A 24-year prospective study of dietary α-linolenic acid and lethal prostate cancer

Abstract

Several meta-analyses have attempted to determine the relations between intake of α-linolenic acid (ALA) and prostate cancer, but results were inconclusive. 47,885 men aged 40–75y without prior cancer in the Health Professionals Follow-up Study were prospectively followed from 1986 to 2010. Intake of ALA was determined from validated food frequency questionnaires every four years. We used multivariate Cox proportional hazards models to estimate hazard ratios (HR) with 95% confidence intervals (CIs) for lethal prostate cancer (distant metastasis or prostate cancer death). 386 lethal prostate cancers were diagnosed in the pre-PSA era (before February,1994) and 403 cancers in the PSA era. Intake of ALA was associated with increased risk of lethal prostate cancer in the pre-PSA era (comparing top to bottom quintile of intake, multivariate-adjusted HR =1.78; 95% CI = 1.22–2.06; p trend = 0.003), but not in the PSA era (HR =0.81; 95% CI = 0.56–1.17; p trend = 0.53), and the difference in associations was statistically significant (p for interaction = 0.02). Mayonnaise, a primary food source of ALA intake in our cohort, was likewise only significantly associated with lethal prostate cancer in the pre-PSA era. Among many other fatty acids that are correlated with ALA due to shared food sources, none was associated with lethal prostate cancer in the pre-PSA era. In conclusion, higher intake of ALA was associated with an increased risk of lethal prostate cancer in the pre-PSA era, but not in the PSA era. Potential reasons for the differential associations warrant further investigation.

Introduction

Prostate cancer remains the second leading cause of cancer death in American men.¹ About 26,000 men are projected to die of prostate cancer in 2016.¹ The large geographic variations in the rates of prostate cancer and changing rates in migrant studies suggest that modifiable environmental factors such as dietary factors play a role.^2–4

Total Dietary fat was frequently studied in earlier years,⁵ but recent interest has focused on specific types of fats. Alpha-linolenic acid (ALA) is an 18-carbon omega-3 (n-3) fatty acid found in some vegetable oils, walnuts, leafy green vegetables, grains and animal fats.⁶ Although commonly regarded as the precursor for long-chain n-3 fatty acids, increasing evidence suggests that ALA has independent and specific effects on some chronic diseases⁷. Several meta-analyses^8–11 have attempted to determine the relationships between intake of ALA and risk of prostate cancer but the interpretations of results have been complicated by substantial inter-study heterogeneity.

Many factors potentially contribute to the inter-study heterogeneity, including variations in the amount of ALA intake, food sources, dietary assessment methods, frequency of dietary assessments, food composition databases, adjustment for confounding factors, and duration of follow-up. For example, studies in Spain (approximate interquartile range (IQR) for intake of ALA, 0.7–2.1g/d)¹² and Uruguay (approximate IQR, 0.7–1.6g)¹³, where ALA intake was derived primarily from meats, showed positive associations, whereas a study in Italy (approximate IQR, 0.6–2.6g)¹⁴ where ALA intake came mainly from olive oil and other vegetable sources, showed an inverse association. The divergent associations of ALA indicate that adequate control for confounding by dietary patterns or other components of the diet is essential in future studies.

Widespread PSA screening may have further added to the heterogeneity. In the pre-PSA era, prostate cancers were generally diagnosed due to urinary symptoms, whereas in the PSA era, many indolent cancers were diagnosed that likely would have remained undiagnosed in the absence of screening. Therefore, it has been argued that lethal prostate cancer (those that develop distant metastases or cause death) is a more specific outcome to evaluate risk factors in the PSA era.¹⁵

In our past prospective analyses^16–18 in the Health Professionals Follow-up Study (HPFS) with repeated dietary assessments and 4–16 years of follow-up, ALA intake was positively associated with risk of prostate cancer, especially for advanced stage disease. However, we lacked power to examine lethal prostate cancers. With an additional 8 years of follow-up that doubled the number of lethal prostate cancers, we sought to provide further insights into the relationship by focusing on lethal prostate cancer and taking PSA screening into consideration.

Methods

Study Population

The HPFS is an ongoing prospective cohort that includes 51,529 male US health professionals aged 40 to 75 years old at baseline in 1986. Cohort participants are followed by questionnaires every 2 years about lifestyle factors and new medical diagnoses, and by food frequency questionnaires (FFQ's) every 4 years to obtain dietary information. At baseline we excluded those who did not adequately complete the baseline FFQ or had a previous diagnosis of cancer (except nonmelanoma skin cancer). 47,885 eligible participants were prospectively followed for prostate cancer incidence, metastasis and mortality until January 31, 2010. The study protocol was approved by the Institutional Review Board at the Harvard T.H. Chan School of Public Health.

Assessment of Dietary Intake

On FFQs, commonly used units or portion sizes were specified for each food item and participants were asked to report how often, on average over the past year, they had consumed each food item (9 possible responses ranging from “≤1 time per month” to “≥ 6 times/day”). The FFQs specifically inquired about the usual kind of fat used for frying, sautéing and baking. The FFQs also inquired about the usual brand and type of margarine using an open-ended question. Such information was taken into account when calculating ALA intake from fried, sautéed, and baked foods prepared at home. The daily nutrient intake was calculated by multiplying the consumption frequency of each food by its nutrient content and then summing across all foods. The nutrient composition data were primarily based on the US Department of Agriculture Nutrient Database supplemented with information from manufacturers and published reports. We adjusted all the nutrient intakes for total energy using the residual method to reflect the composition of the diet.¹⁹

In the recently completed Women's Lifestyle Validation Study,²⁰ an extensive validation study that involved more than 700 women from the Nurses’ Health Studies, two cohorts of women with similar FFQs to those in the HPFS, the Spearman correlation between intake of energy-adjusted ALA from the FFQ and from two 1-week diet records was 0.57 (95% CI = 0.48 – 0.65) after correcting for random within-person error in the diet records. Similarly, the de-attenuated correlation for ALA between the FFQ and four 24-hour dietary recalls was 0.58.

Identification of Prostate Cancer Cases

Diagnoses of prostate cancer were initially self-reported on biennial questionnaires by the participants and then confirmed by review of medical records and pathology reports. Participants with confirmed prostate cancer diagnoses were separately followed by a biennial questionnaire to obtain information on prostate cancer treatment, progression and metastasis. Deaths in the cohorts were ascertained through reports by family members and searches of National Death Index. Underlying causes of death were determined by review of medical records and death certificates by a study physician, and were based on death certificates alone in the rare cases when the primary medical records were not available. The mortality follow-up rate in the cohort was nearly 100%. The primary study outcome was lethal prostate cancer, defined as cancers that caused death or had distant metastases by the end of follow-up.

Statistical Analysis

Each participant contributed person-time to the analysis from the return of baseline questionnaire to the confirmed initial diagnosis of lethal prostate cancer, death, or the end of follow-up, January 31, 2010, whichever occurred first. To best represent long-term intake and minimize measurement error,²¹ we calculated the cumulative average intake of ALA by averaging all available FFQs up to the start of each two-year risk interval. All cumulative averages were categorized into quintiles based on the distribution in the entire cohort for that two-year risk interval. Likewise, we calculated the cumulative average intakes of foods and categorized them into pre-specified groups.

We used time-varying Cox proportional hazards model to estimate the hazard ratios (HR’s) and 95% confidence intervals (CI’s) for lethal prostate cancer. Multivariable models were stratified by age in months and calendar year and were adjusted for known and suspected risk factors previously identified in our cohort and other studies. We also adjusted for PSA testing, which was first inquired in 1994 and biennially thereafter. We lagged the PSA testing by one period in the analysis to avoid counting diagnostic PSA testing as screening. For example, PSA testing during 1994–1996 was used to adjust for the 1996–1998 follow up period. We tested the linear trend across quintiles of ALA intake by modeling the median intake of each category as a continuous variable.

We stratified our analysis by the time period before and after the clinical introduction of PSA screening. The pre-PSA screening era was defined as February 1, 1986 to January 31, 1994, and the PSA screening era, February 1, 1994 to January 31, 2010. Cumulative average intake of ALA was calculated separately in two periods. To test if the risk estimates differed between the two periods, we created an interaction term by multiplying a continuous time-varying ALA variable derived from the median intake of each quintile by the binary indicator variable for time period and used a Wald test to ascertain the statistical significance of interaction.

Results

During 24 years (941,461 person-years) of follow-up, we confirmed 789 lethal prostate cancer cases among 47,885 participants. 386 cases had an initial cancer diagnosis date in the pre-PSA era and 403 cases in the PSA era. Many demographic and lifestyle factors did not vary appreciably across quintiles of ALA intake in 1990 (mid-point of pre-PSA era) or 2002 (mid-point of post-PSA era), except that participants with higher intake of ALA in 1990 were less likely to be never smokers or to engage in vigorous physical activity (Table 1). Intake of ALA was positively related to intakes of LA, marine n-3 fatty acids (only for 2002 intake), coffee (only for 1990 intake), and tomato sauce, but inversely related to multivitamin use (only for 1990 intake). Intake of ALA was also positively related to the prudent dietary pattern score in both years whereas intake of ALA was positively related to western dietary pattern only in 1990.

Table 1.

Age-standardized characteristics of the study population according to energy-adjusted intake of α-linolenic acid in 1990 and 2002

	1990, Quintiles					2002, Quintiles
	Q1	Q2	Q3	Q4	Q5	Q1	Q2	Q3	Q4	Q5
No. of participants	7,009	7,194	7,147	6,814	7,241	5,499	5,479	5,592	5,479	5,508
Age, year	59	58	58	58	58	68	68	68	68	68
BMI, kg/m2	25.0	25.5	25.6	25.7	25.8	25.9	26.3	26.4	26.4	26.3
Height, inches	70	70	70	70	70	70	70	70	70	70
White, %	96	97	97	96	97	96	97	97	96	96
Never smokers, %	46	47	45	45	43	40	41	42	42	43
Current smokers, %	8	7	8	7	8	4	4	4	3	3
Vigorous activity, % top quintile	21	18	17	16	17	16	15	15	15	16
Has diabetes, %	3	3	3	4	4	5	6	8	8	10
Family history of prostate cancer, %	13	14	14	13	14	12	13	13	13	14
‡PSA testing in the prior two years,%	NA	NA	NA	NA	NA	84	86	85	85	84
Daily dietary intake
Multivitamin use, %	41	38	38	38	39	67	65	66	66	67
Total calories, kcal	1,873	1,916	1,959	1,976	1,869	2,027	2,028	2,034	2,022	2,006
α-linolenic acid, mg	749	922	1,048	1,193	1,527	809	1,011	1,163	1,347	1,966
Linoleic acid, g	9.4	10.3	11.0	11.9	14.0	9.1	10.3	11.2	12.3	15.4
Marine omega3 fatty acids, mg	340	331	337	348	443	306	339	360	382	427
trans 18:1, g	2.4	2.8	2.9	2.9	2.9	1.8	2.0	2.1	2.1	2.0
Supplemental vitamin E, mg	36	32	33	33	41	118	115	116	117	126
Calcium, mg	919	919	913	907	901	1,206	1,140	1,135	1,125	1,146
Total coffee, servings	1.8	1.8	1.9	2.0	2.0	1.6	1.6	1.6	1.6	1.5
Tomato sauce, servings	0.11	0.12	0.13	0.13	0.13	0.14	0.17	0.18	0.19	0.19
Dietary pattern score
Prudent	−0.16	−0.08	0.001	0.06	0.14	0.01	0.11	0.18	0.23	0.39
Western	−0.32	−0.09	0.06	0.18	0.16	0.12	0.20	0.23	0.23	0.11

Values are means(SD) or percentages and are standardized to the age distribution of the study population (except for age).

The number of participants in1990 and 2002 does not equal the number of participants at 1986 baseline due to missing ALA intake in 1990 or 2002.

^‡PSA testing was not asked until 1994.

During the follow-up, the age-adjusted intake of ALA increased about 30% in the HPFS from 1986 (mean ± SD: 1,110 ± 303 mg) to 2006 (mean ± SD: 1,435 ± 745 mg).

The associations between cumulative averaged intake of ALA and lethal prostate cancer for each 4-year follow-up period decreased over time (Figure). Since 1994, which we used to mark the widespread clinical introduction of PSA screening, the associations were markedly attenuated and no longer significant. In Table 2, we stratified the analysis by the pre-PSA and PSA era. The HR comparing the top to the bottom quintile after extensive adjustment for lifestyle and dietary risk factors (multivariable model 2) was 1.46 (95% CI = 1.04 – 2.04; p trend = 0.04) in the pre-PSA era and 0.77 (95% CI = 0.55 – 1.09; p trend = 0.31) in the PSA era. Further adjusting for LA intake strengthened the association in the pre-PSA era (HR = 1.78; 95% CI = 1.22 – 2.60; p trend = 0.003). The difference in pre and post associations was statistically significance (p=0.02).

Figure. Multivariable-adjusted hazard ratios for the association between intake of α-linolenic acid and lethal prostate cancer at different follow-up period.

^ap for trend test across quintiles of ALA intake <0.05.

Multivariable model (MV) included: age, calendar time, race (White, African American, Asian American, or other), current BMI (<21, 21 to 23, 23 to 25, 25 to 27.5, 27.5 to <30, or ≥ 30 kg/m2), height (quartiles), vigorous activity (quintiles, MET-hours/wk), smoking (never, former quit > 10 y ago, former quit ≤ 10 y ago, or current), family history of prostate cancer in father or brother (yes or no), diabetes (Type I or II, yes or no), multivitamin use (yes or no), history of PSA testing (yes or no, lagged by one questionnaire cycle), total calories (quintiles) and linoleic acid (quintiles).

2002–2006 and 2006–2010 intervals were combined into 2002–2010 to increase power due to a small number of cases in each time period

Abbreviations, multivariable model (MV); linoleic acid (LA)

^b PSA testing was not asked until 1994.

Table 2.

The associations between intake of α-linolenic acid and lethal prostate cancer in the pre-PSA (1986–1994) and PSA (1994–2010) era

	Quintiles (Q)						P interaction between two eras
	Q1	Q2	Q3	Q4	Q5	p trend
Pre-PSA eraa
Median intake, mg/d	800	950	1,070	1,205	1,445
Cases/person years	63 / 71,601	78 / 71,472	74 / 71,335	88 / 73,909	83 / 71,435
Age-adjusted model	1 (ref)	1.38 (0.98,1.93)	1.32 (0.94,1.86)	1.55 (1.11,2.15)	1.44 (1.03,2.00)	0.04	0.06
Multivariable model 1c	1 (ref)	1.34 (0.96,1.88)	1.27 (0.90,1.80)	1.48 (1.06,2.05)	1.41 (1.01,1.96)	0.06	0.04
Multivariable model 2d	1 (ref)	1.37 (0.98,1.93)	1.32 (0.93,1.86)	1.53 (1.10,2.14)	1.46 (1.04,2.04)	0.04	0.03
+ linoleic acid	1 (ref)	1.43 (1.01,2.02)	1.43 (1.00,2.04)	1.74 (1.22,2.48)	1.78 (1.22,2.60)	0.003	0.02
PSA erab
Median intake, mg/d	813	980	1,090	1,230	1,500
Cases/person years	71 / 107,198	68 / 106,862	95 / 107,187	74 / 106,580	66 / 106,813
Age-adjusted model	1 (ref)	0.94 (0.68,1.31)	1.06 (0.77,1.46)	1.24 (0.91,1.69)	0.83 (0.59,1.17)	0.60
Multivariable model 1c	1 (ref)	0.93 (0.67,1.29)	1.02 (0.74,1.41)	1.17 (0.86,1.60)	0.78 (0.55,1.09)	0.32
Multivariable model 2d	1 (ref)	0.93 (0.66,1.29)	1.02 (0.74,1.41)	1.18 (0.86,1.61)	0.77 (0.55,1.09)	0.31
+ linoleic acid	1 (ref)	0.94 (0.67,1.31)	1.04 (0.75,1.45)	1.21 (0.87,1.68)	0.81 (0.56,1.17)	0.53

^aAnalysis in the pre-PSA era was based on cumulatively updated ALA intake calculated from 1986 to 1990 FFQs

^bAnalysis in the PSA era was based on cumulatively updated ALA intake calculated from 1994 to 2006 FFQs

^cMultivariable model 1 adjusted for age, calendar time, race (White, African American, Asian American, or other), current BMI (<21, 21 to 23, 23 to 25, 25 to 27.5, 27.5 to <30, or ≥ 30 kg/m2), height (quartiles), vigorous activity (quintiles, MET-hours/wk), smoking (never, former quit > 10 y ago, former quit ≤ 10 y ago, or current), family history of prostate cancer in father or brother (yes or no), diabetes (Type I or II, yes or no), multivitamin use (yes or no), history of PSA testing (yes or no, lagged by one questionnaire cycle), total calories (quintiles)

^dMultivariable model 2 includes all variables in model 1 and intakes of calcium (< 500, 500 to 750, 750 to 1000, 1000 to 1250, 1250 to 1500, 1500 to 2000, > 2000 mg/d), tomato sauce, coffee, supplemental vitamin E and marine omega3 fatty acids (all in quintiles).

Among primary ALA-containing foods, in the pre-PSA era only mayonnaise was significantly associated with higher risk of lethal prostate cancer (adjusted HR comparing 5–6 servings/wk to ≤ 3 months = 1.49; 95% CI = 1.06–2.09; p trend = 0.02) (Table 3). In the PSA era, none of the foods was positively associated with the risk of lethal prostate cancer. However, the highest intake of margarine was inversely associated with the risk of lethal prostate cancer.

Table 3.

Hazard ratios of lethal prostate cancer according major food sources of α-linolenic acid

	Servings
	% contribution to ALA intakea	≤ 3/month	1/week	2–4/week	5–6/week	1/day	≥ 2/day	P trend
Pre-PSA erab
Mayonnaise	13.5	1(ref)	1.31 (0.98,1.75)	1.29 (0.97,1.70)	1.49 (1.06,2.09)			0.02
Oil & vinegar dressing	11.1	1(ref)	1.25 (0.93,1.69)	1.08 (0.80,1.44)	1.15 (0.84,1.56)			0.46
Beef, pork or lamb	6.3	1(ref)	1.22 (0.92,1.62)	1.24 (0.94,1.63)				0.23
Margarine	3.6	1(ref)	0.91 (0.67,1.23)	1.06 (0.77,1.46)		0.96 (0.66,1.40)	0.86 (0.62,1.18)	0.39
Cheese	3.5	1(ref)	0.74 (0.53,1.05)	0.79 (0.60,1.05)	1.07 (0.79,1.46)			0.42
PSA erab
Walnutsc	18.2	1(ref)	1.16 (0.66,2.03)	0.83 (0.41,1.66)				0.72
Mayonnaise	5.7	1(ref)	0.99 (0.72,1.36)	0.50 (0.30,0.82)	0.87 (0.54,1.41)			0.08
Light mayonnaise	1.3	1(ref)	1.04 (0.74,1.46)	1.23 (0.86,1.77)	0.59 (0.32,1.06)			0.44
Oil & vinegar dressing	4.1	1(ref)	1.28 (0.92,1.80)	0.85 (0.61,1.19)	0.98 (0.71,1.35)			0.34
Beef, pork or lamb	0.4	1(ref)	0.95 (0.72,1.25)	0.95 (0.68,1.31)				0.76
Margarine	1.9	1(ref)	0.88 (0.66,1.17)	0.69 (0.48,0.98)		0.71 (0.44,1.13)	0.66 (0.46,0.96)	0.02
Cheese	2.5	1(ref)	1.02 (0.74,1.41)	0.89 (0.65,1.21)	0.75 (0.53,1.07)			0.07

^a% contribution to ALA intake on the 1986 FFQ in the pre-PSA era and to ALA intake on the 2006 FFQ in the PSA era.

^bMultivariable models were adjusted for the same variables as multivariable model 1 in the Table 2.

^cWalnuts were not included in the FFQ until 1998.

Prudent and western dietary patterns are two major dietary patterns identified by factor analysis in the HPFS cohort to reflect overall dietary quality,²² and have been found to be associated with the risk of heart disease²³ but not with risk of prostate cancer.²⁴ ALA intake was modestly positively correlated with those two dietary patterns (Supplemental Figure 1). The correlation of ALA with western dietary pattern was higher than that with prudent pattern before 1994 but the pattern reversed thereafter. This finding raised the question if the positive association in the pre-PSA era but the lack of association in the PSA-era is the result of confounding by the "unhealthy" versus "healthy" sources of ALA. We conducted several analyses in the pre-PSA era to address this question. First, we found that neither prudent (comparing the top to the bottom quintile, HR = 1.01; 95% CI = 0.69 – 1.48; p trend=0.90) or western dietary pattern (HR = 1.16; 95% CI = 0.76 – 1.75; p trend=0.44) was associated with the risk of lethal prostate cancer. Second, at baseline, we separated ALA intake into animal and plant sources (such information was not available after 1986). Animal-sourced ALA (mean ± SD: 366 ± 145 mg) and plant-sourced ALA (745 ± 300 mg) were weakly inversely correlated (Pearson r = −0.22). The multivariable-adjusted HR comparing the top to the bottom quintile was 1.31 (95% CI = 0.94 – 1.84; p trend=0.06) for animal-sourced ALA and 1.49 (95% CI = 0.99 – 2.25; p trend=0.04) for plant-sourced ALA. When further mutually adjusting those two sources, the HR was 1.40 (95% CI = 1.00 – 1.97; p trend=0.02) for the animal-sourced ALA and 1.60 (95% CI = 1.05 – 2.43; p trend=0.02) for the plant-sourced ALA. Moreover, we found that none of the many fatty acids that were correlated with ALA, including trans 16:1 (Pearson r=0.26), trans 18:1 (r=0.08), cis 18:1 (r=0.25), trans 18:2 (r=0.17), cis 18:2 (r=0.41), total saturated fatty acids (r=0.30), was associated with lethal prostate cancer (Supplemental Figure 2). However, the association between intake of ALA and lethal prostate cancer persisted when adjusting for each fatty acid (Supplemental Figure 2). Finally, when we restricted to those participants who did not consume walnuts in the PSA era analysis, the HR for intake ALA comparing top to bottom quintile was 0.82 (95% CI = 0.54–1.23; p trend = 0.62; n=307 events) in the fully adjusted model.

Discussion

In the present study with 24 years follow-up, the association between higher intake of ALA and lethal prostate cancer was mainly evident in the pre-PSA era but not in the PSA era. This association in the pre-PSA era could not be attributed to obvious sources of confounding.

Lethal prostate cancer has been increasingly recognized as a more specific outcome to ensure comparability of results in the pre-PSA and PSA era.^15,25–27 For example, in the HPFS we had found that dietary lycopene intake was similarly inversely associated with lethal prostate cancer in the pre-PSA and PSA eras, but results for total prostate cancer were heterogeneous by time period.²⁸ Despite of focusing on lethal prostate cancers, we found that ALA intake was only related to the risk of lethal prostate cancer in the pre-PSA era. This difference could reflect methodological reasons or biological reasons. It is possible that the positive association in the pre-PSA era was largely due to confounding, because ALA intake in our cohorts, especially during early follow-up, was mainly derived from foods that often contained partially hydrogenated vegetable oils (e.g. mayonnaise, salad dressing, margarine), processed baked foods and red meat. However, several analyses could not identify a likely confounding factor. First, although the findings from several studies suggested that high meat rather than high ALA intake is responsible for the positive association with prostate cancer,^12,13 in the present study, both intakes of animal-sourced and plant-sourced ALA were associated with lethal prostate cancer with an even stronger association for the plant-sourced ALA. Second, we found specificity for the positive association for intake of ALA among many other correlated fatty acids. Lastly, we excluded the possibility that intake of ALA is simply a marker for dietary patterns, because neither prudent nor western dietary pattern was associated with lethal prostate cancer in the pre-PSA era.

Alternatively, the reason for no association in the PSA era could be due to altered prostate cancer epidemiology influenced by PSA screening. Prostate cancers with lethal potential are diagnosed much earlier in their natural history and receive curative treatment. The European trial²⁹ and Göteborg trial³⁰ and one observational study that took advantage of a natural experiment in Sweden³¹ all showed a significant reduction in prostate cancer mortality by PSA screening. The PLCO trial had a null finding,³² but this trial compared less intensive with more intensive screening; moreover, contamination by intensive PSA screening in the control group may account for the null effect.^33–35 Thus, the pre-PSA and PSA eras include a diverse mixture of lethal prostate cancers; in the pre-PSA era there are lethal cancers that would have been potentially curable had they been diagnosed early enough, while in the PSA era, these curable cancers would be removed from the pool of lethal cancer. If ALA only increases risk of a subset of lethal prostate cancers that is curable by treatment, the onset of widespread PSA screening could have largely removed the subset of lethal cancers through curative treatment, leaving only those incurable lethal cancers unrelated to ALA.

Another possibility is that the nature of the exposure “ALA” has changed over time. Trans ALA isomers are formed during partial hydrogenation, deep frying and industrial deodorization.^36,37 European scientists found the presence of trans ALA in many foods (e.g. vegetable oils,^36,38 low-calorie spreads³⁹ and infant formulas^40,41) and in human body composition.^42–44 Up to 40% of ALA can be present as trans isomers.^37,38. Pre-PSA era coincides with the same time period when the trans fat level were higher and also presumably a higher level of trans ALA compared to the PSA era. A downward trend in intake of total trans fat⁴⁵ and in plasma levels⁴⁶ over time was found which is largely due to changes made by food manufacturers in reducing partially hydrogenated oils. Trans ALA can be incorporated into plasma lipids and converted to trans long-chain polyunsaturated fatty acids in humans.^44,47 Similarly, long-term feeding rats of a diet high in trans ALA resulted in a significant increase in trans Docosahexaenoic acid (DHA) and a decrease in cis DHA.^48,49 Therefore, it is possible that high ALA intake in the pre-PSA era is a marker for trans ALA that promotes prostatic carcinogenesis by interfering with normal DHA function. The possibility related to trans ALA, coupled with bias related to PSA screening, may explain why the positive association was mainly evident in the pre-PSA era. However, this hypothesis related to trans ALA is novel and warrants further study.

Findings from previous prospective cohort studies that examined ALA intake and prostate cancer are mixed. However, stratifying by intensity of PSA screening and characterizing cancers by aggressiveness may offer more clarity. It has been argued that, in the PSA era, even prostate cancer outcomes such as high-grade or advanced stages are not good predictors for the lethal propensity.^18,25,26,50 Four prospective studies^51–54 that examined non-lethal prostate cancer in the PSA era found no associations, which is not surprising, and consistent with our previous findings in the HPFS.^16–18 However, the large NIH-AARP study⁵⁵ reported a modest (17% higher risk in the highest quintile) but significant positive association with advanced prostate cancer and a nonsignificant positive association with fatal prostate cancer despite widespread PSA screening. It is possible that the use of a single dietary questionnaire collected in the mid 1990’s may partially reflect the effects of pre-PSA diets. One study in Finland⁵⁶ and anther in the Netherlands⁵⁷ had no widespread PSA screening. Neither study found a significant association with incident prostate cancer. Possible explanations could be due to a short 6 years of follow-up and high proportion of non-lethal cancers among all incident prostate cancers, or different sources of ALA in those populations.

In addition to the potential for residual confounding, several other limitations in our study are worth noting. Random measurement error in the ALA exposure is inevitable and this would likely lead to an underestimation of the true association. However, we tried to minimize this bias by using cumulative average of multiple assessments of diets over long-period of time. Finally, our cohort consists of primarily white health professionals and results may not be generalizable to other populations. However, such homogeneity of study population minimized confounding by socioeconomical status and differential access to healthcare, and facilitated the high follow-up rate.

In conclusion, higher intake of ALA was associated with an increased risk of lethal prostate cancer in the pre-PSA era; however, ALA as currently consumed does not appear to be a risk factor for lethal prostate cancer. Our findings are important because ALA is considered an essential fatty acid and has important health benefits. Further studies are warranted to determine the causes for the differential associations by PSA era.

Novelty and Impact.

ALA is a popular and health-beneficial omega3 fatty acid. However, some prior studies reported a positive link between ALA intake and prostate cancer. We studied 48,000 men over two decades and found that higher intake of ALA was only associated with an increased risk of lethal prostate cancer in the pre-PSA era, but not in the PSA era. This means that current consumption of ALA is not related to developing lethal prostate cancer.

Abbreviations

ALA

α-linolenic acid

LA

linoleic acid

DHA

docosahexaenoic acid

HR

hazard ratios

CI's

confidence intervals

n-3

omega-3

HPFS

Health Professionals Follow-up Study

FFQ's

food frequency questionnaires