Is BMI Associated with Cardiorespiratory Fitness? A Cross-Sectional Analysis Among 8470 Apparently Healthy Subjects Aged 18–94 Years from the Low-Lands Fitness Registry

The purpose of the present study was to investigate the association between cardiorespiratory fitness (CRF) measured as peak oxygen uptake (VO2peak, expressed in mL/min) and body mass index (BMI) in a large cohort of apparently healthy subjects. BMI and VO2peak were measured in a cross-sectional study of 8470 apparently healthy adults. VO2peak (mL/min) was determined by an incremental cycle ergometer test to exhaustion. Linear regression analyses were performed to identify predictors of CRF. There was no difference in CRF between adults with a normal weight (BMI between 18.5–24.9 kg/m2) and those who were overweight (BMI 25.0–29.9 kg/m2). Subjects who were underweight (BMI < 18.5 kg/m2) as well as females who were obese (BMI ≥ 30.0 kg/m2) showed a reduced CRF compared to the normal and overweight groups. Age, height, and gender were significant predictors of CRF (R2 = 0.467, P < 0.0001); BMI did not add significantly to this relationship. Our findings indicate that BMI was not associated with CRF in addition to age, height, and gender. In subjects with a BMI < 18.5 kg/m2, CRF was lower compared to subjects with a BMI between 18.5 and 29.9 kg/m2. In obese subjects, CRF was only lower in females compared to females with a BMI between 18.5 and 29.9 kg/m2. Correcting CRF for BMI may be beneficial for subjects with a low BMI, and females with a BMI ≥ 30.0 kg/m2. The outcome of this study might help to improve the interpretation of exercise testing results in individuals with a low or high BMI.


Introduction
Cardiorespiratory fitness (CRF) plays an increasingly important role in treatment and prevention of diseases. A low level of CRF is associated with metabolic dysfunction [13,23], cardiovascular disease [1], and an independent predictor for cardiovascular events [18]. In fact, individuals with a CRF rating in the lowest quartile of the general population are at a 2-5 times greater risk of all-cause mortality [14].
The gold standard for determining cardiorespiratory fitness is the measurement of peak oxygen uptake (VO 2peak ) during a maximal cardiopulmonary exercise test (CPET) [14,25]. CRF is related to sex, age, anthropometric characteristics, and level of habitual physical activity [15,24,30]. According to guidelines for CPET, reference values regarding CRF should be adjusted for height, weight, or BMI [4].
Overweight or obese persons are often expected to have lower levels of CRF compared to normal weight peers because of deconditioning [3]. However, the literature is inconsistent in this regard. Some studies have indicated that individuals with an increased BMI have a normal CRF and others have suggested a lower CRF [9,11,20].
The excess body mass might be accompanied by an increased skeletal muscle mass [17], and increased cardiac stroke volume resulting from an expanded blood volume. However, obese subjects can develop left ventricular dysfunction, which might reduce the cardiac ejection fraction, thus might negatively affect the cardiac output in individuals with a high BMI [2]. In addition a smaller arterio-venous oxygen difference in the overweight and obese population impacts the oxygen transport negatively [28,29].
Obesity and low cardiorespiratory fitness (CRF) have been shown to independently increase the risk of cardiovascular diseases (CVD) and mortality [8]. The relationship between BMI and CRF was investigated in male Indian [19], Finnish [25], Brazilian [6], and female Israeli [7] populations, and a negative association between BMI and CRF was found. However, CRF was expressed relative to body mass in the above-mentioned studies. This would result in an "artificially" lowered CRF because of the excessive body mass in subjects with a high BMI [16]. A high CRF per kilogram body mass is important for the performance of weight bearing physical activities. A high absolute CRF might be important for health purposes as well as for assessing change following an intervention.
In the current study we aim to investigate the relationship between BMI and CRF among a large group of apparently healthy subjects with a large range of BMI categories.

Study Design
This study was conducted using data from the Low-Lands Fitness Registry. This database contains currently cross-sectional "open source" exercise testing data from 11

Study Population
The study population comprised 6914 males and 1556 females in the age group of 18-94 years. Inclusion criteria were a minimum age of 18 years, apparently healthy subjects, and measurement of VO 2peak that was performed on a bicycle ergometer.

Testing Protocol
Only CPETs performed on an electromagnetically braked cycle ergometer were used for this analysis. These ergometers came from distinctive manufacturers, including Lode BV, Groningen, the Netherlands and Ergoline, Bitz, Germany [22]. Individually adjusted step-wise incremental protocols to volitional exhaustion were employed, after the incremental phase a cooldown period was employed. The aim of the protocol was to exhaust the subjects within 8-12 min of exercise. Verbal encouragements were provided throughout the CPET. Tests were performed in a temperature-controlled room (20-23 °C). The VO 2peak was defined as the highest amount of oxygen uptake during maximum effort of the CPET and was expressed in ml/min. Peak values were averaged over 30-60 s, based on the preference of the institution.

Measurements
In order to study the association between BMI and CRF, the following variables were used: VO 2peak (in mL/min), Heart Rate (HR) and Respiratory Exchange Ratio (RER = VCO 2 / VO 2 ), height, and weight. HR was measured using a standard exercise electrocardiogram (XECG).
All tests were performed at room temperature at sea level. Data were obtained by qualified and experienced health care providers.

Statistical Analysis
The baseline characteristics of the study population were calculated using descriptive statistics. One-way ANOVA test was performed to test differences in CRF between BMI categories. P-values adjustment for multiple post hoc tests was done using the Bonferroni method. Furthermore, a multiple linear regression analysis using the backward mode was carried out to study the relationship between age, gender, BMI, and VO 2peak .
Statistical significance was defined as a P value < 0.05 (2 tailed). SPSS software (version 21.0) was used to perform all statistical analyses.
The CRF results are displayed in Table 2.
The differences between CRF in the BMI classes are shown in Fig. 1 for females and males (Males: ANOVA F = 29.87, df = 3.00, P < 0.0001; Females: ANOVA F = 17.329, df = 3.00, P < 0.0001). When BMI was between 18.5 kg/m 2 and 29.9 kg/m 2 , no statistically significant differences in CRF were apparent between the BMI categories.
However, in males and females who are underweight (BMI < 18.5 kg/m 2 ) or in females who are obese a statistically significantly lower VO 2peak compared to their normal weight peers was observed (P < 0.0001, P = 0.0017, P= 0.0013, respectively). As expected, the VO 2peak was lower in females compared in males with the same BMI status. Table 3 presents the multivariable linear regression models for predicting VO 2peak . Age was negatively associated with VO 2peak , males had a higher VO 2peak compared to females, and height was positively associated with VO 2peak . BMI was excluded from the regression model since it did not add significantly to the explained variance next to age and gender (R 2 : 0.354); age, gender, and height (R 2 : 0.467); and age, gender, and BMI (R 2 : 0.365).

Discussion
The primary aim of this study was to determine the association of weight status expressed as BMI with CRF in a large cohort of apparently healthy subjects. Our findings indicated that CRF was negatively related in males and females who are underweight (BMI < 18.5 kg/m 2 ). Further, there was a small difference in CRF between the normal weight, overweight, and obese male subjects, of which the difference between normal weight and overweight reached statistical significance.
In females the underweight and obese groups scored lower than the normal weight and overweight groups. This indicates that the CRF is independent of weight status, when BMI is between 18.5 and 29.9 kg/m 2 . CRF was only lower in obese females, but not in obese males. This gender difference might be related to the fact that our female subjects in the obese group were smaller and older compared to the normal weight group. Care should be taken when CRF is interpreted in terms of VO 2peak related to body mass. Our data found that CRF expressed in absolute values (mL/min) is independent of BMI in subjects with a BMI between 18.5 and 30.0 kg/ m 2 . However, when CRF is divided by body mass in case of overweight or obesity, it will be "artificially" lowered because of the excess in fat mass. This finding was also reported in children by Goran et al., who observed that fat mass did not have any effect on VO 2peak ; however, they found that absolute VO 2peak was significantly higher in obese children compared to lean children [10]. Moreover, Hansen et al. reported a higher than predicted absolute VO 2peak in obese adult subjects [12].
We did not observe a higher VO 2peak in our adult subjects who were overweight or obese. In their classic study Buskirk and Taylor stated that the presence of fat does not interfere with the maximal cardiovascular-respiratory performance of a subject [5]. This in also in accord with the results of our regression analysis. BMI did not add to the explained variance in the relation between VO 2peak and age, gender and height.
Based on our findings, we recommend to express CRF in absolute values (mL/min or L/min), as well as in relative values (mL/kg/min) in persons who are overweight or obese. Males and females with a low BMI had a lower absolute VO 2peak value compared to peers with a normal BMI. Females with a low BMI have comparable relative VO 2peak values compared to females with a normal BMI. This can be (partially) explained by the fact that the females with a low BMI were taller than the females with a normal BMI.
However, males with a low BMI have higher relative VO 2peak values compared to peers with a normal BMI. The latter is described as the "fallacy of the per weight ratio" [26]. Subjects with a higher body mass are penalized when VO 2peak is related to body mass [27]. However, from a performance viewpoint, relative VO 2peak is important for capacity to perform weight bearing activities, such as walking, running, or uphill cycling [5].
The results from our regression analyses confirm the findings of Debeaumont et al. [9]. They observed that next to gender, age and height were the best predictors of VO 2peak in the obese population. Our regression model was able to explain almost 50% of the variance in VO 2peak among the subjects.
A change in absolute CRF would be helpful when interpreting the effectiveness interventions, particularly lifestyle interventions, such as exercise training [31]

Strengths and Limitations
A primary merit of the current study is the large sample size used to analyze the association between body composition expressed in BMI and CRF. Another strength of this study is its multicenter design and the direct measurement of VO 2peak which provided accurate data about CRF. On the other hand, there are some limitations that should be addressed. Firstly, the number of subjects with a BMI < 18.5 kg/m 2 was relatively small. The groups contained of 49 male and 27 female subjects. Secondly, the participants included in the current study were regarded as apparently healthy. However, this was self-reported by the subjects. Also, body composition data (fat mass and fat free mass) were not available.

Recommendations for Future Research
A larger number of females, more participants older than 70 years, smoking, BMI < 18.0 kg/m 2 and > 35.0 kg/m 2 , body composition (fat free mass), and physical activity level should be taken into consideration in a future study. Extensive research on this topic would offer more insight into the influence of excess body weight on CRF.

Clinical Relevance
Our findings indicate that adjustment of the absolute value of VO 2peak in overweight and obese individuals is not necessarily in males, as well as in obese females. Adjustments for CRF in individuals with a BMI < 18.5 kg/m 2 are also required.

Conclusion
Our findings indicate that BMI was not associated with CRF in addition to age, gender, and height. In subjects with a BMI < 18.5 kg/m 2 , CRF was lower compared to subjects with a BMI between 18.5 and 29.9 kg/m 2 . In obese subjects, CRF was only lower in females compared to females with a BMI between 18.5 and 29.9 kg/m 2 . Correcting CRF for BMI may be beneficial for subjects with a low BMI, and females with a BMI ≥ 30.0 kg/m 2 . The outcome of this study might help improve the interpretation of exercise testing results in individuals with a low or high BMI.