The worldwide difference in prostate cancer (PCa) incidence highlights an important role of lifestyle in PCa [1
]. Dietary fatty acid (FA) intake is one dietary factor thought to impact PCa development. Interestingly, Inuit people have a very low incidence of PCa and pre-malignant lesion at autopsy, which may be attributed to their traditional diet rich in omega-3 FA (ω3) [5
]. In addition, preclinical and clinical experimental studies support that high ω3 intake have protective effects against PCa, likely via their anti-inflammatory properties [7
]. Animal studies showed that ω3, particularly long-chain ω3 (LCω3), suppress neoplastic transformation, angiogenesis, and tumor cell growth [8
]. A prospective randomized trial with fish oil supplementation showed that LCω3 decrease PCa cell proliferation in men undergoing radical prostatectomy [10
Epidemiological studies provide mixed results regarding associations between FA intake and PCa. Some studies showed that a high level of ω3 is inversely associated with PCa risk [12
], while other studies demonstrated null [17
] or positive associations [22
]. These studies used various methods to evaluate FA intake, such as food frequency questionnaires (FFQ) and FA composition of plasma circulating lipids or of red blood cell (RBC) membranes. Most of the studies evaluating the association between ω3 and PCa were based on FFQ [12
], while fewer studies were based on FA levels in plasma or in RBC membranes [16
]. The FA composition of RBC is thought to be more reliable than plasma, as it reflects the last 3-month diet [27
]. To our knowledge, only one previous study was based on FA levels in the prostate tissue [28
]. In this cohort of 48 men with low-risk PCa, we measured the level of eicosapentaenoic acid (EPA), a LCω3 subtype, directly in prostatic tissue, and found that it was inversely associated with reclassification to high-grade disease. This association was not observed in RBC, suggesting that information bias may be present when considering only blood FA measures.
We undertook a validation study using the same methodology. We hypothesized that ω3 intake, and particularly LCω3 levels in the target prostate tissue, are inversely associated with high-grade PCa during active surveillance. The ω3 can be assessed through food intake, directly in circulating RBC or in target tissue such as prostate. In this study, we aimed at evaluating the associations between ω3 status assessed through different measurements and high-grade PCa, or grade reclassification, in men under active surveillance for a low-risk PCa.
We identified 157 men meeting inclusion criteria. At first repeat—or confirmation—biopsy, 39 men were reclassified to high-grade PCa and 118 had low-grade disease. Characteristics of the study population are summarized in Table 1
. The mean age of patients was 61 years (± 7) and the mean PSA was 5.0 ng/mL (± 2.7). Of those men, 43% had a university degree and 38% had never smoked. BMI higher than 30 Kg/m2
was observed in 23% of participants. The characteristics of the included and excluded patients were similar, except for age (Supplementary Table S1
presents FA dietary intake, FA profiles in RBC, and in prostatic tissue at the first repeat biopsy session. The mean caloric intake was 2500 kcal (± 759). Participants had an average daily intake of ω3 and LCω3 of 2.17g (± 0.90) and 0.41g (± 0.35), respectively. The mean ω3/ω6 ratio from dietary intake was 0.15 (± 0.05) and was borderline higher in low-grade PCa group (p
= 0.05). This difference was significantly reflected in prostate tissue (p
< 0.01). Finally, LCω3 subtype EPA level in prostatic tissue was higher in low-grade PCa group (p
Associations between FA levels in prostate tissue and reclassification to high-grade PCa are presented in Table 3
. FA levels in prostate tissue are associated with high-grade PCa. Compared to men in the lowest tertile of EPA level in prostatic tissue, men in the highest tertile had a decreased risk of high-grade PCa (OR 0.25; 95% CI 0.08-0.79; p
-trend = 0.03). Weaker and stronger associations were observed, respectively, for the LCω3/ω6 ratio and ω3/ω6 ratio measured in prostatic tissue (OR 0.38; 95% CI 0.14–1.00; p
-trend = 0.05, and OR 0.20; 95% CI 0.07–0.59; p
-trend < 0.01, respectively). Table 4
shows the analysis with finer categories of EPA measured in prostate tissue (Supplementary Figure S1
), in which we found a significant dose-response association between EPA and high-grade (p
-trend < 0.01).
The enzymatic activity for long-chain (LC) FA metabolism in prostate tissue was estimated for ω3 and ω6. As shown in Table 5
, we found that compared to high-grade, low-grade PCa group had a higher activity of ∆5-desaturase, which converts eicosatetraenoic acid (ETA) to EPA, as well as a lower activity of Elovl2 elongase, which converts EPA to docosapentaenoic acid (DPA), altogether suggesting an EPA bioaccumulation in prostate tissue of low-grade PCa (Supplementary Figure S2
Associations between FA levels in RBC and reclassification to high-grade PCa are presented in Table 6
. We observed a similar direction of effect of the LCω3/ω6 ratio measured in RBC than that measured in prostate tissue, but none of the RBC associations were significant (all p
-trends > 0.20).
The associations between dietary FA intake and reclassification to high-grade PCa are presented in Table 7
. Contrasting with FA levels in RBC, men in the highest tertile of LCω3 intake had a decreased risk of high-grade PCa (adjusted OR 0.30; 95% CI 0.11–0.83; p
-trend = 0.02) compared to men in the lowest tertile of LCω3 intake. A similar protective effect was observed with a higher LCω3/ω6 ratio (OR 0.29; 95% CI 0.11–0.76; p
-trend = 0.03).
This is a validation study of the first and only study that evaluated the associations between high-grade PCa and FA content measured in the prostate tissue during active surveillance, moreover linking it to RBC and diet. The opportunity of an ongoing randomized controlled trial in men starting active surveillance was ideally suited to examine that question. We found that high level of EPA, a subtype of LCω3, was associated with a decreased risk of high-grade disease in the prostate tissue, and found a significant dose-response association (p
-trend < 0.01). We also observed that higher ω3/ω6 and LCω3/ω6 ratios decreased risk of reclassification to high-grade disease. This validation study supports our a priori hypothesis, based on our initial study [28
], that higher level of ω3, particularly EPA, is inversely associated with risk of high-grade PCa. This current study represents an important opportunity to re-examine this question under the lens of information bias, given that two recent studies reported conflicting results about the role of ω3 on PCa [18
]. These observational studies were conducted within large-scale randomized trials testing other interventions on PCa incidence. Such studies are subject to prognostic selection bias, potentially limiting the generalizability of study findings.
More importantly, the prostate gland is metabolically active. Indeed, the estimated prostatic FA metabolism was also linked to high-grade PCa. Compared to the high-grade group, the low-grade group had a higher activity of ∆5-desaturase, which converts ETA to EPA, while the activity of Elovl2 elongase, which converts EPA to DPA, was significantly lower. Thus, the prostatic FA metabolism, which translates to elevated EPA levels, decreased the risk of high-grade disease. This supports our hypothesis that EPA may be a key protector against PCa aggressiveness.
In order to evaluate if the FA measurement in RBC can be used as biomarker for the prostate FA content, and hence a PCa biomarker, we also examined its relation with PCa grade. We observed that none of the FA measured in RBC were significantly associated with PCa grade. Several methodologic distinctions of our study are worth mentioning to support our findings. Our bio-banking protocol included drawing blood in the early morning after overnight fasting, as well as treatment of bio-specimens within one hour of collection to prevent FA oxidation, which affects the accuracy of FA profiles. These procedures were not part of any of the previous blood FA biomarker studies. Moreover, FA profiles measured in RBC are much less affected by the recent dietary intake and circadian variation than is the FA profile measure in plasma phospholipids, which was used in the vast majority of LCω3 biomarkers studies of PCa risk, including the two large-scale recent ones [18
]. The RBC FA profile rather reflects the FA content of diet of the last three months. One explanation for the lack of association between blood biomarker and PCa grade might also be that the RBCs, being anucleated, reflect both dietary intake and FA metabolism of the whole body, not that specifically of the prostate. In fact, we observed a regression to the mean effect of the same direction of associations with loss of significance. In addition, the FA level correlations between RBC and prostate tissue were weak, at best moderate (Supplementary Table S2
). Unfortunately, RBC FA profile does not seem to be a good biomarker for PCa compared to FA profile in the prostatic tissue.
Since LCω3 FA are essential, we also verified the association with ω3 measured in the diet. We observed, in line with the prostate content, that a high dietary content in LCω3 and LCω3/ω6 were associated with a deceased risk of high-grade disease. This contrasts with the absence of association with FA levels in RBC. This suggests that there may be other protector elements in the diet in addition to LCω3 alone. Interestingly, while we found no such data in cancer patients, we found a trial in insulin-resistant patients that have compared the source of ω3. Patients consumed equal amounts of LCω3 directly from fish versus from processed fish oil. Only fish-consumed LCω3 reduced C-reactive protein, a marker of systemic inflammation, and improved insulin sensitivity [38
]. Thus, the dietary source of LCω3 may matter more for clinical inflammatory-related effects than their blood levels alone. In our current study, even though small sample size precluded most post-hoc subgroup analyses, we observed that greater fish intake (at least once a week) decreased the risk of high-grade PCa (adjusted OR 0.40; 95% CI 0.16–0.98; p
= 0.04), while intake of plant sources of LCω3, such as flaxseed and other nuts, did not (OR 0.73; 95% CI 0.31–1.72; p
= 0.47). However, we did not find any association with vitamin D and other protein sources (data not shown).
Our findings of a protective association between LCω3 and high-grade PCa are in line with previous experimental studies. In fact, most preclinical experimental studies in mice showed that a LCω3 enriched diet reduces PCa progression [40
]. Also, all clinical trials testing the impact of LCω3 on PCa cell proliferation showed a significant reduction [10
]. However, results from observational studies have been inconsistent. Among 30 cohorts and case-control studies investigating incidence of PCa and FA intake or biomarkers thereof, 11 reported inverse associations, similar to our study; the others showed positive or no association. These inconsistent observations may arise from various important methodological issues. First, various techniques and sample types are used to measure FA. For example, the contradictory studies have measured FA concentration in plasma or serum. Plasma levels of ω3 FA are considered a poor biomarker of long-term ω3 intake, because this measurement may be highly influenced by the last meal’s content and plasma triglyceride levels, especially if patients were not fasting prior to blood collection, which was the case in the two larger recent studies [18
]. These methods can lead to information bias of the exposure measure. Second, also in contrast to our study, most studies did not adjust for many potential confounding factors, such as physical activity, alcohol, and anthropometric measures. All of these may affect systemic and possibly prostate tissue inflammation, a potential causal mechanism of ω3, and other biological pathways leading to PCa progression [37
]. Moreover, none of these studies measured the FA profile of the target prostate tissue, supporting the relevance of our study design.
To our knowledge, this is the first study to examine the relation between the FA profile of prostate tissue and high-grade PCa during active surveillance. We found only one other group who examined how prostate tissue FA profile was related to a risk measure of PCa aggressiveness [46
]. In radical prostatectomy specimens, prostate tissue levels of LCω3 (EPA + DHA) were inversely associated with the risk of locally advanced disease (OR 0.52, 95% CI, 0.30-0.90; p
= 0.02) [46
]. There is currently no non-invasive method to quantify any tissue FA profile, and we are the first group to examine this profile while the prostate is still at risk of cancer progression. With emerging technologies, the concept of measuring metabolic aspects of prostate tissue non-invasively is becoming a reality. For example, magnetic resonance spectroscopy provides a non-invasive method of detecting small molecular biomarkers (choline-containing metabolites, polyamines, and citrate) within the prostate [47
]. This improves diagnosis and staging of PCa [47
], and algorithms are being derived to identify prostate tissue inflammation [49
This prostate tissue chronic inflammation is an increasingly supported causal mechanism of PCa development and progression [52
]. While ω6 are pro-inflammatory, LCω3 are anti-inflammatory, mainly because of their derived metabolites [54
]. These include eicosanoids, which also have anti-neoplastic properties when deriving from LCω3 [55
]. A case-control study in the placebo arm of the Prostate Cancer Prevention Trial (PCPT) observed that, as expected from studies of systemic markers of inflammation, plasma ω3 were inversely associated to prostate tissue inflammation [56
]. This observation supports the findings of our current study. However, another study by Brasky et al. observed in the same PCPT study [18
] found positive associations between plasma ω3 FA levels and risk of PCa diagnosis. Thus, because of conflicting observational evidence, PCa risk cannot clearly be related to blood markers of LCω3 intake.
Some other methodological aspects of our study are worth mentioning. We used a cross sectional design, nested in a randomized trial, in which many methodological aspects were designed to evaluate the role of dietary FA on PCa grade. First, dietary intake was measured before the outcome ascertained, reducing possibility of recall bias, which affects some case-control studies of dietary factors. Second, the widely used and well-validated web-FFQ used in the present study measures various food preparation aspects, not measured in many FFQ, but important in FA biology, such as type of oil dressing and cooking modes. This is a great opportunity to evaluate the impact of FA on PCa. Third, rather than a risk of diagnosis model where causal factors may act during a very long period, we used an active surveillance model. This model provides an opportunity to examine the last factors leading to high-grade, potentially lethal, PCa. Inflammation could play an important role in that transition. However, some data provide uncertainty to this chain of continuity [57
], suggesting that low-grade and higher-grade tumor nodules are different entities. Finally, in this current study, the same specialized ultrasonographist performed all study imaging procedures and tissue sampling, using a specific protocol uniform throughout the study. This eliminates inter-observer variation and contributes to increase internal validity.
In summary, this study validates in a larger cohort our previous findings suggesting that LCω3, the EPA subtype in particular, are protective against high-grade PCa. Specifically, the prostate tissue level of EPA and the LCω3/ω6 ratio were inversely associated with high-grade PCa at initiation of active surveillance in men with low-risk PCa. The dietary intake of LCω3, essential FA, and the LCω3/ω6 ratio showed similarly protective associations. In contrast, we observed both an absence of associations between grade and the RBC FA profile, and weak correlations between RBC and prostate tissue FA profiles, suggesting that blood FA biomarkers may be less useful in their capacity to predict PCa aggressiveness. This study provides a rationale for future PCa prevention studies to increase LCω3 intake from food rather than from processed supplement sources. More studies are needed and justified to decipher the effects of LCω3 on PCa, as well as to identify the best FA biomarker of PCa risk and aggressiveness.