Changes in Serum, Red Blood Cell and Colonic Fatty Acids in a Personalized Omega-3 Fatty Acid Supplementation Trial

Abstract

This study evaluated changes in fatty acids from sera, red blood cells and colonic biopsies from a phase Ib clinical trial of personalized ω-3 fatty acid dosing in 47 healthy volunteers. The trial aimed to reduce colonic prostaglandin E₂ (PGE₂), a pro-inflammatory product of arachidonic acid (AA) oxidation. The personalized doses ranged 2-10 grams/day (54% eicosapentaenoic acid, EPA, 24% other ω-3 fatty acids). In colon, increases in ω-3 highly unsaturated fatty acids (HUFA) and EPA:AA ratios each were correlated with decreases in PGE₂. Changes in either colonic EPA:AA ratios or ω-3 HUFA were significantly correlated with changes in the same fatty acid measures in red blood cells or serum. The only blood-based measure significantly correlated with changes in colonic PGE₂ was change in red blood cell ω-3 HUFA (ρ = − 0.39), and the increase in red blood cell ω-3 HUFA was significantly greater in participants who had at least a median reduction in colonic PGE₂ versus those who did not. In summary, fatty acid changes in blood did reflect fatty acid changes in the colon, but additional factors will be needed for optimizing dosing models that seek to predict the anti-inflammatory effects of ω-3 fatty acids on the colon.

Introduction

Colorectal cancer is the third most common incident cancer in men and women in the United States [1]. Studies in animal models of colon cancer have shown that a diet containing ω-3 fatty acids substantially reduces invasive colon cancer incidence and multiplicity in rodents [2, 3]. It is therefore worrisome that 90% of the population in the United States consumes less than the recommended amount of ω-3 fatty acids [4]. Studies in humans do show preventative effect of fish oil or high fish intake on colorectal cancer risk, although not all studies agree [5-8]. Inconsistencies in the epidemiological data could stem from the difficulty in obtaining high enough doses of ω-3 fatty acids from foods, the absence of a defined cancer preventive dose, and benefits that are only evident in specific subgroups of individuals. In a large U.S. study, fish oil supplement use >4 days/week, but not dietary intake of ω-3 fatty acids from foods, was associated with substantially decreased colon cancer risk, especially in men, and there was variation in results by genotype [5].

The potential use of ω-3 fatty acid supplements for colon cancer prevention would offer an attractive alternative to non-steroidal anti-inflammatory drugs that inhibit cyclooxygenases due to the toxicity profiles of that are unacceptable in healthy individuals for chronic use [9].

We conducted a phase Ib clinical trial with ω-3 fatty acids (the Fish Oil Study), to establish the dose of ω-3 fatty acids needed for reducing colonic prostaglandin (PG)E₂ [10]. PGE₂ is a pro-inflammatory mediator in the colon, and the preponderance of rodent data suggest that a reduction in colonic neoplasia is associated with roughly a 50% reduction in mucosal PGE₂ [11-16]. Data on the role of PGE₂ in colonic homeostasis suggest that a partial, but not complete, reduction in PGE₂ is optimal for achieving anti-inflammatory effects while maintaining gut barrier integrity [17]. Therefore, the aim of the trial was to establish a dose that would reduce colonic mucosal PGE₂ by 50%.

The dosing algorithm used the relationship between serum eicosapentaenoic acid, ω-3 (EPA) to arachidonic acid, ω-6 (AA) ratios and colonic PGE₂ that was first established in rodents and enriched with human data as the trial progressed in a Bayesian design [10]. The results of our trial showed that mean colonic PGE₂ was 45% lower after 12 weeks of individualized dosing; however, colonic PGE₂ decreased to a greater extent in normal weight vs. overweight and obese participants despite the fact that the dosing algorithm resulted in higher doses with higher BMI [10]. Obesity has been reported to affect lipid composition and structure, including relatively low proportions of ω3 FA in obese versus normal weight individuals [18]. In addition, supplementation trial have shown that obesity interferes with the effects of dietary ω-3 fatty acids on changes in plasma and tissue fatty acids [19, 20].

Here we evaluated if changes in ω-3 fatty acids achieved after supplementation were related to the heterogeneity in changes in colonic PGE₂ since this might identify more optimal biomarker targets for personalizing dosing. We focused on the fatty acid variables that appear important for maintaining PGE₂ concentrations based on biochemical data. When dietary ω-3:ω-6 fatty acid ratios are increased, arachidonic acid ω-6 and other fatty acids are supplanted by the ω-3 fatty acids in membrane phospholipids [21]. This reduces the availability of arachidonic acid for cyclooxygenase-2 (COX-2) mediated oxygenation to form PGE₂. Biochemical data show that the binding of EPA to COX-1 inhibits arachidonic acid oxygenation as a second mechanism to reduce PGE₂ [22]. Other long-chain ω-3 fatty acids are poor COX-1 substrates and very modest COX-1 inhibitors [22]. This then justified use of EPA:AA ratios as a target for reducing PGE₂ with ω-3 fatty acid supplementation. It is, however, possible that other ω-3 fatty acids present in the fish oil extract such as docosahexaenoic acid could affect immune cells or signaling pathways to reduce the pro-inflammatory milieu in the colon [23]. Unsaturated fatty acids containing 20 or 22 carbons and multiple double bonds are known to produce many eicosanoid-like molecules with important signaling functions; this led to the development of the ω-3/ω-6 highly unsaturated fatty acid (HUFA) balance as a suggested measure of ω-3 fatty acid status that impacts on inflammation [21, 24].

The goal of the current investigation was to identify whether changes achieved in colonic EPA:AA and ω-3 HUFA were associated with reduction in colonic PGE₂ to better understand the heterogeneity observed in PGE₂ changes. We also evaluated the same fatty acid measures in serum and red blood cells since blood measures potentially could be useful biomarkers for optimizing dosing. Serum fatty acids change more rapidly in response to diet than red blood cell fatty acids [25, 26], and we therefore had utilized serum EPA:AA responses to short-term (2-week) supplementation with low and high ω-3 fatty acids to derive a personalized dose in the trial [10]. However, red blood cell fatty acids might better represent changes in colonic fatty acids since this would better represent equilibration into cellular membranes. Here we show changes in EPA:AA ratios and ω-3 HUFA in red blood cells and serum after short-term dosing with low and high ω-3 fatty acid doses, as well as after 12 weeks of the personalized target dose. The colonic biopsies obtained at baseline and after 12 weeks of the personalized target dose also afforded the opportunity to evaluated associations between changes in colonic fatty acids and PGE₂ with serum and red blood cells fatty acids.

Methods

Clinical trial summary

Data for this project was obtained from a phase Ib clinical trial of fish oil supplementation that was described previously [10]. The goal of the trial was to define a model of personalized ω-3 fatty acid supplementation for reducing colonic mucosal PGE₂ by 50% based on serum EPA:AA fatty acid responses. The study was approved by the University of Michigan Medical IRB, (HUM00051786) and registered at www.clincialtrials.org (NCT# 01860352, Effects of Fish Oil on the Colonic Mucosa). All study participants gave signed, informed consent to participate.

Healthy men and non-pregnant women between the ages of 25 and 75 with a BMI between 18 to 40.0 kg/m² were eligible, and a total of 30 women and 17 men were enrolled. In addition, participants were required to have normal white blood cell counts, hemoglobin and platelet counts, and normal renal and hepatic function within the last 28 days. Participants taking fish oils were eligible if they underwent a two-week wash-out, which was the case for six participants. Participants were also asked to avoid taking non-steroidal anti-inflammatory drugs two weeks before the trial and during the trial, and to take only acetaminophen for occasional pain. Five participants had a two-week wash-out of non-steroidal anti-inflammatory drugs before starting the study. A total of 47 eligible participants completed the trial and there was only one adverse event, a Grade 1 headache and nausea that was relieved by one dose of ibuprofen, and this was likely unrelated to the study agent [10].

The fish-oil as supplement was EPA-xtra from Nordic Naturals (Watsonville, CA). The 1 g capsules contained 781 mg ω-3 fatty acids and lemon flavoring (530 mg EPA, 150 mg DHA, 101 g other ω-3 fatty acids per capsule). The personalized dosing protocol involved first providing short-term (two-week), low and high test doses, holding on low dose while the determination of the personalized target dose could be done, and 12 weeks of target dose (Figure 1). The low dose approximated a dietary EPA:ω-6 fatty acid ratio of 0.1 and the high dose approximated a dietary EPA:ω-6 fatty acid ratio of 0.3. To determine the dose needed to attain this ratio rapidly, the ω-6 fatty acid intakes for each participant were estimated. The estimate used the Harris-Benedict equation for energy intakes and the average U.S. intake of ω-6 fatty acids as 5.9% of energy [27, 28].

Figure 1.

Schema of sample collection and visit timing for the study. Shown is the timing of dosing set by the study protocol as well as the actual timing achieved, given as the mean in days and range.

The serum EPA:AA ratio at baseline and the responses to the two-week, low and high ω-3 fatty acid dosing periods were utilized to establish each participant’s own slope and intercept for the relationship between serum EPA:AA ratio versus the ω-3 fatty acid dose, as described previously [10, 29]. The mathematical model relating serum EPA:AA to colonic PGE₂ initially utilized rodent data and was continually updated as the trial progressed with human data. The target dose was defined as the dose needed to reduce colon mucosal PGE₂ by 50%, and the target dose was provided for 12 weeks as shown in Figure 1 [29]. The initial dosing model that established the serum EPA:AA ratio needed to reduce colonic PGE₂ by 50% included rodent data only. This was updated twice with human data as it became available during the conduct of the trial in a Bayesian fashion [29]. The basic principle utilized is that the serum EPA:AA ratios at baseline and after short-term, low- or high-dose ω-3 fatty acids defined a linear relationship of dose to serum EPA:AA for each individual. This linear relationship could then be utilized to calculate the “target” dose needed to achieve the desired EPA:AA ratio. The personalized target dose was provided for 12 weeks, and we hypothesized that the target dose would reduce colonic PGE₂ by 50% [29]. Changes in colonic PGE₂ with each dosing model were as follows: model 1, mean target dose 6.6 g/day, mean change −1.25 ng PGE₂/mg protein (range −7.47 to 3.76 ng/mg, n=16); model 2 mean target dose 6.0 g/day, mean change −6.61 ng PGE₂/mg protein (range −16.11 to 1.35 ng/mg, n=15), and model 3 mean target dose 4.1 g/day, mean change −1.13 ng PGE₂/mg protein, (range −8.87 to 5.49 ng/mg, n=16).

Dietary assessments and guidance

Dietary intakes of study participants were assessed by 24-hour dietary recalls that were conducted by the five-pass method [30]. Two recalls were done in person, one at the study screening visit and one at study baseline. Three recalls were done by un-announced telephone calls during supplementation: one about four weeks after starting the acute dosing protocol (visit 4), one approximately half-way through the 12-week target dosing period (visit 5), and one a week before the target dosing was concluded (visit 6). The actual number of recalls completed was an average of 4.8 per participant (range 3 – 5 per participant). The recalls were analyzed using the Nutrition Data System for Research (2013 version, University of Minnesota, Nutrition Coordinating Center). Food and nutrient intakes are reported as an average of all recalls obtained for each participant.

Study participants were asked to follow their own usual diet while on study. They were provided with a written instruction that they may consume one 3 to 4-ounce portion of listed seafood types once a week. The types of allowed seafood were selected based on low mercury levels and relatively low ω-3 fatty acid content: carp, clams, cod, crab, flounder, halibut, perch, pike, scallops, sole, red snapper, shrimp, tilapia, light tuna, walleye and whiting. Participants were asked to avoid other seafood and to avoid supplements containing ω-3fatty acids while on study. Participants were asked about any problems with taking the fish oil capsules at a minimum of once every 2 weeks (either at study visits or by telephone), and they were provided with suggestions to minimize any issues they reported.

Since the study imposed limits on seafood intake, usual seafood intake before starting the study was assessed at baseline using a “Omega-3 Food Frequency” questionnaire that asked about intakes of 5 categories of seafood (based on ω-3 fatty acid content) and intake of flax and flax oil. Participants were asked to enter usual serving size and to check off the frequency of consumption (Never or rarely, 1-3 times/ month, Once/week, 2-4 times/ week, 5 or more times/week and 2 or more times/day). The frequency and serving size data was used together with the USDA nutrient values of each food to calculate estimated total weekly intakes of ω-3 fatty acids and EPA.

Study participants were instructed to take the fish oil capsules with meals at about the same time each day. Participants could divide their daily dose of capsules between meals as long as the supplementation was consistent each day. They were provided with a pill diary for tracking doses, and they were asked to return any unused pills to the next study visit. The pill diary and the pill counts allowed for calculation of an actual dose consumed.

Biological samples and fatty acid analyses

Fasting blood samples were collected at six time-points: at screening, baseline, after two weeks of the low dose, after two weeks of the high dose, and after 6 and 12 weeks of the personalized target dose (Figure 1). The protocol called completing study visits after the low dose, high dose, and target dose as shown in Figure 1, and the actual timing of visits shown in the figure did vary depending on scheduling issues.

Serum was separated after centrifugation and aliquots frozen at −80°C. Red blood cells were isolated from blood drawn into tubes with EDTA preservative. Briefly, after separating plasma, red blood cells were washed three times with 0.9% NaCl, and packed red blood cells were frozen at −80°C until analysis. Sigmoid colonic mucosal biopsies were obtained at baseline (visit 2) and after 12 weeks of target dosing (visit 6) using un-prepped flexible sigmoidoscopy, and biopsies were flash frozen as previously described [10].

Fatty acids in serum, red blood cells and a portion of the colonic biopsy (as a homogenate) were analyzed by gas chromatography with mass spectral detection [31]. Prostaglandins were extracted from colonic homogenate aliquots using ethyl acetate:hexane extraction and quantified by HPLC with tandem mass spectral detection (HPLC-MS-MS) [10]. Measures of serum total cholesterol, HDL, and triglycerides were performed using a Cobas Mira Chemistry analyzer from Roche Diagnostics Corporation (Indianapolis, IN).

Fatty acids in serum, red blood cells and colon were expressed as a mole percent of total fatty acids, and the ratio of EPA to AA was calculated using the mole percent values. This trial utilized doses of ω-3 fatty acid supplementation that were higher than in many other studies. We therefore calculated several published indices of ω-3 fatty acid status (Table 1). The Omega-3 index was calculated as the relative content of the sum of EPA and DHA in RBC membranes, expressed as a percent of total fatty acids by weight using the calculation method of Harris et al. [32]. The ω-3 highly unsaturated fatty acids (HUFA) and ω-6 HUFA represent percentages of total HUFAs present as either ω-3 or ω-6, respectively; HUFAs are fatty acids with 20-22 carbons and more than 3 double bonds as described by Lands et al. [33]. The HUFA quantified were 20:3 ω-6, 20:4 ω-6, 22-4 ω-6, 20:5 ω-3, 22:5 ω-3, and 22:6 ω-3. EPA percent was percentage of EPA in RBC total fatty acids, which has been suggested to serve as a biomarker of colon tumor EPA content by Watson et al. [34]. Finally, we calculated fatty acid ratios that represent desaturase activities. These were the ratios of 18:1 to 18:0 and 16:1 to 16:0 to assess stearoyl CoA desaturase (SCD-1, δ-9 desaturase), and ratio of 20:4, ω-6 to 18:2, ω-6 to assess fatty acid desaturase activity (FADS, δ-5 desaturase).

Table 1.

Fatty acid variables at baseline (visit 2) and after 12 weeks of dosing with a personalized dose of ω-3 fatty acids (visit 6).

Fatty acid measure	Visit 2	Visit 6	P-value a
Red Blood Cells
AA (20:4, ω-6), mole %	14.4 (3.8)	11.2 (3.0)	<0.001
EPA:AA mole ratio	0.01 (0.06)	0.41 (0.10)	<0.001
ω-3 HUFA, mole % b	26 (6)	48 (5)	<0.001
20:4, ω-6 to 18:2, ω-6 ratio (δ-5 desaturase)	1.02 (0.31)	0.90 (0.30)	<0.001
18:1 to 18:0 ratio (δ-9 desaturase)	0.94 (0.10)	0.93 (0.10)	0.187
16:1 to 16:0 ratio (δ-9 desaturase)	0.019 (0.009)	0.016 (0.008)	0.001
Omega-3 Index c	5.6 (2.0)	10.3 (3.1)	<0.001
EPA, mole % d	1.29 (0.55)	4.48 (1.42)	<0.001
Serum
AA, mole % (20:4, ω-6)	7.36 (1.94)	6.57 (1.60)	0.003
EPA:AA (20:4, ω-6)	0.14 (0.09)	0.72 (0.23)	<0.001
ω-3 HUFA, mole %	28 (7)	56 (7)	<0.001
20:4, ω-6 to 18:2, ω-6 ratio (δ-5 desaturase)	0.29 (0.09)	0.27 (0.09)	0.222
18:1 to 18:0 ratio (δ-9 desaturase)	2.51 (0.70)	1.99 (0.53)	<0.001
16:1 to 16:0 ratio (δ-9 desaturase)	0.071 (0.025)	0.055 (0.020)	<0.001
Colon
EPA, mole % (20:5, ω-3)	1.71 (0.95)	4.42 (1.66)	<0.001
EPA:AA (20:4, ω-6) ratio	2.82 (1.66)	9.63 (11.84)	<0.001
ω-3 HUFA, mole %	29 (5)	47 (5)	<0.001
20:4, ω-6 to 18:2, ω-6 ratio (δ-5 desaturase)	0.47 (0.12)	0.44 (0.13)	0.045
18:1 to 18:0 ratio (δ-9 desaturase)	2.77 (1.10)	2.43 (0.72)	0.013
16:1 to 16:0 ratio (δ-9 desaturase)	0.099 (0.056)	0.090 (0.046)	0.092

^aP-values are from paired t-tests. After controlling for multiple comparisons using the method of Benjamini and Hochberg, P≤0.041 remained statistically significant across the 20 comparisons made.

^bHUFA are highly unsaturated fatty acids (HUFA) acids containing 20-22 carbons and 3 or more double bonds. The proportion of total ω-6 and ω-3 HUFA present as ω-3 HUFA was calculated as published by Strandjord et al. [33] taking the sum of the mole percent of 20:5, 22:5 and 22:6 as a percentage of total 20 carbon and longer fatty acids with 3 or more double bonds (20:3 n6, 20:4, n6, 20:5, n3, 22:4 n6, 22:5 n3 and 22:5 n3).

^cOmega-3 index is the sum of EPA and DHA as a weight percent of total fatty acids using the formula of Harris et al. [32].

^dEPA is a percent of total fatty acids, by weight, in red blood cells.

Statistical analysis

All statistical analyses were done using IBM SPSS version 24 (Armonk, NY: IBM Corp. 2016). For some of the analyses, the participants were classified into two groups using the median change in colonic PGE₂ (which was −1.18 ng/mg protein), as observed after 12 weeks of target dose, compared to baseline at stud visit 2. The distribution of changes in colonic PGE₂ are in Figure 2 shown by dose in grams of supplement/day or grams of supplement per kg body weight per day. Descriptive statistics of blood lipids and fatty acids were used to visualize changes over the course of the study (Figure 3). Changes in all measures were calculated as the difference between the value at visit 6 (after 12 weeks of target dosing) and the value at visit 2 (study baseline) and as a fold change (visit 6/visit 2).

Figure 2.

Distribution of change in colonic mucosal PGE₂ after supplementation with EPA-enriched fish oil for 12 weeks versus baseline by A. supplement dose given as grams/kg body weight/day, B. change in serum ω-3 highly unsaturated fatty acids (HUFA), and C. change in colonic mucosal ω-3 HUFA . The median change was −1.18 ng PGE₂/mg protein.

Figure 3.

Changes in blood fatty acids and lipids over the course of the dosing study. Visits 1 and 2 were pre-supplementation, visit 3 was after two weeks of low dose (average 2.8 g/day), visit 4 was after two weeks of high dose EPA-enriched fish oil supplement (average 8.8 g/day), visit 5 was after six weeks and visit 6 was after 12 weeks of the target dose (average 5.4 g/day). Shown are fatty acid variables as mean and SD and serum cholesterol and triglycerides as mean and SE. The closed symbols and solid line show data for study participants with a decrease of at least the median in colonic PGE₂ at visit 6 (n=24), and the open symbols and dashed line show data for study participants with a smaller decrease or an increase in colonic PGE₂ at visit 6 (n=23). Shown are A: Serum EPA:AA ratio, B: Percent of serum highly unsaturated fatty acids (HUFA) as ω-3 fatty acids, C. Red blood cell EPA:AA fatty acid ratios, D. Red blood cell ω-3 HUFA percent, E. Serum cholesterol (mg/dl), and F. Serum triglycerides (mg/dl).

Changes in fatty acid variables after 12 weeks of target dosing versus baseline were evaluated using paired sample t-tests (Table 1). Two-sample t-tests or Pearson Chi-square tests were performed to evaluate differences in demographics, dietary intakes, lipid variables and fatty acid variables between participants grouped by being above or below the median for change in colonic PGE₂ (Table 2). The method of Benjamini and Hochberg was used to control for multiple comparisons [35]. Correlations between change variables in red blood cells, serum and colon were evaluated using Spearman correlation coefficients and partial Spearman correlations adjusted for age, gender, and BMI (Table 3).

Table 2 -.

Characteristics of study participants who did or did not display a decrease in colonic PGE₂ by at least the median after 12 weeks of personalized ω-3 fatty acid supplementation.

Variable	Did not decrease PGE2 by at least 1 ng/mg protein (n=23)	Decreased PGE2 by at least 1 ng/mg protein (n=24)	P valuea
Demographics
Female (n, %)	11 (48%)	19 (79%)	0.025
Age (years)	52 (13)	43 (13)	0.037
Average BMI over all time points (kg/m2) b	28.7 (4.9)	25.1 (3.6)	0.006
Number of participants with wash-out of ω-3 fatty acid supplements f	5 (22%)	1 (4%)	0.071
Dosing-related Variables c
Target Dose Provided (g/day)	5.3 (2.0)	5.5 (2.0)	0.822
Target Dose Consumed (g/day)	5.4 (2.4)	5.3 (1.9)	0.935
Target Dose Consumed (g/kg body weight)	0.066 (0.030)	0.076 (0.026)	0.239
Percent of targeted serum EPA:AA ratio reached at week 12 (% of target)	134 (52)	135 (55)	0.917
Dietary Intakes d
Energy (kcal/day)	2182 (628)	1966 (402)	0.171
Fat (g/day)	89 (34)	77 (23)	0.184
Fat (kcal % of energy)	36 (6)	35 (5)	0.514
Fiber (g/1000 kcal)	12.6 (3.7)	11.6 (4.1)	0.385
Total ω-3 fat intake from foods while on study (g/week)	14.8 (4.9)	12.1 (4.5)	0.054
Usual ω-3 fat intake from foods before starting study by questionnaire (g/week) e	5.8 (6.4) f	2.4 (3.6)	0.032
Fatty Acid Measures
EPA:AA serum	0.57 (0.23)	0.59 (0.25)	0.854
ω-3 HUFA serum	26 (8)	31 (9)	0.067
EPA:AA red blood cells	0.29 (0.10)	0.32 (0.10)	0.323
ω-3 HUFA red blood cells	19 (5)	23 (5)	0.014
EPA:AA colon	0.29 (0.12)	0.44 (0.13)	<0.001
ω-3 HUFA colon	15 (5)	20 (6)	0.001

^aP-values are from the Pearson Chi-square tests for gender and number of participants with wash-out, or from two-sample t-tests for the continuous variables. P-values <0.01 are significant after adjustment for multiple comparisons using the method of Benjamini and Hochberg.

^bBody weight was measured at every study visit and average BMI while on study was calculated for each participant. Weight change while on study was minimal and not significant (average 1.5 pound weight gain, SD 4.7)

^cThe dose shown is grams of ω-3 fatty acid supplement/day. The supplement was provided in 1 gram capsules and contained 530 mg EPA, 150 mg DHA, 101 g other ω-3 fatty acids per capsule.

^dDiet data is from the average of five 24-hour recalls that were conducted within 4 days of visits 1, 2, 4, 5 and 6.

^eUsual intake of ω-3 fats from foods before going on study was estimated using a questionnaire that asked about usual frequency, serving size and type of seafood intake. Individuals taking ω-3 fatty acids supplements were eligible for the study if they agreed to a 2-week wash out before going on study.

^fThere was one participant with high ω-3 fatty acid intakes from flax, removing that participant resulted in a mean of 3.2 (SD 4.5), with P = 0.172 for the difference between the study participants who did or did not display at least a median decrease in colonic PGE₂.

Table 3 –

Correlations of changes in fatty acid variables in blood with changes in colonic PGE₂ and with changes in colonic fatty acids ^a. Shown is the Spearman coefficient (ρ) and P-value.

Fatty Acid Change Variables	Spearman Correlations (top) and Spearman Partial Correlations (bottom) of Changes in Fatty Acid Variables with Changes in Colonic PGE2 b	Spearman Correlations (top) and Spearman Partial Correlations (bottom) of Changes in Blood Fatty Acid Variables with Changes in Colonic Fatty Acids b
EPA:AA in Red Blood Cells	−0.229 (p = 0.121) −0.246 (p = 0.095)	0.267 (p = 0.070) 0.214 (p = 0.149)
ω-3 HUFA in Red Blood Cells	−0.388 (p = 0.007) −0.280 (p = 0.057)	0.401 (p = 0.005) 0.340 (p = 0.019)
EPA:AA in serum	0.035 (p = 0.815) 0.034 (p = 0.822)	0.303 (p = 0.038) 0.173 (p = 0.244)
ω-3 HUFA in serum	−0.199 (p = 0.179) −0.110 (p = 0.462)	0.339 (p = 0.020) 0.308 (p = 0.035)
EPA:AA in colon	−0.563 (p = 0.001) −0.425 (p = 0.003)	-
ω-3 HUFA in colon	−0.541 (p < 0.001) −0.475 (p = 0.001)	-

^aChange was expressed as the difference of the value at visit 6 minus that at visit 2. The Spearman correlation coefficient (ρ) is shown with the p-value. The partial Spearman correlations were calculated after adjusting for age, gender, and BMI. P-values ≤ 0.020 were significant after correction for multiple comparisons.

^bThe corresponding changes in fatty acid measures in colon with either red blood cells or serum were evaluated for correlations in each case.

Results

Changes in fatty acids after 12 weeks of target dosing

The increases in measuers of ω-3 fattty acid status after 12 weeks of personalized ω-3 fatty acid supplementation were substantial (Table 1). This included the omega-3 index in red blood cells, calculated by the method of Harris et al. [32], the percentage of EPA in red blood cells that was utilized by Watson et al. in studies of the relationships between red blood cells and colon tumor EPA [34], and the proportion of ω-3 HUFA in red blood cells, serum and colon calculated by the method of Strandjord et al. [33].

While measures of ω-3 fatty acid status increased, the mole percent of AA decreased significantly in all three sample types (Table 1). Changes in total SFA, MUFA, ω-3 PUFA and ω-6 PUFA were also observed. In serum and colon, the mean increases in ω-3 PUFA were accompanied by statistically significant decreases in mean SFA and ω-6 PUFA. In red blood cells, mean ω-6 PUFA decreased to balance the increase in ω-3 PUFA, but changes in SFA and MUFA were not significant (not shown). Fatty acid ratios that represent activities of desaturases involved in synthesis of AA (δ-5 desaturase) and MUFA (δ-9 desaturase) also decreased modestly, and this was statistically significant in some cases for δ-9 desaturase, as shown in Table 1.

Demographic characteristics, ω-3 fatty acid dosing, and usual diet in participants who did or did not achieve at least a median decrease in colonic PGE₂

The mean decrease in colonic mucosal PGE₂ was 2.92 ng/mg protein after 12 weeks of target dosing versus that at baseline (SD 5.3, range 5.5 to −18.1, median −1.18). The distribution of changes in colonic PGE₂ by dose is shown in Figure 2. There were 24 participants with a change of −1.18 to −18.11 ng PGE₂/mg protein (ie. a median decrease or more) and 23 participants with changes ranging from −0.78 to 5.49 ng PGE₂/mg protein. Mean values of colonic PGE₂ in persons who did not exhibit at least a median decrease in PGE₂ were 3.6 (SD 3.5) at baseline and 4.2 (SD 4.0) post-supplementation. Mean values of colonic PGE₂ in persons who exhibited at least a median decrease in PGE₂ were 9.5 (SD 6.3) at baseline and 3.2 (SD 2.3) post-supplementation.

In addition to change as a difference, the data were also evaluated in terms of percent change in colonic PGE₂ since the goal of the clinical dosing model was to achieve a 50% reduction in colonic PGE₂. Analyses using percent change yielded similar results as those using the absolute changes (not shown). Since some individuals had very low levels of ω-3 fatty acids and/or PGE₂ at baseline, a large percent change could be misleading. Changes are therefore shown as a difference of values at visit 6 minus that at visit 2 since this better represents the magnitude of the effect that supplementation with ω-3 fatty acids imparts.

The study participants had a mean age of 47 years (range 25-75 years), mean BMI of 26.9 kg/m² (range 20.4-39.9 kg/m²), 30 of the 47 participants were women, five were smokers, and 45 were Caucasian. Table 2 shows select characteristics of study participant who did or did not achieve at least the median reduction in colonic PGE₂. The study participants who displayed relatively larger decreases in colonic PGE₂ (i.e. below the median) were more likely to be female, younger, and have a lower mean BMI although only the latter was statistically significant. Use of fish oil or other ω-3 fatty acid supplements before going on study also was evaluated since supplement users were eligible if they agreed to a two-week wash-out prior to starting the study. Six subjects had been taking supplements containing ω-3 fatty acids, only one of whom was in the group with a decrease in colonic PGE₂ that was below the median (Table 2). Five study participants had washed-out non-steroidal anti-inflammatory drugs (NSAIDs), two of whom were in the group with at least a median decrease in colonic PGE_2. This difference in proportions of subjects with NSAID washout between groups was not statistically significant (not shown). Weight change while on study was minimal and not statistically significant (mean weight gain of 1.5 pounds, SD 4.7, range −7 to 12 pounds).

Study participants consumed 97% of the target dose that was provided to them, on average, as determined from pill diaries and pill counts. The personalized target dose of ω-3 fatty acids provided and consumed was similar in the two groups defined by whether or not at least a median decrease in colonic PGE₂ was achieved (Table 2). However, the dose of ω-3 fatty acids per kg body weight was slightly higher in the group with the larger decreases in colonic PGE₂ (Table 2). Consistent with this, the correlations of change in colonic PGE₂ with dose was weaker when dose was expressed as grams supplement/day (ρ = −0.05, p = 0.73) versus as grams supplement/kg/day (ρ = −0.30, p = 0.04) (Figure 2). The mean serum EPA:AA ratio achieved at visit 6 after 12 weeks of supplementation was 34% higher than that targeted by the dosing model, and this did not differ between the two groups defined by reduction in colonic PGE₂ (Table 2).

Dietary intakes were evaluated from an average of the 24-hour dietary recalls completed throughout the duration of the study at five time points. Mean dietary intakes were not significantly different in the two groups defined by decreases in colonic PGE₂, but mean calorie, total fat, and total ω-3 fat intakes were relatively higher in the group that did not exhibit the larger PGE₂ decreases. The estimate of usual total ω-3 fatty acid intake from foods before enrolling in the study, based on a questionnaire, also was higher in the participants without the larger decreases, as was the number of subjects who washed out ω-3 fatty acid supplement use. This suggests that supplementation might be more effective in subjects with lower prior intakes of ω-3 fatty acids.

Red blood cell and serum fatty acids during the dosing protocol

The EPA:AA ratios in both serum and red blood cells increased in a dose-dependent manner. Figure 3 shows the changes in mean EPA:AA by visit. Visit 1 was at eligibility screening and visit 2 was at baseline. EPA:AA ratios at these two pre-dosing visits were similar. Visit 3 was after two weeks of the low dose, visit 4 after two weeks of the high dose, and visits 5 and 6 were six and twelve weeks after starting with the individualized target dose, respectively. As shown in Figure 3, mean serum EPA:AA ratios and ω-3 HUFA in the two groups defined by change in PGE₂ (above or below the median change of −1.18 ng/mg protein) were similar across visits. Mean serum ω-3 HUFA in the two groups also were very similar (Figure 3). In red blood cells, fatty acids again were similar in the two groups, but fatty acid change did not appear to have plateaued by visit 6 like it did in serum (Figure 3). There was a suggestion that ω-3 HUFA in serum and red blood cells were slightly higher at baseline and slightly lower at visit 6 in the participants with smaller decreases in colonic PGE₂ (i.e. above the median), but this was not statistically significant. The SD at each time point was smaller for ω-3 HUFA than for EPA:AA ratios.

Mean serum cholesterol was higher in participants with smaller decreases in colonic PGE₂ versus participants who exhibited a decrease below the median (Figure 3), consistent with their higher mean BMI (Table 2). Mean cholesterol concentrations in serum did not change appreciably in the 47 subjects combined although there was a trend for decrease after the high dose period at visit 4 (p=0.02 versus baseline, from paired t-tests, not quite significant after adjustment for multiple comparisons) followed by a rise at visits 5 and 6 during target dosing (Figure 3). No significant changes were observed for LDL and HDL (not shown). Mean triglycerides for the 47 study participants exhibited decreases versus baseline at visits 4, 5 and 6 (p≤0.007 in each case using paired sample t-tests vs. baseline), but there were no significant differences between the two groups defined by extent of change in colonic PGE₂ (Figure 3).

Changes in red blood cell, serum and colonic fatty acids after 12 weeks of personalized dosing

In the colonic mucosal biopsies, increases in both EPA:AA ratios and in ω-3 HUFA were significantly higher in the participants who achieved a reduction in colonic PGE₂versus that in those who did not after supplementation with a personalized target dose of ω-3 fatty acids (Table 2). In red blood cells and serum, the differences in the two groups were smaller and not statistically significant after adjustment for multiple comparisons, although the changes ω-3 HUFA in red blood cells approached significance (Table 2). The mean changes in ω-3 HUFA exhibited a smaller coefficienst of variation than changes in EPA:AA ratios.

We also evaluated correlations between changes in colonic measures and changes in blood fatty acids. As shown in Table 3, the correlations betweeen measures of ω-3 fatty acid status with colonic PGE₂ were negative. As could be expected, the correlation was strongest between changes in colonic fatty acids and colonic PGE₂, and these remained statistically significant when partial correlations were calculated after adjusting for age, gender, and BMI. None of the the correlations in serum were significant while the correlations in red blood cells were modest. Changes in red blood cells fatty acids, however, are problematic since they do not appear to have plateaued by visit 6 (Figure 3). Noentheless, changes in blood measures of ω-3 fatty vacid status were positively correlated with colonic measures of ω-3 fatty acid status, and the strongest correlations were found with ω-3 HUFA (Table 3).

Discussion

The clinical trial in this study utilized personalized dosing of a fish oil supplement highly enriched in EPA. The target dose that was provided for 12 weeks was based on a model that utilized serum EPA:AA ratios achieved after low and high test doses delivered during short, 2-week periods, as we published previously [10]. This resulted in a target dose that was higher than that used in most other studies, with an average of 5.4 g/day of EPA-xtra supplement (range 2-10 g/day of supplement that contained 78% ω-3 fatty acids). Several indices of ω-3 fatty acid status did increase favorably. The Omega-3 Index improved from an average of about 6% to over 10% (Table 1). This was well above the 8% cut-off proposed by Harris et al. to be cardioprotective, although it should be recognized that Omega-3 Index values do vary with the analysis method used [32, 36]. The percentage of EPA in red blood cell fatty acids after 12 weeks of supplementation was a mean of 1.22%, a level that was associated with better clinical outcomes in colorectal cancer patients [34]. Fatty acid ratios that estimate δ−9 desaturase activity were decreased post supplementation, and this could be important for colon cancer prevention given the association of stearoyl-coenzyme A desaturase (SCD) with both adiposity and colon carcinogenesis [37-39].

One measure of −3 fatty acid status that is relevant to eicosanoid balance is the proportion of ω-3 HUFA. The percentage of ω-3 HUFA in all three samples types at baseline was almost 30%, or conversely, just over 70% ω-6 HUFA. This is slightly lower than the average ω-6 HUFA value in U.S. population reported by Lands et al. of 78% [24, 40-42]. At the end of the trial, the ω-6 and ω-3 HUFA in red blood cells were roughly 50% each, a level that is suggested to have clear cardiovascular health benefits in healthy individuals, and better clinical outcomes in chronic pain patients [41, 42]. However, compared with data from a Japanese population where the usual diet incorporates a large amount of seafood, the ω-6 HUFA in the present study still did not reach the Japanese average ω-6 HUFA value of 30% [24]. This could be due to the short, 12-week time frame for target dose supplementation, and/or to other dietary factors, such as relatievly higher intakes of other fats in the U.S. versus than in Japan to dilute out the efffects of the supplemented ω-3 fatty acid dose.

We also evaluated changes in serum cholesterol and triglycerides since fish oils have been suggested to be useful for lowering serum triglcyerides [43-45]. Paired t-tests showed that triglycerides significantly decreased after the high dose supplementation, and the significant decrease persisted to the end of the 12-week trial (Figure 3). The high dose at visit 4 was an average of 8.8 g/day (containing 6.8 g/day ω-3 fatty acids). The decrease in triglycerides that we observed after 2 weeks of this high dose the indicates the dose recommended for triglyceride lowering of 4 g/day of ω-3 fatty acids might not be high enough in some individuals [43]. The lack of effects of ω-3 fatty acid supplementation on LDL and total cholesterol in this study agree with that published previously [46, 47].

Although target dosing was based on personal responses to short-term doses of ω-3 fatty acids, the changes in colonic mucosal PGE₂ achieved in this trial spanned a large range (Figure 2). This observation stimulated the present investigation into whether or not differences in ω-3 fatty acid levels achieved are associated with change in colonic PGE₂. In the dosing model we used, serum EPA:AA ratios were used because of the relatively rapid response of serum fatty acids that would be convenient for gauging a preventive dose quickly while at the same time reflecting inter-individual differences in dose absorption.

Red blood cell fattty acids are a better long-term marker of fat intakes than plasma or serum fatty acids since red blood cells have a lifespan of 100-120 days [48]. There are, however, some studies showing that red blood cell membranes can respond to changes in dietary fatty acid composition as quickly as plasma or serum does, reaching an equilibrium in 2-3 weeks [49-51]. We found that EPA:AA in serum plateaued after about 6 weeks (at visit 5) of target dose supplementation (Figure 3). However, EPA:AA in red blood cells was still increasing from week 6 to week 12 of the target dose period. The time frame needed for a full response in red blood cells could be a disadvantage in establishing methods that seek to determine an optimal preventive dose for each individual, but it is recognized that fatty acid changes in at risk tissues also may require a long time frame for equilibration.

Overall, this study indicates that in addition to gender, age, BMI, and possibly dose per kg body weight, ω-3 fatty acid measures could contribute to a portion of the variability in PGE₂ reduction in the colon. The blood measure most strongly correlated with colonic PGE₂ was ω-3 HUFA in red blood cells. The potential predictive value of ω-3 HUFA is interesting since this measure includes EPA, docosapentaenoic acid and docosahexaenoic acid, indicating that fatty acids other than EPA could have important effects on the inflammatory milieu. Use of ω-3 HUFA as a biomarker also might have advantages in that the coefficient of variation was smaller for ω-3 HUFA than for the EPA:AA ratio.

Limitations of this study include the small sample size and that target dosing was conducted for only 12 weeks, at which time we could not determine whether fatty acids in red blood cells or colon had plateaued. In addition, background diet was not controlled. A low-fat diet can facilitate accretion of ω-3 fatty acids in cells [52, 53]. In our study fat intake was not a significant predictor of responder status, but there were very few study participants who followed a low-fat diet. In addition, the study participants were normal volunteers, not individuals at increased risk of colon cancer who might have elevated concentrations of colonic PGE₂. Baseline PGE₂ concentrations were 3-fold higher in the individuals who had PGE₂ decreases that were below the median change. Finally, it should be recognized that the association of change in fatty acids with colonic PGE₂ could be due to mechanisms other than the direct effects of ω-3 fatty acids on formation of PGE₂ such as effects on immune cell activation, induction of enzymes involved in maintaining PGE₂ homeostasis, or some other unmeasured pathway.

Strengths of the study are that the clinical trial afforded the oportunity to evalaute relationships between serum, red blood cells and colonic musocals fatty acids, and adherence to the dosing regimen was excellent. The ω-3 fatty acid supplementation resulted in large changes in red blood cell, serum and colonic fatty acids, and of those, the changes in colonic fatty acids were most strongly related to changes in colonic PGE₂. Mean blood measures of ω-3 fatty acid status either at baseline or after personalized ω-3 fatty acid supplementation were similar in the study participants who did or did not achieve at least a median decrease in colonic PGE₂. Despite this, the data show that the change in the mole percent of ω-3 HUFA in red blood cells was significantly related to change in PGE₂, unlike EPA:AA ratios in red blood cells or either measure in serum. Our previous reports indicated significant contributions of cyclooxygenase induction and differences in the composition of the intestinal microbiome to reduction in colonic PGE₂ [54, 55]. Taken together, these results indicate that defining the personalized, optimal dose of ω-3 fatty acids for reduction of colonic PGE₂ might require multi-faceted dosing models.

List of Abbreviations

AA

arachidonic acid

BMI

body mass index

COX

cyclooxygenase

EPA

eicosapentaenoic acid

HDL

high-density lipoproteins

HUFA

highly unsaturated fatty acids

LDL

low-density lipoproteins

PGE₂

prostaglandin E₂