Meta-analysis on insulin treatment for type 2 diabetes: rushed conclusions?
KeywordsGlycaemic control HbA1c Hypoglycaemia Insulin therapy Intention to treat Randomised clinical trials Titration
Treating to Target in Type 2 Diabetes
To the Editor: We read the meta-analysis by Lasserson et al. , published in Diabetologia, with interest. However, there are several issues relevant to the conclusions of this paper that are worth pointing out.
Most of the conclusions are based on five studies for the comparison of basal vs biphasic insulin, seven studies for the comparison of prandial vs basal (two of which were included in the previous comparison) and three for the comparison of prandial vs biphasic; certainly not the 22 randomised clinical trials and more than 4,000 participants described in the abstract and methods section. This may be somewhat misleading to readers.
The quality of the studies included in the meta-analysis was low. Given the fact that none was blinded, the maximum expected score for these studies would be four (three out of the five Jadad points , plus one for intention-to-treat analysis), but the authors give five points to two of these studies.
Two of the authors (R. R. Holman and A. J. Farmer) are also co-authors of the largest study (Treating to Target Type 2 Diabetes [4-T])  included in the meta-analysis, and longer follow-up results of this study  have recently been published that are quite different from those given by Lasserson et al. . Indeed, given the weight of the 4-T study in the meta-analysis, if the 3 year results were to be included, the superiority of the other two patterns of insulin administration relative to basal insulin would probably disappear.
The insulin doses used in the basal regimen were consistently lower in the studies reporting this information. Little emphasis is given to this fact in the discussion, even though it is probable that under-titration of basal insulin plays a role in the results obtained. Indeed, this is something that has been pointed out previously for the 1 year results of the 4-T study .
All basal insulins are included in the same analysis, despite the fact that higher doses of insulin detemir [B29Lys(ε-tetradecanoyl),desB30 human insulin] than NPH insulin  or insulin glargine (A21Gly,B31Arg,B32Arg human insulin)  are needed to obtain equivalent effects.
Finally, type 2 diabetes is characterised by the progressive loss of insulin secretion. Thus, the time-point in the development of the disease is an aspect to be taken into consideration when choosing the optimal therapy for each patient. Basal insulin administration would be chosen for individuals with a high level of fasting glycaemia, whereas regimens based on quick-acting insulins would be expected to cover prandial hyperglycaemia. Furthermore, individual patient features, titration and meal distribution are central for achievement of glycaemic optimisation and for avoidance of hypoglycaemia. Individual features have to be taken into consideration in any treatment, but a tailored algorithm is especially important in the context of insulin administration.
A. M. Wägner is being supported by grants from the European Foundation for the Study of Diabetes (EFSD)/ Novo Nordisk/JDRF 2008 Programme for Type 1 Diabetes, Instituto de Salud Carlos III (PI08/1113) and Fundación Canaria de Investigación y Salud (FUNCIS-PI 16/07), and D. Mauricio, from Instituto de Salud Carlos III (PI 061104)
Duality of interest
A. M. Wägner has received support from Novo Nordisk, Eli Lilly and sanofi-aventis for participation in scientific meetings. D. Mauricio has received speaker’s fees or has participated in Advisory Board activities for Eli Lilly and sanofi-aventis. The authors are not aware of any other conflicts of interest related to the contents of the manuscript.
- 7.Rosenstock J, Davies M, Home PD, Larsen J, Koenen C, Schernthaner G (2008) A randomised, 52-week, treat-to-target trial comparing insulin detemir with insulin glargine when administered as add-on to glucose-lowering drugs in insulin-naive people with type 2 diabetes. Diabetologia 51:408–416CrossRefPubMedGoogle Scholar