Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Dipeptydil peptidase-4 inhibitors in type 2 diabetes

Larry H. Bernstein, MD, FCAP, Curator

LPBI

Dipeptydil peptidase-4 inhibitors in type 2 diabetes: A meta-analysis of randomized clinical trials

Oral Dipeptidyl Peptidase-4 (DPP-4) inhibitors sitagliptin [1] and vildagliptin [2], which increase circulating levels of Glucagon-Like Peptide-1 (GLP-1), have recently been approved for use in type 2 diabetes; other molecules of the same class (such as saxagliptin and alogliptin) are under development.

The role of those new drugs in the treatment of type 2 diabetes is debated. The consensus algorithm of the American Diabetes Association and the European Association for the Study of Diabetes [[3], [4]], in its revised version [4], suggests limiting the use of GLP-1 receptor agonists and DPP-4 inhibitors only in some specific cases, without considering those agents in the mainstream (“Tier 1”) of the algorithm. Conversely, DPP-4 inhibitors are not even included as a second choice, although their use is contemplated in selected patients. The reasons for this exclusion are their perceived limited efficacy on HbA1c in comparison with other agents, their poorly defined safety profile, and their cost [[3], [4]].

Efficacy and safety need to be assessed through a comprehensive review of currently available clinical trials. Some detailed reviews of published studies have been recently published [[1], [2], [5]]; furthermore, some meta-analyses have been performed [[1], [6], [7], [8]]. However, currently available meta-analyses included only published studies, without any attempt at retrieving data from completed and publicly disclosed, although not formally published, clinical trials. Furthermore, several trials have been published in the last few months, increasing in a relevant manner the available data base for the assessment of the clinical profile of DPP-4 inhibitors.

The aim of the present study is to offer a comprehensive and updated synthesis of all available clinical data on the safety and efficacy of DPP-4 inhibitors.

The trial flow is summarized in Fig. 1, and the characteristics of the trials included in the meta-analysis are summarized in Table 1. Among the trials included, 32 were described in publications in peer-reviewed journals; results of 9 unpublished trials were disclosed on different websites. Furthermore, 10 unpublished trials, the results of which were undisclosed, could be identified (Table 2). Notably, results could be retrieved for the large majority of trials on currently available DPP-4 inhibitors (sitagliptin and vildagliptin), while only results of preliminary phase II studies were available for products currently under development (saxagliptin).

Figure 1

Trial flow diagram. RCT: randomized clinical trial.

View Large Image | Download PowerPoint Slide

The Begg adjusted rank correlation test (Kendall tau: −74; p=0.13) and the Egger regression approach (intercept, −2.81 [CI, –6.91–1.27]) suggested no major publication bias.

…………….

Figure 2

Standardized differences (with 95% CI) of mean HbA1c at endpoint.

View Large Image | Download PowerPoint Slide

Safety: hypoglycemia

Information on hypoglycemia could be retrieved for 15 out of 18 trials with sitagliptin, and 19 out of 21 with vildagliptin. In some trials (1302 [14], 1303 [14], PN-035 [15], PN-036 [15], and PN-044 [15]), the overall number of hypoglycemic episodes was not reported. Of the remaining 34 studies, 5 (2 with sitagliptin and 3 with vildagliptin) reported that no hypoglycemic event had occurred during the trial. In the remaining studies, hypoglycemia was observed in 184 of 9944 patients treated with DPP-4 inhibitors (103 of 4573, 81 of 5100, and 0 of 271 with sitagliptin, vildagliptin and saxagliptin, respectively) and 293 of 5698 in comparator groups. The incidence of hypoglycemia with DPP-4 inhibitors (both sitagliptin and vildagliptin) was not significantly different from that observed with a placebo, even when those agents were used in combination with sulphonylureas or insulin (Fig. 3). In direct comparison, DPP-4 inhibitors were associated with a significantly lower hypoglycemic risk than sulphonylureas, whereas no significant differences were detected with respect to thiazolidinediones (Fig. 3). The number needed to harm (NNH) for sulphonylureas in comparison with DPP-4 inhibitors, with respect to hypoglycemia, on an yearly basis, was 128.

Figure 3

Mantel–Haenszel odds ratio (with 95% CI) for any hypoglycemia (logarithmic scale).

View Large Image | Download PowerPoint Slide

All trials with DPP-4 inhibitors reported information on severe hypoglycemia, except 10 (6 with sitagliptin, 4 with vildagliptin). In three distinct trials [[16], [17], [18]], severe hypoglycemia was reported by 5 patients treated with sitagliptin, in monotherapy (N=4) or in combination with metformin (N=1), versus 9 cases in comparator groups (all treated with sulphonylureas). No case of severe hypoglycemia was reported in patients treated with vildagliptin, versus 5 with comparator (placebo in all cases); the severe hypoglycemic episodes detected in vildagliptin trials occurred in combination studies with sulphonylureas [19] or insulin [20].

……………………….

DPP-4 inhibitors have been proposed as an alternative to currently available therapies (sulphonylureas, thiazolidinediones or insulin), mainly as an add-on treatment in patients failing with metformin monotherapy. However, even the most recent version of the ADA–EASD consensus algorithm does not consider these drugs a viable option, except for selected cases [4]. The reasons for exclusion from the main treatment algorithm are scarce efficacy, limited amount of available evidence and high cost. With respect to available evidence, it should be recognized that several trials, which had not been included in previous meta-analyses [6], have been recently published [[19], [21], [22], [23], [24], [25]]. Furthermore, there are a relevant number of unpublished trials, the results of which have been disclosed on different websites, and are therefore available. The decision to publish a trial is, in most instances, performed by the sponsor which has a specific interest in pursuing the greater safety and tolerability of the new drug. This bias is unfortunate and limits the reliability of this and other meta-analysis, often based only on data provided from manufacturers; however, the retrieval of all available information should always be attempted, although the possibility of including some information of poorer methodological quality should be taken into account. The overall amount of evidence from randomized clinical trials which can be retrieved using this comprehensive approach is relevant, and probably sufficient for a reliable assessment of the clinical profile of this new class.

The overall efficacy on HbA1c of DPP-4 inhibitors in placebo-controlled trials is similar to that reported in previous meta-analyses [[1], [6], [7], [8]]. However, the greater number of available studies allowed separate analyses of trials in which DPP-4 inhibitors were used either as monotherapy or as an add-on to other agents. In fact, most currently available hypoglycemic treatments show a smaller additional effect on HbA1c when used as an add-on to metformin, in comparison with monotherapy trials [26]. Conversely, DPP-4 inhibitors produce a similar placebo-subtracted reduction of HbA1c either in monotherapy or as an add-on to other agents. This pattern resembles that of other drugs specifically active on post-prandial glucose, such as acarbose or glinides [26]. In fact, DPP-4 inhibitors, as well as GLP-1 receptor agonists, show a relevant effect on post-prandial hyperglycemia. Although data on post-prandial glucose measured through self-monitoring were not available, the results obtained in many trials with meal tests [[7], [27]] support the hypothesis of a specific action of DPP-4 inhibitors on post-prandial hyperglycemia.

Based on the considerations reported above, DPP-4 inhibitors, when used in combination with other drugs, should not be expected to be less effective on HbA1c than other agents (such as sulphonylureas, thiazolidinediones or insulin). Unfortunately, only a small number of head-to-head comparisons with other drugs are currently available. The efficacy of DPP-4 inhibitors on HbA1c, either in monotherapy or in combination with metformin, appears to be somewhat smaller than that of sulphonylureas, and similar to thiazolidinediones; the only two available comparisons with metformin, both in monotherapy, one with vildagliptin [28] and one with sitagliptin (PN-036 on www.merck.com/mrl/clinical_trials/results.html) suggest a smaller effect on HbA1c. It should be considered that most trials are of a relatively short duration and it is possible that sulphonylureas, which are known to produce a less durable effect on glucose than other available agents, [29] could provide less favorable results in the long-term.

Taken together, the present results on efficacy do not support the use of DPP-4 inhibitors in monotherapy as an alternative to metformin. On the other hand, these drugs appear to be effective as add-on treatments in patients failing with metformin monotherapy, with a specific effect on post-prandial glucose, although the short-term efficacy of sulphonylureas on HbA1c could be greater than that of DPP-4 inhibitors.

With respect to body mass index, this meta-analysis confirms the neutrality of DPP-4 inhibitors [[1], [6], [7], [8]]. In direct comparison, DPP-4 inhibitors appear to have an advantage in this respect over thiazolidinediones.

GLP-1 stimulates insulin secretion and inhibits glucagon production in a glucose-dependent manner, i.e. its effects are blunted when blood glucose reaches the lower limits of the normal range [30]. Therefore, DPP-4 inhibitors are expected to reduce glycemia with a low hypoglycemic risk. In fact, DPP-4 inhibitors do not induce any additional risk, in comparison with a placebo, either in monotherapy or in combination with sulphonylureas or insulin. This confirms the results of a recent meta-analysis performed on patient-level data from randomized clinical trials with sitagliptin [31]. Interestingly, in the only trial performed in insulin-treated patients, vildagliptin reduced the incidence of hypoglycemia in comparison with a placebo [20]. The mechanisms underlying this phenomenon need to be further elucidated. As expected, DPP-4 inhibitors do not increase the incidence of hypoglycemic episodes when compared with insulin-sensitizing drugs; on the other hand, they show a markedly reduced risk of hypoglycemia in head-to-head comparisons with sulphonylureas. This difference, which could be partly determined by a marginally greater efficacy of sulphonylureas on HbA1c, is consistent with the different mechanisms of action of the classes of drugs.

No patient experienced severe hypoglycemia during vildagliptin therapy. Unexpectedly, episodes of severe hypoglycemia occurred in five patients treated with sitagliptin, either in monotherapy or in combination with metformin, in three different trials [[16], [17], [18]]. Notably, two of those trials [[16], [18]], although published, did not report those events but since those trials were included in the registration data for drug approval in the US, the information on severe hypoglycemia can be retrieved from the FDA website. Furthermore, episodes of severe hypoglycemia were not considered in a recent meta-analysis of trials with sitagliptin, although a greater number of such events had occurred in comparator groups, which included sulphonylureas [31]. It should also be considered that some of the trials did not report any information on severe hypoglycemia, raising the possibility of a selective reporting bias. The occurrence of cases of severe hypoglycemia with DPP-4 inhibitor monotherapy is difficult to explain on the basis of the current knowledge of the mechanism of action of those drugs, and deserved further investigation.

Among other expected adverse events, the previously reported increased incidence of some infections during DPP-4 inhibitor therapy [[6], [8]] is confirmed, with sitagliptin, but not vildagliptin, associated with nasopharyngitis, and with a nonsignificant trend toward an increased risk of urinary tract infections. These results are consistent with those of a recent meta-analysis on patient-level data from trials with sitagliptin, which included only a fraction of the studies summarized in the present meta-analysis, and which showed a similar trend toward the increase of risk of nasopharyngitis with the DPP-4 inhibitor, although it failed to reach statistical significance [31]. It should be considered that DPP-4 is involved in the interaction between immune cells and that it could therefore modulate immune responses [32]; however, there is no evidence from mechanistic studies that inhibition of DPP-4 with currently available agents has an immunodepressant effect. Consistently, treatment with DPP-4 inhibitors does not appear to increase the risk of infections other than nasopharyngitis and urinary tract infections.

The introduction of a new class of drugs which are designed for long-term use always raises some concerns about safety during prolonged treatment. The possibility of rare, unexpected serious adverse events, which could not be detected in registration trials, should be considered. The number of reported deaths in available trials is still very small; however, there is no evidence suggesting an increase in mortality during treatment with DPP-4 inhibitors. The number of cardiovascular events registered in clinical trials is remarkably greater, although still inadequate to detect minor differences between groups. The two drugs which have been more thoroughly studied (sitagliptin and vildagliptin) do not seem to be associated with increased cardiovascular risk; in fact, the actual risk is lower than with comparators, although differences do not reach statistical significance. In fact, available data do not rule out the possibility of an increase of cardiovascular risk up to 28%, or of a reduction up to 54%. It should be considered that the duration of the available trials (up to one year) is insufficient to detect any effect of treatment (either detrimental or beneficial) on atherogenesis.

The addition of unpublished trials does not substantially modify the estimates of efficacy of DPP-4 inhibitors. However, the retrieval of unpublished, but publicly disclosed, information allowed the identification of some potentially interesting phenomena, such as cases of severe hypoglycemia with DPP-4 inhibitor monotherapy, which could not be detected in published papers.

The limitations of the present meta-analysis should be recognized and considered when interpreting the results. The analysis was performed on summary data, therefore lacking the accuracy of assessment which can be obtained when using patient-level data. For the very same reason, a time-to-event analysis for categorial outcomes (including cardiovascular events) could not be performed; the proportion of patients experiencing at least one event during the trial, which was used for meta-analysis, approximates the actual incidence of events only if this incidence is assumed to be constant throughout the duration of the trial. Furthermore, the number of subject studies and the duration of trials performed is insufficient to draw any definitive conclusion on the long-term cardiovascular safety of DPP-4 inhibitors.

In conclusion, DPP-4 inhibitors are effective in reducing HbA1c and post-prandial glucose; when used as an add-on to metformin, they show a medium-term efficacy on HbA1c similar to thiazolidinediones and marginally inferior to sulphonylureas, with a reassuring short- and medium-term safety profile. In fact, the hypoglycemic risk is low, and there is no evidence of detrimental effects on cardiovascular disease. In comparison with sulphonylureas or insulin, which have been proposed as first-choice agents in patients failing with metformin [4], DPP-4 inhibitors exhibit, at least in the short- and medium-term, a lower hypoglycemic risk and a more favorable action on body weight, at the price of a somewhat smaller efficacy and higher cost. The choice of the drugs to be used as add-ons to metformin in monotherapy failure largely depends on the relative weight attributed to each of these three components (safety, efficacy on HbA1c and cost).