Management of Type 1 Diabetes With a Very Low–Carbohydrate Diet

Mayer-Davis and colleagues in their comment of May 12, 2018 criticize the professional and media attention to our study, but we do not think that suppression of information about a novel treatment for type 1 diabetes is in the public interest.

For decades, the professional diabetes establishment focused almost exclusively on drug and technology development, to the neglect of research into nutritional therapies. Unfortunately, management of type 1 diabetes remains suboptimal, placing many at increased risk for life-threatening complications.

In our study, we document exceptional glycemic control, low rates of complications and high patient satisfaction among a community of children and adults following a very-low-carbohydrate diet. Our study design included extensive review of medical records and survey of diabetes medical care providers to confirm diagnosis and validate reported data.

The study was observational, and we fully acknowledged the limitations of this design in our manuscript. But to document a phenomenon not thought possible by many diabetes professionals, this design is an appropriate next step. The estimates by Drs. Mayer-Davis, Laffel, and Buse regarding potential selection bias may be exaggerated, as a significant number of members in the social media community were likely not active or did not have type 1 diabetes themselves.

In any event, Pediatrics considered the findings of sufficient importance to commission an accompanying commentary.

The American Diabetes Association considered our study of sufficient merit to publish a “DiabetesPro SmartBrief” (1).

The New York Times coverage was balanced, including opinion from two highly regarded diabetes experts with no role in the study (2). In that article, we urged caution, saying “because our study was observational, the results should not, by themselves, justify a change in diabetes management.”

Of special significance, reader comments to the New York Times article included hundreds of testimonials from people with type 1 diabetes who overwhelmingly reported remarkable benefits from a low-carbohydrate diet that were often dismissed by their doctors.

Of course, media hyperbole can be a problem in any research area of interest to the public. Scientists, physicians and public health experts are certainly within their rights to correct misleading stories. However, we should avoid selective enforcement against research that challenges (versus supports) conventional thinking. On that account, we would note that a relatively high carbohydrate diet is actively promoted to people with type 1 diabetes, despite the lack of any high quality clinical trials demonstrating superiority.

One hundred years ago, before the discovery of insulin, a very-low-carbohydrate diet was considered the most effective treatment for diabetes, including type 1 (3)(4). Yet to this day, there have been no major government-funded studies of a very-low-carbohydrate diet in the management of diabetes. It sometimes takes patient activism to stimulate research into neglected treatments, and a very-low-carbohydrate diet for diabetes may be one such area. If the media surrounding our study helps stimulate that research, it will have done a public health service.

Additionally, Dr. Bistrian in his comment of May 9, 2018 raises two valid points regarding our study. Relating to terminology, we agree that our acronym for a very-low-carbohydrate diet (VLCD) has been previous used for very-low-calorie diet, and this may cause confusion. For clarify, we would note that our participants consumed an isocaloric diet with an average of 36 grams carbohydrate per day. Also, per the approach specified by Dr. Bernstein, they did not aim for ketosis. Due to the high protein content of this diet, participants may or may not have been in nutritional ketosis, depending on individual differences in metabolism, physical activity level and other variables. (A conventional ketogenic diet typically limits dietary protein to ≤20% of total energy.)

Regarding the second point by Dr. Bistrian, we agree that a very-low-carbohydrate diet, by inducing nutritional ketosis (ketoacids ≤ 5 mEq/L), could theoretically increase risk for ketoacidosis (ketoacids ≥ 7 to 10 mEq/L). However, we did not observe evidence of increased rates of ketoacidosis in the study, nor did our review of the literature substantiate this concern. Conversely, it is also possible that for someone who maintains such low average glucose levels as is typical of a very-low-carbohydrate diet, that rises in glucose would alert to impending ketoacidosis sooner than might be the case for someone on a convention diet, for whom substantial hyperglycemia may not be unusual. In any event, the issue warrants examination in clinical trials.

REFERENCES:

1. American Diabetes Association. Study links very low-carb diet to better glycemic control in diabetes. (May 7, 2018) [online] Smartbrief.com.

2. O’Connor, A. How a low-carb diet might aid people with type 1 diabetes. (May 7, 2018) [online] Nytimes.com. New York Times.

3. Newburgh LH, Marsh PL. The use of a high fat diet in the treatment of diabetes mellitus. Arch Intern Med (Chic). 1921;27(6):699–705

4. Henderson G. Court of last appeal. The early history of the high-fat diet for diabetes. J Diabetes Metab 2016;7:8 DOI: 10.4172/2155-6156.100696

Dr. Bistrian raises two valid points regarding our study. Relating to terminology, we agree that our acronym for a very-low-carbohydrate diet (VLCD) has been previous used for very-low-calorie diet, and this may cause confusion. For clarify, we would note that our participants consumed an isocaloric diet with an average of 36 grams carbohydrate per day. Also, per the approach specified by Dr. Bernstein, they did not aim for ketosis. Due to the high protein content of this diet, participants may or may not be in nutritional ketosis, depending on individual differences in metabolism, physical activity level and other variables. (A conventional ketogenic diet typically limits dietary protein to ≤20% of total energy.)

Regarding the second point, we agree that a very-low-carbohydrate diet, by inducing nutritional ketosis (ketoacids ≤ 5 mEq/L), could theoretically increase risk for ketoacidosis (ketoacids ≥ 7 to 10 mEq/L). However, we did not observe evidence of increased rates of ketoacidosis in the study, nor did our review of the literature substantiate this concern. Conversely, it is also possible that for someone who maintains such low average glucose levels as is typical of a very-low-carbohydrate diet, that rises in glucose would alert to impending ketoacidosis sooner than might be the case for someone on a convention diet, for whom substantial hyperglycemia may not be unusual. In any event, the issue warrants examination in clinical trials.

The public often looks to nutrition to improve health, and reporting of nutrition findings from the scientific literature in the popular media often highlights unproven benefits (1). Lennerz et al. present data collected via an on-line community and conclude that “exceptional” glycemic control in type 1 diabetes with low risk for adverse events is possible with a very low carbohydrate diet (VLCD), and that research is needed to confirm the generalizability of these findings (2). While it may be true that a VLCD can be useful, we find the study of Lennerz to fall well short of the level of scientific evidence to merit the media and professional attention it seems to have garnered. The on-line community was not a general type 1 diabetes community; rather, this was a community following a very specific type of VLCD as promoted by one book. And of the estimated 1900 community members, only 493 responded to an eligibility survey, with 316 included in analyses (17%) and 148 with confirmed medical data, only 8% of the community (2). Of the small subset of participants with self-reported lipid concentrations (n=82; 4% of the community), 62% had dyslipidemia (2), which clearly is not desirable.

We suspect that only individuals who “believe” in the VLCD approach as promoted by the book would be in the community and respond to this survey. We can appreciate the effort made by the authors to confirm the diagnosis of type 1 diabetes, glycemic control (the HbA1c), and adherence to the diet; however, ultimately those efforts pale in comparison to the problem of selection bias. Further, respondents who report following the VLCD likely have other attributes that are likely contributors to excellent glycemic control, like careful monitoring of blood glucose (blood sugar) levels, meticulous attention to insulin administration, vigilant exercise management, etc., which can confuse or confound attribution of the VLCD to the glycemic outcomes. Nutrition guidance for patients and families living with type 1 diabetes must be made based on appropriate scientific evidence, not on what more closely resembles testimonials. We agree with the authors that VLCD’s may confer benefit for some patients with type 1 diabetes and that rigorous science is needed on this topic. The problem we now face is that it is far too easy for the potential benefit and safety of VLCDs to be publicized broadly based on this report as though findings were definitive, potentially misleading the public and adding to the substantial confusion that exists around whether or not VLCDs should be used in type 1 diabetes. Promulgating such methodologically weak though enticing data broadly through the media creates risk that patients or providers may pursue such plans without adequate insulin adjustment resulting in serious issues with hypoglycemia as well as risk for nutritional deficiencies without adequate monitoring due to the substantially reduced intake of fruits and vegetables on the VLCD (3)(4).

References:
1. O’Connor, A. (2018). How a Low-Carb Diet Might Aid People With Type 1 Diabetes. [online] Nytimes.com. New York Times.
2. Lennerz BS, Barton A, Bernstein RK, et al. Management of Type 1 Diabetes With a Very Low-Carbohydrate Diet. Pediatrics. 2018.
3. Smart CE, Annan F, Bruno LP, Higgins LA, Acerini CL. Nutritional management in children and adolescents with diabetes. Pediatric Diabetes. 2014;15(S20):135-153.
4. Evert AB, Boucher JL, Cypress M, et al. Nutrition therapy recommendations for the management of adults with diabetes. Diabetes care. 2014;37(Supplement 1:S120-S143.

The public often looks to nutrition to improve health, and reporting of nutrition findings from the scientific literature in the popular media often highlights unproven benefits (1). Lennerz et al. present data collected via an on-line community and conclude that “exceptional” glycemic control in type 1 diabetes with low risk for adverse events is possible with a very low carbohydrate diet (VLCD), and that research is needed to confirm the generalizability of these findings (2). While it may be true that a VLCD can be useful, we find the study of Lennerz to fall well short of the level of scientific evidence to merit the media and professional attention it seems to have garnered. The on-line community was not a general type 1 diabetes community; rather, this was a community following a very specific type of VLCD as promoted by one book. And of the estimated 1900 community members, only 493 responded to an eligibility survey, with 316 included in analyses (17%) and 148 with confirmed medical data, only 8% of the community (2). Of the small subset of participants with self-reported lipid concentrations (n=82; 4% of the community), 62% had dyslipidemia (2), which clearly is not desirable.

We suspect that only individuals who “believe” in the VLCD approach as promoted by the book would be in the community and respond to this survey. We can appreciate the effort made by the authors to confirm the diagnosis of type 1 diabetes, glycemic control (the HbA1c), and adherence to the diet; however, ultimately those efforts pale in comparison to the problem of selection bias. Further, respondents who report following the VLCD likely have other attributes that are likely contributors to excellent glycemic control, like careful monitoring of blood glucose (blood sugar) levels, meticulous attention to insulin administration, vigilant exercise management, etc., which can confuse or confound attribution of the VLCD to the glycemic outcomes. Nutrition guidance for patients and families living with type 1 diabetes must be made based on appropriate scientific evidence, not on what more closely resembles testimonials. We agree with the authors that VLCD’s may confer benefit for some patients with type 1 diabetes and that rigorous science is needed on this topic. The problem we now face is that it is far too easy for the potential benefit and safety of VLCDs to be publicized broadly based on this report as though findings were definitive, potentially misleading the public and adding to the substantial confusion that exists around whether or not VLCDs should be used in type 1 diabetes. Promulgating such methodologically weak though enticing data broadly through the media creates risk that patients or providers may pursue such plans without adequate insulin adjustment resulting in serious issues with hypoglycemia as well as risk for nutritional deficiencies without adequate monitoring due to the substantially reduced intake of fruits and vegetables on the VLCD (3)(4).
References:

1. O’Connor, A. (2018). How a Low-Carb Diet Might Aid People With Type 1 Diabetes. [online] Nytimes.com. Available at: https://www.nytimes.com/2018/05/07/well/live/low-carb-diet-type-1-diabet....
2. Lennerz BS, Barton A, Bernstein RK, et al. Management of Type 1 Diabetes With a Very Low-Carbohydrate Diet. Pediatrics. 2018.
3. Smart CE, Annan F, Bruno LP, Higgins LA, Acerini CL. Nutritional management in children and adolescents with diabetes. Pediatric Diabetes. 2014;15(S20):135-153.
4. Evert AB, Boucher JL, Cypress M, et al. Nutrition therapy recommendations for the management of adults with diabetes. Diabetes care. 2014;37(Supplement 1):S120-S143.

The report based on survey data that a very low carbohydrate diet might be effective in some patients with type 1 diabetes mellitus may represent an important advance in the therapy of this condition if randomized clinical trials confirm its potential advantages in terms of better blood glucose control with less risk for hypoglycemia. However there are several issues of concern. One is the general use of the term, very low carbohydrate dieting, to describe the diet when there are in fact two distinct forms. One type of ketogenic diet is the very low calorie diet (VLCD) or semistarvation, ketogenic diet (1,2) which is severely hypocaloric and provides less than 800 kilocalories/day and usually less than 400 kilocalories/day and is intended for the weight loss phase in the medical treatment of obesity. With these diets to allow for starvation ketosis to occur, which reduces hunger, from 0 to less than 50 grams of carbohydrate are provided and dietary fat intake is markedly reduced. A second type of ketogenic diet, a so-called eucaloric ketogenic diet (EKD) (3), which also restricts carbohydrate to a similar degree but contains substantially more total calories as fat is intended to provide sufficient energy to allow growth in children while helping to establish seizure control, or maintain weight and athletic performance in normal weight adults (3), and recently to allow modest weight loss in moderate to severe obesity complicating type 2 diabetes mellitus by reducing hunger with improved glucose control and reduced medication use (4). To avoid confusion the diet described in the present article should be viewed as in the second category and probably should use a different acronym than VLCD which is already spoken for.
A second concern is that although no greater risk of adverse events were noted from the survey, there should be a clinical concern in subsequent trials that patients who have mild starvation ketosis with lower ambient serum insulin could develop diabetic ketoacidosis more rapidly with the onset of intercurrent illness and the development of insulin resistance, even though there is some experimental evidence in animals that beta-hydroxybutyrate has substantial anti-inflammatory activity (5).

1. P Vertes V. Very low calorie diets--history, safety and recent developments. Postgrad Med J. 1984;60 Suppl 3:56-8.
2. Palgi A, Read JL, Greenberg I, Hoefer MA, Bistrian BR, Blackburn GL. Multidisciplinary treatment of obesity with a protein-sparing modified fast: results in 668 outpatients. Am J Public Health. 1985;75:1190-4.
3. Phinney SD, Bistrian BR, Evans WJ, Gervino E, Blackburn GL.
The human metabolic response to chronic ketosis without caloric restriction: preservation of submaximal exercise capability with reduced carbohydrate oxidation. Metabolism. 1983;3:769-76.
4. Hallberg SJ, McKenzie AL, Williams PT, Bhanpuri NH, Peters AL, Campbell WW, Hazbun TL, Volk BM, McCarter JP, Phinney SD, Volek JS. Effectiveness and Safety of a Novel Care Model for the Management of Type 2 Diabetes at 1 Year: An Open-Label, Non-Randomized, Controlled Study. Diabetes Ther. 2018;9:583-612.
5. Nandivada P, Fell GL, Pan AH, Nose V, Ling PR, Bistrian BR, Puder M. Eucaloric Ketogenic Diet Reduces Hypoglycemia and Inflammation in Mice with Endotoxemia. Lipids. 2016;51:703-14.