Strengths and limitations

We lack data on dietary habits and physical activity in the present study. Health conscious behaviour not only affects fish consumption, but also physical activity and intake of other nutrients considered healthy. This might lead to an overestimation of the benefits of marine n-3 PUFAs. On the other hand, in a Norwegian study of dietary habits, fish consumption was only associated with geographical region, not other life-style factors (150).

We performed a single measurement of plasma marine n-3 PUFA levels at ten weeks post-transplant in the present study. A single measurement does not adjust for temporal changes in fatty acid levels. A second measurement would have increased the precision in determining plasma marine n-3 PUFA levels that truly reflected the patients’ habitual levels. Thus, a single measurement might lead to random error and most likely underestimate the benefits of marine n-3 PUFAs. However, individual plasma phospholipid fatty acids were determined by gas chromatography, which in contrast to food questionnaires correlate very well with actual marine n-3 PUFA intake (203). Moreover, a recent report from a Norwegian population showed a strong coherence in plasma marine n-3 PUFA levels measured three years apart and a good correlation with marine food intake (204). This indicates that a single measurement might be adequate in an epidemiological study.

We had access to information about the number of anti-hypertensive drugs, but not the type or dose. We had no data on the use of statins, anti-thrombotic and glucose-lowering drugs.

Marine n-3 PUFAs may lower the risk of diabetes, hypertension, obesity and atherosclerosis, which in turn will lead to a reduced probability of receiving treatment for these conditions. As optimal treatment reduce their risk of mortality, beneficial effect of marine n-3 PUFAs on mortality might be diluted. On the other hand, the additional effect of marine n-3 PUFAs to optimal care is more interesting from a clinical point of view.

Competing risk from other causes of death might influence associations between plasma marine 3 PUFA levels and cause-specific mortality. Associations between plasma marine n-3 PUFA levels and mortality could possibly also influence associations with death censored graft loss.

The observational time started at ten weeks post-transplant. We can therefore not make assumptions about associations between plasma marine n-3 PUFA levels and mortality or graft loss endpoints within the first ten weeks after transplantation. However, we assume that

the risk of reaching these endpoints in the early phase after transplantation was less likely modified by nutritional factors. Since patients necessarily had to be alive with a functional graft at ten weeks post-transplant to be included in the study, immortal time bias will influence results, overestimating the benefits of marine n-3 PUFAs.

Seventy-five patients under the age of 16 years were not included in the study. About half of the 769 adult RTRs who were not included in the study were eligible for inclusion. They were not included due to reduced laboratory capacity. Revision of medical records in 100 patients transferred early to their local hospital revealed a myriad of reasons for transferal, not necessarily associated with increased comorbidity or mortality risk. Consequently, there were only minor differences in baseline characteristics between patients who were included and adult patients not included in the present study. Nonetheless, by design, there must be some degree of selection bias in the present study and the results are therefore not necessarily representative for all Norwegian RTRs. In recent years, tacrolimus has been the calcineurin inhibitor of choice at our center, whereas most patients in the present study received treatment with cyclosporine A. Since associations between plasma marine n-3 PUFA levels and various outcomes were similar in the early and recent transplant era (1999 – 2006 versus 2007 – 2011), this change in immunosuppressive therapy probably had no impact on associations between plasma marine n-3 PUFA levels and various outcomes.

Insights provided in paper II (cross-sectional study design) are more limited than in paper I and paper III (cohort study design). The time-point at which measurements were performed in paper II (ten weeks after renal transplantation) might not be ideal for assessment of

associations between plasma marine n-3 PUFA levels and cardiovascular risk markers.

During the first ten weeks after transplantation, dietary habits could possibly have been influenced by post-operative complications as well as intended dietary changes motivated and made possible by the renal transplantation. Uremic autonomic neuropathy would probably still affect resting heart rate at this early phase after transplantation. Plasma HDL cholesterol levels are low during uremic dyslipidemia and new lipoproteins must be synthesized for HDL cholesterol levels to normalize.

The major limitation in the present study is the observational study design, which does not allow for causal inference. In paper III, we speculate that inflammatory and / or anti-fibrotic mechanisms could possibly explain better renal graft survival with higher plasma marine n-3 PUFA levels, but present no data to substantiate this assumption. In paper II, we

infer that beneficial associations between plasma marine n-3 PUFA levels and cardiovascular risk markers, measured at a single time-point early after renal transplantation, is sufficient evidence to suggest that marine n-3 PUFAs might improve patient survival, based on previous reports from non-transplant populations. On the other hand, for the purpose of generating hypothesis about mechanisms by which marine n-3 PUFAs might improve patient and renal graft survival, we argue that the study design is adequate.

Since we advised RTRs to eat fish at our center, we cannot rule out an association between plasma marine n-3 PUFA levels and adherence to medical advice in general, which might influence results.

The majority of patients in this Norwegian study were Caucasians and the findings may not apply to other ethnic groups. Theoretically, the results may also not apply to regions with low consumption of fish and seafood.

This study also have several strengths, including fatty acid analysis, a 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 graft loss risk factors included as confounders in the multivariable Cox regression models.