Introduction
Heart failure with preserved ejection fraction (HFpEF) and right ventricular (RV) dysfunction are common in symptomatic obesity and can make a significant contribution toward exercise symptoms. Both predispose obese patients to further impairment in left ventricular (LV) filling on exercise, greater risk of biventricular remodelling and more severe haemodynamic derangement (Alpert et al.
2014). Independent of HFpEF and sleep disordered breathing, obesity has also been associated with increased RV mass, higher RV end-diastolic volumes and reduced RV systolic function (Chahal et al.
2012; Wong et al.
2006). This suggests that obesity itself may carry specific predisposition to RV dysfunction and that, in cases where resting haemodynamics do not explain the level of exercise intolerance, pulmonary haemodynamic evaluation during exercise may be used to unmask obesity-related pulmonary vascular and RV dysfunction (Chahal et al.
2012).
We hypothesised that obesity itself invokes a direct negative influence on the RV exercise contractile response through greater thoracic mechanical loading and higher exercise RV afterload. To measure RV contractile and afterload responses, we evaluated RV end-systolic elastance (Ees) and pulmonary arterial elastance (Ea) respectively, to derive RV–arterial coupling (Ees/Ea) ratios. RV and pulmonary arterial elastance were derived from resting and exercise RV pressure–volume relationships drawn directly from invasive cardiopulmonary exercise test data of obese patients undergoing investigation of unexplained exercise intolerance. To account for the influence of impaired LV filling on exercise in obesity which may increase RV exercise afterload, RV elastance, pulmonary arterial elastance and Ees/Ea ratios were measured in two groups of symptomatic obese patients demonstrating either normal or elevated LV filling pressures on exercise. Their exercise data were compared to a third group of non-obese controls also undergoing evaluation of unexplained dyspnoea. In a larger cohort of 37 obese patients and 18 controls drawn from the same referral pool, we examined relationships between BMI and exercise haemodynamics.
Discussion
The principle observations from this study were of exercise uncoupling between the RV and pulmonary artery and a reduction in RV contractile reserve in symptomatic obese patients both of which occurred irrespective of changes in exercise LV filling pressure. RV–arterial uncoupling was driven primarily by impaired RV contractile responses to higher exercise RV afterload in obesity. This suggests that even in the absence of elevated LV filling pressures on exercise, the RV in obesity may undergo maladaptation at higher flow. Potential explanations for this include any combination of higher RV exercise afterload specific to obesity, metabolic dysregulation of pulmonary vascular tone or the presence of RV myocardial dysfunction intrinsic to obesity.
In the larger cohort, the obesity exercise response in patients demonstrating raised LV filling pressures on exercise was characterised by lower peak cardiac output, a steeper mean pulmonary artery pressure-flow response and lower pulmonary vascular compliance. This same group also harboured lower FEV1 values which may have also contributed to lower pulmonary vascular compliance and high exercise afterload. The positive relationship between higher BMI and lower pulmonary vascular compliance at peak exercise, when all pulmonary capacitance vessels are maximally recruited, further suggests that increasing severity of obesity may predispose to increased RV afterload on exercise.
Reduced RV–arterial coupling ratios in obesity signify a loss in energetic efficiency between forward blood flow from the RV to pulmonary artery. This lends further insight into the pathophysiology of exercise RV dysfunction in obesity in that elevated LV filling pressures, which did not influence RV exercise coupling, appear to play a less critical role in moderating RV exercise responses. This as well as the finding of lower RV contractile reserve supports the hypothesis of intrinsic RV dysfunction in obesity suggesting an increasing predisposition to obesity-associated pulmonary vascular dysfunction may occur at upper extremes of BMI.
One potential mechanism of RV contractile impairment in obesity may be higher circulating plasma volume which on exercise predisposes to exercise RV dilatation through higher venous return (Alpert et al.
2014; Obokata et al.
2017). Increased sympathetic nervous system activation and metabolic dysregulation may also drive higher filling pressures in obesity (Ketabchi et al.
2009; Noble et al.
1981). Both obese groups in our study developed higher right atrial pressure with exercise suggesting excess RV preload. In the Obese+ePVH group, right atrial pressure rose even further perhaps through higher LV filling pressures and greater atrial septal interaction. In support of this finding, Obokata et al. recently showed higher ventricular filling pressures led to greater ventricular interdependence in obesity through greater pericardial restraint (Obokata et al.
2017). Although we could not directly measure RV volume in our study, higher RV filling pressures are likely to drive greater exercise RV dilatation in turn increasing RV wall stress and mechanical inefficiency (Alpert et al.
1989). Thus, greater metabolic fatigue within a RV operating at higher ventricular volumes may account for reduced contractile responsiveness such as we observed in obesity.
Exercise pulmonary vascular responses in obesity demonstrated significant elevation in RV exercise afterload. Peak exercise LV filling pressures were highest in the Obese+ePVH group (by study design); however, the Obese−ePVH group who also demonstrated high exercise afterload (high Ea), had similar exercise pulmonary arterial wedge pressure values to controls. This suggests upstream transmission of high LV filling pressure on exercise insufficiently accounts for the higher afterload observed in this group. One plausible explanation for higher exercise afterload in obesity may be inadequate pulmonary vasodilatation. This has a number of potential origins including enhanced sympathetic signalling associated with obesity, increased predisposition to exercise-associated pulmonary vasoconstriction via the metaboreflex and older age (Lykidis et al.
2008). Alongside this, obesity phenotypes associated with the metabolic syndrome may also predispose to impaired pulmonary endothelial function via reduced nitric oxide availability driven by vasoactive adipokines (Lai et al.
2016; Yudkin et al.
2005). It seems likely therefore that either age-related, mechanical or metabolic factors exert the predominant influence over exercise RV afterload in obesity with a lesser contribution from increased LV filling pressures.
Obese patients with and without high exercise LV filling pressures differed in several important characteristics. The Obese+ePVH group were older, had greater exposure to vasoactive medications particularly beta blockers and had lower values for FEV
1. In keeping with more severe exercise haemodynamic derangement and the older age of this group, LV compliance may have been reduced by increased prevalence of systemic hypertension and a longer duration of obesity (Alpert et al.
2014). Beta blockers may also have conceivably reduced RV contractile responses in four out of eight patients. Against this conclusion however was the finding of similar RV contractile reserve (ΔEes) in the Obese−ePVH group, where only one of eight patients took beta blockers. This makes a significant drug contribution of beta blockade in the Obese+ePVH group unlikely. Recent evidence suggests that reduction in tachycardia, observed in the Obese+ePVH group on exercise, may act to reduce RV pulsatile loading and thus total afterload (Metkus et al.
2016). We cannot therefore discount the possibility of both detrimental and beneficial effects of beta blockade on RV and pulmonary vascular exercise responses in our Obese+ePVH cohort.
Lower FEV
1 values in the Obese+ePVH group gave rise to marked differences in ventilatory response with a greater proportion of patients in this group developing a pulmonary mechanical limitation to exercise. In contrast to typical gas exchange responses in HFpEF without obesity, the Obese+ePVH group developed lower O
2 saturations and raised PaCO
2 levels at peak exercise compared to controls, driven by inadequate compensatory hyperventilation. The net result is a lowering of the VE/VCO
2 slope in obesity allowing for preserved VCO
2 at a lower level of alveolar ventilation. We did not undertake an isoWork analysis to enable meaningful gas exchange comparisons between groups; however, higher arterial and, by implication,
alveolar pCO
2 levels have been shown to provoke greater pulmonary vasoconstriction (Barer and Shaw
1971; Kregenow and Swenson
2002; Nishio et al.
2001; Noble et al.
1981; Sweeney et al.
1998; Viitanen et al.
1990) which may have added to RV afterload in obesity. Dempsey and Wagner have previously highlighted the influence of a raised PaCO
2 on exercise-induced hypoxaemia during maximal exercise (Dempsey and Wagner
1999). We found that exercise hypoxaemia was mild in both obese groups in our study, hence it is unlikely that gas exchange observations carried significant influence on hypoxic pulmonary vasoconstriction. Instead, lower FEV
1 in the Obese+ePVH group may reflect either an underlying restrictive lung deficit or presence of occult airflow obstruction not captured by exclusion of patients with a reduced FEV
1/FVC ratio < 70%. Restrictive lung function and dynamic hyperinflation are both prevalent in obesity and can increase RV afterload through increased pulmonary vascular resistance (Pinsky
2016) and higher cardiac filling pressures. However given lack of available lung volume data, we used electronic averaging of haemodynamic pressures throughout the respiratory cycle to standardise against respiratory variation when analysing group differences in RV contractile response (Boerrigter et al.
2014).
Our study’s findings of exercise RV–arterial uncoupling and impaired contractile reserve in obesity complements recent reports of exercise RV dysfunction in HFpEF and obesity-associated HFpEF in which RV exercise performance may be compromised by both reduced RV contractile reserve and higher exercise afterload (Borlaug et al.
2016; Borlaug and Obokata
2017; Obokata et al.
2017). Compared to these populations, our obese groups maintained a high/normal pulmonary vascular resistance and lower pulmonary vascular compliance suggesting exercise elevation in mean pulmonary artery pressure was more likely an independent obesity effect than simply a passive response to high LV filling pressure on exercise (Oliveira et al.
2016). The development of HFpEF on exercise is still an emerging concept with no current agreed haemodynamic definition in place. This is despite evidence for both ‘passive’ and ‘reactive’ forms recently shown to carry prognostic relevance (Huang et al.
2017). Nevertheless, our data show exercise RV–arterial uncoupling in obese patients constitutes an abnormal pulmonary vascular response to exercise with the most significant contributor to RV afterload being a reduction in exercise pulmonary vascular compliance.
Limitations
Our study was retrospectively conducted and so represented a highly selected group of symptomatic obese patients with relatively low levels of co-morbidity which may limit extrapolation to all obese patients, especially those with fewer symptoms. Specifically, the administration of medication to treat systemic hypertension may have modified RV contractile and afterload responses in an unpredictable manner. Our single-beat model also included presumptive use of mean pulmonary artery pressure as the end-systolic point of the RV pressure–volume relationship, which may lead to overestimation of contractility and afterload. As right atrial pressure also increases with exercise, the presumptive use of mPAP/mean SV likely underestimated the RV–arterial coupling ratios reported (Spruijt et al.
2015). Finally, the cross-sectional design of this study precludes insight into time-varying pathophysiology at different disease stages with the associated risk of Type 1 error from multiple statistical hypotheses.