How might individualized medicine apply to the treatment of type 2 diabetes? Type 2 diabetes is heterogeneous. We know that for many patients the principal problem is insulin resistance. Many are hyperinsulinemic. For those patients, while insulin or insulin secretagogues are often given to try and achieve glycemic control, such treatment is not physiologic. It promotes weight gain in patients who are already obese and it may be atherogenic.
To address this issue researchers from Hong Kong have been treating DM 2 patients based on metabolic phenotype. Those who are truly insulin deficient, based on low C-peptide levels, are treated with insulin. Those who are normoinsulinemic or hyperinsulinemic (their C-peptide levels are normal or high) are not given insulin.
Several years ago those investigators showed that such an approach produced improved glycemic control. More recently they have published a study looking at clinical outcomes. In that longitudinal cohort study patients treated based on C-peptide levels had better outcomes than patients whose treatment was not based on phenotype. In particular, patients who had elevated C-peptide levels and were treated with insulin had the worst outcomes. Although that difference in outcome lost statistical significance after adjustment for other variables it fit the authors’ physiologic rationale and led them to suggest that treatment based on C-peptide phenotype is promising. Hyperinsulinemia promotes macrovascular disease, which may explain why many of the treatments for type 2 diabetes are macrovascular neutral or, worse yet, cause macrovascular harm. That’s why this paper is so important despite the editorial in the same issue of CMAJ expressing the view that this is an unproven strategy for the treatment of type 2 diabetes.
The paper is also of interest to me for personal reasons. In the early 1970s as a Vanderbilt medical student I was privileged to spend my summers on a research project in the laboratory of Oscar B. Crofford, M.D., where we had a similar idea. Even better, the method was simple and more readily applicable to patient care. The problem was, the timing was wrong, as I’ll explain.
Dr. Crofford, who would later become the principal investigator of the landmark Diabetes Control and Complication Trial, at that time headed up Vanderbilt’s Diabetes and Metabolism center. His lab was involved in diabetes research on multiple levels from bench to bedside. He firmly believed that research at the molecular level should inform clinical medicine. He was truly the physician-scientist. Every year his lab took in students to work on various projects. My project, along with several other students, was to explore clinical applications of the measurement of breath acetone by gas chromatography. Specimen collection was simple (the patient would blow into a ground glass syringe) and results were available almost immediately, the acetone retention time on the column being just under 3 minutes. The high sensitivity of the apparatus enabled detection of acetone concentrations down to 1 nM, far below the threshold for detection of ketonuria or ketonemia in the clinical laboratory. This enabled detection of subtle changes in insulin effect.
In those days diabetes was classified into juvenile onset and adult onset type diabetes. The terms “type 1” and “type 2” did not yet exist and we were just beginning to understand the distinctions. Although most of the patients we evaluated in the clinic had what we would now call type 2 diabetes we gained a new appreciation of their heterogeneity when we started measuring breath acetone. Although patients with type 2 diabetes do not develop ketoacidosis in the basal state they do have small fluctuations in breath acetone due to the varying action of insulin and counter-regulatory hormones. These small fluctuations could be detected by the highly sensitive gas chromatographic method. For those patients who were uncontrolled we noticed two patterns: hyperglycemia with and without elevated breath acetone. I believe these patterns correspond to the two C-peptide phenotypes (low and high, respectively) noted by the Hong Kong investigators.
About the time this work was completed the interest in diabetes control was shifting away from acetone metabolism to carbohydrate metabolism. Finger stick blood sugar measurement was just emerging and the debate over optimal glycemic targets was heating up. No one outside Dr. Crofford’s lab seemed very interested in breath acetone. Several years after I left the project their paper was finally accepted in, of all places, Transactions of the American Clinical and Climatological Association.
Although not measuring C-peptide (they were measuring something much simpler) Crofford and colleagues recognized the two patterns described in the Hong Kong paper:
For clinical purposes we classify diabetics with hyperglycemia as being in one of two categories depending upon the breath acetone concentration. The scheme is shown in Table VI. It is undoubtedly oversimplified but nevertheless useful. Patients with hyperglycemia and a normal breath acetone (less than 50 nM) are considered to have hyperglycemia of overeating and are treated with more vigorous dietary measures. Although it had been our practice in the past to treat hyperglycemic patients with "more insulin," the results were usually disappointing if overeating was the major problem. More insulin led to more overeating and to more obesity. More obesity led to more insulin resistance and, in most cases, the net result was that the hyperglycemia was not improved. This is generally referred to as the "diabetic clinic cycle." Our emphasis today is: "you don't treat overeating with insulin but with better dietary management." Our general policy in well nourished adult diabetics is not to increase the insulin dose unless the breath acetone is elevated. If the patient has hyperglycemia and an elevated breath acetone he is considered to be inadequately treated with insulin and the dose is adjusted appropriately.
Employing “more vigorous dietary measures” may seem simplistic, and is certainly easier said than done, but Crofford’s words were prescient. Thirty years later, our failed experiment with diabetes drugs to ameliorate macrovascular disease in type 2 diabetes has taught us that the best methods may be non-pharmacologic, i.e. diet and exercise, in those who have normal insulin secretion or are hyperinsulinemic. (Those of you inclined to consider me a pharma shill mark those words!). I believe the best explanation for the disappointing results of ACCORD can be found in that paper from over 30 years ago, as expressed in Crofford’s simple words: you don't treat overeating with insulin…
The appeal of breath acetone was ease of collection and availability of results at the point of care. That appeal was overshadowed by the emerging finger stick glucose technology, heightened interest in glycemic targets and the development of glycosylated hemoglobin assays. Breath acetone was all but forgotten.
Now that macrovascular benefits are such an elusive treatment goal in type 2 diabetes and given that it matters how we treat hyperglycemia in these patients breath acetone measurement deserves renewed interest. It is inexpensive, easily performed and available at the point of care. We need more studies to validate the method and correlate breath acetone with C-peptide levels and other metabolic parameters and establish normal values. Then it will be time for a randomized trial comparing breath acetone guided treatment with “standard” treatment of patients with type 2 diabetes.