Risk prediction with blood pressure during physical activity: A METter of exercise?

According to the 2016 European Society of Cardiology (ESC) guidelines on cardiovascular disease prevention in clinical practice, regular aerobic physical activity is recommended for the prevention and treatment of cardiovascular disease, including hypertension, and has been shown to lower cardiovascular risk including mortality.1,2 A meta-analysis of randomised, controlled trials indeed documented significant reductions in blood pressure (BP) (systolic blood pressure (SBP) and diastolic blood pressure (DBP)) in subjects with regular, self-reported aerobic endurance training (3.5/2.5 mmHg), dynamic resistance training (1.8/3.2 mmHg) and isometric training (10.9/6.2 mmHg).3 In addition, it has been reported that regular aerobic exercise can substantially impact cardiovascular risk: 3 hours of at least moderate intensity activity per week reduced mortality by 27%.4 Twenty-five years ago it was shown that SBP of middle-aged, apparently healthy men with a casual resting SBP of 140 mmHg or greater, measured during moderate bicycle ergometer exercise carried information on cardiovascular mortality. In this study, an increase of SBP to 200 mmHg or greater during 6 minutes of a starting exercise was associated with a two-fold higher risk of death within 16 years (relative risk (RR) 2.0, 95% confidence interval (CI) 1.1–4.0; P = 0.025) when compared with a smaller increase in BP during the same exercise.5 The current ESC hypertension guidelines recommend that hypertensive patients should be advised to participate in at least 30 minutes of moderate intensity dynamic aerobic exercise (walking, jogging, cycling or swimming) on 5–7 days per week. However, a quantification of exercise-induced changes in BP, indexed to workload during exercise testing, and its impact on long-term mortality has yet been missing. In this issue of the journal, Hedman and coworkers6 provide an interesting and detailed analysis of 7542 male US veterans from the prospective Veterans Exercise Testing Study, in which they assessed the association between fitness and BP changes during exercise stress testing. The exercise tests were performed between 1987 and 2007, participants were followed for a mean of 18.4 years. The authors identified the ratio of SBP and the metabolic equivalent of task (MET), as a measure of workload-indexed SBP response to exercise, to be more predictive in terms of overall mortality when compared with peak SBP alone. A subgroup of 709 non-smoking subjects with fewer risk factors were defined as a ‘lower risk’ cohort. In these lower risk men, the median SBP/MET slope was 4.9 mmHg/MET, representing an increase in SBP of 4.9 mmHg for each additional MET during treadmill exercise. One of the major findings was a 20% increase in all-cause mortality when achieving a lower peak SBP (<210 mmHg) or a higher SBP/MET slope (>10 mmHg) (adjusted hazard ratio (aHR) 1.20, 95% CI 1.08–1.32 and aHR 1.20, 95% CI 1.10–1.31, respectively). A total of 1028 subjects in the highest quartile of the SBP/MET slope died during the follow-up of almost 20 years, compared with 703 in the lowest quartile (57% vs. 37%, overall (log rank) P < 0.0001), which was not influenced by co-medication, including beta-blockers or other cardiovascular comorbidities. Interestingly, the association between the SBP/MET slope and all-cause mortality showed a J-shaped association with the lowest mortality (nadir) around 4 mmHg/MET, which is in line with evidence indicating lowest risk at office SBP values between 120 and 130 mmHg in high-risk patients included in the ONTARGET/TRANSCEND study, also showing a J-shaped association between BP and risk.7 Of note, subjects in the lower risk group were more often non-smokers, with lower rates of diagnosed hypertension and diabetes mellitus and had a significantly lower body mass index (BMI) compared with the higher risk group. Hedman et al.6 should be congratulated for conducting a thought-provoking observational study appraising the association between fitness and BP changes during exercise stress testing in a large cohort of US veterans. The strengths include the high number of subjects, the long-term follow-up of close to 20 years, and the utilisation of a standardised exercise protocol. As BP during exercise is variable, METs were used as a physiological model for expressing the energy cost of physical activities. Hence, the calculation and clinical utility of METs have be put into context: 1 MET equals the amount of oxygen the body uses at rest (VO2), which was approximated by 3.5 ml oxygen (O2), in relation to body weight in kilograms (kg) per minute. This number was originally derived from the resting O2 consumption of a 40-year-old man with a body weight of 70 kg and was published back in 1985.8 However, resting oxygen uptake (VO2) as well as peak oxygen uptake (VO2max) differ significantly depending on body fat mass,9 with a VO2 and VO2max of 2.8 ± 0.2 and 36.3 ± 3.6 ml/kg per minute in obese men (113.2 ± 9.1 kg) compared with a VO2 and VO2max of 3.6 ± 0.3 and 50.9 ± 7.6 ml/kg per minute in normal weight men (69.7 ± 10.5 kg), respectively (Figure 1). When calculating the MET value at peak oxygen uptake according to the standard formula, the obese group would have achieved 10.4 METs (36.3/3.5 ml/kg/min) and the normal weight group 14.5 METs (50.9/3.5 ml/kg/min). According to the real physical VO2, however, the obese group would have achieved 13.0 METs (36.3/2.8 ml/kg/min) and the normal weight group 14.1 METs (50.9/3.6 ml/kg/min), leading to a difference of only 1.1 METs (8%) in contrast to the difference of 4.1 METs (38%) between groups using the conventional estimate. In line with this, when the SBP/MET slope at an SBP of 200 mmHg is then calculated, the obese group would reach a real SBP/MET slope of 15.4 mmHg/MET compared with 19.3 mmHg/MET using the conventional estimate (with a difference of 4 mmHg/MET, 21%). This represents a significant limitation of this method because an exact calculation of weight-dependent METs according to resting VO2 is impossible due to lacking VO2 values at rest and maximum exercise level. In the present study, the lower risk group had a lower BMI (27.6 ± 4.5 vs. 28.6 ± 5.2 kg/m2, P < 0.001). However, even though this difference of 1 kg/m2 in BMI was statistically significant due to the large sample size, the difference in absolute body weight was most likely clinically irrelevant. When acknowledging that MET values in overweight subjects were likely to be underestimated, one may speculate that the SBP/MET slope could be less steep with increasing body weight. Therefore, the direct comparison of METs appears to be clinically feasible yet should be interpreted with caution. Difference in oxygen uptake at rest (VO2) and at maximum exercise (VO2max) in a normal weight (a) and an obese (b) man. Accordingly, peak METs are lower in obese subjects using the conventional estimate with the standard formula for 1 MET of 3.5 ml O2/kg per minute as compared with the real physical VO2, which is lower due to higher fat mass. The SBP/MET slope differs less between different body weights, when using the real calculation. Kg: kilograms; MET: metabolic equivalent of task; ml: millilitre; mmHg: blood pressure in millimetre of mercury; O2: oxygen; SBP: systolic blood pressure. As all BP values during exercise testing were measured by auscultation, some inaccuracies have to be accepted. There is no reliable information on DBP by nature of the measurement.10 In a study published in 1999, indirectly measured DBP using a conventional sphygmomanometer correlated well with the intra-arterial measurement at rest; however, during exercise the correlation coefficient between the direct and indirect method was rather low (r = 0.36). Indirect SBP on the other hand was correlated to the direct measurement both at rest (r = 0.684) and during all-out anaerobic exercise (r = 0.87). The authors concluded that this may be related to a better distinction in identifying Korotkoff sounds in higher (systolic) BPs, making them louder for auscultation. Detailed information, though, on the BP measurement devices and techniques used in the study by Hedman et al.6 were not provided, challenging the reproducibility of the findings. Despite the added prognostic precision provided by the findings that a higher SBP/MET slope was associated with worse survival, the underlying cause for this association remains elusive. The study included male veterans only and, as stated by the authors, the findings may not necessarily be applicable to a broader population. Indeed, in normal weight women VO2 is slightly lower than in normal weight men, with approximately 0.2 ml/kg/min at rest and 9.8 ml/kg/min at VO2max.9 A low peak SBP may be indicative of subclinical coronary artery disease (CAD), causing a lesser increase in BP during exercise, possibly due to ischaemic activation of mechanoreceptors or impaired left ventricular function.11 Herein, data on the incidence of the development of CAD over time based on baseline exercise testing results are lacking. Of note, the higher SBP/MET slope may also be a predictor of undiagnosed underlying hypertension. In summary, the authors should be congratulated for further enhancing our understanding of BP during physical activity. Along with the well-known association between peak METs and mortality, the workload-indexed BP response to SBP may represent a novel parameter for risk assessment, which may be helpful in identifying individuals at risk. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. The author(s) received no financial support for the research, authorship, and/or publication of this article.