ObjectivesThe objective of this study was to establish and validate an in vitro-in vivo correlation (IVIVC). To investigate the safety of a fixed-dose combination (FDC) versus the reference formulations (Januvia® 100 mg Filmtabletten co-administered with Glucophage® SR 1000 mg prolonged-release tablets), a bioequivalence study was conducted in the fasted and fed states, and the data generated were used to establish the correlation.Materials and MethodsThe formulations used in the bioequivalence study were a FDC (sitagliptin hydrochloride equivalent to 100 mg of sitagliptin and metformin hydrochloride 1000 mg prolonged release) and Januvia® 100 mg co-administered with Glucophage® SR 1000 mg. The plasma profiles from the bioequivalence study and respective dissolution data were then utilized to establish "level A" IVIVC. The procedure comprises pharmacokinetic modeling to derive the empirical constants for further use in deconvolution and convolution procedures. Levy plots were constructed to understand the relationship between in vitro and in vivo properties. The internal and external predictabilities were evaluated by comparing the predicted pharmacokinetics with the observed values from the bioequivalence study.ResultsThe formulations showed approximately 91%-95% and 89%-91% dissolution, respectively in fasted and fed-state dissolution media for sitagliptin. The dissolution of metformin was 96%-98% and 89%-95%, respectively, in fasted and fed-state media. The regression coefficients of all the Levy plots were more than 0.900, indicating a linear correlation between in vitro release and in vivo parameters. The prediction error value of internal and external predictabilities was below 10 and met the US Food and Drug Administration criteria. Therefore, it can be stated that the correlation models are validated and can be used for predictions and to setting the dissolution specifications. The safety and tolerability of the FDC was found to be superior to those of the reference formulations, as fewer adverse events occurred following administration of the FDC.ConclusionCorrelation models can be useful for the prediction of FDCs during the management life cycle of the product. The models can also serve as a surrogate for in vivo studies. The FDC was tolerable, and the adverse events were mild and similar to those observed with the reference products. Therefore, the FDC is safe for use in human subjects.