Insulin resistance (IR) is the inability of the insulin hormone to facilitate glucose uptake from peripheral tissues to meet metabolic demands (Abdul-Ghani and DeFronzo, 2010). IR precedes type 2 diabetes mellitus (T2DM; Abdul-Ghani and DeFronzo, 2010) and is a hallmark of the disease. In 2017, the overall public spending on T2DM reached $ 237 billion in the United States alone (Wang et al., 2017). Among the different available therapies to treat T2DM and its complications, exercise training is a unique, non-pharmacological intervention that improves several physical inactivity-related metabolic disorders including IR (Jenkins and Hagberg, 2011; Slentz et al., 2011; Matos et al., 2018) and is a therapy for T2DM (Little et al., 2018). The major beneficial metabolic effects of exercise training are associated with structural adaptations (i.e., changes to adipose, skeletal muscle, bone tissue, and vessels), and with residual effects (hours or days post-exercise) in at-risk populations (Andersen and Høstmark, 2007; Short et al., 2012; Burke et al., 2017).

It is well known that some training regimens can promote improvements in several cardiometabolic markers (American Diabetes Association, 2017). High-intensity interval training (HIIT) and resistance training (RT) have been recommended for people with poor glucose control (Hayes et al., 2009; Colberg et al., 2016; Colberg, 2017). HIIT promotes greater cardiorespiratory fitness (CRF) and metabolic benefits in populations IR (Jelleyman et al., 2015) and T2DM (Little et al., 2011). Additionally, RT also promotes similar glucose control improvements to HIIT (Ross et al., 2021), by increasing skeletal muscle (Brooks et al., 2007) and bone mass (Wood and O’Neill, 2012) in patients with T2DM (Dunstan et al., 2002; Ross et al., 2021). The combination of moderate-intensity continuous training (MICT) plus RT, or also as HIIT plus RT (Da Silva et al., 2020), resulting in a training regimen known as concurrent training (CT), has also relevant evidence in favor of IR patients (Timmons et al., 2018; Álvarez et al., 2019). Indeed, we recently reported beneficial cardiometabolic outcomes in women with hyperglycemia after a 20-week CT intervention, including −4mg/dl decreases in fasting plasma glucose (FPG), and other physiological adaptations in body composition (−4cm waist circumference, +400g increases in lean mass), cardiovascular system (−6/−3mmHg reduction in systolic/diastolic blood pressure), and plasma lipoproteins (−11mg/dl LDL-cholesterol reduction) and greater increases in endurance performance (+56m in the 6-min walking test; Álvarez et al., 2019). While numerous studies have reported improved cardiometabolic health after short- or long-term exercise training with HIIT, RT, or CT, these exercise adaptations are typically short-lived and are reversed after training cessation (detraining), as is the case for metabolic outcomes (Bajpeyi et al., 2009; Del Vecchio et al., 2020). At the level of physiological adaptations from HIIT, RT, and CT, the results are scarce at level of post-exercise cessation, and few studies, and few studies have monitored the residual effects on glucose control markers in IR cohorts (Short et al., 2012).

In the case of CT, some studies have reported “interference effects” when performing HIIT plus RT in the same exercise routine (Wilson et al., 2012; Vechin et al., 2021), while others have found no detrimental interactions in health-related and performance outcomes (Lundberg et al., 2013; Villareal et al., 2017). Specifically, the “interference effect” has been described when MICT, or RT alone promotes a specific molecular profile that generates mitochondrial biogenesis and thus increases fatigue tolerance or hypertrophy, and strength at the skeletal muscle level, but when both regimens are applied concurrently (in the same exercise session), other more specific training adaptations can occur, and endurance training can attenuate the muscle hypertrophy and strength gains (Coffey and Hawley, 2017). Some of the discrepancies in the literature likely arise because of the order of the sessions of CT [i.e., RT+HIIT or HIIT+RT (Bagheri et al., 2020)], or in the comparisons of CT vs. RT or HIIT alone, in which no differences were reported in strength performance increases, but minor cardiorespiratory adaptations are evident when comparing CT vs. MICT or RT alone (Glowacki et al., 2004). In addition, while HIIT has a strong capacity for improving CRF, and marked evidence for reducing adiposity impacting overall adiposity markers, by contrast, RT has a high capacity for increasing skeletal muscle mass and other tissues as bone.

It has been proposed that the beneficial adaptations to exercise training are not related to the training itself, where most of these are linked to the recovery period, or after exercise cessation, which is known as “residual effect.” For example, most events in which glucose control is improved by exercise training (i.e., FPG decreases) are after exercise cessation, major body fat decreases operate similarly after HIIT, or the greater skeletal muscle mass increases are resulting from the sum of physiological events during recovery time. Thus, particularly, the residual effect is the post-exercise cessation time in which the beneficial exercise effects are extended without exercise participation until these physiological adaptations are lost, but at the same time other adaptations are assembled at different systems (tissues, cardiovascular, metabolic among others).

Based on previous reports about some detrimental adaptations to increase strength from CT of endurance plus RT exercise in which RT alone promote more strength increases (Bajpeyi et al., 2009), considering the relevance of increasing skeletal muscle mass for glucose control in IR and T2DM patients (Dunstan et al., 2002; Brooks et al., 2007; Ross et al., 2021), we hypothesized that both HIIT or RT alone, but not CT, would have long-term residual effects on FPG and FI in women with IR. Thus, the present study aimed to assess the residual effects (post-72-h training cessation) on FPG and FI after 12-weeks of HIIT, RT, or CT in adult women with IR.

We studied physically inactive adult women with insulin resistance but with no diagnosis of T2DM. The study was performed in accordance with the ethical standards established by the Declaration of Helsinki (2013) and was approved by the local ethical committee of the Family Healthcare Center TRV of Los Lagos, Chile (no. 0204015). The study was not registered in a database. All participants signed a written informed consent.

Eligibility criteria included the following: (a) age between 25 and 60years; (b) diagnosis of insulin resistance [homeostasis model assessment of insulin resistance (HOMA-IR) ≥2.6] (Garmendia et al., 2009); (c) to be physically inactive (i.e., a total of <150min/week of low-moderate physical activity or<75min/week of vigorous activity according to an International Physical Activity Questionnaire; O’Donovan et al., 2010); (d) to do not report exercise training participation the last 3months; (e) family history of T2DM; (f) to live in urban areas (related to modern life); (g) absence of musculoskeletal disorders; (h) absence of bone inflammatory, ischemic, and/or cardiac diseases; (i) absence of asthma and/or chronic obstructive pulmonary disease; (j) not to be under pharmacological treatment that modulates metabolic and/or respiratory control; and (k) free from hypertension and/or hypothyroidism diagnosis.

The present study aimed to assess the residual effects (post-72-h training cessation) on FPG and FI after 12-weeks of HIIT, RT, or CT in women with IR. We also aimed to determine the training-induced, post-training residual impact of CT. The main findings were as follows: (i) HIIT decreases FPG and RT decreases FI 24-h post-training, (ii) both HIIT and RT exercises alone show improved residual effects on FPG and FI (post-72h) compared with the CT intervention where at this time their effects are worsened, and (iii) adiposity-related markers [model 2: ∆WC+∆T_SF, R² 0.62 (62%)] together with body composition/anthropometric adaptations [model 4: ∆SMM+∆HumD+∆FemD+∆ArmP, R² 0.28 (28%)] models mostly explain the poorer residual effects promoted by the CT intervention. These results highlight that both adiposity reduction particularly, together with SMM, and bone increases play a particular role in the residual effects after CT.

While there is a large body of evidence on the training-induced changes in metabolic markers after HIIT (Little et al., 2011; Jelleyman et al., 2015; Alvarez et al., 2016; Matos et al., 2018), RT (Dunstan et al., 2002; Wood and O’Neill, 2012), and CT (Timmons et al., 2018; Álvarez et al., 2019), evidence is sparse concerning their residual effects post-exercise, particularly for CT. For instance, Marcus et al. (2013) showed that despite 16-weeks of RT in patients with T2DM, which promoted an improvement in SMM (∆ 8%) and regional insulin sensitivity based on a euglycemic-hyperinsulinemic clamp (∆ 33.9%), there was a loss in whole-body insulin sensitivity (∆ –12.4%) 7days after exercise csessation even though SMM was maintained. Also, when highly trained athletes and young people were participating under MICT, these showed showed an increase in FI and a decrease in insulin sensitivity after 5 to 7days post-training (Vukovich et al., 1996; Goulet et al., 2005). Contrastingly, the present study shows that after 12-weeks of HIIT, RT, or CT, only HIIT and RT groups showed improve ∆FPG_{24 h}, and the RT group showed improved ∆FI_24h. Nevertheless, our data reveal that improvements from HIIT in ∆FPG_24h were then worsened after 72h of exercise cessation (∆FPG_72h+1.2mg/dl), from RT these worsened was from −5.5 at 24h to +1.0 at 72h, but more worryingly in CT from −2.0 at 24h to +6.6mg/dl at post-72h, which is similar to previous reports (Vukovich et al., 1996; Goulet et al., 2005). Accordingly, our data along with the aforementioned evidence suggest that HIIT, and RT alone, but also CT can decrease (i.e., significant or not) FPG particularly 24-h post-exercise cessation, but however is the CT regime that shows a minor efficiency for maintaining these beneficial residual effect post-72h of exercise cessation. These results open the door for speculating that the positive metabolic adaptations promoted during exercise training participation could be critically dependent on maintaining this regular exercise routine in this precise metabolic topic of IR patients, but it is a matter of future and more complex studies to corroborate these findings.

Regarding the physiological mechanisms related to anthropometric/body composition on metabolic control, it is well known that SMM plays an important role in glucose control through peripheral glucose uptake (Richter et al., 1988). In the present study, we observed minimal increases in SMM in the RT (∆SMM; +0.4%) and CT (∆SMM; +0.6%) groups (Table 1 and Figure 4A). Additionally, other outcomes could be involved in the beneficial residual effects on glucose control post-exercise, such as WC or subcutaneous fat decreases, as previously reported in IR cohorts. Other anthropometric effects on metabolic health are shown for example after 14-weeks of a) exercise plus weight loss or b) exercise without weight loss that led to a reduction in both ∆FPG −5.8 and−1.5mg/dl, and ∆FI –44.4 and −13.7%, respectively, in each strategy (Ross et al., 2004). However, considering the increases in SMM in RT and CT, but also the higher decreases in adiposity (i.e., WC and subcutaneous fat) markers after HIIT, it is possible to presume that the residual effect and thus their prolonged beneficial effects of exercise on glucose control in FPG can be highly dependent more on fat reduction than SMM increases post-exercise cessation. In brief, part of these presumptions has been previously described in the physiopathology of the IR (Abdul-Ghani and DeFronzo, 2010), and the athlete paradox, which is relevant to this exercise intervention in our IR cohort (Goodpaster et al., 2001). Following this, exercise training could decrease both adiposity markers as can be seen in Table 1 and Figure 5, and potentially also at intramyocellular content as previously (Dube et al., 2008), and thus indirectly prolong insulin sensitivity from the last exercise cession in both HIIT and RT.

The possibility of “interference effects” from CT adaptations has been proposed from the combination of RT (i.e., increased SMM and bone mass) or moderate-intensity CT (i.e., decrease in body fat markers) regimens, in contrast to the individual application of RT and moderate-intensity CT (Wilson et al., 2012; Lundberg et al., 2013; Fyfe et al., 2014). However, our data reveal that CT promoted a remarkable metabolic beneficial effect. Indeed, 12-weeks of CT modified FPG; however, this improvement was rapidly lost at 72-h post-training, increasing ∆FPG_72h (Δ +6.6mg/dl). Based on these results, CT exercise appears to be the intervention that most rapidly loses its acquired benefits post-training compared with the HIIT and RT regimens, showing the poorest residual capacity to maintain FPG from 24h to post-72h of exercise cessation. There is evidence that CT could compromise the adaptations induced by each HIIT and RT alone (Fyfe et al., 2014; Coffey and Hawley, 2017); it consists in that signaling responses mitochondrial biogenesis adaptations [i.e., adenosine monophosphate (AMP)-activated protein kinase (AMPK), Ca2+/calmodulin-dependent kinase II, and peroxisome proliferator-activated receptor-c coactivator-1] seem to diminish the muscle anabolic pathways activated by RT [i.e., mechanistic target of rapamycin complex 1 (mTORC1)] and downstream effectors the 70kDa ribosomal S6 protein kinase (S6K) and eukaryotic initiation factor 4E binding protein (4E-BP1; Ronnestad et al., 2012; Jones et al., 2016; Coffey and Hawley, 2017), and although we do not measure these molecular proteins, but apparently, it seems not at this moment to affect the metabolic and cardiovascular adaptations from CT (Álvarez et al., 2019; Delgado-Floody et al., 2020, 2021).

Training volume (quantity) has been proposed as one of the principles of training along with quality (intensity; Da Silva et al., 2020) and offers a possible explanation for our results. Our study used low frequency and volume training for RT and HIIT modalities, in contrast to the higher frequency and volume training for the CT modality (RT group: two sessions/week, 80min/week; HIIT group: three sessions/week, 60min/week; and CT group: three sessions/week, 180min/week). It is therefore difficult to speculate on the possible mechanisms for the low maintenance (i.e., low residual capacity) of FPG_72h in the CT intervention. The physiological effects of CT (combining RT and HIIT in this specific order) were then presumably neither synergic nor additive, as was shown by worsened in FPG from 24h to 72h (+6.6mg/dl). Moreover, we observed relative maintenance of both FPG and FI at 72h after the RT cessation but not after the combination with HIIT in CT. For example, Tokmakidis et al. (2014) reported that 9months of CT resulted in an ∆FPG −12% in patients with T2DM, whereas this outcome was increased slightly (+0.7%) 3months after exercise cessation; however, improvements in glycosylated hemoglobin were promoted by 9months of CT were completely reversed after 3months of physical inactivity. This information can also suggest that metabolic adaptations following a period of physical training are critically dependent on being regular and consistent.

Considering that CT is a combination of RT and HIIT, our data revealed that exercise adaptation for secondary endpoints of adiposity markers, the results for WC, T_SF, SE_SF, IC_SF, and Calf_SF revealed that HIIT alone showed greater effect sizes for improving (i.e., decreasing) these markers, followed by RT, whereas CT showed the lowest effect size for these outcomes (Figure 5). The role of HIIT in stimulating molecular mechanisms that improve fat oxidation is well established (Astorino and Schubert, 2018), and catecholamines are highly activated after intermittent exercise, such as HIIT (Boutcher, 2011). Thus, one may speculate that additionally to those adiposity markers that were decreased post-exercise, other potential mechanisms underlying the residual effects post-HIIT and RT at least can be related with those acute hormonal roles from post-exercise cessation (Boutcher, 2011) that mediate some of the post-exercise glucose control improvements (Dube et al., 2011). These results of improving adiposity markers in HIIT and RT were displayed despite the higher energy expenditures per week from CT (90kcal·kg⁻¹·min⁻¹) than from HIIT and RT (~45kcal·kg⁻¹·min⁻¹ per session). However, it is well known that improvements in oxidative metabolism (i.e., decreases of carbohydrate oxidation and increases in fat oxidation during exercise) after ~10min of HIIT can be similar to ~40–60min of MICT (Burgomaster et al., 2008).

Concerning predict the residual effects of CT at ∆FPG_72h and ∆FI_72h by anthropometric/body composition or adiposity, stronger associations were found for body composition/anthropometry outcomes (model 4) with ∆FPG_72h (Table 2) and with ∆FI_72h (model 5; Table 3). Thus, as ∆FPG_72h was increased in the CT group with a larger effect size (+6.6mg/dl, η²: 0.76, Figure 3G), and the ∆FI_72h showed no real change, we presume that the residual effects for FPG and FI after 72-h post-training cessation are more related to body composition outcomes (i.e., decreases at adiposity outcomes) than increases at SMM that was also shown in both RT and CT groups (Figure 4A). Additionally, the ∆FPG_72h was significantly different to CG, as well as HIIT, and RT, but only the CT results showed a larger effect size, revealing the magnitude of difference to loss the residual effects from HIIT or RT alone, than by CT. As expected, the RT group showed significantly improved anthropometric parameters, whereas a significant loss of adipose tissue was mainly observed in the HIIT group. Accordingly, our data reveal that the metabolic adaptations after CT then can be explained by RT and HIIT, independently. However, considering that there may be an interference effect promoted by RT and HIIT, future studies of the cellular and molecular mechanisms might help to explain in part the mechanisms underlying the residual effects of RT and HIIT exercises on CT adaptations.

Our study presents some strengths that are important to highlight, such as (i) we used FPG, FI, and HOMA-IR indices, which are frequently used in the public health setting; (ii) we report RT and HIIT adaptations related to anthropometric/body composition outcomes regularly used in exercise training interventions; and (iii) the residual effects after CT training were not sufficient to improve the metabolic outcome in insulin-resistant patients. Our study is not without its limitations. For example, there was a lack of dietary control during the intervention, especially during the 24h to 72h in which metabolic outcomes were studied at pre- and post-test; however, the participants were reminded to maintain their current lifestyle habits. During the CT intervention, we first applied the RT component and then the HIIT exercises, and it is known that there is a potential interaction in the effects of exercise according to their order in the session. The sessions per week were different (HIIT 3, RT 2, and CT 3s/wk); however, these were displayed following international exercise recommendations per week (Bull et al., 2020). Also, the recovery period was different between HIIT and RT; however, the cardiorespiratory demands are higher in HIIT, with a need to re-establish the muscle and cardiorespiratory capacity using a double recovery period time. We also acknowledge that there are issues with the study design related to women’s reproductive health since we did not control for the menstrual cycle phase when performing pre-post testing, nor did we stratify groups by menopausal status. Also, assuming that participants maintained their energy intake patterns during the 72-h post-testing retention period, the CT group would have nearly twice as much energy excess compared to the RT or HIIT groups since the CT group expended more energy in the exercise sessions. This excess energy intake may explain part of the worsened FPG results in the CT group. Future studies could randomly assign the order of the intervention to avoid bias, include more sample size, to increase the control of the energy expenditure similarly at pre- and post-exercise cessation conditions, and to apply more than FPG the oral glucose tolerance test in which peripheral tissues key for glucose control as skeletal muscle mass is key more than FPG in which is the liver this role. Lastly, we did not perform any cellular and molecular analysis, which would improve the interpretation of our data and reveal potential mechanisms for more explanations.

In summary, our results reveal that in insulin-resistant women HIIT decreases FPG, RT decreases FI 24-h post-training, and both types of exercise interventions when performed independently have a noteworthy residual effect on FPG and FI (post-72h) when compared with the CT intervention. However, it would be necessary to increase the residual effect study with larger sample sizes in clinical cohorts.

The original contributions presented in the study are included in the article/supplementary material, and further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Family Healthcare Center Tomas Rojas Vergara (TRV) of Los Lagos, Chile (no. 0204015). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

CA, EC, GG, DA, and MI: conceptualization and visualization. CA, EC, GG, DA, MM-A, PD-F, and MI: methodology. CA, DA, MM-A, PD-F, and MI: software. CA, EC, MM-A, AA-M, and MI: validation. CA, EC, GG, DA, MM-A, and MI: formal analysis. CA, EC, GG, DA, MM-A, PD-F, FF, AA-M, and MI: investigation and writing—review and editing. CA: data curation and project administration. CA, EC, GG, and MI: writing—original draft preparation and supervision. CA, EC, and MI: funding acquisition. All authors have read and agreed to the published version of the manuscript.

CA was funded partially by privates. MI was funded by a research grant PI17/01814 of the Ministerio de Economía, Industria y Competitividad (ISCIII, FEDER). EGC was funded by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq #303399/2018-0) during this project.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.