Welcome to the second article in the ongoing series of an attempt to partially vet the claims made in “The Carnivore Code,” by Dr. Paul Saladino. If you want to learn more about the motivations and methods of this venture, then I recommend you read the first article, which also covers research and claims about the diets of the Nicoyans. To briefly recap the key takeaways of that article, the Nicoyans seem to eat a diet similar to that of other Costa Ricans, albeit with slightly less processed food and slightly more protein. Thus, Dr. Saladino seems correct that the Nicoyans’ diet isn’t an exemplar of a plant-based “blue zones” diet. Yet the study Dr. Saladino cited didn’t support the claim he made about the Nicoyans eating lots of meat and animal fat.
In this article, I want to focus on the claim made in the book that the triglyceride to HDL ratio (TG:HDL ratio for short) is indicative of insulin sensitivity. If true, then it might have a noteworthy clinical impact given a) how often people have these lab tests done and b) how infrequently people have tests that comprehensively and accurately assess insulin sensitivity. Additionally, it seems that many proponents of the carnivore and animal-rich ketogenic diets claim that the TG:HDL ratio is a more valuable marker of cardiovascular health than is LDL cholesterol. Some proponents, including Dr. Saladino, also claim that LDL cholesterol is irrelevant in the context of insulin sensitivity. All that to say, I think this claim about the TG:HDL ratio is worthy of investigation. The approach I intend to take for this article is to randomly pick a citation within the passage in which Dr. Saladino makes this claim about the TG:HDL ratio. Thus, it’s not an entirely random process (as was the case in first article), though it seems random enough to satisfy my standards for this undertaking.
After scanning the chapter about the effects of red meat consumption on cardiovascular health, I found one referenced claim about HDL decreasing and triglycerides increasing in people with insulin resistance. (I also found one unreferenced claim about triglyceride and HDL levels being a good indication of insulin sensitivity.) There are a few referenced claims about HDL levels correlating with insulin sensitivity, and I may randomly pick one of those references to vet in another post.
In the referenced study, the researchers tried to test the hypothesis that defects in the insulin signaling pathway downstream of the receptor are responsible for metabolic dyslipidemia (i.e., elevated triglycerides and low HDL) and hepatic steatosis. To do so, they examined the biochemical profiles of a few different groups of people: people with genetic or acquired defects in the insulin receptor, people with defects in the downstream protein kinase AKT2 pathway, people with lipodystrophy, and people with idiopathic insulin resistance.
As an aside, though it’s unclear what the researchers’ definition of “insulin resistance” was, the average fasting insulin of the participants with idiopathic insulin resistance was 453 pmol/L, which is well-above the “normal” value, <174 pmol/L, that I found online. Additionally, the average fasting blood glucose of these participants was 115. mg/dL, which is in the prediabetic range, though some of these participants had been diagnosed with diabetes.
The graphic below illustrates the paper’s hypothesis, which is that in most people with insulin resistance, insulin seems to be able to promote de novo lipogenesis, yet it can’t adequately suppress hepatic gluconeogenesis. The net effect of elevated hepatic lipogenesis and gluconeogenesis, so the authors assert, is hepatic steatosis.
To clarify an important point, the authors contend that most people with insulin resistance have a defect in the pathway downstream of the insulin receptor. In support of this hypothesis, the researchers found a slew of differences between people with defects in the insulin receptor (whether genetic or acquired defects) and people with defects downstream of the insulin receptor or idiopathic insulin resistance. Pertinent to this article, the former group had “normal” lipid panels, whereas the latter group had elevated triglycerides and low HDL (i.e., metabolic dyslipidemia). Abdominal and liver fat was measured in subsets of participants from each group, and the former group (i.e., people with defects in the receptor) had relatively little liver fat, whereas the latter group (i.e., people with defects downstream) had a relatively large amount of liver fat.
I’m not sure if this study supports the claim Dr. Saladino made, however, since the statistical analysis of this study seems vague and to inadequately address the topic Dr. Saladino discusses. For example, the authors write, “Patients with lipodystrophy or with severe insulin resistance of undefined etiology had an exaggerated form of the highly prevalent metabolic dyslipidemia, with high TG and low HDL cholesterol levels.” However, the figure below shows that the participants with idiopathic severe insulin resistance (SIR) have an HDL level above the WHO threshold for the diagnosis of metabolic syndrome (represented by the dashed line). On the other hand, the two lipodystrophy subgroups, PLD and GLD, have HDL levels below the WHO threshold. It’s therefore wondering if the researchers grouped the data from the lipodystrophy and idiopathic insulin resistance when performing the analysis. If they did group the data, then it seems unreasonable to extrapolate the findings to idiopathic insulin resistance, which is what most people have. Regardless, the group with idiopathic insulin resistance seemed to have “healthy” levels of HDL, which seems at odds with the claim Dr. Saladino made about HDL being a marker of insulin sensitivity.
Below is another relevant graph from the study, and this figure shows that participants with idiopathic insulin resistance had elevated fasting triglycerides, which supports the claim Dr. Saladino made. (The dashed line represents the WHO threshold for metabolic syndrome.)
Lastly, below is a graph showing the lipid panels of diabetic and nondiabetic participants with idiopathic severe insulin resistance. Of note, the triglyceride levels are higher in the diabetic group, though this difference is not statistically significant. Perhaps if the study had a larger sample size, then this difference might reach statistical significance—I’ll save a rant on p-values and p-hacking for another time. There were no statistically significant differences in HDL or LDL, either.
To return to the question of how well the study supports the referenced claim made in “The Carnivore Code,” the answer seems mixed. HDL was not lower in the group with idiopathic insulin resistance, though triglycerides were elevated. Additionally, it’s unclear what the researchers’ definition of idiopathic insulin resistance was, though it seems to include people with prediabetes, diabetes, and elevated fasting insulin.
More broadly, however, this study was primarily trying to elucidate the effects of disrupting different steps in the insulin signaling pathway in a variety of populations with different causes of insulin resistance. This study’s statistical analysis of the relationships between insulin resistance, HDL, and triglycerides in the most prevalent types of insulin resistance was simplistic and leaves much to be desired. I have a rudimentary understanding of statistics at best, though it seems paramount to have a Pearson correlation coefficient or another assessment of correlation when trying to make claims about the direction and strength of a correlation between variables. (If anyone adept with statistics wants to chime in, then by all means!) For example, perhaps there is a direct correlation between the TG:HDL ratio and insulin resistance, though if the r value is 0.1, then it seems unlikely to be of much clinical utility. Still yet, I don’t know if there’s a consensus definition/measurement of insulin resistance, so maybe conversations about correlations with insulin resistance are somewhat moot until it insulin resistance a quantifiable definition. I’m not especially familiar with the literature on the relationships between HDL, triglycerides, and insulin resistance, though suffice it to say that I hope there are studies better designed than this one to address the question.
On that point, I hope that Dr. Saladino referenced studies better-suited to support the claims he makes about HDL and triglycerides because I want to continue to explore this topic in future posts by reviewing other referenced claims. HDL and triglycerides may be excellent indicators of insulin resistance, though this study hasn’t proven it, at least to me.
Saladino Paul. Carnivore Code: Unlocking the Secrets to Optimal Health by Returning to Our Ancestral Diet. S.l.: Houghton Mifflin Harcourt; 2020.
Semple RK, Sleigh A, Murgatroyd PR, et al. Postreceptor insulin resistance contributes to human dyslipidemia and hepatic steatosis. J Clin Invest. 2009;119(2):315-322. doi:10.1172/JCI37432