Cardiorespiratory fitness in individuals with type 2 diabetes mellitus: a systematic review and meta-analysis

ABSTRACT Objective: To conduct a systematic review and meta-analysis assessing the cardiorespiratory fitness (CRF) among individuals with and without type 2 diabetes Materials and methods: The current review was registered in PROSPERO under the number CRD42018082718. MEDLINE, EMBASE, and Cochrane Library databases were searched from inception through February 2022. Eligibility criteria consisted of observational or interventional studies that evaluated CRF through cardiopulmonary exercise testing or six-minute walk test in individuals with type 2 diabetes compared with individuals without type 2 diabetes. For data extraction, we used baseline CRF assessments of randomized clinical trials or follow-up CRF assessments in observational studies. We performed a meta-analysis using maximal oxygen consumption (VO2max), and distance walked in the 6MWT as primary outcomes. They were extracted and expressed as mean differences (MDs) and 95% CIs between treatment and comparator groups. The meta-analysis was conducted using Review Manager (RevMan) software. Results: Out of 8,347 studies retrieved, 77 were included. Compared with individuals without type 2 diabetes, individuals with diabetes achieved a lower VO2max (−5.84 mL.kg−1.min−1, 95% CI −6.93, −4.76 mL.kg−1.min−1, p = <0.0001; I2 = 91%, p for heterogeneity < 0.0001), and a smaller distance walked in 6MWT (−93.30 meters, 95% CI −141.2, −45.4 meters, p > 0.0001; I2: 94%, p for heterogeneity < 0.0001). Conclusion: Type 2 diabetes was associated with lower cardiorespiratory fitness, as observed by lower VO2max on maximal tests, and smaller distance walked in 6MWT, however the quality of studies was low.


INTRODUCTION
C ardiorespiratory fitness (CRF) appraises an individual's exercise capacity, it is directly linked to the integrated function of several body systems and may be a marker of total body health (1).Low CRF is associated with an increased risk of cardiovascular disease among patients with type 2 diabetes (2).Balducci and cols.(3) observed that increasing maximal oxygen consumption (VO 2 max) by approximately 2 mL.kg −1 .min −1 can significantly reduce 10-year risk of coronary heart disease in these individuals.Moreover, a 9% lower relative risk of all-cause mortality was shown among adult men with VO 2 max of 1 mL.kg −1 .min−1 higher (4).The annual cost savings per person were $5,193 in type 2 diabetes for each 1-metabolic equivalent (MET) higher fitness (5).
The cardiopulmonary exercise test -by gas analysis -is the gold standard assessment of CRF.It evaluates the VO 2 max or peak oxygen uptake (VO 2 peak) during an incremental exercise test (6).Several protocols use a cycle ergometer or a treadmill (7), but these devices are expensive and require a trained team, being unfeasible in some situations such as population-based studies and in clinical practice.Therefore, other tests, such as sixminute walk test (6MWT), are also useful as they can estimate oxygen consumption (8).
Previous studies have shown controversial results when comparing CRF between individuals with and without diabetes: some showed comparable results (9)(10)(11), whereas others showed lower CRF in individuals with diabetes compared to those without diabetes (12)(13)(14).These differences could be methodological and derive from different protocols used for the evaluations.However, there are physiopathological mechanisms to justify the lower levels of exercise capacity observed among individuals with type 2 diabetes, which may occur from insulin action, mitochondrial dysfunction, skeletal muscle microvasculature, and cardiac dysfunction (15).Moreover, poor glycemic control can reduce CRF (15) because of diabetes itself or diabetes-associated sedentary behavior (16).Thus, it is essential to understand the magnitude of VO 2 max impairments observed in these individuals during planning of appropriate interventions to improve exercise performance and avoid increasing disability in this population.However, it is uncertain whether the magnitude of this difference and age, sex, body mass index (BMI), diabetes duration and control of the disease could negatively affect exercise capacity.
We aimed to conduct a systematic review with metaanalysis to summarize studies that assessed CRF measured by VO 2 peak or VO 2 max in individuals with and without type 2 diabetes.We also evaluated the differences in distance walked in the 6MWT among them.

MATERIALS AND METHODS
A systematic review and meta-analysis was conducted according to the Cochrane Handbook for Systematic Reviews of Interventions (17) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (18).This review was registered in the international prospective register of systematic reviews (PROSPERO: CRD42018082718).

Eligibility criteria
The eligibility criteria were as follows: 1) participants: adults with type 2 diabetes, > 18 years old; 2) outcomes: CRF measured by maximal exercise tests and expressed as VO 2 (peak or maximal), or distance walked evaluated by the 6MWT; and 3) control group: individuals without type 2 diabetes; 4) study design: observational design (i.e., cohort or cross-sectional studies) and baseline data from quasi-experimental, randomized, or non-randomized clinical trials.Only studies in English, Portuguese, and Spanish were included.Studies were excluded if the participants had peripheral arterial disease, heart failure, chronic neurological diseases.Also, studies reporting that individuals without diabetes took any medication were excluded as well as studies when groups were matched by CRF.

Outcomes definition
The primary outcome was VO 2 (peak or maximal) measured by direct expired gas analysis.The secondary outcome was distance walked evaluated by the 6MWT.

Databases and search strategy
Three electronic databases (i.e., PubMed/MEDLINE, EMBASE and Cochrane Library) were searched using a combination of MeSH headings, keywords and related entry terms, such as "type 2 diabetes" and "cardiorespiratory fitness".The search strategies are presented in Supplementary File 1. Besides, the reference list of studies was manually searched.The search strategy was conducted from inception until December 2017, updated in March 2021 and February 2022.

Selection process
Two pairs of authors (ACPM/MBP and PMB/CEB) independently evaluated the titles and abstracts of all studies based on eligibility criteria.All studies with abstracts lacking enough information regarding the eligibility criteria were included to full text evaluation.Finally, the full-text studies were evaluated by the same reviewers according to the inclusion and exclusion criteria and any disagreement between them was resolved by a third reviewer (DU).

Data collection process
Data were extracted independently by two pairs of authors (ACPM/CWS and PMB/CEB) using a

Risk of bias and publication bias assessment
The risk of bias of the included studies was assessed by two pairs of authors (PMB/CEB and CWS/MBP), previously trained and qualified.The Newcastle-Ottawa Scale (NOS) version for cohort studies was adapted and used (19).The quality score was calculated by assessing three domains: selection of the study groups (0-3 points); comparability, which represent the quality of adjustment for confounding factors (0-2 points); evaluation of the outcomes of interest (0-3 points).The maximum score was eight and the classification of the studies were: (1) good quality: 2-3 points in the selection domain, 1-2 points in the comparability domain and 2-3 points in the outcome domain; (2) fair quality: 1 point in the selection domain, 1-2 points in the comparability domain and 1-2 points in the outcome domain; and (3) poor quality: 0 points in any domains.Disagreement between reviewers were resolved by consensus, and, in cases of persistent disagreement, the assessment was made by a third reviewer (ACPM).
Publication bias was assessed using a contourenhanced funnel plot with each study effect size against the standard error of the estimate.

Synthesis methods
The quantitative assessment of the included studies was performed by meta-analysis using the Review Manager (RevMan) software (Cochrane Review Manager, version 5.3).Each outcome (VO 2 max/peak, and distance walked) was expressed as mean differences (MDs) and 95% confidence interval (CI) between individuals with and without type 2 diabetes.The results were pooled using a random-effects model.
Statistical heterogeneity was assessed by the Cochran's Q test, at 0.1 significance level, and inconsistency I 2 test.Considerable heterogeneity was indicated when I 2 value was > 75%, according to the Cochrane Handbook for Systematic Reviews of Interventions (17).Heterogeneity among studies was investigated based on two strategies: (1) the metaanalysis was re-run by removing each study to check if one specific study explained the heterogeneity and (2) stepwise meta-regression analyses were conducted.Univariate meta-regression models were performed in STATA software (version 20) to assess clinical and methodological variables associated with CRF, i.e., BMI, age, HbA1c, and diabetes duration, based on R 2 values and statistical significance p < 0.05.Subgroup analysis was conducted by type of ergometer (i.e., cycle ergometer and treadmill) and sex.

Data treatment
In studies that presented the results as standard deviation (n = 31), the conversion to standard error was made by the equation SD = SEM.√samplesize.The VO 2 max unit was converted from absolute (mL/min) to relative weight values (mL.kg −1 .min−1 ) in six studies.The metabolic equivalents were converted into relative weight values (mL.kg−1.min−1) in three studies, based on the standard equation (VO 2 = METS × 3,5) (20).
The data were combined in an unique group in studies with more than one group of individuals with and without type 2 diabetes (e.g., men and women), as suggested by the Cochrane's handbook.

Study selection
In total, 77 out of 8,347 studies identified in the data search (databases 7,146 + manual searching 13 + update 1,188) met the eligibility criteria and were included in our review.Figure 1 shows the flowchart of inclusion and exclusion criteria of studies.Meta-analysis for the VO 2 max and distance walked in the 6MWT included 72 and 5 studies, respectively.

Study characteristics
The included studies were published from 1984 to 2022 and the sample sizes ranged from 10 (21,22) to 3,770 participants (23).A total of 8,725 individuals were included in the meta-analysis, 2,007 in the diabetes group and 6,718 in the group without diabetes.The participants were aged < 60 years in 89% of the studies.Twenty-two studies included only men, eight studies included only women, and 42 studies included both men and women.The baseline HbA1c ranged from 5.8% to 12.2% in individuals with diabetes (data available in 64 studies) and the duration of the disease ranged from 2.5 to 12.5 years (data available in 54 studies).Most of the included studies (n = 43) presented matched groups by age, sex and/or BMI.Tables 1 and 2 show the characteristics of the studies included in the VO 2 max and distance walked metaanalyses, respectively.
A total of 73 included studies reported the BMI.Among the type 2 diabetes group, 1.3% (n = 1), 43.8% (n = 32), and 54.8% (n = 40) were classified as normal weight, overweight and obese, whereas lean and obese individuals with diabetes were pooled to be analyzed in four studies.We observed a high prevalence of patients classified as overweight (59.2%, n = 40) and obese (22.2%, n = 17) in the group without diabetes.
We evaluated the publication bias using a funnel plot for the VO 2 max (Supplementary File 3).The points for the missing studies would be on the bottom of the plot.Since most of this area contains regions of small sample size, publication bias is unlikely to be the cause of this asymmetry.The analyzed studies did not run further tests to distinguish chance from real asymmetry.

DISCUSSION
To our knowledge, this systematic review with metaanalysis was the first study comparing CRF between individuals with and without diabetes, in which we observed that individuals with type 2 diabetes presented lower CRF evaluated by VO 2 max.This is essential because VO 2 max is a measure associated with health and this review included studies with different designs to broadly analyze this variable in diabetes and non diabetes groups.The lower VO 2 max values indicated may be useful to qualify future studies about physical rehabilitation and physical activity for individuals with type 2 diabetes.
Cardiac, respiratory, and skeletal muscular systems determine VO 2 max (66).This assumption is supported by studies that indicate that VO 2 max reduction is associated with diastolic dysfunction and/or impaired myocardium perfusion during exercise (57), as well as with abnormalities in skeletal muscle morphology (67), VO 2 kinetics (O 2 uptake/use) (54,56), endothelial Absolute changes in VO 2 max in studies conducted with type 2 diabetes patients compared with group without type 2 diabetes.Squares represent study-specific estimates; diamonds represent pooled estimates of random-effects meta-analyses dysfunction (43,68,69), blood viscosity (55), and glycemic profile/control (69).Changes in these components and in their integration can lead to VO 2 max impairment.Our study only assessed glycemic control based on HbA1c.However, meta-regression results did not associate HbA1c with VO 2 max in patients with diabetes.
We observed lower values of VO 2 max in individuals with diabetes compared with the other group, however, 22 out of the 70 included studies did not show differences between the groups.This could be explained by: 1.One study had a diabetes group with higher levels of physical activity than the group without diabetes (61); 2. Another study included only individuals with obesity in the group without diabetes with VO 2 max values lower than predicted for normal weight individuals (25); 3.One study included a 10 times smaller sample size in the diabetes group than in the group without diabetes (70); and 4. Four studies did not control for comorbidities, such as cardiovascular, endocrine or renal diseases, in groups with and without diabetes (14,25,71,72).
The 6MWT has been used to estimate exercise capacity in adults with diabetes.However, results have shown moderate correlation between estimated VO 2 max and 6MWT (73), which suggest that 6MWT can be used to assess the patients' ability to maintain the exercise, but not to estimate VO 2 max.We observed a mean reduction of 93 meters in the distance walked among individuals with diabetes compared to the group without diabetes.Studies showed that reductions of 25-30 meters in distance walked in patients with coronary artery diseases and pulmonary diseases were associated with increased risk of death (74,75).However, the minimal clinically significant difference values of distance walked are not established among patients with diabetes.Although our study found significant reduction in distance walked evaluated by 6MWT in patients with diabetes, our results have limited generalizability due to few included studies.
The first strategy adopted to explore heterogeneity was to remove each study from the analyses, which did not cause changes.Despite the group without diabetes being restricted to individuals without diabetes, we cannot assert their health status.Moreover, factors such as obesity and/or age could explain similar VO 2 max between the diabetes group and the group without diabetes in one third of the analyzed studies.
The high heterogeneity in the VO 2 max metaanalysis was explored by sensitive analyses considering the ergometer and sex and also by performing a metaregression analysis.The type of ergometer used in the tests did not change the results of the meta-analysis, as well as the subgroup analyses based on sex.Besides, meta-regression analysis applied to VO 2 max showed that age, HbA1c and diabetes duration could not explain the high heterogeneity presented in VO 2 max meta-analysis, but BMI partly explained it.We believe that the high I 2 in the VO 2 max analysis is related to different magnitudes of these effects in the different studies shown.However, most of them present the same direction of effects.Therefore, the practical/clinical implications are that despite the high heterogeneity, the direction of effect shows lower cardiorespiratory fitness in individuals with diabetes, but we cannot estimate the exact magnitude of the difference between type 2 diabetes and control groups.
Higher levels of CRF may coexist with higher BMI (76).Hemmingsson and cols.observed a reduction of the normal weight and high CRF category (relative change −30%) when analyzing time trends combinations between CRF and BMI (1995-2020), an increase in overweight and low CRF (relative change +34%) and in obesity and low CRF (relative change +154%) categories.Studies show that the risk of cardiovascular disease and all-cause mortality in individuals with obesity varied by CRF (77), such as among individuals with diabetes (78).
Our study has some limitations.Although the search was not limited by language, the included studies were only in Portuguese, English, and Spanish.It was a challenge to summarize the results of this review, since different protocols were used to evaluate CRF.Moreover, there was a considerable variation among the studies about pharmacological treatments that diabetes patients received, some studies did not mention the drugs used to treat health conditions other than diabetes, and most studies did not mention the drug doses used.Most of the included studies were carried out in participants with a mean age of less than 60 years, and the highest prevalence of type 2 diabetes is found in older adults.Another challenge was the wide range of duration of diabetes, because the CRF can change along with diabetes duration.Therefore, these are limitations that could affect the generalization of our outcomes.Additionally, the overall quality of the studies was low, indicating increased risk of bias in many of them, however, there are fewer instruments to evaluate risk of bias of observational studies and they are less accurate compared to those evaluating clinical trials.
The strength of this systematic review is that we could summarize how lower the VO 2 max is reduced in individuals with type 2 diabetes compared with the group without type 2 diabetes, due to many included studies (n = 77).Furthermore, the exploratory analyses to explore the high heterogeneity followed all guidelines for systematic reviews.
In conclusion, individuals with type 2 diabetes showed lower CRF than the group without diabetes.CRF was evaluated by the VO 2 max individuals attained in maximal cardiopulmonary exercise testing, and was partially influenced by the BMI, but not influenced by age or sex of this population.Moreover, lower distance walked was observed in the group with diabetes.This review emphasizes exercise as a component to treat and to control diabetes that should be evaluated and prescribed individually.

Figure 1 .
Figure 1.Flow diagram of included studies.

Figure 2 .
Figure 2. Forest plot of the maximal oxygen consumption (VO 2 max) evaluated in maximal cardiopulmonary exercise tests.

Figure 3 .
Figure 3. Forest plot of the distance walked in the six minute walk test (6MWT).Absolute changes in the distance walked in the 6MWT in studies conducted with type 2 diabetes patients compared with control group.Squares represent study-specific estimates; diamonds represent pooled estimates of random-effects meta-analyses

Supplemental file 2 .
Data availability: data are available on request from the authors.Contribution statement: BDS had full access to all of the data in the study, supervised study and takes responsibility for the integrity of the data and the accuracy of the data analysis.ACPM, PMB, CEB, DU and BDS designed study.ACPM, CWS, PMB, MBP and CEB acquired data.ACPM and CWS analysed data.ACPM, CWS, PBM, CEB, DU and BDS interpreted data.ACPM and CWS drafted the manuscript.ACPM, CWS, PBM, MBP, CEB, DU and BDS revised the manuscript for important intellectual content and approved the version to be published.Funding: we thank the Research Incentive Fund (Fipe) of the Hospital de Clínicas de Porto Alegre and National Council for Scientific and Technological Development (CNPq) for partly funding this study.This study was partly supported by the Coordination for the Improvement of Higher Education Personnel -Brazil (Capes) -Finance Code 001.Disclosure: no potential conflict of interest relevant to this article was reported.Cardiorespiratory fitness in diabetesArch Endocrinol Metab, 2023, v.67(5), 1-19, e230040.Risk of bias by Newcastle-Ottawa Scale (n = 77 studies)

Table 2 .
Characteristics of the six-minute walk test studies included in the meta-analysis (n = 5) We classified all 77 included studies as poor quality and they achieved a mean score of 2.4/8 in the modified NOS (Supplementary File 2).No study has reached the maximum score (3 points) in the selection domain, although seven studies scored 2 points and 46 studies scored 1 point.Eight studies reached maximal score (2 points) and 33 scored 1 point in the comparability domain.Finally, 22 studies reached the maximal score (2 points) and 35 studies scored 1 point in the outcome domain.Mac Ananey and cols.(