| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
COMMENTARY |
Division of Endocrinology, Diabetes, & Clinical Nutrition, Oregon Health & Science University, Portland, Oregon 97239-3098
Address all correspondence and requests for reprints to: Matthew C. Riddle, Professor of Medicine, Division of Endocrinology, Diabetes, & Clinical Nutrition, Oregon Health & Science University L-345, Portland, Oregon 97239-3098. E-mail: riddlem{at}ohsu.edu.
Injected insulin is the best-tested and most powerful treatment for hyperglycemia. Clinical trials show that it can limit microvascular and, perhaps, cardiovascular complications of diabetes. Even so, in clinical practice the use of insulin for type 2 diabetes continues to be delayed, allowing hyperglycemia to persist on oral therapy, and once started is often disappointingly ineffective. Failure to begin or optimize insulin treatment is often attributed to the patient’s fear and incapacity, but success in studies and in specialized diabetes centers argues that other barriers are more important. Attention is increasingly focused on developing suitable algorithms, i.e. standardized protocols for starting and adjusting insulin, to address this problem. For example, continuing oral therapies while starting insulin can improve the results of simple insulin regimens such as a single daily injection of longer-acting (basal) insulin (1). In addition, systematic titration of insulin dosage, seeking a defined, self-measured glucose target, is gaining acceptance.
These concepts came together in the Treat-to-Target Trial, which tested the full ability of initiation and systematic titration of neutral protamine Hagedorn (NPH) or glargine insulin, taken at bedtime and with continuation of prior oral antihyperglycemic agents, to achieve the evidence-based 7.0% A1c target (2). Of the 756 patients studied, 58% reached this target (from a mean baseline of 8.6%) by adding either NPH or glargine. This level of success has since been matched in other "treat-to-target" studies, using similar designs and titration schemes, with NPH, glargine, detemir, and premixed insulins (3, 4, 5, 6).
However, questions remain. Can prandial or premixed insulins be used just as successfully as basal insulin for initial therapy? Can the treat-to-target method be applied as effectively in long-term clinical practice as in trials? What should be done when any initial insulin regimen fails to reach the A1c target? An ambitious new trial directly addressed these questions (7). It is 4T (Treat-To-Target in Type 2 diabetes), sponsored by NovoNordisk but, laudably, conducted and analyzed by the investigators at Oxford University. The 708 patients enrolled were slightly older and less obese than those in the original Treat-to-Target Trial but had similar duration of diabetes, prior oral therapies, and mean baseline A1c. They were randomized to three treatment methods: once or twice-daily detemir, twice-daily premixed 30:70 aspart/protamine-aspart, or aspart before each meal. Each arm had a specific glucose testing schedule and a computerized algorithm for adjusting dosage based on the results of these tests. In the basal insulin arm, detemir was given only in the evening unless high glucose values persisted in the late afternoon, in which case a second daily injection was to be added each morning. The primary endpoint of the first phase of the trial was the A1c level at 1-yr treatment, after which a 2-yr phase of further intensification of treatment for each arm will follow.
The 1-yr results confirmed many implications of the earlier studies. All three regimens improved glycemic control significantly, and a low 1-yr dropout rate (
6%) indicated good acceptance of insulin therapy. Severe hypoglycemia was uncommon, and few serious adverse events occurred. However, hypoglycemia was more frequent with the premixed and prandial regimens than with basal insulin. Similarly, the mean weight gain with premixed and prandial insulins (4.7 and 5.7 kg, respectively) was much greater than with basal insulin (1.9 kg). Referring to these findings, an accompanying editorial (8) concluded that "prandial and biphasic insulin formulations are suboptimal choices for insulin initiation" and that "the best approach is to continue metformin and add a basal insulin," as currently advised by American Diabetes Association/European Association for the Study of Diabetes consensus guidelines. Unfortunately, the results of 4T shed no light on whether continuing a sulfonylurea (>90% of the participants did so) contributed to or limited either glycemic control or hypoglycemia in any of the arms.
Other results were more surprising and, thus, informative. First, all the regimens failed to achieve mean glycemic control at 7.0% A1c or better. Mean A1c with prandial aspart was 7.2%, with aspart/protamine-aspart 7.3%, and with detemir 7.6%. All of these values were disappointing, but especially the A1c with detemir, which was much higher than observed in previous trials with detemir itself (4), and with NPH (2, 3, 4) and glargine (2, 3, 5). The stated 6.5% goal in 4T was seldom reached. In retrospect it was overly ambitious and, currently based only on epidemiological evidence, of uncertain relevance to clinical practice.
What went wrong? The authors rightly conclude that "most patients are likely to need more than one type of insulin to achieve target glucose levels," but this conclusion seems insufficient. The results of 4T suggest a flaw in its design, which in turn reflects a more general problem in the management of diabetes. The systematic titration algorithm was described only generally in the report and its online appendix. How dosing adjustments were electronically calculated on each occasion was not described in detail, and it is not clear how well the process was understood by the study site personnel and the participants themselves. Perhaps this computerized approach had an effect opposite to that intended. That is, rather than "empowering" the patient and reducing the burden on study staff, assignment of titration decisions to an obscure electronic system may have limited the ability of patients and providers to make appropriately individualized decisions when needed, leading to mediocre results from a one size fits all scheme.
This difficulty, compounded by incomplete information and incorrect assumptions about the properties of detemir, probably explains the poor results with this new insulin. The regimen called for once-daily injection of detemir, with an increase to twice daily only if it was proven necessary. When the trial began, detemir was thought to have a 24-h effect for most patients. However, later studies in the best available model do not confirm this assumption (9). In type 1 diabetic persons entirely lacking endogenous insulin, the mean duration of effect of substantial sc doses (0.35 U/kg), measured by the glucose infusion rate needed to maintain near-normal glucose levels, was about 16 h with detemir vs. over 24 h with glargine. Thus, twice-daily injection would be appropriate for many if not most patients taking detemir in 4T, but only 34% of them ended with two injections. Moreover, the titration algorithms assumed that all insulins would have similar bioavailability, or effect per unit injected. This too was incorrect. Detemir’s bioavailability is lower than that of other insulins. In two head-to-head clinical studies, 47 and 58% more units of detemir were needed to obtain the same level of glycemic control as with twice-daily NPH (4) or once-daily glargine (10). In 4T, the total daily dosages used were 0.49 U/kg detemir, 0.53 U/kg aspart/protamine-aspart, and 0.61 U/kg aspart. For effects equivalent to the other insulins, the mean dosage of detemir should have been closer to 0.7–0.8 U/kg. Detemir itself was not the problem, for it has been effective when used properly and, in addition, causes slightly less weight gain than other insulins (4, 10). But, because 4T’s algorithm for detemir began with wrong assumptions, too few participants used two injections, and the total dosage was too low. With dose titration assigned to an electronic system, human decisions failed to correct these problems.
The inability of the 4T algorithm to improve control to 7.0% A1c with detemir contrasts with the success in other studies of basal insulins. Yki-Jarvinen et al. (3) treated a Finnish population starting with higher A1c (mean 9.5%) with either NPH or glargine added to oral therapy, yet obtained mean A1c 7.2 and 7.1% with these insulins. Their titration scheme, like that in 4T, included a specific algorithm and an electronic system for handling glucose test results, but it also provided individualized advice by experienced personnel (3, 11). Similarly, in the original Treat-to-Target Trial, the site investigators were experienced endocrinologists who with their personnel were expected to interpret and modify the algorithm provided to optimize results and protect participants against severe hypoglycemia (2). Both NPH and glargine titrated in this way led to mean A1c 7.0%, albeit with more mild or moderate hypoglycemia with NPH.
How has 4T advanced our knowledge? It answered the first question posed previously: glycemic control can indeed be improved and sustained by any of the three simple insulin regimens, at least for some patients. Regarding the other questions, i.e. whether long-term treatment can be successful in a clinical practice setting with a simple algorithm and what to do for those patients who do not have a good response, the initial phase of 4T does not give us clear and favorable guidance. The problems with detemir as basal insulin highlight the need for more than just a basic, standardized algorithm that can be applied by an electronic system or by personnel with limited training. When the standardized procedure does not succeed, either because of inherent flaws in its design, or due to idiosyncrasies of an individual patient, it must be adapted and individualized by a competent provider. Apparently, the transition to insulin can begin with an algorithm but may not end with it. The need for decisions and teaching by providers, supplementing the algorithm, will be even more apparent when the next step of intensifying insulin therapy is taken, as in the next 2 yr of 4T. That is, when prandial insulin is added to basal insulin, or basal to prandial, or an additional dose of premixed or rapid-acting insulin to a twice-daily premixed regimen, further complexities of dosing adjustments will occur, and a one size fits all algorithm is likely to fall short. We can look forward to further insights on this point from 4T, but for the moment we must conclude that standardized protocols with electronic support cannot entirely replace professional expertise. We will have to find the resources to teach many patients to use insulin the old-fashioned way, using specialized diabetes providers to individualize treatment when algorithms alone fail.
 |
Footnotes |
|---|
First Published Online December 11, 2007
Abbreviation: NPH, Neutral protamine Hagedorn.
Received November 21, 2007.
Accepted November 29, 2007.
|
References |
|---|
 Top References |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
