Is the Efficacy of Milnacipran in Fibromyalgia Predictable? A Data-Mining Analysis of Baseline and Outcome Variables

Minalcipran has been approved for the treatment of fibromyalgia in several countries including Australia. Australian agency considered that the overall efficacy is moderate, although clinically significant, and could be translated into a real and strong improvement in some patients. The determination of the characteristics of patients who could benefit the most from milnacipran (MLN) is the primary objective of this manuscript.

Data from the 3 pivotal phase 3 clinical trials of the Australian submission dossier were assembled into a database. A clustering method was implemented to exhibit natural groupings of homogeneous observations into clusters of efficacy outcomes and individual patients. Next, baseline characteristics were investigated using a data-mining method to determine the clinical features that may be predictive of a substantially improved effect of MLN on a set of efficacy outcomes.

The clustering analysis reveals 3 symptom domains: "Pain and global," "Mood and central status," and "Function." We show that improvement in "Fatigue" goes with improvement in "Function." Furthermore, the predictive data-mining analysis exhibits 4 single baseline characteristics that are associated with a substantially improved effect of MLN on efficacy outcomes. These are high pain intensity, low anxiety or catastrophizing level, absence of major sleeping problems, and physical limitations in the daily life effort.

Clustering and predictive data-mining methods provide additional insight about fibromyalgia, its symptoms, and treatment. The information is useful to physicians to optimize prescriptions in the daily practice and to regulatory bodies to refine indications.