Abstract
The mean pulmonary artery pressure (P̄pa) achieved on mild-to-moderate exercise is age related and its haemodynamic correlates remain to be documented in patients free of pulmonary hypertension (PH).
Our retrospective study involved patients free of PH investigated in our centre for possible pulmonary vascular disease between January 1, 2007 and October 31, 2009 who underwent right heart catheterisation at rest and during supine exercise up to 60 W. The 38 out of 99 patients aged <50 yrs were included and a P̄pa of 30 mmHg was considered the upper limit of normal on exercise.
The 24 subjects who developed P̄pa>30 mmHg on exercise had higher resting P̄pa (19±3 versus 15±4 mmHg) and indexed pulmonary vascular resistance (PVRi; 3.4±1.5 versus 2.2±1.1 WU·m2; p<0.05) than the remaining 14 subjects. Resting P̄pa >15 mmHg predicted exercise P̄pa >30 mmHg with 88% sensitivity and 57% specificity. The eight patients with resting P̄pa 22–24 mmHg all had exercise P̄pa >30 mmHg.
In subjects aged <50 yrs investigated for possible pulmonary vascular disease and free of PH, patients with mild-to-moderate exercise P̄pa >30 mmHg had higher resting PVRi and higher resting P̄pa, although there was no resting P̄pa threshold value that could predict normal response on mild-to-moderate exercise. The clinical relevance of such findings deserves further long-term follow-up studies.
For the last 30 yrs, the diagnosis of pulmonary hypertension (PH) depended on either a resting mean pulmonary artery pressure (P̄pa) of >25 mmHg or an increase in P̄pa on exercise to >30 mmHg, with the pulmonary capillary wedge pressure ≤15 mmHg in the subgroup of pre-capillary PH. Since the 4th World Conference on PH, new guidelines have recommended that the exercise criterion should be eliminated [1, 2], given both the marked age-dependency of “normal” P̄pa threshold on exercise [3] and the paucity of robust data supporting its clinical relevance [1, 2]. The age-dependency of P̄pa is much less at rest [3–7], such that a common 20.6 mmHg upper limit of normal (ULN) was suggested in supine healthy subjects [3]. Though a P̄pa of ≥21 mmHg is beyond the normal range (mean +2 standard deviations) and may be suspicious of pulmonary vascular disease, a small but significant proportion of apparently normal individuals will have a P̄pa ≥21 mmHg and they will outnumber the previously documented proportion of patients with PH [3]. As a result, new guidelines have defined pulmonary hypertension by a P̄pa at rest ≥25 mmHg (mean +3 standard deviations), and have also highlighted the fact that studies focusing on patients with resting P̄pa of 21–24 mmHg are especially needed [1–3].
Numerous studies have documented the high percentage of patients at high risk for PH exhibiting elevation of P̄pa on exercise >30 mmHg while their P̄pa was normal at rest [8–16], and this may be considered as an early manifestation of pulmonary vasculopathy [12, 14, 17, 18]. Most of these studies had been carried out in middle-aged patients, at a time when normal P̄pa values during exercise have not yet been defined. Recently, the review of the range of pulmonary haemodynamic responses to exercise in normal subjects by Kovacs et al. [3] was timely in alerting the community to the huge amount of available data supporting the “classical” definition of exercise induced PH in patients aged <50 yrs, while the ULN of 30 mmHg was not always supported by the available data in older patients. Thus, the precise relationship between resting P̄pa and the age-related P̄pa responses during mild-to-moderate exercise still deserve further studies in patients free of PH, and this may have implications for improving our understanding of PH pathophysiology.
The present study examined the range of haemodynamic responses in individuals aged <50 yrs with resting P̄pa <25 mmHg being investigated for possible pulmonary vascular disease in our institution who underwent measurement of pulmonary haemodynamic responses to mild-to-moderate exercise while supine.
METHODS
This was a retrospective study. We extracted the catheter laboratory records of all patients who underwent diagnostic right heart catheterisation at the Centre National de Référence de l'Hypertension Pulmonaire Sévère, Hôpital Antoine Béclère, Assistance Publique Hôpitaux de Paris, Université Paris-Sud, Paris, France over a 34-month period (January 1, 2007–October 31, 2009).
We included patients with a P̄pa at rest of <25 mmHg and a pulmonary artery occlusion pressure (Ppao) ≤15 mmHg who underwent progressive supine exercise test during the right heart catheterisation procedure. Patients with unexplained exertional dyspnoea or an abnormal screening echocardiogram P̄pa were included, as well as patients with a history of probable or possible pulmonary thromboembolic disease being investigated for chronic thromboembolic pulmonary hypertension. As our aim was to examine a “real life” patient population undergoing diagnostic right heart catheterisation, we did not exclude patients with significant comorbidities, including diseases known to carry a risk of PH [1, 2]. We also excluded patients in whom acceptable quality Ppao could not be obtained. On exercise, Ppao >20 mmHg was considered as abnormal but the corresponding patients were not excluded a posteriori. The 6-min walking distance and respiratory and biological tests were obtained according to our routine protocol. Our retrospective study was compliant with requirements of the French Commission Nationale de l'Informatique et des Libertés (CNIL), and right heart catheterisation with exercise is part of the usual care at our institute.
Amongst the 99 eligible patients, only patients aged <50 yrs (n=38) were included in our final analysis and 30 mmHg was considered the ULN on mild-to-moderate exercise [3]. Patients aged ≥50 yrs were excluded, given that Kovacs et al. [3] have suggested that an upper limit of 30 mmHg could not be supported by the available data in such subjects. The main risk factors and comorbidities were a previous history of thromboembolic pulmonary disease (n=14, 37%), connective tissue disease (n=8, 21%; namely two lupus and six systemic sclerosis) and anorexigens intake (n=5, 13%) (table 1).
Patients had baseline haemodynamic measurements showing resting P̄pa <25 mmHg and Ppao ≤15 mmHg. They then carried out supine bicycle exercise ergometry [19, 20], including baseline measurements with feet in the pedals but no dynamic exercise followed by a stepwise increase of load. The number of steps and the pattern of increase in load were determined for each individual by the operator's judgement based on the patient's age, comorbidities and clinical response to initial load. As we have concentrated on examining the haemodynamic response at mild-to-moderate exercise, we examined the data obtained <60 W [3]. As our standard protocol was developed before the new guidelines from the 4th World Conference on PH [1, 2], the exercise was terminated in cases where the P̄pa was noted to be >30 mmHg. In patients whose exercise was terminated prior to 60 W, either due to symptoms or reaching a P̄pa >30 mmHg, we analysed their response at their highest workload.
Statistics
Data are presented as mean±sd. Comparisons at baseline (rest) were performed by using one-way ANOVA followed by unpaired t-test. The haemodynamic effects of exercise were compared between patients with normal and abnormal P̄pa on mild-to-moderate exercise by using a two-way ANOVA (group × time interaction). Correlations were tested by using the least squares method. Frequency distribution of both sex and P̄pa responses between subgroups were compared using the Chi-squared test. Receiver operating characteristic (ROC) curves (with 95% confidence interval) were constructed for testing the ability of the resting P̄pa to predict mild-to-moderate exercise P̄pa >30 mmHg. A p-value <0.05 was considered statistically significant. The statistical analysis was performed using StatView 512 software (Abacus concepts, Berkeley, CA, USA), except for ROC curves analysis which was performed by using MedCalc8.1.0.0 software (Mariakerke, Belgium).
RESULTS
The study population (n=38) comprised 30 females and eight males (age = 40±8 yrs); their clinical characteristics are listed in table 1. Median workload was 40 W (mean±sd 41±16 W). Overall, 24 out of 38 (63%) of patients developed P̄pa >30 mmHg on mild-to-moderate exercise. As compared with the remaining 14 subjects, the 24 patients who developed P̄pa>30 mmHg on mild-to-moderate exercise had lower body surface area, lower forced expiratory volume in 1 s, and lower diffusing capacity of the lung for carbon monoxide (table 1). Differences in risk factors and comorbidities were also observed between the two groups (table 1). The two groups had similar sex ratio, age, systolic and diastolic arterial pressure, cardiac frequency and haemoglobin and brain natriuretic peptide blood content. The 6-min walking distance was 478±99 m in the 24 patients who developed P̄pa >30 mmHg and 550±100 m in the remaining 14 patients (p=0.062) (table 1).
The haemodynamic characteristics of the study population are listed in table 2. Individual haemodynamic data are presented as online supplementary material. The 24 patients who developed P̄pa >30 mmHg on mild-to-moderate exercise had higher resting P̄pa (19±3 versus 15±4 mmHg; p<0.01) and higher indexed pulmonary vascular resistance (PVRi) at rest (3.4 ±1.5 versus 2.2 ±1.1 WU·m−2; p<0.05) compared with the remaining 14 subjects, (table 2). They also had similar Ppao and cardiac index at rest (table 2), and similar right atrial pressure at rest (4±3 versus 5±3 mmHg; p = NS).
In the overall study population, there was a weak positive relationship between resting P̄pa and mild-to-moderate exercise P̄pa (r2=0.44; p<0.001) (fig. 1). There was no relationship between age and either resting P̄pa or mild-to-moderate exercise P̄pa.
Linear relationship between mean pulmonary arterial pressure whilst exercising (P̄pa,ex) and at rest (P̄pa,rest). n=38; r2=0.44; p<0.001.
Haemodynamic responses to mild-to-moderate exercise in the two subgroups are detailed in table 3 and individual P̄pa, Ppao and cardiac index values are presented as online supplementary material. Cardiac index increased in a similar way and PVRi remained unchanged in the two subgroups (table 3). Mild-to-moderate differences in Ppao changes (p=0.047) were documented between the two subgroups. On exercising, two patients had Ppao>20 mmHg (25 and 21 mmHg, see online supplementary material) with >12 mmHg transpulmonary pressure gradient (24 and 13 mmHg, respectively), and both had P̄pa >30 mmHg.
Exercising P̄pa exceeded 30 mmHg in 15 out of 27 (55%) of the patients with resting P̄pa <21 mmHg and in nine out of 11 (82%) of the patients with resting P̄pa between 21 and 24 mmHg (p = NS) (table 4). The eight patients with resting P̄pa 22–24 mmHg all had an exercising P̄pa >30 mmHg. ROC curve analysis (fig. 2) indicated that a resting P̄pa >15 mmHg predicted an exercising P̄pa >30 mmHg with 88% sensitivity (95% CI 68–97%) and 57% specificity (95% CI 29–82%).
Receiver operating characteristic curve showing mean pulmonary arterial pressure at rest (P̄pa,rest) >15 mmHg and predicted exercise P̄pa >30 mmHg with 88% sensitivity (95% CI 68–97%) and 57% specificity (95% CI 29–82%).
DISCUSSION
Our retrospective study was performed in 38 patients aged <50 yrs, free of PH (resting P̄pa <25 mmHg), being investigated in our centre for possible vascular disease between January 1, 2007 and October 31, 2009. The main results were as follows: 1) patients with mild-to-moderate exercise P̄pa >30 mmHg had higher resting PVRi and higher resting P̄pa; 2) it was not possible to reliably set a lower limit of resting P̄pa that guarantees normal P̄pa at mild-to-moderate exercise loads; and 3) all eight patients with resting P̄pa 22–24 mmHg had exercising P̄pa >30 mmHg. The clinical relevance of such findings deserves further long-term follow-up studies.
The present study was undertaken following recent articles and editorials that stressed the necessity of further research in the area of haemodynamics in patients with pulmonary vascular diseases, with special focus on the potential link between resting and exercising pulmonary haemodynamics and on the significance of resting P̄pa 21–24 mmHg [1–3, 12, 17, 18]. Numerous studies [8–16] have documented the so-called “exercise-induced pulmonary hypertension” [12, 13, 15, 17] frequently observed in various populations carrying a high risk of PH while their P̄pa was normal at rest. To the best of our knowledge, our study is the first to take into account the recent recommendations made by Kovacs et al. [3], namely that the 30 mmHg ULN for the P̄pa achieved on mild-to-moderate exercise fairly applies only in patients aged <50 yrs. Thus elderly patients (61 out of 99) were not included in our final analysis, given that an ULN of 30 mmHg could not be supported by the available data in such patients [3].
Our study focused on mild-to-moderate exercise only, and this was based on the following rationale. First, the literature review made it possible to define reliable ULN for P̄pa during mild-to-moderate exercise [3, 5]. Secondly, reliable and consistent P̄pa, Ppao and cardiac output data have been published during mild-to-moderate exercise [4, 5, 21], thus allowing pathophysiological interpretation of our data. Finally, light exercise may reflect the daily life physiological stress put on the pulmonary circulation and right ventricle more accurately than maximal exercise [1, 2].
In healthy subjects aged <50 yrs, the resting P̄pa is ∼14 mmHg on average [1–3], and the haemodynamic changes on mild-to-moderate exercise while supine slightly differ according to the research team, with either unchanged PVR [5, 22], or slightly decreased PVR [23]. Pulmonary capillary pressure may increase slightly [3, 5, 6, 24], although other studies and reference textbooks often indicate unchanged pulmonary capillary pressure during exercise. The 18 mmHg resting P̄pa value documented in our study (table 2) is consistent with that previously reported in populations similar to ours [12–15] and reflects the fact that patients were investigated in our centre for possible vascular disease. It has been suggested that age, sex and resting systolic blood pressure significantly influence P̄pa responses to exercise, but all were similar in the two groups (table 1). The underlying risk factors and comorbidities (table 1) may contribute, at least in part, to explaining the high percentage (24 out of 38; 63%) of patients exhibiting abnormal P̄pa responses [8–16].
The stress put on the right ventricle is minimal at rest and this may in part explain why resting pulmonary haemodynamics do not correlate highly with exercise pulmonary haemodynamics in patients with established PH [6, 8, 12, 19, 20, 25, 26]. In our patients at risk for PH and exhibiting normal P̄pa at rest, ROC curve analysis indicated that resting P̄pa >15 mmHg predicted exercise P̄pa >30 mmHg with 88% sensitivity and 57% specificity. Interestingly, Saggar et al. [15] have suggested that resting P̄pa ≥14 mmHg was associated with abnormal P̄pa responses on exercise in patients with systemic sclerosis. However, in our study, it was not possible to reliably set a lower limit of resting P̄pa that guarantees normal P̄pa at mild-to-moderate exercise loads
As far as the upper limit that guarantees abnormal P̄pa at mild-to-moderate exercise loads is concerned, it may be expected that the closer the resting P̄pa lies to the ULN on exercise (30 mmHg) the more likely the exercise P̄pa threshold is to be breached. Resting P̄pa was consistently higher in the 24 patients who developed P̄pa >30 mmHg on mild-to-moderate exercise, and this was explained by the 55% higher levels for resting PVRi as compared with the remaining 14 patients who did not develop P̄pa >30 mmHg on mild-to-moderate exercise (table 3). This could also explain why exercise P̄pa was >30 mmHg in 82% (nine out of 11) of the patients with resting P̄pa 21–24 mmHg and in all eight patients with resting P̄pa of 22–24 mmHg.
Significant differences in Ppao changes were also documented and contributed to explaining differences in exercising P̄pa in the two subgroups (table 3). Amongst the 24 patients with P̄pa >30 mmHg on exercise, two (8%) had an exercising Ppao of >20 mmHg (see online supplementary material), thus confirming that acute left ventricular dysfunction could also contribute to the rise in P̄pa, e.g. diastolic dysfunction [27, 28]. Conversely, similar cardiac output responses on exercise were documented in the two subgroups and gave similar PVRi responses (table 3). In summary, in patients aged <50 yrs and free of PH, the increased PVRi at rest resulting in higher resting P̄pa was the main factor likely to explain abnormally high P̄pa on mild-to-moderate exercise. Additionally, further exercise-related increases in capillary wedge pressure also played a role.
Our study did not involve healthy subjects, but patients with symptoms and a certain risk of PH. Accordingly, the results cannot be used to create novel thresholds of physiological changes during exercise and may not be compared with studies examining healthy individuals. The clinical heterogeneity of the study group reflects the current “real-life” experience of a reference PH centre. Other limitations include the retrospective study design and the lack of extensive assessment of left ventricular function at rest (e.g. detailed echocardiography to detect diastolic and/or systolic dysfunction). The intrinsic limitations related to the exercise protocol must also be discussed. We have examined P̄pa responses on mild-to-moderate exercise as best as we could, with the understanding that we do not measure oxygen consumption during our right heart studies. We could not determine the slope and pressure axis intercept of the P̄pa–cardiac output relationship, as the number of data points and pattern of exercise varied between individuals. The determination of multipoint P̄pa–cardiac output plots provides a more accurate insight into the nature of PVR than the single-point PVR, as the intercept may be higher than pulmonary capillary pressure [6, 7, 19, 20, 29, 30]. Thus, the observed pattern of increased transpulmonary pressure gradient and increased cardiac output together with unchanged or decreased single-point PVR does not necessarily reflect unchanged or decreased resistive properties of the pulmonary circulation [7, 20, 29, 30]. Similarly, we cannot exclude the possibility that our results reflect averaging patients with various patterns of P̄pa–cardiac output relationship on exercise [20]. The two patients with exercise Ppao >20 mmHg were included in our final analysis as our aim was to study the relationship between resting and mild-to-moderate exercise P̄pa in patients at risk of PH but free of PH at rest (P̄pa <25 mmHg) and with normal filling pressure at rest (Ppao ≤15 mmHg). Finally, elderly subjects could not be studied for the above-mentioned reasons and further studies focusing on this population are thus needed.
The implications of our study must be carefully considered. First of all, we wish to emphasise the fact that our study did not intend to challenge the 4th World Conference proposal that exercise testing must be abandoned in the definition of PH [1, 2]. However, we remain concerned by the fact that the new consensus does sometimes leave clinicians faced with a patients that have symptoms suggestive of pulmonary vascular disease but with resting P̄pa <25 mmHg [12, 14, 18]. Interestingly, our study pointed to a major redundancy between resting and mild-to-moderate exercise P̄pa values in the subgroup of patients <50 yrs with a resting P̄pa of 22–24 mmHg. The 22–24 mmHg range of resting P̄pa may help clinicians to recognise patterns consistent with abnormal haemodynamic responses on mild-to-moderate exercise. Elsewhere, our study demonstrates a lack of tight correlation between resting and exercise haemodynamics in non-PH patients. In other words, our data would suggest that it is not possible to reliably set a lower limit of resting P̄pa that guarantees that pulmonary haemodynamic responses to exercise will be normal at mild-to-moderate exercise loads in patients <50 yrs.
In conclusion, in subjects aged <50 yrs and free of PH, patients with mild-to-moderate exercise P̄pa >30 mmHg had higher resting PVRi and higher resting P̄pa. Although all patients with resting P̄pa 22–24 mmHg had an exercising P̄pa >30 mmHg, there was no resting P̄pa threshold value that could reasonably predict normal/abnormal response on mild-to-moderate exercise. The clinical relevance of such findings deserves further long-term follow-up studies.
Acknowledgments
The authors thank the nurses of the catheterisation laboratory for their helpful work. Part of the study has been presented at the 2010 European Respiratory Sociey Congree held in Barcelona, Spain.
Footnotes
For editorial comments see page 231.
This article has supplementary material available from www.erj.ersjournals.com
Statement of Interest
Statements of interest for K. Whyte, S. Hoette, D. Montani, L. Savale, D.S. O’Callaghan, G. Garcia, O. Sitbon, G. Simonneau and M. Humbert can be found at www.erj.ersjournals.com/site/misc/statements.xhtml
- Received February 3, 2011.
- Accepted June 23, 2011.
- ©ERS 2012
REFERENCES
