This systematic review and meta-regression covered 50 years of data from 5973 participants and investigated the impact of exercise training variables on skeletal muscle mitochondrial biogenesis and capillarization. Our main findings were as follows: (1) larger training volumes (higher training frequency per week and larger number of training intervention weeks) and higher training intensities (per hour of training, SIT > HIT > ET) are associated with greater increases in mitochondrial content and \(\dot\)O2max. (2) SIT is effective in improving mitochondrial content and \(\dot\)O2max in the early stages of exercise training. In contrast, ET and HIT show slower initial increases but continue to improve steadily over a greater number of training weeks. (3) Gains in capillarization occur primarily in the early stages of training (< 4 weeks) and are observed only in untrained to moderately trained participants. Capillaries per fiber (CF) increase similarly with ET, HIT, and SIT, while capillaries per mm2 fiber area (CD) increase only after ET and HIT, with ET showing larger increases compared to HIT and SIT. (4) Generally, responses to exercise training are largely determined by the initial fitness level and are not influenced by age, sex, disease, and menopause. However, women display larger percentage gains in VO2max compared with men.

In the past five years, a debate coined “the mitochondrial content contest” by Fiorenza and Lemminger [39] has emerged in the scientific community focusing on whether exercise training intensity or volume is the most important exercise training variable for promoting increases in mitochondrial content [8, 10]. Granata and colleagues [11] pooled the results from 58 exercise training studies and found no significant association between exercise training intensity (% of Wmax) and change in mitochondrial content, as assessed by either CS activity or MvD. Instead, they highlighted a positive relationship between total exercise training volume, which they defined as the exercise intensity multiplied by the total training duration (which in the current study is defined as training load, see explanation below), and changes in mitochondrial content. This led them to advocate for exercise training volume as the primary driver for increasing mitochondrial content [10, 11]. In response, MacInnis et al. [8] contend that exercise training intensity is the most important variable mediating exercise-induced increases in mitochondrial content within a fixed training duration. In support of their view, they emphasize that (1) findings from acute exercise studies where mRNA expression of peroxisome proliferator activated receptor ƴ coactivator 1α (PGC-1α), a key regulator of mitochondrial biogenesis, increases to a larger extent following HIT than work-matched low-intensity exercise [40], and (2), findings from longitudinal studies comparing changes in mitochondrial content show larger increments in mitochondrial content after HIT compared with work-matched, moderate-intensity continuous exercise training (e.g., [41]). Furthermore, they point out studies comparing low-volume SIT with higher volume of moderate-intensity continuous exercise, where similar mitochondrial adaptations were observed (e.g., [42]). In the present study, we adopted a slightly different and more nuanced statistical approach compared with previous research. Our aim was to investigate the distinct contributions of each factor constituting the training load: training intensity, training frequency per week, and the number of training intervention weeks. In each analysis, we controlled for the other factors when investigating the factor of interest. Additionally, we conducted analyses to assess training responses per hour of training across the three training intensity categories, aiming to investigate training efficiency. Our findings provide evidence that supports both sides of the debate, underscoring the importance of both training intensity and volume. Specifically, we observed that ET, SIT, and HIT did not promote differential changes in mitochondrial content after adjustment for the total number of training sessions. However, the increase in mitochondrial content per hour of exercise training followed an exercise intensity-dependent pattern (SIT > HIT > ET). This highlights the role of exercise training intensity in enhancing mitochondrial content. Furthermore, our study also shows greater elevations of mitochondrial content with higher training frequency per week and, for ET and HIT, greater number of training intervention weeks, supporting the argument for training volume. Notably, we observed that high-intensity training, and especially SIT, may reach a plateau in mitochondrial adaptations sooner than lower intensity training. Our understanding of the accumulated evidence is therefore simple; a high exercise intensity will compensate for a low training volume in exercise training interventions with a short duration, but a larger exercise training volume is necessary to promote further increments in mitochondrial content over time, as indicated by SIT not leading to a further increase in mitochondrial content after the initial 2 weeks of exercise training. However, for a given exercise training volume (i.e., number of exercise training hours), training with a high exercise intensity is the most efficient way of increasing mitochondrial content, as SIT in this context showed larger increments than both HIT and ET, and HIT was more efficient than ET.

The Journal of Physiology crosstalk about the mitochondrial content contest [8, 10, 39] was probably influenced by semantic differences among researchers in how to interpret the exercise training volume and intensity terms. In the current study we have, consistently with other researchers [12,13,14], defined exercise training volume as the total exercise training time (i.e., minutes per training session × training session frequency). However, other researchers, such as Granata and colleagues [11], interpret this term differently. As mentioned, they define exercise training volume as the exercise intensity multiplied by the total training duration. This interpretation is the same as Banister and colleagues [12] in 1975 defined as exercise training load, a term which has been widely used with the same interpretation in later years [13,14,15,16]. However, in the recent crosstalk debate, the semantic issues of distinguishing between exercise training volume and -load were not properly addressed. Albeit belatedly, we consider it reasonable to put forward the argument that two exercise training variables (i.e., exercise training volume and intensity) against one (i.e., exercise training intensity) is an unfair match for evaluating the most important variable mediating exercise-induced increases in mitochondrial content. The term training load may be used to describe the physiological stress a certain amount of exercise training poses on an individual, which, as expected, leads to a given magnitude of adaptation. However, the training load for an individual is largely affected by their fitness level. For example, 1 h of ET will pose a larger physiological stress on an untrained individual than a well-trained individual, and typically lead to greater training adaptations. In the present study, to elucidate this discrepancy in exercise training-induced adaptations in response to training volume and intensity between individuals with different fitness levels, we estimated the percentage improvement per hour of ET, HIT, and SIT for the variables mitochondrial content and \(\dot\)O2max (Figs. 4, 8, respectively). When planning training, such an approach can be used as a training load factor for predicting adaptations in relevant physiological variables to a given training stimulus.

Exercise training-induced skeletal muscle capillary growth appears to be influenced by at least two physiological factors: (1) increased blood flow and thereby shear stress on endothelial cells [43], and (2) mechanical stretch of the tissue [44]. It is also likely that factors related to increased muscle metabolism and/or hypoxia with exercise add an additional angiogenic stimulus [45]. In the present study, we show that C/F is increased to a similar extent by ET, HIT, and SIT when number of training intervention weeks and initial fitness level of the participants are controlled for. This finding contrasts with the prevailing view that the most potent type of exercise for inducing capillary growth is long-duration training sessions conducted at a low- to moderate exercise intensity [24]. This perspective is based on the finding of attenuated VEGF levels after HIT compared with low-intensity exercise [27]. Nor is it in line with the indications of more pronounced capillary adaptations after SIT compared with HIT in a recent meta-analysis [29], although that finding, in particular, should be interpreted with caution owing to the low number of studies included in the analysis (two SIT studies and one HIT study [29]). However, ET in the current study did indeed lead to a greater increase in CD compared to both HIT and SIT, which may be used as further evidence for low- to moderate intensity training sessions to be the superior type of exercise in promoting muscle capillary growth. Importantly, the present finding of superior change in CD with ET was not due to greater capillary growth per se, but rather a consequence of numerically lower muscle hypertrophy response (i.e., change in muscle fiber CSA) compared with HIT and SIT. We cannot rule out the possibility that methodological decisions may have influenced the capillarization results, such as adjusting for covariates, instead of only displaying the mean changes as in Fig. 5J. Moreover, the ET exercise intensity category in the current study must be considered as quite heterogeneous, spanning all types of exercise activities and a wide range of minutes per training session conducted in a continuous fashion with an exercise intensity below the second ventilatory threshold/4 mmol/L blood lactate concentration/87% of HRmax/87% of \(\dot\)O2max /75% of Wmax. The heterogeneity of training interventions within the different exercise intensity categories may thus have diluted the exercise intensity-dependent capillarization effects. More research is therefore needed to further elucidate the etiology of exercise-induced capillary growth. Common for both C/F and CD was that the increase occurred in the early stages of exercise training onset (i.e., ≤ 4 weeks), with greater number of training intervention weeks not associated with further improvements. This is strikingly similar to the time course of exercise training-induced angiogenesis that has been demonstrated in rodent skeletal muscle [46]. Additionally, this agrees with the finding of Jensen et al. [47] who sampled multiple muscle biopsies across a seven-week training intervention but observed no further increase in capillarization beyond 4 weeks of exercise training. This phenomenon is speculated to be owing to the fact that the initial exercise training-induced increase in vascularization possibly reduces the shear stress on the endothelial cells, and thus provides a weaker angiogenic stimulus. Longer-duration training interventions are therefore requested to fully uncover the evolution of vascularization in human skeletal muscle and to explain the major differences in mean baseline C/F values between untrained, moderately trained, and well-trained participants (1.7, 2.6, and 3.0, respectively) as described in the present analysis (Table 2).

It is widely reported that \(\dot\)O2max gradually decreases with advancing age [22, 48, 49] owing to associated reductions in central factors (e.g., decreases in blood volume, maximal heart rate, and stroke volume) as well as peripheral factors (reduced muscle capillarization, mitochondrial content and oxidative capacity) [50]. However, exercise training has previously been reported to improve \(\dot\)O2max by a similar magnitude across age-groups [51, 52], which generally is in line with the percentage change findings of the present study. Across age groups, the percentage improvements in \(\dot\)O2max, mitochondrial content and capillarization were not different, indicating that trainability is largely maintained throughout the lifespan although participants < 35 years tended to display larger percentage \(\dot\)O2max gains than their older counterparts. However, in terms of the absolute change of body weight-normalized \(\dot\)O2max, participants < 35 years demonstrated larger increases than participants between 35–55 years and > 55 years, which as probably partially related to the lower baseline values in these age groups. Furthermore, absolute values of \(\dot\)O2max are known to be closely related to factors, such as total lean body mass [53, 54], and cardiac size [55], both of which are known to decrease with increasing age [56, 57]. In theory, such relationships between dimensions of different body parts and \(\dot\)O2max also suggest that the absolute changes in \(\dot\)O2max to exercise training are scaled to the absolute proportions of factors such as lean body mass and cardiac size, thus giving persons with lower initial levels of these characteristics, as older individuals, lower absolute changes in \(\dot\)O2max. Moreover, increases in \(\dot\)O2max were not significantly different following ET, HIT, and SIT. This is in agreement with previous meta-analyses showing no difference for changes in \(\dot\)O2max with HIT compared with SIT [58, 59] and studies displaying no difference in effect between SIT and ET [60, 61]. However, it is contrary to the previous finding of HIT to be more potent than ET in improving \(\dot\)O2max [62], although there was a tendency toward greater \(\dot\)O2max increases with HIT compared with ET in the present study. The discrepancies across studies could potentially be partly explained by different inclusion criteria, i.e., HIT criteria being 90–95% of HRmax in the aforementioned meta-analysis [62]). Interestingly, we found that women in general have larger percentage increase in \(\dot\)O2max than men in response to exercise training, while displaying similar increases to males expressed in mL/kg/min. This is, to a certain extent, in contrast to what is previously observed for cardiac adaptations, where males seem to improve more than females [63], and in contrast to a recent meta-analysis finding a 2 mL/kg/min larger increase in \(\dot\)O2max after exercise training in males than females [64]. However, it must be emphasized that the mentioned meta-analysis only included eight studies (eight training groups of men and eight training groups of women), while model 17 in the present study included 24 and 183 training groups of women and men, respectively. Moreover, since men in general have a higher baseline \(\dot\)O2max, it can be calculated that the mean percentage increase in body mass-normalized \(\dot\)O2max was exactly 16% for both men and women for the eight included studies in Diaz-Canestro and Montero [64]. Furthermore, the above analysis only included studies in which both male and female participants were recruited in the same study and completed the same training program, but failed to include data from a few other studies with these criteria, such as Hoppeler et al. [65] (10% and 19% increase in \(\dot\)O2max for men and women, respectively) and one of the largest exercise training intervention studies conducted on both sexes, the HERITAGE study [52]. That study included 633 participants (287 men and 346 women) who carried out 20 weeks of ET, with men and women increasing \(\dot\)O2max by 5.5 and 5.2 mL/kg/min, respectively, and the percentage increase being significantly lower for men (15.9% and 19.5% for men and women, respectively; P < 0.01). This sex difference persisted even when normalizing \(\dot\)O2max to fat-free mass instead of body mass (14.6% and 17.9% increase for men and women, respectively; comparison between sexes: P < 0.01) [52], which, in our opinion, is the fairest comparison between sexes both for baseline and trainability comparisons owing to initial differences in body size and composition. Furthermore, a recent meta-analysis also concluded that there are no clear indications of sex-specific differences in \(\dot\)O2max trainability [66]. Hence, we argue that the summative information indicates similar or even better \(\dot\)O2max trainability for (premenopausal) women than men.

Recently, the magnitude of both central and peripheral (i.e., vascular) training adaptations has been reported to differ in pre- and recently post-menopausal women (< 5 years) compared with women initiating exercise training later after the menopause [22]. This finding is thought to be related to the estrogen status and changes in estrogen receptor signaling [67]. In support of this, the present data show indications (P = 0.060) of greater percentage increases in \(\dot\)O2max in young women compared with older women above 55 years of age (significantly larger in mL/kg/min). However, these analyses must be interpreted with caution since only six and five training groups of untrained, healthy young and old women were included in this dataset, respectively. Moreover, exercise training-induced changes in mitochondrial content and capillarization are still very similar between sexes and age groups, and remain potent if the exercise intensity is high, at least when comparing untrained young (< 35 years) and old (> 55 years) women and men. This suggests, despite an age-related decrease in physical performance, that regular exercise training can oppose this decrement, that mitochondrial and \(\dot\)O2max trainability is largely maintained throughout life and highly affected by the total training load (training intensity × volume), and that trainability is primarily determined by the initial fitness level. Interestingly, our data did not show any conditional effect of disease status. This is somewhat contrary to the view that common comorbidities with a range of diseases, such as systemic inflammation, insulin resistance, low capillarization, and mitochondrial content and function, in addition to the use of certain medications, are associated with impaired responses to exercise training [68,69,70,71,72]. This finding was further strengthened in our post hoc analyses which sub-analyzed different disease groups: in terms of percentage changes in mitochondrial content, capillarization or \(\dot\)O2max, people with metabolic diseases, CVD, or COPD did not respond differently to exercise training compared with healthy, age-matched, and initial fitness level category-matched participants (i.e., untrained < 45 mL/kg/min). This may, therefore, be related to initial lower levels of fitness amongst different disease groups, since we did not differentiate by specific \(\dot\)O2max, and thus may have a greater potential for adaptations, as previously discussed [72]. Importantly, this also reiterates the beneficial effects of exercise training across all groups in the population. The finding is very consistent with the common understanding of magnitude of training-induced adaptations being inversely proportional to the initial fitness level [51, 73, 74], regardless of sex and age. Of note, in terms of absolute changes in C/F, this variable was found to increase less with exercise training in young individuals with metabolic diseases and older individuals with COPD compared to their age-matched healthy controls. This differentiates from the percentage change results, implying that lower baseline values in these groups had an impact. COPD subjects are clearly limited during exercise by their low cardiopulmonary capacity, particularly when conducting whole-body exercises [75, 76]. Feasibly, the lower blood flow to exercising muscles in COPD participants may have provided a weaker angiogenic stimulus for participants with COPD compared with the healthy participants.

In this study we also show that exercising with a small amount of muscle mass (such as during one-legged cycling or knee extension exercises), which provides a higher capacity for muscle mass-specific energy turnover compared with whole-body exercises owing to less oxygen-delivery (central) limitations [75, 77], does not translate into greater increases in mitochondrial content in the stimulated muscle. This suggests (1) that the degree of local muscle metabolic perturbations (i.e., accumulation of metabolites such as hydrogen ions and inorganic phosphate) during exercise is the primary stimulus for mitochondrial growth, and (2) that the mass-specific energy turnover in the muscle is not predictive of mitochondrial adaptations, and hence, not a suitable measure of training load. The association between central and peripheral adaptations does not occur in a 1:1 ratio. For instance, one-legged endurance training is associated with minor stress on the cardiovascular system and thus has little potency for central cardiovascular adaptations in healthy individuals, as evident by whole-body \(\dot\)O2max being unchanged after one-legged knee extension endurance training [78] and only small changes occurring after one-legged cycling endurance training [79, 80]. Moreover, since small amounts of muscle mass are stimulated during isolated training models, it must be emphasized that the favorable mitochondrial adaptations only take place in the stimulated muscle and lead to less expansion of the total body’s “mitochondrial pool” than for whole-body exercises [81].

In the current study, the ratio of change in mitochondrial content compared to change in \(\dot\)O2max was larger with SIT compared with ET and HIT (3.0, 2.2, and 2.2, respectively), which could indicate that the balance between central to peripheral adaptations is lowest for SIT. Indicatively, Vigelsø et al. [82] observed a moderate correlation of r = 0.42 between change in CS activity and change in \(\dot\)O2max for ET interventions, which is consistent with our observation (r = 0.46). Surprisingly, the change in \(\dot\)O2max was not associated with the change in MvD, opposing the seemingly accepted dogma of a near and direct relationship between \(\dot\)O2max and MvD [65, 83, 84]. The current study also demonstrates a positive correlation between mitochondrial (CS content) and capillary (C/F) adaptations to exercise training (Fig. 9E). The cooccurrence of these peripheral adaptations may partly be explained by shared signaling pathways. For instance, PGC-1α, a major regulator of mitochondrial biogenesis in response to exercise training, can also regulate VEGF expression and angiogenesis [85]. Taken together, mitochondrial growth seems primarily to be determined by local muscle metabolic perturbations during exercise, and the ratio between training-induced change in \(\dot\)O2max and mitochondrial content are different between exercises engaging small amounts of muscle mass and whole-body exercises, as well as between exercise intensity categories. Exercise training-induced increases in mitochondria and capillarization seem to be modestly correlated.

As a secondary aim of this study, we compared the effects of exercise training on muscle fiber type I proportion in studies that were already included in the qualitative synthesis of this review. This was done with the aim of elucidating a debated topic using a rather high number of exercise training studies: whether the proportion of the more oxidative and fatigue-resistant muscle fiber type I may be influenced by exercise training. Although numerous studies have clearly demonstrated that prolonged exercise training promotes transformation within the fast-twitch fiber types (from type IIx to type IIa; e.g., [86, 87]), it remains to be experimentally confirmed if exercise training-induced transformation from musc