The Association between Marine n-3 Polyunsaturated Fatty Acid Levels and Survival after Renal Transplantation

Abstract

Background and objectives

Several studies have reported beneficial cardiovascular effects of marine n-3 polyunsaturated fatty acids. To date, no large studies have investigated the potential benefits of marine n-3 polyunsaturated fatty acids in recipients of renal transplants.

Design, setting, participants, & measurements

In this observational cohort study of 1990 Norwegian recipients of renal transplants transplanted between 1999 and 2011, associations between marine n-3 polyunsaturated fatty acid levels and mortality were investigated by stratified analysis and multivariable Cox proportional hazard regression analysis adjusting for traditional and transplant-specific mortality risk factors. Marine n-3 polyunsaturated fatty acid levels in plasma phospholipids were measured by gas chromatography in a stable phase 10 weeks after transplantation.

Results

There were 406 deaths (20.4%) during a median follow-up period of 6.8 years. Mortality rates were lower in patients with high marine n-3 polyunsaturated fatty acid levels (≥7.95 weight percentage) compared with low levels (<7.95 weight percentage) for all age categories (pooled mortality rate ratio estimate, 0.69; 95% confidence interval, 0.57 to 0.85). When divided into quartiles according to marine n-3 polyunsaturated fatty acid levels, patients in the upper quartile compared with the lower quartile had a 56% lower risk of death (adjusted hazard ratio, 0.44; 95% confidence interval, 0.26 to 0.75) using multivariable Cox proportional hazard regression analysis. There was a lower hazard ratio for death from cardiovascular disease with high levels of marine n-3 polyunsaturated fatty acid and a lower hazard ratio for death from infectious disease with high levels of the marine n-3 polyunsaturated fatty acid eicosapentaenoic acid, whereas there was no association between total or individual marine n-3 polyunsaturated fatty acid levels and cancer mortality.

Conclusions

Higher plasma phospholipid marine n-3 polyunsaturated fatty acid levels were independently associated with better patient survival.

Introduction

Dietary factors, like the essential marine n-3 polyunsaturated fatty acids (PUFAs), may exert potent biologic effects (1). Cross-cultural epidemiologic studies in the 1970s reported a strong negative association between fish consumption and the incidence of cardiovascular disease (CVD) (2). Several beneficial metabolic effects of marine n-3 PUFAs have been reported, including lipid modulation (3), plaque stabilization (4), reduced BP (5), less artery calcification (6), and improved endothelial (7) and myocardial function (8). Furthermore, anti-inflammatory (9), antiarrhythmic (10), antiatherosclerotic (11), and antithrombotic effects (12) have been reported, which may influence the incidence of CVD and mortality rates. However, recent prospective cohort studies and randomized, controlled trials (RCTs) have shown mixed results, with only moderate beneficial or even neutral effects on major cardiovascular event and mortality rates (13–15).

In patients with ESRD, favorable effects of marine n-3 PUFAs on cardiovascular morbidity and mortality have been shown (16,17). Although renal transplantation reduces the risk of CVD in patients with ESRD, CVD remains the leading cause of death in recipients of renal transplants (RTRs) (18).

No cohort study and only small RCTs have assessed the effects of marine n-3 PUFAs in RTRs, with insufficient data to evaluate the effects on mortality (19–21). Because intake of marine n-3 PUFAs in Scandinavia is generally high, with considerable variation between individuals (22,23), our cohort is ideal for epidemiologic studies of associations between marine n-3 PUFA levels and mortality in RTRs. The aim of this study was to assess whether plasma phospholipid levels of marine n-3 PUFAs, which derive from consumption of fish, seafood, or marine n-3 PUFA supplements, were associated with overall and cause-specific mortality in RTRs.

Materials and Methods

Study Design and Population

In Norway, all organ transplantations are performed at Oslo University Hospital, Rikshospitalet. From a population of approximately 5 million inhabitants, 2978 renal transplantations were performed in 2837 patients with ESRD between September 30, 1999 and October 13, 2011. Patients under the age of 16 years old (n=78) and patients who were transferred to their local hospitals (n=335), suffered graft loss (n=58), or died (n=21) within the first 10 weeks post-transplant were not eligible for participation in the study (Figure 1). Most patients attended our outpatient clinic for the first 10 weeks after renal transplantation. Informed consent was obtained from 2001 of 2345 eligible patients, in whom blood samples were drawn, measurements were taken, and clinical information was recorded at a final appointment 10 weeks post-transplant. From eligible patients, 344 patients did not attend the final appointment because of reduced capacity at our laboratory during 2007 and 2008. They were not included in the study because of the lack of laboratory and clinical data. In 11 patients, the amount of plasma was too small for individual fatty acids to be adequately analyzed. The remaining 1990 patients were included in this study.

Figure 1.

Flowchart of the inclusion of study participants.

Data Collection and Registry

In study participants, blood samples were drawn in the fasting state at the clinical appointment 10 weeks post-transplant. Some samples were analyzed at a central biochemical department, and laboratory test results were entered uniformly into a database. Other blood samples were immediately frozen, stored at −80°C, and later on sent to The Lipid Research Center, Aalborg University Hospital for analysis of plasma phospholipid fatty acid composition by gas chromatography.

End point data were collected from The Norwegian Renal Registry. The registry is on the basis of annual reports from all nephrologists working in Norway and includes all patients on RRT. Mortality is continuously registered. Surviving patients were censored on February 1, 2014. Mortality end points were defined according to the European Renal Association–European Dialysis and Transplant Association causes of death and included the broad categories cancer, infectious disease, and cardiovascular mortality (from the latter category, death from stroke, myocardial infarction [MI], and sudden cardiac death [SCD]).

In a random selection of 200 patients, all variables included in the Cox models were checked for consistency between data continuously registered at The Norwegian Renal Registry and data retrieved from medical records. There was excellent correspondence of data, with the exception of smoking status. After reviewing the medical records, correcting smoking status data, and obtaining missing data of all variables in 1990 study participants, there was 0.6% missing data on smoking status and <0.5% for other variables, and for most variables, there were no missing data. In total, there were only 14 study participants (0.7%) with missing data. They were not included in the multivariable Cox regression analysis. Baseline characteristics of patients with missing data did not differ significantly from patients without missing data. Description of the standard treatment protocol, clinical characteristics of adult RTRs not included in the study, and fatty acid analysis are described in Supplemental Appendix. In short, individual fatty acids were identified by gas chromatography and quantitated as the weight percentage (wt%) of total fatty acids. In this study, marine n-3 PUFAs were defined as the sum of plasma phospholipid levels of three individual marine n-3 PUFAs: eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA), and docosahexaenoic acid (DHA).

Statistical Analyses

Differences in baseline characteristics by quartiles of marine n-3 PUFAs according to plasma phospholipid levels were evaluated using the Mantel–Haenszel test of linear trend for categorical data, the nonparametric Kruskal–Wallis test for time in dialysis, and linear regression for other continuous data. The associations between total and individual marine n-3 PUFAs and mortality end points were studied by stratified and Cox proportional hazard regression analyses. A stratified analysis adjusting for recipient age was used for estimation of mortality rates. Identification of confounders and fitting of Cox models and graphing are described in Supplemental Appendix. In short, recipient age was identified as a strong confounder and an effect-measure modifier for marine n-3 PUFAs; hence, a product term was included in the survival analysis. Two Cox models were fitted. In addition to recipient age and a product term of recipient age and marine n-3 PUFA level, model 1 included the following traditional and transplant-specific mortality risk factors: sex, eGFR according to the Modification of Diet in Renal Disease formula, time in dialysis before transplantation, preemptive transplantation (no dialysis treatment before transplantation), choice of calcineurin inhibitor (treatments with cyclosporin A or tacrolimus were included as two separate categorical variables), smoking status (current smoker, former smoker, or lifelong nonsmoker as a categorical variable), body mass index, albumin and total plasma cholesterol concentrations, number of antihypertensive drugs, diabetes mellitus before transplantation, pretransplant coronary artery, and cerebrovascular and peripheral vascular disease. Model 2 added n-6 PUFA levels as a covariate. A review of the rationale for the two models is found in Supplemental Appendix. Model assumptions were checked by inspection of the log-log survival time plots and formal hypothesis tests (Schoenfeld residuals). A two-sided P value of <0.05 was considered statistically significant. PASW Statistics, version 17.0 (IBM, New York, NY) and STATA, version 13.0 (Stata Corp, College Station, TX) were used for the statistical analysis.

The study was approved by the Regional Committees for Medical and Health Research Ethics in Norway and performed in accordance with the Declaration of Helsinki (https://clinicaltrials.gov/ no. NCT02017990). The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the Declaration of Istanbul on Organ Trafficking and Transplant Tourism.

Results

Baseline characteristics of the study participants grouped according to marine n-3 PUFA levels are shown in Table 1. The median level of marine n-3 PUFAs in plasma phospholipids was 7.95 wt%. Patients with high marine n-3 PUFA levels (≥7.95 wt%) were older than patients with lower levels (<7.95 wt%). From 2007, most patients under the age of 50 years old were treated with tacrolimus, whereas older patients received cyclosporin A. When adjusted for age and transplant era, neither choice of calcineurin inhibitor nor eGFR differed between high and low levels of marine n-3 PUFAs. We found a negative association between marine n-3 PUFA levels and both current smoking and n-6 PUFA levels, even after adjustments for age and transplant era, and a trend toward less living donor transplantations and lower prevalence of diabetes mellitus with high marine n-3 PUFA levels. Adult RTRs not included the study were older (mean age of 55.1 years old) than the study participants (mean age of 51.6 years old). Because there were more patients not included in the study after 2007, the groups differed with regards to choice of calcineurin inhibitors (Supplemental Table 1). When adjusting for age, other baseline characteristics for the two groups were similar.

Table 1.

Baseline characteristics of the study participants according to levels of marine n-3 polyunsaturated fatty acids

Variables	All Patients	Quartile 1	Quartile 2	Quartile 3	Quartile 4	P Value (Trend)
Marine n-3 PUFA level, wt%	1.35–23.87	≤6.20	6.21–7.94	7.95–10.02	≥10.03
No. of patients	1990	499	499	495	497
Recipient age, yr	51.6 (14.6)	43.3 (14.0)	51.6 (14.1)	54.2 (13.3)	57.3 (13.1)	<0.001
Sex (men), %	66.9	69.0	64.8	65.6	68.3	0.90
Tacrolimus, %	23.1	35.5	22.2	19.4	15.3	<0.001
Cyclosporin A, %	74.2	11.9	17.3	15.9	16.0	0.13
Cerebrovascular disease, %	5.2	4.2	5.8	6.0	4.8	0.66
Peripheral vascular disease, %	8.0	6.2	9.3	8.2	8.2	0.37
Diabetes mellitus, %	18.2	23.5	21.1	15.3	13.0	<0.001
Current smoker, %	16.0	23.5	18.0	12.9	9.4	<0.001
Former smoker, %	36.2	30.2	37.7	36.5	40.4	0.002
No. of antihypertensive drugs, %						0.22
None or one	54.4	56.3	55.1	53.5	52.7
Two or three	41.2	39.0	41.6	41.9	42.3
Four or more	4.4	4.6	3.2	4.6	5.0
Body mass index, kg/m2	24.8 (3.9)	24.6 (4.2)	24.7 (3.9)	25.0 (3.7)	24.9 (3.5)	0.09
Total cholesterol, mg/dl	244.4 (59.0)	238.2 (60.6)	243.2 (56.4)	249.8 (59.1)	246.7 (59.5)	0.01
Albumin, g/dl	4.00 (0.37)	4.06 (0.40)	3.98 (0.36)	3.98 (0.34)	3.98 (0.35)	0.001
Time in dialysis, mo	9 (0–19)	8 (0–18)	9 (1–20)	8 (0–18)	9 (1–19)	0.21
Preemptive transplantation, %	25.2	27.0	23.9	25.6	24.2	0.41
First renal transplantation, %	90.3	88.2	88.8	92.0	92.2	0.01
eGFR, ml/min per 1.73 m2	56.9 (18.8)	64.5 (20.4)	56.5 (18.3)	54.2 (17.3)	52.6 (16.9)	<0.001
Delayed graft function, %	6.3	6.8	6.6	8.6	3.6	0.26
Early rejection episode, %	27.7	27.8	29.7	27.3	26.3	0.44
Donor age, yr	47.2 (16.1)	44.2 (15.7)	46.1 (16.7)	49.3 (15.3)	49.5 (16.1)	<0.001
Living donor transplantation, %	37.8	42.1	33.2	41.4	34.7	0.08
No. of HLA mismatches, %						0.23
None or one	15.7	20.5	11.5	14.9	16.0
Two	23.4	20.3	24.1	25.6	23.2
Three	32.9	31.2	34.6	32.2	33.5
Four or more	28.0	28.0	29.8	27.4	27.1

Baseline characteristics of the study population according to marine n-3 polyunsaturated fatty acid (PUFA) levels defined as the sum of plasma phospholipid eicosapentaenoic acid, docosapentaenoic acid, and docosahexaenoic acid levels in weight percentage (wt%) of total plasma phospholipid fatty acids. Results are presented as proportions for categorical data, medians and interquartile ranges for time in dialysis, and means and SDs for other continuous data. Differences in baseline characteristics were evaluated using the Mantel–Haenszel test of linear trend for categorical data, the Kruskal–Wallis test for time in dialysis, and linear regression for other continuous data. Pretransplant diabetes mellitus, coronary artery, cerebrovascular and peripheral vascular disease were recorded before first renal transplantation. Recipient and donor age, deceased or living donor, time in dialysis, number of HLA mismatches, and smoking status were recorded at the time of transplantation. Choice of calcineurin inhibitor, number of antihypertensive drugs, delayed graft function, early rejection episodes, body mass index, eGFR using the Modification of Diet in Renal Disease formula, total cholesterol, and albumin values were recorded at a clinical appointment 10 weeks post-transplant.

During the study period, the total number of deaths was 406 (20.4%). In 164 patients, death was caused by CVD (40.4% of deaths). There were 95 deaths by cancer (23.4%) and 101 deaths by infectious disease (24.9%). The median follow-up time for study participants was 6.8 years.

The crude analysis showed a small positive association between levels of marine n-3 PUFAs and mortality. However, in an age-stratified analysis reducing the confounding effect of recipient age, the mortality rate was lower in patients with marine n-3 PUFA levels at or above the median value of 7.95 wt% compared with the patients with lower marine n-3 PUFA levels for all age categories (Figure 2, Table 2). The pooled estimate for the mortality rate ratio was 0.69 (95% confidence interval [95% CI], 0.57 to 0.85). Consistent with this finding, we found a negative association between marine n-3 PUFA levels and mortality in multivariable-adjusted Cox proportional hazard regression analysis using both models 1 and 2. The association between marine n-3 PUFA levels and mortality was similar in periods with high and low inclusion rates (Supplemental Table 2), and adjustment for transplant eras had no effect on results.

Figure 2.

Kaplan–Meier survival curve in recipients of renal transplants. Shown are the proportions of surviving patients grouped according to age (age <60 and ≥60 years old) and marine n-3 polyunsaturated fatty acid (PUFA) levels (the sum of eicosapentaenoic, docosapentaenoic, and docosahexaenoic acid levels in weight percentage [wt%] of total plasma phospholipid fatty acids <7.95 wt% [low] and ≥7.95 wt% [high]).

Table 2.

Mortality rates by marine n-3 polyunsaturated fatty acid levels and age category

	Age Category, yr
	16–44		45–59		60–80
Marine n-3 PUFA level	High	Low	High	Low	High	Low
Mortality rate, patients per 1000 person-yr	4.1	10.1	16.6	26.7	59.2	69.0
Mortality rate difference	−6.0		−10.1		−9.8
Mortality rate ratio (95% confidence interval)	0.41 (0.17 to 0.88)		0.62 (0.40 to 0.85)		0.86 (0.63 to 1.05)
Pooled estimate (95% confidence interval)	0.69 (0.57 to 0.85)		0.69 (0.57 to 0.85)		0.69 (0.57 to 0.85)

The mortality rate according marine n-3 polyunsaturated fatty acid (PUFA) levels (the sum of eicosapentaenoic, docosapentaenoic, and docosahexaenoic acid levels in weight percentage [wt%] of total plasma phospholipid fatty acids; ≥7.95 wt% [high] and <7.95 wt% [low]) in different age groups. Mortality rate ratios for each age category were obtained by dividing the mortality rate of patients with high marine n-3 PUFA levels by the mortality rate of patients with low levels.

We found lower levels of n-6 PUFAs with higher levels of marine n-3 PUFAs (Supplemental Table 3). There was no interaction between marine n-3 PUFA and n-6 PUFA levels. The mean ratio of n-6 PUFA to n-3 PUFA was 4.75. When using marine n-6 PUFA to n-3 PUFA ratio (Supplemental Table 4) or arachidonic acid (AA) to EPA ratio as the primary exposure, we found similar results to using marine n-3 PUFA levels alone. Although there were differences in hazard ratio (HR) estimates between models 1 and 2, similar trends were found for all causes of death with both models (Table 3).

Table 3.

Estimated mortality risk according to quartiles of marine n-3 polyunsaturated fatty acid levels using multivariable Cox proportional hazard regression analysis

Cox Models	Quartile 1	Quartile 2			Quartile 3			Quartile 4
Model 1
wt%	≤6.20	6.21–7.94			7.95–10.02			≥10.03
No. of patients	495	498			492			491
Mortality	HR	HR	95% CI	P Value	HR	95% CI	P Value	HR	95% CI	P Value
All cause	1.0	0.82	0.60 to 1.12	0.22	0.59	0.40 to 0.86	0.01	0.44	0.26 to 0.75	0.003
Cardiovascular	1.0	0.56	0.34 to 0.91	0.02	0.36	0.20 to 0.65	0.001	0.18	0.08 to 0.42	<0.001
MI	1.0	0.55	0.14 to 2.08	0.38	0.63	0.14 to 2.83	0.55	0.51	0.07 to 3.76	0.51
SCD	1.0	0.58	0.29 to 1.16	0.12	0.31	0.13 to 0.73	0.01	0.08	0.02 to 0.35	0.001
Stroke	1.0	0.50	0.16 to 1.54	0.22	0.30	0.08 to 1.20	0.09	0.14	0.02 to 1.09	0.06
Infectious disease	1.0	0.78	0.41 to 1.46	0.43	0.47	0.21 to 1.04	0.06	0.51	0.18 to 1.50	0.22
Cancer	1.0	1.47	0.71 to 3.06	0.31	1.30	0.56 to 3.04	0.54	1.37	0.45 to 4.18	0.58
Model 2
wt%	≤6.20	6.21–7.94			7.95–10.02			≥10.03
No. of patients	495	498			492			491
Mortality	HR	HR	95% CI	P Value	HR	95% CI	P Value	HR	95% CI	P Value
All cause	1.0	0.77	0.55 to 1.06	0.11	0.48	0.32 to 0.73	0.001	0.33	0.19 to 0.58	<0.001
Cardiovascular	1.0	0.51	0.31 to 0.85	0.01	0.29	0.15 to 0.55	<0.001	0.12	0.05 to 0.30	<0.001
MI	1.0	0.70	0.17 to 2.85	0.61	1.00	0.17 to 6.04	>0.99	0.62	0.06 to 6.13	0.68
SCD	1.0	0.55	0.27 to 1.12	0.10	0.29	0.11 to 0.73	0.01	0.07	0.02 to 0.32	0.001
Stroke	1.0	0.48	0.15 to 1.54	0.22	0.26	0.06 to 1.12	0.07	0.08	0.01 to 0.70	0.02
Infectious disease	1.0	0.70	0.37 to 1.34	0.28	0.36	0.15 to 0.84	0.02	0.39	0.13 to 1.19	0.10
Cancer	1.0	1.42	0.67 to 3.05	0.36	1.15	0.45 to 2.93	0.77	1.08	0.32 to 3.59	0.90

The estimated risk of total and cause-specific mortality using multivariable-adjusted Cox proportional hazard regression models 1 and 2. Results are presented as multivariable-adjusted hazard ratios for developing mortality end points relative to the lower quartile of marine n-3 polyunsaturated fatty acid (the sum of eicosapentaenoic acid, docosapentaenoic acid, and docosahexaenoic acid in weight percentage [wt%] of total plasma phospholipid fatty acids) levels. In addition to marine n-3 polyunsaturated fatty acids levels, model 1 included the following variables: recipient age, a product term of recipient age and marine n-3 polyunsaturated fatty acid levels, sex, eGFR using the Modification of Diet in Renal Disease formula, time in dialysis before transplantation, preemptive transplantation, body mass index, number of antihypertensive drugs, diabetes mellitus, coronary artery, cerebrovascular and peripheral vascular disease, and albumin and total plasma cholesterol concentrations. Model 2 also includes n-6 polyunsaturated fatty acids levels as a covariate. HR, hazard ratio; 95% CI, 95% confidence interval; MI, myocardial infarction; SCD, sudden cardiac death.

When grouped according to marine n-3 PUFA levels comparing the upper with the lower quartile, there was a 56% lower risk of death (multivariable-adjusted HR, 0.44; 95% CI, 0.26 to 0.75) using model 1 (Table 3) and 67% lower risk of death (multivariable-adjusted HR, 0.33; 95% CI, 0.19 to 0.58) using model 2 (Figure 3, Table 3). Similar results were found for EPA and DHA alone (Supplemental Table 5), whereas no association was found between DPA levels and mortality. The risk of death from CVD was markedly lower in patients with high marine n-3 PUFA levels compared with low levels. In particular, death from stroke and SCD showed a strong negative association with marine n-3 PUFA levels (Table 3). Patients with high levels of EPA were also less likely to die from infectious disease and equally associated with nonsepticemia infectious disease mortality as death from septicemia (Supplemental Table 5). No association was found between marine n-3 PUFA levels and death by cancer.

Figure 3.

Estimated survival probability curve in recipients of renal transplants in multivariable–adjusted Cox proportional hazard regression model 2. Shown are the survival probabilities of patients belonging to quartiles 1–4 according to marine n-3 polyunsaturated fatty acid (PUFA) levels (the sum of eicosapentaenoic, docosapentaenoic, and docosahexaenoic acid levels in weight percentage [wt%] of total plasma phospholipid fatty acids) after adjustment for the following variables: recipient age, sex, n-6 PUFA levels, eGFR using the Modification of Diet in Renal Disease formula, time in dialysis before transplantation, preemptive transplantation, body mass index, number of antihypertensive drugs, diabetes mellitus, coronary artery, cerebrovascular and peripheral vascular diseases, and albumin and total plasma cholesterol concentrations. Quartile 1, marine n-3 PUFA ≤6.20 wt%; quartile 2, marine n-3 PUFA =6.21–7.94 wt%; quartile 3, n-3 PUFA =7.95–10.02 wt%; quartile 4, marine n-3 PUFA ≥10.03 wt%.

In 908 of 984 patients (92.2%) with functional renal grafts 5 years post-transplant, we looked at the decline in renal function over time according to marine n-3 PUFA levels. Creatinine values increased more in patients with low marine n-3 PUFA levels than in patients with higher marine n-3 PUFA levels (Table 4).

Table 4.

Change in renal function within the first 5 years after transplantation according to levels of marine n-3 polyunsaturated fatty acids

Serum Creatinine	All patients	Quartile 1	Quartile 2	Quartile 3	Quartile 4	P Value (Trend)
Marine n-3 PUFA level, wt%	1.35–23.87	≤6.20	6.21–7.94	7.95–10.02	≥10.03
No. of patients	908	196	216	234	262
Serum creatinine at 10 wk post-transplant, mg/dl	1.33 (0.42)	1.27 (0.37)	1.34 (0.39)	1.36 (0.47)	1.34 (0.41)	0.10
Serum creatinine at 5 yr post-transplant, mg/dl	1.43 (0.56)	1.48 (0.63)	1.42 (0.57)	1.41 (0.55)	1.42 (0.53)	0.26
Change between 10 wk and 5 yr, mg/dl	0.10 (0.50)	0.21 (0.55)	0.08 (0.49)	0.06 (0.48)	0.08 (0.46)	0.01

Change in renal function in the first 5 years post-transplant in study participants with functional renal grafts transplanted between 1999 and 2008 according to marine n-3 polyunsaturated fatty acid (PUFA) levels defined as the sum of plasma phospholipid eicosapentaenoic acid, docosapentaenoic acid, and docosahexaenoic acid levels in weight percentage (wt%) of total plasma phospholipid fatty acids. Results are presented as means and SDs for serum creatinine values. Statistic trend was evaluated by linear regression analysis.

Discussion

The major finding of our study was an independent and negative association between plasma phospholipid marine n-3 PUFA levels and overall and cardiovascular mortality after renal transplantation. Deaths from infectious disease were strongly associated with low levels of EPA.

Three separate meta-analyses of RCTs focusing on the effect of marine n-3 PUFA supplementation after renal transplantation have reported a significant reduction in triglyceride levels, a minor reduction of diastolic BP, and slightly increased concentrations of HDL cholesterol, with possible beneficial effect on patient survival (19–21). However, effects on mortality rates could not be evaluated because of a low number of events (only seven deaths in 846 patients) and short follow-up (19,21). The much higher number of patients and longer follow-up time in our study provide a possibility to evaluate mortality.

CVD is prevalent in RTRs (18). Along with traditional CVD risk factors, post-transplant immunosuppressive treatment (24), pretransplant uremia (25), and renal graft function (18) constitute important risk factors for CVD in RTRs. Favorable effects of marine n-3 PUFA supplementation have been shown in diabetic nephropathy and IgA nephritis (26,27), and renal functions were better preserved with marine n-3 PUFA supplementation in patients with a history of MI (28). We found a steeper decline in renal function during the first 5 years after transplantation in patients with low levels of marine n-3 PUFAs, indicating potential renoprotective effects of marine n-3 PUFAs. It is possible that this could have influenced subsequent mortality.

Beneficial effects of marine n-3 PUFAs have also been reported in patients with ESRD. Whether these effects are applicable to RTRs is unclear. In patients who are uremic and treated with hemodialysis, high intake of marine n-3 PUFAs was associated with a lower risk of SCD (17), and an RCT found a lower cumulative incidence of MI after marine n-3 PUFA supplementation (16). In RTRs, there have been reports of improved renal function with marine n-3 PUFA supplementation (29). Most studies, however, reported no effect (19–21).

The assumption that increased intake of marine n-3 PUFAs significantly reduces the risk of cardiovascular morbidity and mortality has been challenged in recent meta-analyses of RCTs (13–15,30–32). The divergent results of early and recent studies may have several explanations, including more patients on optimal antihypertensive, antithrombotic, and statin treatment (33) and an increased background consumption of fish and seafood (34). In this study of a CVD high-risk population, even with optimal conventional treatment, we found an association between high marine n-3 PUFA levels and a lower risk of CVD-related death in RTRs.

A recent meta-analysis of prospective cohort studies found a negative association between cerebrovascular disease incidence and fish consumption (35). Although hampered by a low event rate, our findings indicate negative association between marine n-3 PUFA levels and death from stroke (Table 3).

Impaired glucose metabolism increases the risk of CVD, serious infections, and mortality after renal transplantation (18). Because reduction of total dietary fat in the Western diet during the last decades has led to increased consumption of carbohydrates, with possibly negative health effects (36), there is now more emphasis on the beneficial effects of PUFAs and the composition of various nutrients. High levels of marine n-3 PUFAs were associated with a lower prevalence of diabetes mellitus at the time of transplantation, despite a higher age group. A meta-analysis of RCTs reported a positive effect of marine n-3 PUFA supplementation on insulin sensitivity (37), possibly preventing or delaying onset of diabetes. A large RCT in patients with diabetes found no effect on cardiovascular mortality with low-dose marine n-3 PUFA supplementation (38).

In patients with septicemia, reduced mortality rates have been found with both a high-dose marine n-3 PUFA (>0.1 g/kg per day) supplementation (39) and an inflammation-modulating diet consisting of marine n-3 PUFAs, borage oil, and antioxidants (40). Our study indicates that moderate to high levels of EPA are associated with a lower risk of death from infectious disease in RTRs. EPA competes with the proinflammatory n-6 PUFA AA as substrate in the cyclooxygenase pathway and may, therefore, limit the inflammatory response (9).

A typical Western diet has a high content of n-6 PUFA relative to marine n-3 PUFA or the n-3 PUFA α-linolenic acid found in some green plants, rapeseed, and nuts (41). A study performed in the United States reported a mean n-6:n-3 PUFA ratio of 10:1 (42). In contrast, the mean n-6:n-3 PUFA ratio was 4.75 in this Norwegian population, and 42% of the study participants met with dietary recommendations of an n-6:n-3 PUFA ratio of 4:1 or less.

DPA, an elongation metabolite of EPA, may exert similar effects as EPA (43). We found no association between DPA levels and mortality. The difference between high and low levels of DPA was lower than for EPA and DHA and did not correlate well with either EPA or DHA level, indicating that DPA was a less-sensitive parameter of fish intake than EPA and DHA.

We found no association between marine n-3 PUFA levels and overall or type-specific cancer mortality. This finding is consistent with most studies reporting neutral effects from marine n-3 supplementation in patients with cancer (44). The potential influence of competing risk on cancer mortality is discussed in Supplemental Appendix.

The dietary consumption of fish and seafood varies between regions. In Norway, the average marine n-3 PUFA intake is more than five times higher than in the United States (34). The median value for patients belonging to the upper marine n-3 PUFA quintile in the Cardiovascular Health Study performed in the United States (45) was within the lower quartile in this study. A positive association between fish consumption or marine n-3 PUFA levels and survival has been reported in populations with high, moderate, and low consumption of fish and seafood, indicating beneficial effects, regardless of level (45–51).

Major limitations in our study include lack of dietary data to adjust for the full matrix of nutrients and the fact that we performed only a single measurement of plasma phospholipid marine n-3 PUFA levels 10 weeks post-transplant. This ignores temporal changes in fatty acid levels, and hence, we assume that a single measurement of plasma marine n-3 PUFAs was a unique correlate with mortality. Although an obvious limitation, a recent report from a Norwegian population showed a highly significant coherence in plasma phospholipid levels for all marine n-3 PUFAs measured 3 years apart, and plasma marine n-3 PUFA levels correlated well with marine food intake (52). Nonetheless, repeated measurements of plasma fatty acids would have been ideal to check for consistency in dietary habits and exclusion of measurement errors.

Several lifestyle factors may be associated with intake of fish and act as confounders not adjusted for in the analyses. We did not register information on education level, socioeconomic status, or physical activity. A recent Nordic study showed a positive association between the intake of fish and education level, physical activity, and various dietary components considered healthy (22). We have no data on the consumption of marine n-3 supplements versus fish consumption, and we performed no measurements of mercury or other contaminants found in fish. We have data on the number but not the type of antihypertensive drugs. We also lack data on antidiabetic treatment and statin use. Because the majority of patients in this Norwegian study were white, the results may not be applicable in other ethnic groups or regions with different dietary intake of marine n-3 PUFAs. We found an inverse association between marine n-3 PUFA levels and current smoking and diabetes mellitus before transplantation. Although adjustments were made in the Cox regression analysis, the effect of these confounders may not have been fully corrected. Competing risk may influence on cause-specific mortality HRs. Finally, it should be noted that the observational study design does not allow causal inference.

This study also has several strengths. Most importantly, individual plasma phospholipid fatty acids were measured by gas chromatography, which in contrast to dietary questionnaires, correlates very well with actual marine n-3 PUFA intake (53). Moreover, a very low percentage of missing data, quality-assured data, a relatively long follow-up period, a large number of events, a well defined and large study population with a high inclusion rate from a single center, and several traditional and transplant-specific mortality risk factors included as confounders in the Cox models constitute strengths of this study.

In summary, this observational cohort study found lower overall and cardiovascular mortality risk after renal transplantation with high plasma phospholipid marine n-3 PUFA levels, indicating that RTRs may benefit from marine n-3 PUFA supplementation. Future RCTs focusing on the effects of marine n-3 PUFAs on traditional and transplant-specific mortality risk factors with adequate sample sizes, follow-up periods, and dosages of marine n-3 PUFA supplements are warranted in renal transplantation. Hopefully, the ongoing Omega-3 Fatty Acids in Renal Transplantation Trial (ClinicalTrials.gov no. NCT01744067) can provide important information.

Disclosures

None.