Our study found that the IR patients show a reduced CP performance compared with the normal population. Most importantly with respect to the primary endpoint physical exercise treatment alone produce an improvement in oxygen consumption compared with MET therapy alone, whereas MET in association with physical activity does not improve O2 consumption in comparison to therapy with physical exercise alone. Moreover only physical exercise treatment alone produce an improvement in the Aerobic Threshold with respect to MET therapy alone and MET in association with physical activity.
The data referring to the secondary endpoint evidenced that physical exercise, alone or in combination with MET, is superior to MET treatment alone in improving all indices of HRQoL, whether physical or mental. We found no differences on the indices of HRQoL between physical exercise alone and in combination with MET.
In addition, our study confirmed that the treatment with MET decreases the peak VO2 and the ability to work, as already demonstrated by our group in a previous study . Exercise not only improved the CP performance when used alone, but it was able to cancel the negative effect of MET, when used in combination with the drug.
Peak VO2, the maximum capacity of the body to use oxygen, identifies the highest potential for an individual to perform aerobic work. This parameter is influenced by age, sex and level of training, as well as by the presence of disease or drugs capable of affecting any one of its components.The use of exercise capacity as a powerful prognostic factor in normal subjects is widely acknowledged. After adjustment for age, peak exercise capacity measured in metabolic equivalents was the strongest predictor of the risk of death in healthy populations : each metabolic equivalents increase in exercise capacity conferred a 12 percent improvement in survival. A reduced peak VO2 at baseline in the subjects investigated in the present study may be interpreted as a depressive effect induced by IR per se on the CP function, as evidenced in our previous experience . In agreement with this hypothesis, a recent community-based study showed that IR predicts the incidence of heart failure, independently of the known risk factors, including diabetes .
Several animal models have shown a myocardial metabolic incompetence which could lead to an IR cardiomyopathy . In fact, IR determines in myocytes an energy inefficiency as a result of an increased use of fatty acids, energetically less efficient when compared to glucose. Furthermore, myocytes are unable to utilize glucose in situations of stress, as occurs in healthy subjects. These mechanisms, together with the endothelial dysfunction seen in IR subjects , can contribute to myocardial dysfunction [25, 26] and the reduction of CP performance that we observed.
In the Diabetes Prevention Program, MET showed the ability to slow the progression from a state of impaired glucose tolerance to T2DM [27, 28]. However, several adverse events were attributed to the drug. MET was recognized responsible for a reduction in oxygen consumption in healthy individuals without IR in comparison with controls treated with placebo . In vitro studies found that MET exerts its anti-diabetic effects through inhibition of complex 1 of the mitochondrial respiratory chain (electron transfer from NADH to coenzyme Q 10) [29, 30]. This inhibition can slow down the transfer of reducing equivalents during the Krebs cycle and limit the ability of oxidative metabolism. In large muscle groups, the mitochondrial reserve may be used to some extent during exercise. However, if an inhibition of complex 1 reduces this reserve, a critical decrease in CP performance can result. Our previous data showed that MET reduces the average consumption of oxygen for a small but significant extent in IR subjects . It is of particular interest to note that this depressive effect does not manifest itself in all subjects, being present only in individuals with lower degree of IR, while the CP performance is improved in patients with high and very high IR.
Results from the DPP  and other similar studies demonstrated that changes in lifestyle, such as a proper diet and increased physical activity, play a crucial role in preventing or slowing the progression from a state of IR to a frank diabetes [31, 32]. In the presence of this evidence, a combination of lifestyle changes and MET therapy has been suggested as the best strategy to control IR and prevent diabetes. The data currently available suggest a benefit of the association in terms of weight loss , but a little if any effect in reducing IR , These findings, although preliminary, demonstrate how the clinical effects of a combination of exercise and MET are complex and difficult to determine in advance. They confirm the general assumption that exercise-drug interactions cannot be predicted from their individual effects and should be considered systematically to provide information usable in the field of public health. Malin et al. indeed demonstrated that exercise and MET are both able to improve insulin sensitivity after 12 weeks of therapy, in men and women with pre-diabetes. But the addition of MET to physical training did not improve IR and, rather, could have obscured the full benefit of exercise .
In the present study, both the exercise associated with MET and MET alone led to a significant weight loss, while the exercise used alone did not have this effect. These data agree with those previously published in the literature, according to which MET is able to determine by itself a weight loss. Our findings demonstrate, to our knowledge for the first time, that treatment with MET in combination with physical exercise is not superior to exercise alone, in terms of improving CP performance. Conversely, the addition of a tailored and supervised training program in subjects with IR is crucial to reverse the adverse effects of MET on the oxygen consumption and allow a greater adherence to therapy.
Various and mostly little-known mechanisms underlie the MET/exercise interaction and make the effects of this therapeutic association unpredictable. Both exercise and MET act on protein kinase activated by the AMP (AMPK) expression  and on the transduction pathway of the AMPK–eNOS signal. Metformin, an AMPK activator which can act as an exercise mimetic , had been shown to improve exercise in women with angina  and to improve the endothelial flow reserve .
In our study, as previously observed in the study of Malin et all , BMI was significantly lower in the MEx group and in the M group and not in the Ex group. This result may be related to a different effect of metformin and exercise on fat mass and lean body mass distribution.
The last, but not less important remark concerns the net effect of these therapies on HRQoL related to physical, mental and general well-being. The data obtained allow us to confirm the benefit of exercise, without further advantages obtained by its combination with MET. It is our belief that this latest finding is extremely important in a population of relatively young subjects, which, although at high risk of developing diabetes and cardiovascular disease, must however be considered healthy. When MET is employed, it should always be associated with physical exercise to counteract the negative effects of drug therapy on CP performance, contribute to improving the HRQoL and increase patient compliance, which, as is known, is a crucial component in their clinical management.
The main limitations of the study are related to the study design. A single blinded design was chosen due to the difficulties of concealing the exercise. Moreover since the primary endpoint of our study was to verify the effect of exercise alone or in combination to metformin on the maximal aerobic capacity in comparison with the patients not performing exercise a “per protocol analysis” was chosen as more appropriate. Although removing the patients who were unable to regularly follow the physical training program from the final analysis introduces potential bias.