Regarding intervention period, there were studies with short periods (eight weeks) [22] , as well as longer ones, with 52 weeks [14] [18] [19] [20] . Associating intervention period to the results found, a superiority of the long-term trainings is not perceived in comparison to the short-term ones, because similar reductions in HbA1c levels were shown after interventions with short periods (0.5% - eight weeks) [22] and medium terms (0.5% to 0.8% - 12 to 20 weeks) [15] [17] , with high variability in the results with long term interventions (0.27% to 1.21% - from 20 weeks) [14] [16] [18] [19] [20] [23] [24] [25] .

Although the field of T2DM and physical training does not provide studies comparing progressive combined training versus non-progressive combined training, we believe that studies with long term interventions may not expressively impact on glycaemic control because progressions were not conducted in the volume and intensity throughout the interventions. In this sense, it is important to highlight that two [13] [21] of the three studies [12] [13] [21] without

AT: aerobic training, RT: resistance training, CT: combined training group, CG: control group, n: number of participants, RM: maximum repetitions, 1RM: one maximum repetition test, rep: repetitions, ↓: reduction, HR_max: maximum heart rate, HR_res: reserve heart rate, HR_LT: heart rate at lactate threshold, HbA1c: glycated hemoglobin, VO_2peak: peak oxygen consumption, VO_2max: maximal oxygen consumption. %HbA1c: reduction in percentage, m.m.g.: main muscle groups.

significant alterations in HbA1c levels did not perform progression of the trainings, whereas Tessier et al. [12] carried out a slight increase in the intensity of the aerobic component. As these studies performed medium-term interventions periods of 12 [21] and 16 [12] [13] weeks, we speculate that the lack of training progression seems to be a failure in training prescription, violating the scientific principle of training overload. Also, in order to combine both aerobic and resistance stimuli in the same session, some studies avoid performing long training sessions, and therefore have a limited time available to perform both components (i.e. 20 min, 2 sets). Therefore, it seems that with 12 [21] and 16 [12] [13] weeks of follow-up, the training dosage of these interventions seemed to be low to elicit HbA1c adaptations. It is also important to highlight that another possible explanation for the absence of alterations in the study of Cuff et al. [13] , is that the participants had low baseline levels of HbA1c, which limited their amplitude of improvement in this outcome.

The importance of training frequency has been well documented, especially in aerobic training studies. In this direction, a meta-regression analysis conducted by Umpierre et al. [6] , evaluating the association between physical training intensity and volume and the changes in HbA1c levels in T2DM patients found that weekly frequency is the main determining factor in the glycaemic control by means of aerobic training for this population. Moreover, it is recommended that aerobic training is performed in at least three days of the week, with no more than two consecutive days between training sessions, due to the short duration nature of the improvements in insulin action [3] [4] . On the other hand, for combined training, the association between weekly frequency and glycaemic control is still not elucidated.

Regarding weekly frequency of the selected studies, two of them performed CT with aerobic and resistance trainings performed in the same days, twice a week [19] [20] ; eight studies performed resistance and aerobic trainings in the same days, three times per week [12] [13] [14] [16] [17] [21] [23] [25] ; one study performed both trainings in the same days, four times per week [22] ; one study performed resistance and aerobic trainings in alternated days, being twice a week for each one [15] ; one study performed one weekly session of aerobic training, one weekly session of resistance training and two weekly sessions combining both training types in the same session [24] ; and one study conducted aerobic training in three weekly sessions and resistance training twice a week [18] .

According to ADA [3] , physical exercise recommendations for T2DM patients are at least three weekly sessions of aerobic training and two weekly sessions of resistance training. All the interventions performed the minimum recommendation for resistance training. Eleven of the interventions [12] [13] [14] [16] [17] [18] [21] [22] [23] [24] [25] are in accordance with the minimum weekly frequency for aerobic training. Of these, only three studies did not find reductions in HbA1c levels [12] [13] [21] . On the other hand, although the other studies did not reach the minimum weekly frequency recommended for aerobic training [15] [19] [20] , all of them found reductions in HbA1c levels. These reductions could have resulted from other interventions characteristics performed, as the three studies performed progressions in volume and/or intensity throughout the interventions, which, according to Fleck and Kraemer [28] , can optimize the results, that is, these interventions may have generated new physiological stimuli across the interventions, consequently resulting HbA1c reductions.

Additionally, it is important to highlight that the studies which found the most expressive reductions in HbA1c levels (decreases of 0.95% to 1.21%, Table 1) were those conducting both aerobic and resistance trainings in the same days, in three and four weekly sessions [14] [19] [23] [25] . All these studies had progression of the training variables. Moreover, in two of these interventions [23] [25] a significant volume of exercises were performed for the resistance training component (14 exercises each, with 3 sets of 8 - 12 sets of 7 - 9 maximum repetitions, respectively), which is in accordance with the results found in the meta-regression of Umpierre et al. [6] , attributing the weekly volume of resistance training exercises to the HbA1c reductions resulting from CT.

Over the years, well designed studies have been attributing the success of physical training interventions not only to the modality but also to some training variables. In a meta-analysis conducted by Boulé et al. [9] , physical training intensity was associated to reductions in HbA1c levels in a greater extent than training volume. Later, Umpierre et al. [5] demonstrated that weekly duration was fundamental for HbA1c reductions, independently of the training modality. On the other hand, Umpierre et al. [6] showed that in aerobic training the only training variable associated to glycaemic reduction was weekly frequency, whereas any strength training variable was associated to glycaemic reduction and, in CT, only the number of sets in the resistance component presented this association.

Regarding the management of AT duration and intensity for T2DM patients, it is recommended that is must be performed for at least 150 minutes per week with moderate to high intensity, disseminated into three weekly sessions, with no more than two consecutive days between training sessions [4] . Of the 14 studies selected, four did not report session durations [13] [18] [20] [21] and only three studies did not find reductions in HbA1c levels [12] [13] [21] . Thus, we verified that the reality of these studies which reported the training durations does not contemplate ADA [3] guidelines for aerobic training duration. These findings corroborate the meta-regression analysis of Umpierre et al. [6] , which did not evidence association between AT duration and HbA1c levels reduction resulting from CT, being the RT number of sets the variable associated to glycaemic control through this training modality. Faced with that, it becomes necessary a further reflection about training guidelines for T2DM people, as other training characteristics (and not the training duration of 150 minutes) may be attributed to reductions in HbA1c levels of this population.

We believe that finding positive results with weekly durations shorter than 150 minutes in AT can positively impact on training adherence, which is fundamental for the success of physical training interventions. With the advancement of modern and busy life, with often long working hours, in which people have increasingly less free time to exercise, it is important that physical training is time-efficient to help individuals to reach their goals and keep maintaining an active and healthy life.

In the present review, the studies adopted different intensity markers. In AT, seven studies prescribed training intensity using percentages of maximum heart rate (HR_max) [12] [15] [16] [22] [23] [24] [25] , three used percentages of heart rate reserve (HR_res) [13] [14] [17] , three used percentages of maximum oxygen consumption (VO_2max) [18] [19] [20] and one study conducted it through the heart rate corresponding to lactate threshold (HR_LT) [21] . Of the selected studies, those that used the intensities prescribed between 60 and 75% of HR_max (15 to 45 min/session) [25] , 70% to 75% of HR_max (25 to 45 min/session) [23] , 70% to 80% of VO_2max (40 min/session) [19] , and the intensities between 40% and 80% of HR_res (30 min/session) [14] for AT were those that found the greater reductions in HbA1c levels.

It is important to highlight that, among the articles included in the present review, 10 of them [12] [15] [16] [18] [19] [20] [22] [23] [24] [25] used percentages of HR_max and VO_2max for AT prescription. It is well known that the exercise performed in a certain percentage of VO_2max or HR_max can induce different physiologic responses for different individuals, and that submaximal parameters as anaerobic threshold are better indicators for aerobic training prescription [29] . Moreover, the anaerobic threshold determination method through heart rate deflexion point, which has low cost, easy application and was associated to the second ventilatory threshold in T2DM patients [30] , was not used in any of the reviewed studies. Therefore, we emphasize that there are other possibilities with greater metabolic precision, which have not been used in studies with T2DM patients.

In relation to the management of RT intensity, seven studies performed it by means of percentages of one repetition maximum (1RM) test [14] [15] [17] [19] [20] [22] [24] , two used 7 to 9 maximum repetitions [16] [25] , one adopted 10 to 12 maximum repetitions [18] , one used 8 to 12 maximum repetitions [23] and three studies did not specify the way that RT intensity was conducted [12] [13] [21] . Most of the studies maintained the intensity fixed throughout all the interventions periods, whereas five studies progressed RT intensities at some point of the intervention [12] [15] [17] [20] [23] . Of these, only one [12] did not find significant alterations in HbA1c levels. Regarding the best results found [14] [19] [23] [25] , the intensities between 7 to 9 maximum repetitions, 8 to 12 maximum repetitions, 80% of 1RM and from 40% to 60% of 1RM were respectively used.

The selected studies for the present review adopted different numbers of exercise for the resistance training component of the CTs. The number varied from two exercises (one for upper and one for lower limbs) [17] , six exercises for the big muscle groups [14] [22] to fourteen exercises [23] [24] [25] , also for large muscle groups.

Although the ADA indications are of 8 to 12 resistance exercises prioritizing the big muscle groups [3] [4] , an analysis of this parameter in the reviewed studies demonstrates that this is not a determining factor for glycaemic control, because studies with similar number of exercises induced different responses. It becomes clear by analyzing the number of RT exercises in the four studies with the best results in HbA1c, because whereas Stegen et al. [25] and Larose et al. [23] performed 14 exercises in their interventions, Balducci et al. [14] performed six exercises and Balducci et al. [19] performed five exercises. Thus, we believe that other training variables, as intensity, number of sets and repetitions may have greater relevance in the combined training programs structuring, as well as training variables progression, aiming to periodically induce new physiological adaptations in T2DM individuals, and consequently the glycaemic control for this population.

In the reviewed studies, different manners of performing the order of resistance and aerobic trainings were applied. Of these, six studies [13] [14] [16] [19] [22] [25] performed AT before RT in the same training sessions; three studies [12] [17] [21] conducted both training modalities in the same session, in circuit; one study [15] performed AT and RT in separate days; one study [24] used one weekly session for AT, one weekly session for RT and two weekly sessions with both training types, without specifying the order; one study [23] allowed the participants to choose the training order; whereas the other studies [18] [20] did not report the training order. Of the four studies that showed the best results in the glycaemic control, three of them performed AT before RT in the same sessions [14] [19] [25] and one of them allowed the participants to choose the order [23] .

CT has been widely explored in the literature, and in this context, the interference effect has also emerged, which explains that the expected results from RT suffer interference of AT, that is, the strength development is impaired when AT precedes RT [31] , whereas cardiorespiratory fitness does not suffer the same impairment when it is performed after RT [32] . The order effect has not yet been investigated in T2DM population, therefore, there is a doubt whether the training order could interfere in the interventions results, if some order could maximize glucose uptake and possibly result in a better glycaemic control. In addition, we also emphasize that the training order could be investigated regarding training adherence, which is a very important issue to be considered in the clinical population. Besides the training order be pre-established by physical exercise professionals, there is also the possibility of the patient to choose the order, which could also positively influence on adherence to interventions.

In the present review, three studies compared CT with AT [13] [17] [19] , one study compared CT with RT [23] , and five studies compared CT with both aerobic and resistance trainings separately performed [16] [18] [21] [24] [25] . Of these, only one study did not find significant results in HbA1c levels in both aerobic and resistance trainings [13] .

In the intervention conducted by Lambers et al. [17] , comparing AT and CT groups, only CT found HbA1c reductions, whereas the other studies [16] [18] [19] [21] [24] [25] found greater reductions in HbA1c levels with CT in comparison to other groups that performed AT or RT alone. These results show that every time that CT was compared with AT or RT isolated, CT induced greater results. Other important issue is that in the study of Church et al. [18] performed with three intervention groups (AT, RT and CT) only CT group participants underwent reduction in HbA1c values. Although this decrease was not very expressive, this reduction was significant, showing that in a long training period (52 weeks), in which there was not progression of the training variables, it seems that only the combination of aerobic and resistance stimuli was able to sustain the gains reached in the first months of training.