The problem with Thermo Fisher Scientific's Kinetica bioequivalence software, which is used during clinical trials to show whether different drugs with the same active ingredient have the same effect in the body, was highlighted in a paper published last month in the journal of the American Association of Pharmaceutical Scientists (AAPS). The EMA told Reuters that its Committee for Medicinal Products for Human Use (CHMP) would investigate the issue after the authors of the paper brought it to their attention.
Commenting on the findings, one of the authors of the paper, Anders Fuglsang, said it was "an unprecedented situation with unclear implications". Fuglsang, a former medicines regulator, now runs a pharmaceutical consultancy in Denmark.
The CHMP's Pharmokinetic Working Party will be meeting next week to decide on the EMA's "next steps", according to Reuters.
Kinetica is one of the most widely-used statistical software packages for analysing bioequivalence in clinical trials, and is used by both pharmaceutical companies and contract research organisations. Around 80% of bioequivalence studies relate to the approval of cheaper, generic versions of medicines after the patents on the originals expire.
According to Reuters, Fuglsang and his colleagues found that Kinetica gave results that were inconsistent with those from other software packages when there were more test subjects in one sequence of treatments than another. Although the inconsistency was not itself a cause for concern, the authors went on to find that the Kinetica software package was producing unreliable "confidence intervals". These are used as an indication of how a clinical trial result would be reflected in the general patient population outside of a clinical trial.
The extent of the potential issue is not yet known, but could mean that suitable drugs were rejected or others approved on mistaken information. The software manufacturer, Thermo Fisher, told Reuters that it was "currently conducting a careful review" of the issues raised in the paper.