George Henderson

Can insulin response patterns predict metabolic disease risk in individuals with normal glucose tolerance

To the Editor: We commend the authors of the recent publication in Diabetologia entitled ‘Glucose patterns during an oral glucose tolerance test and associations with future diabetes, cardiovascular disease and all-cause mortality rate’ [1] and would encourage them to repeat their investigation using insulin response patterns in people with normal glucose tolerance. Hyperinsulinaemia is becoming well-established in the aetiology of many metabolic diseases [2], yet many people will be unaware that they are at risk. Insulin levels are posited to be elevated for many years, possibly even decades, before changes in blood glucose are noted [3, 4]. Although fasting insulin is used inmany studies as a disease riskmarker, it has a high coefficient of variation leading to concerns about repeatability [5]. There has been little emphasis on the use of insulin response patterns to assess the risk of metabolic disease, yet the patterns observed when tracking blood insulin levels following an OGTT may provide an earlier opportunity to assess disease risk compared with changes in blood glucose levels. Hayashi and colleagues noted that cumulative incidence of type 2 diabetes in Japanese-American men over 10 years of follow-up was between 38% and 48% in those who reached a peak insulin level at 120 min following a 75 g OGTT, which was higher than those whose insulin levels peaked at 30 or 60 min (cumulative incidence, <16%) [6]. However, these response patterns were based on relative insulin serum levels in a cohort that included a high proportion of participants with pre-existing impaired fasting glucose and/or impaired glucose tolerance [5]. Between 1970 and 1990, Dr J. R. Kraft collated a large database holding data from multiple-sampled OGTTs with insulin assays. Further analysis of this database suggests that up to 75% of people with normal glucose tolerance may have hyperinsulinaemia [7], but there are no longitudinal outcomes to support risk calculations. There is a distinct paucity of data on insulin response patterns and subsequent risk of cardiovascular disease and we believe the authors would advance the field with further data analysis.

Poor quality data from diet studies may dilute better quality evidence

The hypothesis behind Farvid and colleagues’ study,1 that cancer risk is modified early in life by micronutrients and other factors, is plausible. However, the data collection method—nostalgic re-creation of adults’ childhood diet—invites respondents to invent data. I vaguely remember vegetables I ate in adolescence, in meals my mother cooked, but I can’t accurately recall fruit consumption. Yet the exclusion criteria …

Meat, eggs, full-fat dairy, and nutritional boogeymen: Does the way in which animals are raised affect health differently in humans?

ABSTRACT Background: Food recommendations to improve cancer prevention are generally based on epidemiologic data and remain inconsistent. These epidemiologic studies, while controversial, have generally produced results that caution against the consumption of high-fat foods, including eggs, red meat, and full-fat dairy, such as butter and cheese. Yet, limited data exist assessing the quality of individual sources of these foods and the effect each has after its consumption. This study set out to assess the impact sources of food within the same groups from animals raised differently on variables associated with health in human studies. Methods and Materials: A search was conducted through MEDLINE, Embase, and PubMed. In total, twenty-nine studies met inclusion criteria, measuring physiologic changes in humans after consuming animal products following animal diet manipulation. A meta-analysis was attempted to assess the differences between the cohorts in these studies, but was aborted due to poor study quality, vast differences in study design, and a limited number of studies. Results: Studies varied by animal, animal diet manipulation, food product, and overall design. Significant differences were present between groups eating the same food (cheese, beef, eggs, and butter) from animals raised differently, including levels of: conjugated linoleic acid, omega-3 fatty acids (alpha linoleic acid [ALA], docosahexaenoic acid [DHA], and eicosapentaenoic acid [EPA]), and inflammatory factors (triacyl glycerol [TAG], interleukin-6 [IL-6], interleukin-8 [IL-8], tumor necrosis factor [TNF], and C-reactive protein [CRP]). Lipid levels were minimally affected. Conclusions: This work highlights differences in human health markers after consumption of the same foods from animals raised differently. Overall, lipid levels remained relatively neutral, but significant changes in inflammatory and other serum markers and phospholipids were present. Future studies and dietary recommendations should consider how animals are raised, as this can produce different effects on health markers.

Coffee consumption and health: we need randomised controlled trials

The umbrella review of coffee consumption and health by Poole and colleagues includes a gentle plea to consider a randomised controlled trial of coffee for chronic liver disease as well as the outcomes of a focus group, which indicate interest in this particular usage.1 I hope that such a trial does result from this publication. I was able to find one …

Lingering Questions Concerning Specific Dietary Fats and Mortality

factors,1,2 but no consistent relation with the risk of CHD has been found.3 Thus, in our study4 we did not think about ferritin level as a potential confounder. An important interpretation of our results is that the risk in women seems to be unrelated to menopausal status or to any factors related to menopausal status. There appears to be no protection of being premenopausal vs postmenopausal above what can be explained by a difference in age. The declining relative risk for sex with increasing age seems to be a consequence of a more pronounced flattening of risk level changes in middle-aged men, approaching the risk level for women. This is in contrast to previous beliefs that the risk in women approximate the risk level for men at a certain age. As Mascitelli and Goldstein point out, the absolute risk of MI in women increases about a decade after normal menopausal age. However, an even steeper increase in risk is seen in men, leading to increasing sex differences in absolute risk with increasing age. It is thus difficult to explain the age incidence patterns in view of sex differences in ferritin levels. In a Danish population study5 with follow-up through 1991, standardized age incidence curves were rather parallel for men and women (log-linear scale), in apparent contrast to the ageand sexrelated changes in body iron stores. In addition to increasing oxidative stress, increased iron stores have been suggested to promote CHD by alteration of endothelial function, decreased vascular reactivity, and reperfusion injury by iron-induced free radicals.6 Different markers of body iron status have been used for testing the ironheart hypothesis in humans. Despite initial positive findings from experimental and preclinical research, conflicting results have emerged from epidemiological studies.3,6 The only large randomized clinical trial that assessed the effect of ironstore reduction by phlebotomy in patients with peripheral arterial disease found no significant association with allcause mortality or MI.7 Mascitelli and Goldstein refer to results from age-specific analyses of data from this study. A positive effect of reduction in body iron stores were found only among persons in the lowest age quartile. Although the ferritin hypothesis is fascinating and has generated much research, there is no strong evidence for an association between body iron stores and risk of CHD. It is therefore less likely that body iron can explain the contrast in risk between sexes. However, body iron has been suggested to act synergistically with hypercholesterolemia.3,6 Thus, it is possible that ferritin levels act as effect modifier on the association with established CHD risk factors, and that sex differences in ferritin levels lead to sex heterogeneity in the association with these factors unless adjusted for in the analysis. Interaction effects involving body iron are not yet fully explored.