Recently Elizabeth Snouffer published an interview with me at Diabetes24-7. During the discussion after, Elizabeth noted that a number of people liked the Smart Insulin approach, and she asked me to talk about how it compared with the Islet Sheet. Today I feel I was not completely fair to Smart Insulin and wanted to write a bit more. In particular, I want to explain what a truly Smart Insulin would look like.
The idea is attractive: formulate insulin so that the rate of insulin release increases as glucose goes up.
Islets of Langerhans store insulin in the form of crystals, which contain both the protein insulin and the ion zinc (yes, the same zinc found in mineral supplements that is essential to immune system function.) When the islets release insulin into blood they are actually releasing small crystals of zinc and insulin that dissolve quickly in the blood. When your islets secrete a burst of insulin they release a burst of zinc at the same time.
Classic, “regular” insulin is the same natural, slow-release formulation. It is made up of these natural co-crystals of insulin and zinc, which lodge in the fat beneath the skin and slowly dissolve. Once the insulin dissolves, it is absorbed into the blood. The dissolution of the crystal takes about an hour, and the absorption also takes about an hour. Thus, in a way, nature itself provides a slow release insulin formulation.
Fast-acting insulins such as NovoLog work by using genetic engineering to sabotage the zinc binding site. So the formulation contains insulin but no zinc, so no zinc crystals. And that is why fast-acting insulin works in just an hour. There is no need for decrystalization; the insulin has no zinc bound, and is ready to go.
So how does Smart Insulin work?
The only publication I can find describing Smart Insulin is a patent (United States Patent 7,531,191). The matrix chemistry is designed to accelerate insulin release as glucose levels go up. Unfortunately, the degree it goes up is not great (Fig. 30 see below). 
Let me explain Figure 30. The formulation is exposed to four different levels of glucose, identified in the bottom axis. Insulin release by the glucose trigger is shown on the left axis. The conversion of mg/ml on the glucose axis to the familiar mg/dl is 100. The left-most point is pretty close to normal blood sugar, 80 mg/dl (0.8 mg/ml). So from normal blood glucose (0.8 in Figure 30) to high blood sugar (3 in Figure 30) the insulin release goes up by ~2X, from 1.2 to 2.4, in the most favorable (triangle point) formulation. That is much smaller than you would see from islet insulin secretion over the same change.
I would assume that subsequent work has improved the release acceleration— but it would have to improve it a lot, in order to mark a real step forward.
A bigger problem is the insulin release rate at low glucose. Islets naturally secrete very little insulin below 80 mg/dl glucose. Based on extrapolating Figure 30 the insulin secreted by Smart Insulin at that sort of glucose level appears to be too high.
It would be great to see the data that supports Smart Insulin claims made by the CEO, Todd Zion. “The drug is apparently working really well in the large animal model.” Since the technology is in the formulation, which need not be released with the animal data, there is no reason the company cannot share their results with the world. I have written to them, asking to see their results, but so far I have heard nothing in response.
The ideal study for a fast-acting insulin, to my mind, would be a glucose tolerance test. Make a large animal diabetic by total pancreatectomy. Prove it is diabetic with glucose stimulation followed by demonstration that there is no remaining insulin secretion. Then, inject the Smart Insulin formulation. Wait for glucose to normalize, and do a standard IVGTT using the Bergman protocol measuring glucose and insulin at multiple points. A single graph could demonstrate that Smart Insulin can automatically control blood glucose.
I look forward to seeing these data.
An interesting question underlying these considerations is what functionally constitutes euglycemia? How close to normal would the outcome of any glucose stabilization strategy have to be before it could prevent complications? It has been determined, for example, that even increases in blood sugar below the normal range increase a person’s risk of cardiovascular disease. (Z. Punthakee, et al, “Diabetes and Cardiovascular Disease,” Review of Cardiovascular Medicine, vol. 8, no. 3, p. 145 (2007)) Even a single elevated HbA1c value has been linked to a slowing of peroneal nerve conduction velocities. (H. Dorchy, “Screening for Subclinical Complications in Young Type 1 Diabetic Patients,” Pediatric Endocrinology Review, vol. 1, no. 3, p. 380 (2005)) A non-physiologic variability in the body’s glucose levels, such as would come with any less than perfectly engineered blood sugar control system, would also cause complications just from its abnormal variability. (J. Braqd, et al, “Can Glycaemic Variability Predict the Development of Complications?” Diabetes and Metabolism, vol. 34, no. 6, pt. 1, p. 612 (2008)) Pre-diabetics having an abnormal response to a glucose tolerance challenge but no demonstrable blood glucose abnormalities already have an increased incidence of ‘diabetic’ peripheral neuropathy, but where this comes from, nobody knows. (M. Polydefkis, et al, “New Insights into Diabetic Polyneuropathy,” Journal of the American Medical Association, vol. 290, p. 1321 (2003)) Given the difficulty of building an answer to all these problems into a technical or chemical system by some complex algorithm, I would prefer the most natural solution possible in the hopes that it would take care of these issues on its own, even if we don’t understand exactly how.
There is also increasing evidence now that a decisive element missing in the treatment of diabetes is c-peptide, which is normally present in the same amounts as insulin and is produced along with it. Since diabetics with no surviving insulin production are less able to control their blood sugar, the hypothesis that lack of blood sugar control causes complications would appear to be confirmed by the same data which showed that lack of c-peptide causes complications, with the former hypothesis masking the latter. Independent tests of the effect of c-peptde suggest that it addresses all the classic complications of diabetes: K. Ekberg, et al, “C-Peptide Replacement Therapy and Sensory Nerve Function in Type 1 Diabetic Neuropathy,” Diabetes Care, vol. 30, no. 1, p. 71 (2007); F. Panero, et al, “Fasting Plasma C-Peptide and Micro- and Macrovascular Complications,” Diabetes Care, vol. 32, no. 2, p. 301 (2009); L. Nordquist and J. Wahren, “C-Peptide: The Missing Link in Diabetic Nephropathy?” Review of Diabetes Studies, vol.. 6, no. 3, p. 203 (2009). Thus I predict that any approach to type 1 which does not ensure that an adequate amount of c-peptide is supplied — whether in the form of ‘Smart Insulin’ containing c-peptide in the package, or encapsulated porcine islets harvested so as to retain their c-peptide producing capacity — provided porcine c-peptide is demonstrated to work in humans as well as human c-peptide — will not achieve the desired results.
I don’t see much therapeutic value in C-peptide. (In spite of the fact that the Islet Sheet will deliver C-peptide along with insulin, so this is a potential product advantage.) Aside from not being impressed by the studies, I have a basic objection. One of the sure ways to determine which amino acids that are important in a protein’s function is to compare the sequences of the protein from different species. For instance, all known insulin’s (covering hundreds of millions of years of evolution) have several amino acids in common. This proves that those amino acids are essential to insulin’s function.
This is not true of C-peptide. The sequence drifts rapidly in evolutionary time. Based on that alone C-peptide has no important function other than to guide insulin folding.
You raise a fascinating and grossly under-researched topic when you discuss natural blood sugar management in other species, with or without human levels of c-peptide — and in some cases, such as in certain insects, even entirely without insulin.
Since normalizing blood glucose levels in diabetics is such a difficult and dangerous project with current therapies, and is proving so elusive a goal in new treatment research, it is always tempting to look to models from other species — many of which normally and without harm to themselves endure very high blood sugar levels — to see if that natural neutralization of the effects of hyperglycemia can somehow be replicated in humans. Birds, for example, generally have very high normal blood sugar levels with no apparent harm to themselves, and among these the highest levels are found in hummingbirds, which endure blood sugars ranging from 600 to 800 (post-prandially), without evidence of complications. Although avian vasculature shows the cross-linkages found in aging humans, which are also thought to be promoted by hyperglycemia in diabetics, it appears that hyperglycemia in birds does nothing to intensify cross-linkage formation. If by suppressing certain enzymes or otherwise altering the physiology of humans we could increase their tolerance for excess glucose to match that found naturally in birds, blood sugar control might have to be no more strict than sufficient to maintain weight, avoid infections, and avoid hyperosmolar coma, which would allow patients to avoid all the stress of intensive glucose management and the threat of hypoglycemia. (C. Beuchat and C. Chong, “Hyperglycemia in Hummingbirds and Its Consequences for Hemoglobin Glycation,” Comparative Biochemistry and Physiology: Part A, vol. 20, no. 3, p. 400 (1998); M. Iqbal, et al, “Protein Glycolyzation and Advanced Glycation Endproduct Accumulation: An Avian Solution?” Journal of Gerontology: Series A, vol. 54, no. 4, p. 617 (1999))
The case of animals relatively or absolutely immune to the effects of excess blood sugar raises the interesting question whether similarly exceptional human cases depend on similar physiological mechanisms. Thus while 40% of type 1 diabetics with careful blood sugar control in one report were found to have developed neuropathy (M. Centofani, “Diabetes Complications: More than Just Sugar?” Science News, vol. 149, no. 26/27, p. 421 (1995)), in the DCCT 40% of type 1 diabetics with an HbA1c greater than 9.8% had no retinopathy after 10 years’ follow-up. (J. Estenes, “Absence of Diabetic Retinopathy,” Diabetology and Metabolic Syndrome, vol. 1, no. 13, Oct. 3, 2009) The data from long-term type 1 diabetics show similarly surprising results, with half-century survivors showing minimal complications despite their average HbA1c of over 10%.(G. Gill, et al, “insulin-Dependent Diabetes Over 50 Years’ Duration,” Practical Diabetes International, vol. 10, no. 2, p. 60 (2005)) Another study found half-century survivors to have an average HbA1c of 7.6%, which suggests nothing special in terms of their blood sugar control to explain why they have done so much better than others. (S. Bain, et al, “Characteristics of Type 1 Diabetes of Over 50 Years’ Duration,” Diabetes Medicine, vol. 20, no. 10, p. 809 (2003))
Since most studies which have looked at why some patients escape complications despite poor blood sugar control, and why others succumb early despite meticulous control, have pointed to certain genes as the factors separating the saved from the damned, it would be useful to study what these genes are doing to see if they might be producing some simple substance which can safely be substituted in patients lacking these genes. If it is, for example, the presence of a gene causing an excessively rapid degradation of nitric oxide which is the culprit, which seems plausible, since NO is known to play an important role in the health of the vasculature, then something as simple and harmless as giving diabetics l-arginine as an NO donor might make a world of difference. (A. Mollsten, et al, “The Endothilial Nitric Oxide Synthase Gene and the Risk of Diabetic Nephropathy,” Molecular Genetics and Metabolism, vol. 97, no. 1, p. 80 (2009))
.. then factor in people like me, who were not screened, and then mis-diagnosed, and then self-diagnosed for hereditary hemochromatosis. I have type 1-ish diabetes which was 100.000% preventable. And my diabetes is outta control. It always will be. The sheet or island would be welcome strategies for someone like me whose outcome is probably poor in the medium term.