Associations for Gleason grade and both LDLR and SQLE were linear and consistent through all four cohorts. Associations for Gleason grade and SOAT1 were linear and consistent between HPFS/PHS and TCGA.
Table 2.
Difference in mRNA expression (in standard deviations) of Gleason 9–10 tumors versus Gleason 5–6 tumors (Δ) with test for linear trend across all Gleason grades (Ptrend)
| HPFS and PHS | SWWS | TCGA | ||||
|---|---|---|---|---|---|---|
| Δ (95% CI) | P trend | Δ (95% CI) | P trend | Δ (95% CI) | P trend | |
| ABCA1 | 0.27 (–0.09 to 0.63) | 0.10 | 0.12 (–0.19 to 0.44) | 0.29 | 0.09 (–0.35 to 0.52) | 0.27 |
| IDOL | –0.20 (–0.53 to 0.16) | 0.49 | Not measured | 0.44 (0.01 to 0.87) | 0.006 | |
| LDLR | –0.94 (–1.29 to –0.60) | <0.001 | –0.25 (–0.55 to 0.06) | 0.039 | –0.68 (–1.11 to –0.25) | <0.001 |
| PCSK9 | –0.10 (–0.49 to 0.26) | 0.42 | Not measured | 0.22 (–0.21 to 0.66) | 0.12 | |
| SCARB1 | 0.22 (–0.14 to 0.58) | 0.32 | 0.16 (–0.05 to 0.37) | 0.33 | –0.06 (–0.49 to 0.38) | 0.89 |
| SOAT1 | –0.30 (–0.66 to 0.06) | 0.049 | Not measured | –0.32 (–0.75 to 0.10) | 0.001 | |
| SQLE | 0.65 (0.31 to 1.00) | <0.001 | 0.38 (0.07 to 0.69) | 0.019 | 0.78 (0.36 to 1.20) | <0.001 |
All P-values are nominal; the Bonferroni-corrected significance levels for seven and four comparisons are P < 0.007 and P < 0.013, respectively.
There were no consistent, significant differences across Gleason grade for ABCA1, SCARB1 and PCSK9. IDOL expression was higher in higher Gleason grades in TCGA, while associations were null in HPFS/PHS.
Lethal cancer
Consistent with their association with Gleason grade, LDLR, SOAT1 and SQLE expressions were associated with lethal cancer (Table 3). The magnitude of risk associated with LDLR was similar between HPFS/PHS and SWWS, while the association with SQLE was attenuated in SWWS. SOAT1 expression data was only available for HPFS/PHS.
Table 3.
Associations of mRNA expression and lethal cancer
| HPFS and PHS | SWWS | |||
|---|---|---|---|---|
| OR (95% CI) | P trend | OR (95% CI) | P trend | |
| ABCA1 | 0.95 (0.47–1.96) | 0.85 | 1.59 (0.82 to 3.32) | 0.08 |
| IDOL | 0.74 (0.37–1.49) | 0.56 | Not measured | |
| LDLR | 0.37 (0.18–0.76) | 0.006 | 0.41 (0.21 to 0.83) | 0.031 |
| PCSK9 | 1.02 (0.50–2.06) | 0.73 | Not measured | |
| SCARB1 | 0.89 (0.44–1.78) | 0.72 | 1.77 (0.89 to 3.52) | 0.08 |
| SOAT1 | 0.41 (0.21–0.83) | 0.079 | Not measured | |
| SQLE | 4.64 (2.27–9.42) | <0.001 | 2.03 (1.01 to 4.08) | 0.22 |
A contrast between high expression (5th quintile of expression) and low expression (1st quintile) is expressed as odds ratio for lethal cancer (OR), and a test for linear trend across all quintiles (Ptrend) is shown. All P-values are nominal; the Bonferroni-corrected significance levels for seven and four comparisons are P < 0.007 and P < 0.013, respectively.
Compared to the lowest quintile of expression, HPFS/PHS patients in the highest quintile of LDLR expression had almost threefold lower odds of lethal disease (OR, 0.37; 95% CI 0.18–0.76; Ptrend = 0.006). Similarly, HPFS/PHS patients in the highest quintile of SOAT1 had 2.4-fold lower odds of lethal disease (OR, 0.41; 95% CI 0.21–0.83; Ptrend = 0.079). In contrast to LDLR, the association between SOAT1 expression and lethal outcome was not linear (Supplementary Figure 1, available at Carcinogenesis Online); the results were mostly driven by relatively higher risk of lethal disease in the lowest quintile of expression.
When adjusting for other prognostic factors for lethal cancer, including stage, Gleason grade and SQLE, the ORs for lethal cancer associated with high LDLR (0.42; 95% CI 0.14–1.25; Ptrend = 0.30) and SOAT1 (0.50; 95% CI 0.19–1.32; Ptrend = 0.47) were still low but did not remain significant. Results were similar when additionally mutually adjusting for SOAT1 or LDLR. SQLE remained an independent prognostic factor for lethal cancer when adjusting for all prognostic factors including LDLR and SOAT1 (OR for 5th versus 1st quintile, 5.98; 95% CI 2.01–17.7; Ptrend = 0.002).
While higher SQLE expression was correlated with higher LDLR expression in TCGA (r = 0.25; P < 0.001), no nonlinear associations or significant correlations were observed in HPFS/PHS (r = 0.08; P = 0.09) and SWWS (r = 0.02; P = 0.66). SOAT1 and LDLR were positively correlated in HPFS/PHS (r = 0.18; P < 0.001) and TCGA (r = 0.27; P < 0.001).
The association of LDLR with lethal cancer did not differ by PCSK9 expression (Pinteraction = 0.73) in HPFS/PHS; the risk was similar for the lowest tertile of PCSK9 expression (OR for 5th versus 1st quintile of LDLR, 0.50; 95% CI 0.14–1.79; Ptrend = 0.063) compared the highest two tertiles (OR 0.32; 95% CI 0.13–0.77; Ptrend = 0.037). In contrast, there was evidence of significant interaction between IDOL and LDLR expressions (Pinteraction = 0.015); the association of LDLR and lethal cancer was null for the highest tertile of IDOL expression (OR for 5th versus 1st quintile of LDLR, 0.95; 95% CI 0.30–3.04; Ptrend = 0.71) in contrast to the two lowest tertiles (OR, 0.21; 95% CI 0.08–0.53; Ptrend < 0.001). The risk associated with SQLE expression (Pinteraction = 0.54) and SOAT1 expression (Pinteraction = 0.96) did not significantly differ by IDOL expression.
Discussion
In this study, we characterized if alterations in intratumoral cholesterol metabolism suggested to occur in prostate cancer are associated with features of aggressive disease. Surprisingly, dedifferentiated tumors and those with lethal outcomes had lower expression of LDLR, the major cellular receptor for cholesterol uptake, and of SOAT1, which esterifies cholesterol for storage in lipid droplets. In contrast, aggressive tumors had high expression of SQLE, the second rate-limiting enzyme of cholesterol biosynthesis. Aggressive tumors were not consistently characterized by intratumoral mRNA expression changes in cholesterol efflux or uptake suggested by prior preclinical studies (Table 4). These results from large patient populations with a broad range of tumor characteristics and outcomes may help guide further mechanistic studies and clinical trials, given that prostate cancer prevention and treatment should focus on high-risk tumors.
Table 4.
A synopsis of findings from preclinical studies, mostly comparing mRNA and protein expression in prostate cancer cell lines or tumor tissue to normal tissue, and findings of mRNA expression in clinical specimens in the current study
| Gene | Function | Preclinical studies | Pathoepidemiology (this study) | |
|---|---|---|---|---|
| Association | Reference | Association | ||
| ABCA1 | Cholesterol efflux pump | Decreased | (5,10–12) | Null |
| IDOL | LDLR regulation | — | — | Potential effect modification of LDLR (see discussion) |
| LDLR | Uptake receptor for low-density lipoprotein (LDL) cholesterol | Increased | (7,11) | Decreased in higher-risk cancer |
| PCSK9 | LDLR regulation | — | — | Null |
| SCARB1 | Uptake receptor for high-density lipoprotein (HDL) cholesterol | Increased | (2,28) | Null |
| SOAT1 | Cholesterol esterification | Increased | (7) | Decreased in higher-risk cancer |
| SQLE | Cholesterol synthesis | Increased | (29) | Increased in higher-risk cancer |
Cholesterol has long been known to accumulate in prostate cancer cells, as reviewed by Schaffner (30). Our results demonstrate variable contributions of individual cholesterol regulators. LDLR was consistently downregulated and no significant mRNA upregulation of the cholesterol uptake receptor SCARB1 or downregulation of cholesterol efflux (as evidenced by ABCA1 expression) was observed. While this may be unexpected at a first glance and in contrast to other, generally more aggressive cancer types (14,15), prostate tumors may rely more on intracellular synthesis of cholesterol than on uptake from the bloodstream, as supported by the strong association of higher SQLE expression with higher Gleason grade and higher risk of lethal outcome. The inverse association of LDLR expression and lethal cancer was not fully independent from tumor grade and stage and did not modify the association of SQLE and lethal cancer.
LDLR expression is endogenously inhibited by IDOL and PCSK9. In TCGA, tumors with higher Gleason grade had significantly higher expression of IDOL, which is induced by cholesterol abundance sensed by the nuclear receptor LXR (4). Among HPFS/PHS tumors with high IDOL, LDLR expression did not remain associated with risk of lethal disease, while SQLE expression did. This finding might suggest that lethal prostate cancers retain feedback inhibition on cholesterol uptake (via LDLR) in the setting of high cholesterol synthesis (via SQLE). However, correlations between SQLE and LDLR expression were positive or null in our cohorts, not directly supporting feedback inhibition of LDLR expression. Alternatively, more aggressive cancers might lose expression of a surface protein such as LDLR. Experimental studies are needed to understand how SQLE upregulation and LDLR downregulation are intertwined as prostate cancer progresses.
Monoclonal antibodies against PCSK9 have recently been approved for hypercholesterolemia treatment. Hence, we explored the potential consequences of anti-PCSK9 treatment on the development and progression of prostate cancer given the observed association of LDLR expression and aggressive cancer. PCSK9 expression levels were not associated with Gleason grade or progression to lethal disease in predominantly surgically treated patients. The risk associated with LDLR expression was not different in patients with low prostatic PCSK9 expression, which would be the expected consequence of anti-PCSK9 treatment. However, truly low PCSK9 expression at least due to a germline mutation is rare (3). Our study may therefore not be able to determine the effect of therapeutic PCSK9 lowering, and further mechanistic studies are necessary.
Yue et al. (7). described intracellular cholesterol ester storage in lipid droplets as a unique feature of aggressive prostate cancers, governed by SOAT1 activity. This comprehensive in vitro study demonstrated cholesterol uptake via LDLR and its esterification via SOAT1. Pharmacologic inhibition of SOAT1 abrogated prostate tumor proliferation. In our study, SOAT1 expression was lower in tumors with higher Gleason grade. Although the association of SOAT1 and lethal outcome was not linear and could only be assessed in the HPFS/PHS cohorts, results suggested that tumors that became lethal despite primary treatment were more likely to have low SOAT1 expression. These results are in line our observations of downregulation of LDLR in more aggressive tumors, since LDLR and SOAT1 are functionally coupled, as Yue et al. (7). demonstrated. The interrelationship between SOAT1 expression and intratumoral cholesterol ester contents remains to be elucidated.
Complex roles of ABCA1 have been implicated in prostate cancer. Prostate cancer cell lines lack ABCA1 expression in contrast to normal prostate cell lines (5,11). A study of 33 human prostate cancers reported increasing methylation of the ABCA1 promoter as well as decreasing ABCA1 protein expression with higher Gleason grade (5). However, increased ABCA1 expression has been found in rodent castration-resistant prostate cancer (10). ABCA1 has also been reported to mediate HDL-induced proliferation of prostate cancer cell lines (12). The null associations in our large study do not exclude that ABCA1 contributes to cholesterol accumulation but lend no additional support for it being the primary mechanism. Similarly, in vitro results suggesting decreased viability of prostate cancer cells upon SCARB1 knockdown (2) and those of higher SCARB1 with higher Gleason grade (28) were not mirrored by significant associations of SCARB1 expression and markers of clinical progression in our study.
The main limitation of this study is its reliance on transcriptome data. For this reason, we were unable to study SREBF2, the main transcription factor of cholesterol metabolism, which is mainly regulated on posttranscriptional level (3,9). Post-transcriptional regulation has been described for SQLE (6), SOAT1 (8) and other cholesterol regulators (31). However, mRNA expression changes for both LDLR and SQLE are mirrored by IHC changes in the same direction in prostate cells (11,29). Previous studies have indicated that tumor gene expression varies according to the anatomic zone of the prostate in which the tumor is located, but tumors in this study have an unknown zone of origin (32). Nevertheless, the similar results in a cohort of men with TURP specimens, which are more likely to reflect transition zone tumors, suggest that gene expression of cholesterol regulators is important in both peripheral and transitional zones. Further, our study was large and encompassed a broad spectrum of prostate cancers, but only 8% of HPFS/PHS patients and no SWWS patients had clinical stage IV cancers at diagnosis. Thus, while being adequate to study progression to lethal disease among patients with localized and locoregional cancers, our study cannot exclude effects in metastatic or castration-resistant prostate cancers. Additional studies are needed in patients with these tumors. Finally, interactions between LDLR, IDOL and PCSK9 in prostate cancer need confirmation in separate cohorts and by mechanistic studies.
In summary, this study underscores the importance of intratumoral cholesterol synthesis as a key mechanism in progression of localized prostate cancer to a lethal disease and its role as a potential drug target. The prime importance of intracellular de novo synthesis is in line with data from pharmacoepidemiologic studies suggesting that use of statins, cholesterol synthesis inhibitors, may be associated with decreased risk of mortality from prostate cancer (33). With more intracellular cholesterol synthesis, more aggressive tumors express less of the cholesterol uptake receptor LDLR, possibly mediated through IDOL, and of the cholesterol-esterifying enzyme SOAT1. A concerted effort of experimental, pathoepidemiologic and pharmacoepidemiologic research is needed to better understand how cholesterol metabolism is linked to molecular heterogeneity in prostate cancer and how it can be exploited for cancer prevention and therapy.
Supplementary material
Supplementary data are is available at Carcinogenesis online.
Funding
The Health Professionals Follow-up Study was supported by National Institutes of Health grant UM1 CA167552. The Physicians’ Health Study was supported by grants CA097193, CA34944, CA40360, HL26490 and HL34595. Research was further supported by the Dana-Farber/Harvard Cancer Center Specialized Programs of Research Excellence program in Prostate Cancer (5P50CA090381-08) and the National Cancer Institute (CA141298; CA133891, E.L. Giovannucci). J.R. Rider and L.A. Mucci are Prostate Cancer Foundation Young Investigators.
Supplementary Material
Acknowledgements
We would like to thank the participants and staff of the Health Professionals Follow-up Study and the Physicians’ Health Study for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data.
Conflict of Interest Statement: None declared.
Abbreviations
- HPFS
Health Professionals Follow-up Study
- PHS
Physicians’ Health Study
- SWWS
Swedish Watchful Waiting Study
- TURP
transurethral resection of the prostate
References
- 1. Pelton K., et al. (2012)Cholesterol and prostate cancer. Curr. Opin. Pharmacol., 12, 751–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Twiddy A.L., et al. (2012)Knockdown of scavenger receptor class B type I reduces prostate specific antigen secretion and viability of prostate cancer cells. Prostate, 72, 955–965. [DOI] [PubMed] [Google Scholar]
- 3. Goldstein J.L., et al. (2015)A century of cholesterol and coronaries: from plaques to genes to statins. Cell, 161, 161–172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Zhang L., et al. (2012)Feedback regulation of cholesterol uptake by the LXR-IDOL-LDLR axis. Arterioscler. Thromb. Vasc. Biol., 32, 2541–2546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Lee B.H., et al. (2013)Dysregulation of cholesterol homeostasis in human prostate cancer through loss of ABCA1. Cancer Res., 73, 1211–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Gill S., et al. (2011)Cholesterol-dependent degradation of squalene monooxygenase, a control point in cholesterol synthesis beyond HMG-CoA reductase. Cell Metab., 13, 260–273. [DOI] [PubMed] [Google Scholar]
- 7. Yue S., et al. (2014)Cholesteryl ester accumulation induced by PTEN loss and PI3K/AKT activation underlies human prostate cancer aggressiveness. Cell Metab., 19, 393–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Seo T., et al. (2001)Differential modulation of ACAT1 and ACAT2 transcription and activity by long chain free fatty acids in cultured cells. Biochemistry, 40, 4756–4762. [DOI] [PubMed] [Google Scholar]
- 9. Ettinger S.L., et al. (2004)Dysregulation of sterol response element-binding proteins and downstream effectors in prostate cancer during progression to androgen independence. Cancer Res., 64, 2212–2221. [DOI] [PubMed] [Google Scholar]
- 10. Leon C.G., et al. (2010)Alterations in cholesterol regulation contribute to the production of intratumoral androgens during progression to castration-resistant prostate cancer in a mouse xenograft model. Prostate, 70, 390–400. [DOI] [PubMed] [Google Scholar]
- 11. Murtola T.J., et al. (2012)The importance of LDL and cholesterol metabolism for prostate epithelial cell growth. PLoS One, 7, e39445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sekine Y., et al. (2010)High-density lipoprotein induces proliferation and migration of human prostate androgen-independent cancer cells by an ABCA1-dependent mechanism. Mol. Cancer Res., 8, 1284–1294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Li J., et al. (2016)Integration of lipidomics and transcriptomics unravels aberrant lipid metabolism and defines cholesteryl oleate as potential biomarker of prostate cancer. Sci. Rep., 6, 20984. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Guillaumond F., et al. (2015)Cholesterol uptake disruption, in association with chemotherapy, is a promising combined metabolic therapy for pancreatic adenocarcinoma. Proc. Natl. Acad. Sci. USA, 112, 2473–2478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Villa G.R., et al. (2016)An lxr-cholesterol axis creates a metabolic co-dependency for brain cancers. Cancer Cell, 30, 683–693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Stopsack K.H., et al. (2016)Cholesterol metabolism and prostate cancer lethality. Cancer Res., 76, 4785–4790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Giovannucci E., et al. (2007)Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int. J. Cancer, 121, 1571–1578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Hennekens C.H., et al. (1996)Lack of effect of long-term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease. N. Engl. J. Med., 334, 1145–1149. [DOI] [PubMed] [Google Scholar]
- 19. Christen W.G., et al. (2000)Design of physicians’ health study II–a randomized trial of beta-carotene, vitamins E and C, and multivitamins, in prevention of cancer, cardiovascular disease, and eye disease, and review of results of completed trials. Ann. Epidemiol., 10, 125–134. [DOI] [PubMed] [Google Scholar]
- 20. Setlur S.R., et al. (2008)Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J. Natl. Cancer Inst., 100, 815–825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Cancer Genome Atlas Research Network (2015). The molecular taxonomy of primary prostate cancer. Cell, 163, 1011–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Gao J., et al. (2013)Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal., 6, pl1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Cerami E., et al. (2012)The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov., 2, 401–404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Stark (Rider) J.R., et al. (2009)Gleason score and lethal prostate cancer: does 3 + 4 = 4 + 3? J. Clin. Oncol., 27, 3459–3464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Hoshida Y., et al. (2008)Gene expression in fixed tissues and outcome in hepatocellular carcinoma. N. Engl. J. Med., 359, 1995–2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Penney K.L., et al. (2015)Association of prostate cancer risk variants with gene expression in normal and tumor tissue. Cancer Epidemiol. Biomarkers Prev., 24, 255–260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Sboner A., et al. (2010)Molecular sampling of prostate cancer: a dilemma for predicting disease progression. BMC Med. Genomics, 3, 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Schörghofer D., et al. (2015)The HDL receptor SR-BI is associated with human prostate cancer progression and plays a possible role in establishing androgen independence. Reprod. Biol. Endocrinol., 13, 88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Holzbeierlein J., et al. (2004)Gene expression analysis of human prostate carcinoma during hormonal therapy identifies androgen-responsive genes and mechanisms of therapy resistance. Am. J. Pathol., 164, 217–227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Schaffner C.P. (1981)Prostatic cholesterol metabolism: regulation and alteration. Prog. Clin. Biol. Res., 75A, 279–324. [PubMed] [Google Scholar]
- 31. Sharpe L.J., et al. (2013)Controlling cholesterol synthesis beyond 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR). J. Biol. Chem., 288, 18707–18715. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Sinnott J.A., et al. (2015)Molecular differences in transition zone and peripheral zone prostate tumors. Carcinogenesis, 36, 632–638. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Raval A.D., et al. (2016)Association between statins and clinical outcomes among men with prostate cancer: a systematic review and meta-analysis. Prostate Cancer Prostatic Dis., 19, 151–162. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.