One of the early consequences of a compromised liver is often a drift toward diabetes. The pancreas and the liver work together to manage insulin and the use and processing of glucose so absent a specific pancreas disease liver function and fat processing are fundamental to health. This discussion focuses on insulin and how it responds to different dietary fats but the conductor of the dance is the offstage liver. If you read this carefully it shows the hazards of saturated fats and the benefit of extra virgin olive oil which is advocated by this foundation. When you think about your diet understanding the bio-chemistry will help you as you consider how to change your lifestyle to be kind to your liver. As a practical matter a diet that prevents diabetes will be just fine for your liver as well. The smart plan is to not let your liver look like this one day.
Adding fats to carbohydrate containing meals is a common recommendation to diabetics to make meals “healthier” by reducing the glycemic response to the meal. The primary mechanism through which fat does this is by slowing the rate of gastric emptying, which leads to a slower appearance of glucose into the blood. Given that postprandial glycemia is an important risk factor for many diabetic complications, it makes complete sense to want to minimize post-meal blood glucose excursions.
However, to focus solely on the blood glucose response of a meal misses the forest for the trees. There is a considerable amount of evidence to suggest that consuming starchy carbohydrates in combination with excessive dietary fat, especially saturated fat, causes an acute state of insulin resistance that may last for hours after the meal. This has been known since at least 1983, when Collier and O’Dea published research showing that adding butter to a potato meal significantly blunted the rise in blood glucose without significantly affecting insulin in young and healthy men and women. Thus, the amount of insulin required to handle a similar amount of glucose in the blood was 3-fold greater when butter was added to the potato compared to eating the potato alone.
So while postprandial glycemia was reduced, more insulin was required to dispose of the glucose in the blood and insulin levels remained elevated for a longer period of time. Collier and O’Dea conclude,
These changes found after the co-ingestion of fat may indicate an acute insulin insensitivity or at least a potentiation of insulin secretion which could form the basis of the insulin resistance associated with the chronic consumption of high fat diets.
One year later, Collier et al expanded upon their initial findings by showing that the same effects occur when butter is consumed alongside lentils, a food that contains far more fiber than potatoes. Although this early research could not establish causality, more recent investigations have.
Recently, researchers from Canada used intravenous infusions of different fat types (lard, olive oil, or soybean oil) or a saline control in rodents to study the effects of elevated blood fatty acids on insulin sensitivity. The infusion aspect is important because consuming a high-fat diet can lead to other metabolic alterations which may influence the results. When interested specifically in the effects of blood lipid levels on insulin sensitivity, infusions remove a considerable number of confounding variables. Moreover, the infusion level of fatty acids was designed to reflect the short-term changes related to meals.
The glucose infusion rate was significantly reduced in all fat groups relative to the saline control, indicating reduced whole-body insulin sensitivity. Although there were no statistical differences between the fat conditions, the reduction was about 54% with olive and soybean oil and 74% with lard. It is worth mentioning that the blood fatty acid levels achieved during the infusion were about 20% less in the lard group compared to the olive and soybean oil groups, which may have underestimated the size of the effects. Nonetheless, endogenous glucose production was significantly elevated and insulin clearance significantly reduced in all fat conditions to a similar extent, indicating insulin resistance of the liver. This study clearly shows that physiological elevations of fatty acids reduce whole-body insulin sensitivity (muscle, fat, and liver). It also suggests that the effects may be more pronounced with SFAs as the lard condition, which is 40% SFAs compared to 15% in the olive and soybean oil, tended to have lower insulin sensitivity.
In another study, Dube et al had athletic, lean but sedentary, and obese individuals undergo a euglycemic insulinemic clamp with and without a simultaneous lipid infusion. It was shown that while the athletic participants were able to increase fatty acid oxidation in response to the lipid infusion so as to help offset the adverse effects on glucose disposal, all three groups showed significant insulin resistance. Additionally, Ferrannini et al showed an inhibition of insulin-stimulated glucose utilization in the presence of physiological elevations of NEFAs.
These studies used fat infusions for consistency and increased experimental control, but they only serve to mimic what occurs naturally when we eat dietary fat. For instance, Griffiths et al demonstrated that when fat is ingested alongside carbohydrates, there is a significantly greater increase in blood levels of triglycerides and non-esterified free fatty acids (NEFA). Other work has shown that there is a clear dose-response relationship between the change in blood NEFA and the amount of fat in the meal.
Is there a “safe” level of fat intake to prevent insulin resistance?
This question was first addressed by Gannon et al in 1993 when they conducted a dose-response study on the effects of consuming 50 grams of carbohydrate from a potato without (control) or with 5, 15, 30, or 50 grams of fat from butter in individuals with type-2 diabetes. Over the 5 hours after meal consumption, insulin concentrations were greater in every butter condition compared to the potato alone, maximizing and reaching statistical significance with 15 or more grams of fat.
These findings are supported by the work the work of Norman et al, who recruited healthy young women to consume wheat pasta without (low fat; LF), or with 15 (medium fat; MF) or 40 (high fat; HF) grams of sunflower oil. Over the entire 7-hours after meal consumption, blood glucose concentrations were similar among all groups but the insulin required to dispose of that glucose was greater in the MF and HF groups compared to the LF group. However, in the HF group only was blood glucose and insulin concentrations significantly greater than the LF and MF conditions during the last 4 hours. In other words, the insulin response was similar between the 15 and 40 gram conditions but the 40 gram condition led to a prolonged elevation of blood glucose and insulin to the extent that no return to basal levels was observed even 7 hours after meal consumption.
In both studies the insulin response was similarly increased among conditions with 15 or more grams of fat added, suggesting that this may be an upper limit for carbohydrate-heavy meals where insulin is increased initially, but will return to basal levels normally. In contrast, consuming more fat than this appears to extend hyperglycemia and hyperinsulinemia for hours afterwards.
Prolonged hyperglycemia is very problematic as it only serves to perpetuate the problems caused by the fatty acids to begin with. This is demonstrated by Kolka et al, who showed that even modest hyperglycemia (~120 mg/dL) impairs the dispersion of insulin throughout muscle cells. Interestingly, a study in obese women also demonstrates that postprandial hyperglycemia have impaired delivery of insulin to both adipose tissue and skeletal muscle.
Insulin (and all hormones for that matter) must reach a cell to exert its effects. Insulin does not do anything while circulating around in the blood, and must bind to insulin receptors on the surface of target cell membranes to exert its effects. Insulin transport from the pancreas and capillaries to the muscle cells and its subsequent dispersion through them is the rate-limiting step for insulin signaling. Therefore, not only does the fat co-consumed with the starchy carbohydrates directly cause insulin resistance, but it also indirectly results in prolonged hyperglycemia that acts to perpetuate this fat-induced insulin resistance.
Interestingly, 15 grams of fat corresponded to 20% of calories in the butter trial and 33% of calories in the sunflower oil trial, and this discrepancy may be owed to the type of fat.
The type of fat
Butter is primarily saturated fatty acids (SFA) and sunflower oil is primarily polyunsaturated fatty acids (PUFA). The other major fatty acid category is monounsaturated (MUFA). In order to discern the extent to which the relative amounts of SFAs and MUFA impacted insulin sensitivity, researchers from Spain recruited young and healthy men to consume a low-glycemic-load 800 kcal pasta meal containing about 44g of carbohydrates and 50g of fat from butter, high-palmitic sunflower oil (HPSO), refined olive oil (ROO), or a mixture of vegetable and fish oils (VEFO) that contained a MUFA:SFA ratio of 0.48, 2.42, 5.43, and 7.08, respectively. The subjects also consumed the same test meal containing no fat as a control meal. With no differences among the conditions for the glycemic response, the insulin response was found to be inversely proportional to the ratio of MUFAs to SFAs. Thus, the butter condition led to the greatest insulin resistance, followed by HPSO and then ROO and VEFO (which were not significantly different from one-another).
Additionally, researchers from France had healthy subjects consume whole-food meals (47% fat, 38% carb, 14%) based on steak, cheese, potato, and rice with one of two fat sources: a high-oleic sunflower and canola oil mixture (high-MUFA) or a mixture of sunflower and soybean oils (high-ω6 PUFA). The amounts of SFAs and ω-3 PUFAs were kept constant among test meals to remove potential confounding variables, and it was found that the overall insulin response was not significantly different among the two conditions.
Similarly, Radulescu et al showed that 25 grams of lard, olive oil, or safflower oil added to a potato meal do not differ from one another in their effects on glucose and insulin, with all three blunting the initial peak but prolonging the elevation above basal levels. Importantly, lard contains about 45% MUFA and 40% SFA, which is markedly less SFA than butter (~69%) and suggests that animal fats may not have similar effects as butter. Finally, Burdge et al showed that there were no differences between linoleic acid, α-linolenic acid, EPA+DHA, or MUFA regarding post-prandial changes in glucose or insulin.
Very little research has looked at medium-chained SFA (MCSFAs), which are primarily coconut oil, with regards to its effects on insulin. However, two early studies (here and here) do suggest that MCSFAs potentiate insulin secretion.
Finally, it does appear that the type of carbohydrate matters. Based on the knowledge that blood glucose levels after a meal are determined primarily by the amount of available glucose for absorption, Gannon et al compared meals containing 20% protein, 40% fat, and 40% carbohydrate with the only difference being the source of the carbohydrates. The control American meal was a hamburger with a bun and baked potato whereas the experimental meal was a hamburger with fruit and cheese. As would be expected, the experimental meal had significantly lessened blood glucose and insulin responses and ultimately required 45% less insulin to handle an equivalent amount of glucose.
The role of diet
As could reasonably be expected, the habitual diet of the individual plays a large role in mediating the insulin resistant effects of dietary fat. This is best exemplified by the KANWU study, where 162 healthy, middle-aged men and women were recruited to undergo a 90-day controlled dietary intervention that was either a high-SFA (17% kcal SFA; 14% MUFA; 6% PUFA) diet or a high-MUFA (8% kcal SFA; 23% MUFA; 6% PUFA) diet containing 15% kcal as protein, 46% kcal as carbohydrates, and 37% kcal as fat. The fats used in the diet were combinations of butter, margarines, and oils such as high-oleic sunflower oil. Insulin sensitivity decreased by 10% in the SFA group but did not change in the MUFA group. However, since free-living subjects don’t like doing what they are told despite having their dietary fats provided to them by the researchers for consumption, not everyone consumed the assigned 37% dietary fat, although on average the groups did. When the researchers re-analyzed the insulin sensitivity data based on fat intakes above and below the assigned 37%, it was found that the beneficial effects of the MUFA diet disappeared in those who consumed more, suggesting that when dietary fat intake is greater than 37% of energy intake, all fats are equally detrimental with regards to insulin sensitivity.
However, Kien et al recruited young and healthy men and women consumed one of several diets for one week before switching to another diet for another week until all diets were consumed by all subjects. Initially, all subjects consumed a low-fat, low-SFA diet (20% protein, 50% carb, 30% total fat, 5% SFAs, & 16% MUFAs). Subsequently, the subjects consumed either a high-fat, high-SFA diet (17% protein, 43% carb, 40% fat, 16% SFAs, 16% MUFAs) or a high-fat, high-MUFA diet (17% protein, 43% carb, 40% fat, 2.4% SFAs, 30% MUFAs). Only in women did the high-MUFA diet significantly improved insulin sensitivity relative to the high-SFA diet, suggesting that women may benefit more from paying attention to the type of fatty acid.
In a separate study by researchers from Japan, young and healthy women were instructed to consume one of two identical diets differing only in the SFA:PUFA ratio. Both diets were isocaloric with 30% fat, 54% carb, and 16% protein. However, the high-SFA diet had a SFA:MUFA:PUFA ratio of 5:4:1, whereas the low-SFA diet had a ratio of 3:4:3. This was achieved by using different amounts of butter and soybean oil in food preparation of the standard Japanese meals. On this note, the sugar content is very low, with 93% of the carbohydrates coming from starch (probably white rice). The meals were not only compared to one another, but also to a control meal that was also isocaloric and composed of 20% fat, 64% carb, and 16% protein with a fatty acid ratio of 3:4:3.
After consuming their diets for one week, the participants consumed their respective test meals that consisted of just over 500 kcal, ~23g of protein, 18g of fat (14g in control meal), and 67g of carbohydrates (87g in control meal). No differences were observed in the glycemic response for any of the meals, but insulin was significantly elevated above the control meal in the high-SFA meal only. The KANWU study suggested that SFA were more detrimental than MUFA when total fat intake was less than 37% of total calories, and the current study suggests that SFA are more detrimental than PUFA as well, at least when total fat intake is 30% of calories.
Why is it that SFA appear more detrimental both in the short-term meal and when consumed habitually? It may be because SFA induce a larger post-meal increase in NEFA than MUFA or PUFA, which is the culprit behind insulin resistance in the short-term. Regarding habitual intake, there is a direct association between the amount of saturated fat stored within the muscle and the proportion of SFA in muscle cell membranes and insulin resistance, and both of these reflect the fat composition of the diet.
Other research has since confirmed and expanded upon the above findings. In a separate study of healthy young males, it was shown that consumption of a diet containing 42% fat for two weeks significantly increased small intestinal motility, thus speeding the transit of fat through the intestinal tract. There is also a ton of rodent studies that lend support, but since it is unclear whether observations from studies in animals are applicable to humans, there is no point in mentioning them.
However, it must be noted that the high-fat diets were also contained significantly more calories. In fact, in both studies mentioned above, the low- and high-fat diet groups consumed the same diet with the high-fat group simply adding in concentrated fat sources. Owed to the marked difference in the total energy content of the diets, it is impossible to determine whether the observed changes in gastric emptying and small intestinal transit were related to the fat content of the diet, per se, or to the high energy intake. That said, some research suggests that increasing energy density of meals actually slows gastric transit, which would only make the above results even more impressive.
So the takeaway here is that adding fat to a starchy carbohydrate may indeed lower the glycemic response – but only if the habitual diet is not high in fat, as research in humans suggests that anything above 40% may lead to gastrointestinal adaptations that ultimately speed the gastric transit time of fatty meals which would consequently remove the beneficial reduction of post-prandial glycemia.
So eating fat with starchy carbohydrates only benefits glycemia in low- to moderate-fat diets?
Perhaps in the first meal, as Ercan et al investigated how the effects carried over from one meal to the next. In this study, plasma glucose and insulin responses to a potato ingested with or without butter, given in various combinations as two meals 4 hours apart were determined, again in healthy subjects. After the first meal, blood glucose peaked at 30 min regardless of the composition of the meals, although there was a significant attenuation with the buttered potato, as expected. Yet, when the subjects consumed fat in both the first and second meals, the glucose rise was not attenuated after the second meal as it had been for the first meal, suggesting that the beneficial effects are a one-time deal for the day, at least when meals are consumed within four hours of one another. Ercan eventually concludes (emphasis mine),
These data indicate that when a second meal contained butter with potato but the first meal consisted of only potato, a smaller glucose rise occurred with the second meal. That is, the presence of butter resulted in an attenuated glucose area response just as when it was present in the first meal. However, if both meals contained fat, the glucose area was not decreased with the second meal.
The carry-over effect was confirmed in a later study of healthy post-menopausal women that found a breakfast meal with 40g of fat (48% kcal) and 78g of carbohydrate (42% kcal) to reduce insulin sensitivity both at breakfast and during a low-fat (6g) lunch that was consumed five hours later. Rest assured, however, that the effects do not carry on forever. Traianedes et al demonstrated that the consumption of a high-fat (44% kcal) dinner containing fat from either butter, safflower oil, olive oil, or medium-chained triglycerides has no effect on glucose tolerance or insulin to a standard breakfast consumed 12 hours later after an overnight fast and a good night’s sleep.
The evidence thus far suggests that consuming fat alongside starchy carbohydrates such as potatoes and rice acts to reduce and delay the initial blood glucose spike that would occur if no fat was consumed. However, ultimately the amount of glucose entering the blood will be the same because the fat only acts to slow the absorption of glucose into the bloodstream. Therefore, fat may lower the postprandial glucose spike, but it will cause the body to be in a state of hyperglycemia for a longer time period.
Fat also appears to cause a temporary and acute state of insulin resistance whereby more insulin is required to dispose of the same or in some cases less glucose than if no fat was consumed alongside a starchy carbohydrate. Moreover, the insulin response is prolonged, meaning that blood insulin levels remain elevated for a longer period of time.
The type of carbohydrate also plays a large role in determining the extent of insulin resistance. Starchy carbohydrates may be the main offender regardless of fiber content, as an interaction with fat has been observed with regular potatoes, cooked and cooled potatoes (resistant starch), legumes such as lentils, and grain breads. On the other hand, foods with low available glucose such as fruits and fibrous vegetables appear to not be affected by dietary fat to the same extent that starches are.
The insulin resistance effects are most pronounced with SFAs, as evidenced primarily by research using butter. Oils containing predominantly MUFA and PUFA appear to act similar to one-another, both of which are less insulinemic than SFAs. Additionally, animal fats appear to act similar to oils, perhaps because at least half of their fat content is unsaturated. Regardless of the fat, a reasonably “safe” upper limit of consumption appears to be about 15 grams of total fat in a meal if the meal contains high glycemic carbs so avoiding sugar is confirmed as a proper strategy.
The background diet of an individual plays a large role in moderating the fat-induced effects on insulin sensitivity. If the habitual diet is greater than 35-40% calories from fat, then any potential benefit of fat on hyperglycemia may not occur and all types of fat are equally detrimental to insulin sensitivity. If the diet is less than 35% fat, then SFAs cause greater insulin resistance than MUFA or PUFA. Regardless of total fat intake, if SFA intake makes up a large proportion of fat intake then the skeletal muscle phospholipid and triglyceride composition will reflect this, which ultimately leads to reduced insulin sensitivity.
In 1963, Philip Randle and colleagues proposed a “glucose-fatty acid cycle” that described the biochemical competition between glucose and fatty acids for use as energy within cells. It added a much needed nutrient-mediated fine tuning to the crude hormonal regulation of substrate utilization. The Randle Cycle has been reviewed in depth by Hue and Taegtmeyer, and anyone interested for a far more scientific explanation for how fatty acids cause insulin resistance should check out their work.