British Journal of Anaesthesia 112 (1): 57–65 (2014) Advance Access publication 4 November 2013 . doi:10.1093/bja/aet370
Comparison of oxygen uptake during arm or leg cardiopulmonary exercise testing in vascular surgery patients and control subjects L. Loughney 1, M. West1,2, S. Pintus 3, D. Lythgoe 5, E. Clark 3, S. Jack 1,2 and F. Torella 4* 1
Critical Care Research Area, NIHR Respiratory BRU, University of Southampton and University Hospital Southampton, Southampton, UK Institute of Ageing and Chronic Disease, Department of Musculoskeletal Biology II, University of Liverpool, Liverpool, UK 3 Department of Anaesthesia and 4 Liverpool Vascular and Endovascular Service, University Hospital Aintree, Liverpool, UK 5 Cancer Research UK, Liverpool Cancer Trials Unit, UK 2
* Corresponding author. E-mail: [email protected]
† Cardiopulmonary exercise testing by cycle ergometry has limited utility as a preoperative assessment tool in patients with lower limb dysfunction. † Leg ergometry was compared with arm ergometry in vascular surgery and healthy patients. † Oxygen uptake using the two methods was correlated, but arm ergometry was a poor predictor of leg ergometry. † Further study is required to establish the perioperative utility of arm ergometry.
Background. Cardiopulmonary exercise testing by cycle ergometry (CPETleg) is an established assessment tool of perioperative physical fitness. CPETutilizing arm ergometry (CPETarm) is an attractive alternative in patients with lower limb dysfunction. We aimed to determine ˙ 2 ) obtained by CPETleg could be predicted by using CPETarm whether oxygen uptake (Vo alone and whether CPETarm could be used in perioperative risk stratification. ˙ 2 obtained Methods. Subjects underwent CPETarm and CPETleg. To evaluate the ability of Vo ˙ 2 from CPETleg, we calculated prediction intervals (PIs) at lactate from CPETarm to predict Vo threshold (uˆL ) and peak exercise in both groups. Receiver operating characteristic (ROC) curves were used to risk stratify patients into high and low categories based on published criteria. Results. We recruited 20 vascular surgery patients (17 males and three females) and 20 ˙ 2 at uˆL and peak healthy volunteers (10 males and 10 females). In both groups, PIs for Vo ˙ 2 at uˆL CPETarm ranged from 55% were wider than clinically acceptable (patient group—Vo to 108% of CPETleg and from 54% to 105% at peak; healthy volunteers—37– 77% and ˙ 2 in patients was 0.84 [95% 41– 79%, respectively). The area under the ROC for CPETarm Vo confidence interval (CI): 0.66, 1.0] at uˆL , and 0.76 (95% CI: 0.54, 0.99) at peak. ˙ 2 values for CPETarm and CPETleg, this Conclusions. Although a relationship exists between Vo is insufficient for accurate prediction using CPETarm alone. This however does not necessarily preclude the use of CPETarm in perioperative risk stratification. Keywords: arm ergometry; cardiopulmonary exercise testing; cycle ergometry; preoperative assessment; vascular surgery Accepted for publication: 18 August 2013
Cardiopulmonary exercise testing (CPET) provides an integrated assessment and quantification of the cardiorespiratory system at rest and under stress of maximal exercise. CPET is gaining popularity as a physical fitness assessment tool before major elective surgery, including vascular, with evidence that variables ˙ 2 ) at peak exercise and at estimated such as oxygen uptake (Vo lactate threshold (uˆL ) predict short- and long-term outcomes.1 – 5 Recently, CPET to maximal exercise has been suggested as identifying patients at risk of early perioperative death after elective abdominal aortic aneurysm (AAA) repair.6 CPET in the clinical setting is conducted on a cycle ergometer (CPETleg), but many vascular surgery patients are unable to perform this due to lower limb dysfunction such as joint arthritis, peripheral vascular disease, neurological disease, or previous
amputations. CPETon an arm ergometer (CPETarm) is an attractive alternative as it provides data on physiological responses similar to those obtained by CPETleg. This has been validated in healthy subjects;7 however, it is currently not an accepted test in routine preoperative assessment. The maximum oxygen uptake obtained byarm ergometry is 60–80% of that measured by leg ergometry in healthy individuals,8 – 13 with a recent study suggesting ˙ 2 at uˆL and Vo ˙ 2 at peak.14 The lower Vo ˙ 2 values 34% less Vo obtained during CPETarm could be due to smaller muscle mass, distribution of fast twitch muscle fibres, recruitment pattern of motor units,13 15 and greater glycolytic enzyme activity.15 Literature around arm ergometry is mainly on healthy individuals11 – 13 with no literature available comparing arm and leg exercise testing in a patient population.
& The Author [2013]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Editor’s key points
BJA The primary aim of this study was to determine whether oxygen uptake derived from CPETleg could be accurately predicted by using CPETarm alone in both healthy volunteers and perioperative patients. We considered an a priori predictive ˙ 2 at uˆL and error for CPETarm of +1.5 ml kg21 min21 for Vo ˙ 2 at peak to be clinically acceptable. +3.0 ml kg21 min21 for Vo We also aimed to explore whether patients could be risk strati˙ 2 obtained by CPETarm in a similar fashion as for fied using Vo CPETleg (with known cut-off values of 10.2 ml kg21 min21 for ˙ 2 at peak).6 Finally, we ˙ 2 at uˆL and 15 ml kg21 min21 for Vo Vo aimed to explore the validity of a previously published simple ˙ 2 obtained percentage proportionality relationship between Vo from CPETarm and CPETleg.8 – 11
Subjects We prospectively recruited two cohorts: (i) preoperative patients being assessed before elective abdominal aortic surgery (patient group), and (ii) healthy volunteers. We chose these two groups in order to uncover potential differences in response to arm and leg exercise in elderly patients in comparison with younger healthy subjects. After ethical approval (09/H1001/94) and written informed consent, preoperative patients were requested to undergo CPET twice: CPETleg, as part of their normal preoperative assessment process, and a CPETarm, as part of this study. Healthy volunteers also underwent the same two tests using the protocols detailed below. Eligible patients were free of acute illness or clinically evident peripheral vascular disease, and did not have any disability precluding arm or leg exercise. Healthy volunteers were untrained and free from illness. CPET was reported by an experienced clinical scientist (S.J.) who was blinded to the mode of exercise testing.
Cardiopulmonary exercise testing Symptom-limited CPET was conducted in accordance with American Thoracic Society/American College of Chest Physicians recommendations.16 17 CPET was performed on calibrated electromagnetically braked cycle and arm ergometers. Gas and flow calibration was performed before each test. Both leg and arm CPET were carried out using similar ramped protocols set to 10–25 W min21 based on a calculation described by Wasser˙ 2 at unloaded pedalman and colleagues18 using predicted Vo ˙ 2 at peak exercise, height, and patient age. ling, predicted Vo Both protocols consisted of 3 min of rest, followed by 3 min of unloaded exercise, then the loaded ramp increased until volitional termination. This was followed by 5 min of recovery. Subjects were monitored throughout each test using pulse oximetry, 12-lead electrocardiography, and non-invasive arterial pressure monitoring. Ventilation and gas exchange variables were measured using a metabolic cart (Geratherm Respiratory GmbH (Love Medical Ltd, UK) and a cycle ergometer (Love Medical Ltd, Ergoline) or arm ergometer (Love Medical Ltd, Ergoline). For CPETleg, seat height was adjusted to ensure that full knee extension was achieved when the pedal was in the down position and handlebars raised to maintain full weight supported exercise. Subjects were instructed to cycle at a speed of
58
55–65 RPM during the test. This was monitored by the participant through a light-emitting diode display. The test was ended if RPM decreased below 45 or symptoms were encountered. For CPETarm, the arm ergometer was adjusted to ensure that participants sat on the chair with their arms slightly flexed, and maintaining their feet flat on the floor. They were asked to grasp handles in front of them, and ‘pedal’ with their arms in a circular motion, maintaining 55–65 RPM during the test until they could no longer push against the resistance or if the RPM decreased below 45. Breath-by-breath data were collected through a face mask and flow sensor that were appropriately fitted to each participant.
Measurements Subject characteristics, including age, gender, height, weight, and clinical details, were recorded. Before the first CPET, resting flow-volume loops were measured in the patient group to derive forced expiratory volume over 1 s (FEV1) and forced vital capacity (FVC). CPET variables measured included: ˙ 2 , carbon dioxide output, tidal expired ventilatory volumes, Vo volume, minute ventilation, work rate, respiratory exchange ˙ 2 (ml kg21 min21) at uˆL and at peak ratio, and oxygen pulse. Vo exercise were the primary outcome variables recorded. uˆL was ˙ 2 − Vo ˙ 2 relaestimated conventionally (breakpoint in the Vco tionship), with increases in ventilatory equivalent for oxygen ˙ 2 ) and end-tidal (PE′ ) oxygen but no increase in ventilatory (V˙ E /Vo ˙ 2 ) or decrease in PE′CO 19 by equivalent for carbon dioxide (V˙ E /Vco 2 18 ˙ 2 was averaged an experienced, blinded, assessor. The peak Vo over the last 30 s of exercise.
Data analysis Continuous variables are summarized using the median and inter-quartile range (IQR). Paired t-tests were used to compare CPETarm and CPETleg values within patient groups. Linear regression was used to assess whether CPETleg could be predicted using CPETarm alone by estimating 95% prediction intervals (PIs). The PIs should be interpreted as a measure of how accurate our prediction of CPETleg would be for a new patient given only a CPETarm measurement. In order to meet the assumptions of linear regression, it was necessary to natural log transform CPETleg. For the ease of interpretation, the linear regression results were presented on the original scale, with the consequence that the resulting PIs were asymmetric and non-constant. For simplicity, the predictive ability of CPETarm was evaluated against the predefined acceptable widths at the CPETleg sample mean. Arm and leg measurements were compared using paired t-tests. The assumption of normality for the mean of the paired differences was assessed using the Q–Q plots. Scatter plots of CPETleg vs CPETarm were constructed and boundaries representing CPETarm as a percentage of CPETleg were added. The purpose of these plots was to establish whether a simple relative relationship rule such as ‘CPETarm is typically x–y% of CPETleg’ could be advised. Finally, nonparametric receiver operating characteristic (ROC) curves were used to determine whether CPETarm could discriminate
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Methods
Loughney et al.
BJA
Oxygen uptake in arm or leg ergometry
between high- and low-risk patients and hence become a useful perioperative risk stratification tool. Risk groups (high/ low) were established by dichotomizing CPETleg at 10.2 ml ˙ 2 at uˆL and 15 ml kg21 min21 for Vo ˙ 2 at peak kg21 min21 for Vo based on published cut-off points (≥cut-off—low risk and ,cut-off—high risk).6 All results were considered as statistically significant at the 5% level. All statistical analyses were carried out using Stata version 12.1 (StataCorp., 2011. Stata Statistical Software: Release 12, College Station, TX, USA).
Results
obtained from CPETleg (Fig. 5). The area under the ROC (AUROC) ˙ 2 at uˆL in the patient group was 0.84 [95% confidence for Vo ˙ 2 at peak, the AUROC was interval (CI): 0.66, 1.0], while for Vo 0.76 (95% CI: 0.54, 0.99).
Discussion This study clearly showed that it is not possible to predict ˙ 2 obtained by CPETleg from CPETarm, in patients or healthy Vo ˙ 2 at uˆL and peak were unacceptvolunteers. The PIs for both Vo ably wide. We also found that CPETarm has a discrete ability to ˙ 2 obtained by CPETleg at predict CPETleg when dichotomizing Vo the cut-off points described by Hartley and colleagues;6 however, it must be emphasized that in the proposed model, ˙ 2 at uˆL cut-off was not used in isolation but as part of a the Vo more complex risk prediction model. Furthermore, we must also emphasize that although this might be deemed useful in perioperative risk stratification (especially when extremes of ˙ 2 are obtained by CPETarm), we are unable to justify the use Vo of CPET by arm ergometry in this setting without further robust studies. Finally, when exploring the proportionality rela˙ 2 obtained from CPETarm and CPETleg, we tionship between Vo ˙ 2 values of CPETarm as a profound very high variability in the Vo portion of CPETleg, such that any linear relationship would appear to be overly simplistic and would not accurately reflect the relative relationship between the two measurements. It is not uncommon, in our practice, to encounter patients who are unsuitable for CPET on a cycle ergometer due to co-morbidity affecting the lower limbs. In these patients, arm ergometry is, intuitively, a potential alternative which, however, has not yet been clinically validated. We decided to test our hypothesis in two different groups of subjects (patients and healthy volunteers) because we suspected that elderly vascular patients could show a higher degree of variability between the two tests due to subclinical co-morbidity (e.g. mild joint arthritis in shoulder/hip/knee or COPD). In general, ˙ 2 measured by the we found a strong correlation between Vo two tests, at uˆL and at peak exercise, with CPETarm being systematically lower than CPETleg. Although our findings in patients were mirrored by those in volunteers, it was notable ˙ 2 at uˆL during CPETarm was never meathat, in all volunteers, Vo sured at more than three-quarters of the value recorded during CPETleg, while in patients, higher proportional values were seen
59
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
We prospectively screened 26 and recruited 23 (19 men, four women) vascular surgery patients. Three patients dropped out due to withdrawal of consent, aneurysmal rupture, and expedited elective surgery. Twenty patients were included with the median (IQR) age of 74 (70 –77) yr and with body mass index of 27 (25– 29) kg m22. Baseline pulmonary function tests showed a generalized mild –moderate obstructive pattern [according to GOLD spirometric criteria for chronic obstructive pulmonary disease (COPD) severity] with a median (IQR) FEV1 of 93% (75.5 –105.3), FVC 110% (97.8 –128.5), and FEV1/FVC ratio of 63.5 (59 –69) of their predicted values. Seventeen out of 20 patients had a median FEV1/FVC ratio of 61 (59 –69); of which, six had an FEV1 of ,80%. Patients were undergoing preoperative evaluation before elective AAA repair (n¼16), elective thoraco-abdominal aortic aneurysm repair (n¼1), combined AAA repair and cancer resection (n¼2), and elective aorto-mesenteric bypass (n¼1). Twenty healthy volunteers (10 men and 10 women) were also recruited, median (IQR) age of 31 (24–42) yr and median (IQR) body mass index of 26 (24–28) kg m22. The order of the CPET was not randomized. Arm and leg CPET was not carried out on the same visit and not more than 14 days apart [median 7 (range 6–14) days]. ˙ 2 and workload For both patients and healthy volunteers, Vo values were lower during CPETarm compared with CPETleg (P,0.0001; Table 1). One patient experienced chest pain with electrocardiographic changes during CPETleg, but no signs or symptoms were reported during CPETarm, while another experienced significant oxygen desaturation (to 85%) during CPETleg, but not CPETarm. In both groups, there was a strong correlation between ˙ 2 measured by the two tests, at uˆL (patient group—r¼0.72, Vo P¼0.0003; healthy volunteers—r¼0.64, P¼0.0024) and at peak exercise (patient group—r¼0.75, P¼0.0001; healthy ˙ 2 CPETarm being sysvolunteers—r¼0.77, P¼0.0001), with Vo ˙ 2 CPETleg. After fitting a simple tematically lower than Vo ˙ 2 at uˆL of 11.1 linear regression model, at the mean CPETleg Vo ml kg21 min21 for the patient group, the PI’s width was 8.0 ˙ 2 at peak, the total width of (4.0) ml kg21 min21 (Fig. 1A). For Vo the PI was 14.0 ml kg21 min21 at the mean of 16.6 ml kg21 min21; approximately +7.0 ml kg21 min21 (Fig. 1B). In the ˙ 2 at uˆL of 21.7 ml kg21 healthy volunteer group, at the mean Vo 21 min , the total PI width was 19.9 (10.0) ml kg21 min21. For ˙ 2 at peak, the total width of the PI was 25.3 ml kg21 min21 Vo
at the mean of 35.8 ml kg21 min21; approximately +12.5 ml kg21 min21 (Fig. 2A and B). The scatter plots in Figures 3 and 4 depict the relationship ˙ 2 at uˆL and Vo ˙ 2 at peak for CPETarm and CPETleg between Vo with overlaid percentage boundaries for CPETarm as a propor˙ 2 at uˆL values ranged tion of CPETleg. In the patient group, Vo ˙ 2 at peak values from 55% to 108% (median 79%) andVo ranged from 54% to 105% (median 80%). In the healthy volun˙ 2 at uˆL , the range was 37–77% (median 56%) and teer group, Vo ˙ 2 at peak, the range was 41 –79% (median 64%). We used for Vo ROC curve analysis to test the ability of CPETarm to discriminate ˙ 2 at uˆL between high- and low-risk patients by dichotomizing Vo ˙ 2 at peak at 15 ml kg21 min21 as at 10.2 ml kg21 min21 and Vo
BJA
Loughney et al.
˙ 2 ) and workload during CPET at estimated lactate threshold (uˆL ) and peak exercise (Vo ˙ 2 peak). Data presented as Table 1 Oxygen uptake (Vo median (IQR) Patients
P-value
Arm ˙ 2 at uˆL (ml kg21 min21) Vo ˙ 2 at peak (ml kg21 min21) Vo
Leg
Healthy
P-value
Arm
Leg
8.9 (6.9 –9.6)
10.9 (8.9 –12.8)
,0.00005
11.5 (10.7 –13.6)
20.8 (18.2 –27.9)
,0.00005
12.1 (11.1 – 14.7)
16.2 (12.5 –20.6)
,0.00005
20.6 (19.9 –24.2)
34.5 (29.4 –40.7)
,0.00005
Workload at uˆL (W)
42.0 (32 – 52)
50.0 (46 – 66)
0.0001
58.0 (50 –68)
121 (100 – 155))
,0.00005
Workload at peak (W)
73.0 (52 – 89)
92.0 (69 – 117)
0.0001
101 (86 –137)
214 (188 – 256)
,0.00005
· ^ Vo2 at qL leg (ml kg–1 min–1)
25
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
A
Patient Linear fit Prediction interval Line of equality
20
15 -1
10
5 4
Vo2 at peak leg (ml kg–1 min–1)
B
40
6
10 8 · ^ Vo2 at qL arm (ml/kg–1min–1)
12
14
Patient Linear fit Prediction interval Line of equality
35 30 25 20 15
·
10 5 8
10
12 14 16 · Vo2 at peak arm (ml kg–1 min–1)
18
20
Fig 1 Linear regression modelling of the relationship between CPETarm and CPETleg for the vascular surgery patient group. The curvature in the linear fits and PIs is a result of fitting a linear regression model to the log-transformed CPETleg. The limits of the 95% PIs give an indication of our uncertainty in predicting CPETleg using CPETarm.
60
BJA
Oxygen uptake in arm or leg ergometry
A
Healthy Linear fit Prediction interval Line of equality
50
· ^ Vo2 at qL leg (ml kg–1 min–1)
45 40 35 30 25 20 15 10
4
B 80
Vo2 at peak leg (ml kg–1 min–1)
8
10 12 14 · ^ –1 Vo2 at qL arm (ml/kg min–1)
16
18
20
Healthy Linear fit Prediction interval Line of equality
70
·
6
60 50 40 30 20 10 8
10
12
14
16 18 20 22 24 26 28 · Vo2 at peak arm (ml kg–1 min–1)
30
32
34
Fig 2 Linear regression modelling of the relationship between CPETarm and CPETleg for the healthy volunteer group. The curvature in the linear fits and PIs is a result of fitting a linear regression model to the log-transformed CPETleg. The limits of the 95% PIs give an indication of our uncertainty in predicting CPETleg using CPETarm.
in half of the cases, with one even showing marginally higher ˙ 2 during CPETarm. Vo ˙ 2 obtained from CPETarm to When evaluating the ability of Vo ˙ 2 obtained from CPETleg using the width of the estipredict Vo mated PIs at the CPETleg sample mean, we have shown that ˙ 2 at uˆL and peak were prohibitively in both groups PIs for Vo wide. This coupled with a lack of proportionality relationship indicates that CPETarm should not be used to determine ˙ 2 obtained from CPETleg. Vo Although CPETarm is unable to substitute CPETleg in perioperative risk stratification, it might still be a useful adjunct in risk assessment. Estimated AUROCs indicated that CPETarm may be useful in discriminating between low- and high-risk
patients (since they both exceed 0.75); however, the CIs are wide, suggesting that a larger study is needed to confirm our findings. CPETarm may be useful in identifying patients at low risk of developing postoperative complication when patients ˙ 2 values. Furthermore, it also has the ability to attain high Vo detect subclinical co-morbidity (ECG changes, cardiac or ventilatory limitation) which might only become apparent when the cardiorespiratory system is stressed, albeit to a lesser degree than by using CPETleg. Cycle ergometry has been used in perioperative risk stratification since the late 1990s. Findings by Older and colleagues20 in elderly patients undergoing major intra-abdominal surgery ˙ 2 at uˆL ,11.0 ml kg21 min21 suggest that preoperative Vo
61
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
5
BJA
Loughney et al.
· ^ Vo2 at qL leg (ml kg–1 min–1)
A
40–50%
20
15
50–60%
60–70%
70–80%
90–100%
100%
5
B
30
Vo2 at peak leg (ml kg–1 min–1)
0
25
5
10 · ^ Vo2 at qL arm (ml/kg–1min–1)
15
20
40–50% 50–60% 60–70% 70–80% 80–90% 90–100%
100%
10 5 Line of equality 0 0
5
10
15 20 · Vo2 at peak arm (ml kg–1 min–1)
25
30
˙ 2 at uˆL (A) and Vo ˙ 2 at peak (B). Fig 3 Percentage plots for the vascular surgery patient group Vo
was associated with increased cardiovascular mortality. In a later study and review,21 22 Older investigated the effects of ˙ 2 at uˆL ,11 ml kg21 min21 and triaging patients having a Vo concludes that using a small number of important variables obtained from CPETleg, we can accurately predict future response to perioperative stress. This was then replicated in various observational studies,1 – 6 23 which conclude that variables derived from CPETleg can be used in accurate risk prediction modelling with reasonable accuracy. To date, CPETusing arm ergometry in a perioperative setting has mainly focused on the ability of detecting ECG changes in patients with coronary artery disease.24 25 Arm exercise testing has also been utilized to determine physical fitness of vascular amputee patients before commencement of rehabilitation programmes.26 Traditionally, it has been suggested that if a person is healthy and has not undergone specific upper
62
˙ 2 for arm cycling will be 50– extremity training, the peak Vo 9 13 27 70% of that for leg cycling. We have now demonstrated that the relationship between the two tests is too weak for ˙ 2 values during arm such a rule. In healthy subjects, lower Vo exercise are expected due to a smaller muscle mass of the arms compared with the legs and the lack of experience with rhythmic arm exercise which can result in fatiguing sooner.13 However, more recently, the interaction between limb blood flow and cardiac output (which limits the maximum rate in a given muscle group), the amount of work done, and its ATP cost, together with the ancillary work done by other muscles ˙ 2 measured at the mouth, also which contribute to the Vo need to be considered when accounting for arm-to-leg differences.28 29 This difference is also apparent in COPD patients, where other reasons are suggested for the lower ˙ 2 values.30 Interestingly, baseline pulmonary function Vo
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Line of equality
0
·
80–90%
BJA
Oxygen uptake in arm or leg ergometry
A
40–50%
35
50–60%
60–70%
70–80%
80–90% 90–100%
· ^ Vo2 at qL leg (ml kg–1 min–1)
30 100%
15 10 5
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Line of equality 0 0
Vo2 at peak leg (ml kg–1 min–1)
B
·
5
10
15 20 · ^ –1 Vo2 at qL arm (ml/kg min–1) 40–50%
60
50–60%
25
60–70%
30
70–80%
35
80–90% 90–100%
50 100% 20 10 Line of equality 0 0
5
10
15
20
25 30 35 40 · –1 –1 Vo2 at peak arm (ml kg min )
45
50
55
60
˙ 2 at uˆL (A) and Vo ˙ 2 at peak (B). Fig 4 Percentage plots for the healthy volunteer group Vo
tests in our vascular patient group presented with a generalized obstructive pattern. Strengths of this study include the homogeneous nature of the study population, the blinded reporting of objectively measured CPET variables, and the prospective nature of the study. An inherent limitation is that we used the width of PIs as a measure of the predictive ability of CPETarm since this width is partly determined by the sample size (i.e. a larger sample size would result in a reduced width). However, the width of a PI does not approach zero as the sample size increases. Using our results, the PI widths were re-estimated for a large (infinite) sample size (results not shown), but the reduction in PI widths was only modest and still remained prohibitively wide. Another limitation of our study is that perhaps CPETleg
could be more accurately predicted if other variables, alongside CPETarm, were considered as part of a multivariable model; however, this was beyond the scope of this project. In conclusion, although a relationship exists between ˙ 2 values for CPETarm and CPETleg, this was insufficient for Vo ˙ 2 obtained from CPETleg to be accurately predicted using Vo CPETarm alone. Similarly, any relative or absolute relationship rule between the two tests would be overly simplistic and not encouraged. While these results do not necessarily preclude ˙ 2 values, subclinical ECG changes, the use of high or low Vo or the discovery of cardiac or ventilatory limitation in the perioperative risk stratification of patients with lower limb dysfunction, further evaluation of CPETarm as a perioperative risk assessment tool is necessary.
63
BJA
Loughney et al.
A 1.00
Sensitivity
0.75
0.50
0.25
0.00
0.25
0.50
0.75
1.00
0.75
1.00
1–Specificity Area under ROC curve = 0.8385
B
1.00
Sensitivity
0.75
0.50
0.25
0.00 0.00
0.25
0.50 1–Specificity
Area under ROC curve = 0.7677 ˙ 2 at peak (B) (cut-off Vo ˙ 2 at uˆL 10.2 ml kg21 min21 and Vo ˙ 2 at peak of 15 ml kg21 min21). ˙ 2 at uˆL (A) and Vo Fig 5 ROC for CPETarm Vo
Authors’ contributions
Declaration of interest
L.L., M.W., S.P., D.L., E.C., S.J., and F.T. contributed to the design of the study and interpretation of these data. L.L., M.W., S.P., and E.C. contributed to data collection. D.L., F.T., and S.J. provided clinical support and helped with data interpretation within their specialist area. D.L. performed all statistical analysis. D.L., F.T., M.W., and S.J. produced the figures and tables. L.L., M.W., and S.P. wrote the first draft. L.L., M.W., S.P., D.L., E.C., S.J., and F.T. contributed to manuscript revisions. L.L., M.W., S.P., D.L., E.C., S.J., and F.T. provided critical revisions and produced the final manuscript. All authors have approved the final version.
None declared.
64
Funding Vascular Research Fund at University Hospitals Aintree, Liverpool, UK.
References 1 Smith TB, Stonell C, Purkayastha S, et al. Cardiopulmonary exercise testing as a risk assessment method in non cardio-pulmonary surgery: a systematic review. Anaesthesia 2009; 64: 883– 93
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
0.00
BJA
Oxygen uptake in arm or leg ergometry
17
18
19
20
21
22
23
24
25 26
27
28
29
30
Committee to Revise the 2002 Guideline). J Am Coll Cardiol 2007; 50: 159– 241 Palange P, Ward SA, Carlsen KH, et al. Recommendations on the use of exercise testing in clinical practice. Eur Respir J 2007; 29: 185– 209 Wasserman K, Hansen JE, Sue DY, et al. Principles of Exercise Testing and Interpretation: Pathophysiology and Clinical Applications, 4th Edn. Lippincott: Williams & Wilkins, 2005; 33 –89 Sue DY, Wasserman K, Moricca RB, et al. Metabolic acidosis during exercise in patients with chronic obstructive pulmonary disease: use of the V-slope method for anaerobic threshold determination. Chest 1988; 94: 931– 8 Older P, Smith R, Courtney P, et al. Preoperative evaluation of cardiac failure and ischemia in elderly patients by cardiopulmonary exercise testing . Chest 1993; 104: 701–4 Older P, Hall A, Hader R. Cardiopulmonary exercise testing as a screening test for perioperative management of major surgery in the elderly. Chest 1999; 116: 355–62 Stringer W, Casaburi R, Older P. Cardiopulmonary exercise testing: does it improve perioperative care and outcome? Curr Opin Anaesthesiol 2012; 25: 178– 84 Colson M, Baglin J, Bolsin S, et al. Cardiopulmonary exercise testing predicts 5 yr survival after major surgery. Br J Anaesth 2012; 109: 735– 41 Hanson P, Pease M, Berkoff H, et al. Arm exercise testing for coronary artery disease in patients with peripheral vascular disease. Clin Cardiol 1988; 11: 70 –4 Ilias NA, Xian H, Inman C, et al. Arm exercise testing predicts clinical outcome. Am Heart J 2009; 157: 69 –76 Priebe M, Davidoff G, Lampman RM. Exercise testing and training in patients with peripheral vascular disease and lower extremity amputation. West J Med 1991; 154: 598– 601 Casaburi R, Barstow TJ, Robinson T, et al. Dynamic and steady-state ventilatory and gas exchange responses to arm exercise. Med Sci Sports Exerc 1992; 24: 1365– 74 Boushel R, Gnaiger E, Calbet JAL, et al. Muscle mitochondrial capacity exceeds maximal oxygen delivery in humans. Mitochondrion 2011; 11: 303–7 Helge JW. Arm and leg substrate utilization and muscle adaptation after prolonged low-intensity training. Acta Physiol (Oxford, England) 2010; 199: 519–28 Carter R. Peak physiologic responses to arm and leg ergometry in male and female patients with airflow obstruction. Chest 2003; 124: 511– 8
Handling editor: H. C. Hemmings
65
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
2 West M, Jack S, Grocott MPW. Perioperative cardiopulmonary exercise testing in the elderly. Best Pract Res Clin Anaesthesiol 2011; 25: 427– 37 3 Snowden CP, Prentis JM, Anderson HL, et al. Submaximal cardiopulmonary exercise testing predicts complications and hospital length of stay in patients undergoing major elective surgery. Ann Surg 2010; 251: 535– 41 4 Wilson RJT, Davies S, Yates D, et al. Impaired functional capacity is associated with all-cause mortality after major elective intraabdominal surgery. Br J Anaesth 2010; 105: 297– 303 5 Hennis PJ, Meale PM, Grocott MPW. Cardiopulmonary exercise testing for the evaluation of perioperative risk in noncardiopulmonary surgery. Postgrad Med J 2011; 87: 550–7 6 Hartley RA, Pichel AC, Grant SW, et al. Preoperative cardiopulmonary exercise testing and risk of early mortality following abdominal aortic aneurysm repair. Br J Surg 2012; 99: 1539–46 7 Schrieks IC, Barnes MJ, Hodges LD. Comparison study of treadmill versus arm ergometry. Clin Physiol Funct Imaging 2011; 31: 326–31 8 Lyons S, Richardson M, Bishop P, et al. Excess post-exercise oxygen consumption in untrained men following exercise of equal energy expenditure: comparisons of upper and lower body exercise. Diabetes Obes Metab 2007; 9: 889–94 9 Bar-Or O, Zwiren LD. Maximal oxygen consumption test during arm exercise reliability and validity. J Appl Physiol 1975; 38: 424– 6 10 Franklin B. Exercise testing, training and arm ergometry. J Sports Med 1985; 2: 100– 19 11 Martin T, Zebalos R, Weisman I. Gas exchange during maximal upper extremity exercise. Chest 1991; 99: 420–5 12 Reybrouck T, Heigenhauser GF, Faulkner JA, et al. Limitations to maximum oxygen uptake in arms, leg, and combined arm–leg ergometry. J Appl Physiol 1975; 38: 774–9 13 Davis J, Vodak P, Wilmore J, et al. Anaerobic threshold and maximal aerobic power for three modes of exercise. J Appl Physiol 1976; 41: 544– 50 14 Orr JL, Williamson P, Anderson W, et al. Cardiopulmonary exercise testing: arm crank vs cycle ergometry. Anaesthesia 2013; 68: 497– 501 15 Schneider DA, Wing AN, Morris NR. Oxygen uptake and heart rate kinetics during heavy exercise: a comparison between arm cranking and leg cycling. Eur J Appl Physiol 2002; 88: 100– 6 16 Fleisher L, Beckman JA, Brown KA, et al. ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing
Comparison of oxygen uptake during arm or leg cardiopulmonary exercise testing in vascular surgery patients and control subjects L. Loughney 1, M. West1,2, S. Pintus 3, D. Lythgoe 5, E. Clark 3, S. Jack 1,2 and F. Torella 4* 1
Critical Care Research Area, NIHR Respiratory BRU, University of Southampton and University Hospital Southampton, Southampton, UK Institute of Ageing and Chronic Disease, Department of Musculoskeletal Biology II, University of Liverpool, Liverpool, UK 3 Department of Anaesthesia and 4 Liverpool Vascular and Endovascular Service, University Hospital Aintree, Liverpool, UK 5 Cancer Research UK, Liverpool Cancer Trials Unit, UK 2
* Corresponding author. E-mail: [email protected]
† Cardiopulmonary exercise testing by cycle ergometry has limited utility as a preoperative assessment tool in patients with lower limb dysfunction. † Leg ergometry was compared with arm ergometry in vascular surgery and healthy patients. † Oxygen uptake using the two methods was correlated, but arm ergometry was a poor predictor of leg ergometry. † Further study is required to establish the perioperative utility of arm ergometry.
Background. Cardiopulmonary exercise testing by cycle ergometry (CPETleg) is an established assessment tool of perioperative physical fitness. CPETutilizing arm ergometry (CPETarm) is an attractive alternative in patients with lower limb dysfunction. We aimed to determine ˙ 2 ) obtained by CPETleg could be predicted by using CPETarm whether oxygen uptake (Vo alone and whether CPETarm could be used in perioperative risk stratification. ˙ 2 obtained Methods. Subjects underwent CPETarm and CPETleg. To evaluate the ability of Vo ˙ 2 from CPETleg, we calculated prediction intervals (PIs) at lactate from CPETarm to predict Vo threshold (uˆL ) and peak exercise in both groups. Receiver operating characteristic (ROC) curves were used to risk stratify patients into high and low categories based on published criteria. Results. We recruited 20 vascular surgery patients (17 males and three females) and 20 ˙ 2 at uˆL and peak healthy volunteers (10 males and 10 females). In both groups, PIs for Vo ˙ 2 at uˆL CPETarm ranged from 55% were wider than clinically acceptable (patient group—Vo to 108% of CPETleg and from 54% to 105% at peak; healthy volunteers—37– 77% and ˙ 2 in patients was 0.84 [95% 41– 79%, respectively). The area under the ROC for CPETarm Vo confidence interval (CI): 0.66, 1.0] at uˆL , and 0.76 (95% CI: 0.54, 0.99) at peak. ˙ 2 values for CPETarm and CPETleg, this Conclusions. Although a relationship exists between Vo is insufficient for accurate prediction using CPETarm alone. This however does not necessarily preclude the use of CPETarm in perioperative risk stratification. Keywords: arm ergometry; cardiopulmonary exercise testing; cycle ergometry; preoperative assessment; vascular surgery Accepted for publication: 18 August 2013
Cardiopulmonary exercise testing (CPET) provides an integrated assessment and quantification of the cardiorespiratory system at rest and under stress of maximal exercise. CPET is gaining popularity as a physical fitness assessment tool before major elective surgery, including vascular, with evidence that variables ˙ 2 ) at peak exercise and at estimated such as oxygen uptake (Vo lactate threshold (uˆL ) predict short- and long-term outcomes.1 – 5 Recently, CPET to maximal exercise has been suggested as identifying patients at risk of early perioperative death after elective abdominal aortic aneurysm (AAA) repair.6 CPET in the clinical setting is conducted on a cycle ergometer (CPETleg), but many vascular surgery patients are unable to perform this due to lower limb dysfunction such as joint arthritis, peripheral vascular disease, neurological disease, or previous
amputations. CPETon an arm ergometer (CPETarm) is an attractive alternative as it provides data on physiological responses similar to those obtained by CPETleg. This has been validated in healthy subjects;7 however, it is currently not an accepted test in routine preoperative assessment. The maximum oxygen uptake obtained byarm ergometry is 60–80% of that measured by leg ergometry in healthy individuals,8 – 13 with a recent study suggesting ˙ 2 at uˆL and Vo ˙ 2 at peak.14 The lower Vo ˙ 2 values 34% less Vo obtained during CPETarm could be due to smaller muscle mass, distribution of fast twitch muscle fibres, recruitment pattern of motor units,13 15 and greater glycolytic enzyme activity.15 Literature around arm ergometry is mainly on healthy individuals11 – 13 with no literature available comparing arm and leg exercise testing in a patient population.
& The Author [2013]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Editor’s key points
BJA The primary aim of this study was to determine whether oxygen uptake derived from CPETleg could be accurately predicted by using CPETarm alone in both healthy volunteers and perioperative patients. We considered an a priori predictive ˙ 2 at uˆL and error for CPETarm of +1.5 ml kg21 min21 for Vo ˙ 2 at peak to be clinically acceptable. +3.0 ml kg21 min21 for Vo We also aimed to explore whether patients could be risk strati˙ 2 obtained by CPETarm in a similar fashion as for fied using Vo CPETleg (with known cut-off values of 10.2 ml kg21 min21 for ˙ 2 at peak).6 Finally, we ˙ 2 at uˆL and 15 ml kg21 min21 for Vo Vo aimed to explore the validity of a previously published simple ˙ 2 obtained percentage proportionality relationship between Vo from CPETarm and CPETleg.8 – 11
Subjects We prospectively recruited two cohorts: (i) preoperative patients being assessed before elective abdominal aortic surgery (patient group), and (ii) healthy volunteers. We chose these two groups in order to uncover potential differences in response to arm and leg exercise in elderly patients in comparison with younger healthy subjects. After ethical approval (09/H1001/94) and written informed consent, preoperative patients were requested to undergo CPET twice: CPETleg, as part of their normal preoperative assessment process, and a CPETarm, as part of this study. Healthy volunteers also underwent the same two tests using the protocols detailed below. Eligible patients were free of acute illness or clinically evident peripheral vascular disease, and did not have any disability precluding arm or leg exercise. Healthy volunteers were untrained and free from illness. CPET was reported by an experienced clinical scientist (S.J.) who was blinded to the mode of exercise testing.
Cardiopulmonary exercise testing Symptom-limited CPET was conducted in accordance with American Thoracic Society/American College of Chest Physicians recommendations.16 17 CPET was performed on calibrated electromagnetically braked cycle and arm ergometers. Gas and flow calibration was performed before each test. Both leg and arm CPET were carried out using similar ramped protocols set to 10–25 W min21 based on a calculation described by Wasser˙ 2 at unloaded pedalman and colleagues18 using predicted Vo ˙ 2 at peak exercise, height, and patient age. ling, predicted Vo Both protocols consisted of 3 min of rest, followed by 3 min of unloaded exercise, then the loaded ramp increased until volitional termination. This was followed by 5 min of recovery. Subjects were monitored throughout each test using pulse oximetry, 12-lead electrocardiography, and non-invasive arterial pressure monitoring. Ventilation and gas exchange variables were measured using a metabolic cart (Geratherm Respiratory GmbH (Love Medical Ltd, UK) and a cycle ergometer (Love Medical Ltd, Ergoline) or arm ergometer (Love Medical Ltd, Ergoline). For CPETleg, seat height was adjusted to ensure that full knee extension was achieved when the pedal was in the down position and handlebars raised to maintain full weight supported exercise. Subjects were instructed to cycle at a speed of
58
55–65 RPM during the test. This was monitored by the participant through a light-emitting diode display. The test was ended if RPM decreased below 45 or symptoms were encountered. For CPETarm, the arm ergometer was adjusted to ensure that participants sat on the chair with their arms slightly flexed, and maintaining their feet flat on the floor. They were asked to grasp handles in front of them, and ‘pedal’ with their arms in a circular motion, maintaining 55–65 RPM during the test until they could no longer push against the resistance or if the RPM decreased below 45. Breath-by-breath data were collected through a face mask and flow sensor that were appropriately fitted to each participant.
Measurements Subject characteristics, including age, gender, height, weight, and clinical details, were recorded. Before the first CPET, resting flow-volume loops were measured in the patient group to derive forced expiratory volume over 1 s (FEV1) and forced vital capacity (FVC). CPET variables measured included: ˙ 2 , carbon dioxide output, tidal expired ventilatory volumes, Vo volume, minute ventilation, work rate, respiratory exchange ˙ 2 (ml kg21 min21) at uˆL and at peak ratio, and oxygen pulse. Vo exercise were the primary outcome variables recorded. uˆL was ˙ 2 − Vo ˙ 2 relaestimated conventionally (breakpoint in the Vco tionship), with increases in ventilatory equivalent for oxygen ˙ 2 ) and end-tidal (PE′ ) oxygen but no increase in ventilatory (V˙ E /Vo ˙ 2 ) or decrease in PE′CO 19 by equivalent for carbon dioxide (V˙ E /Vco 2 18 ˙ 2 was averaged an experienced, blinded, assessor. The peak Vo over the last 30 s of exercise.
Data analysis Continuous variables are summarized using the median and inter-quartile range (IQR). Paired t-tests were used to compare CPETarm and CPETleg values within patient groups. Linear regression was used to assess whether CPETleg could be predicted using CPETarm alone by estimating 95% prediction intervals (PIs). The PIs should be interpreted as a measure of how accurate our prediction of CPETleg would be for a new patient given only a CPETarm measurement. In order to meet the assumptions of linear regression, it was necessary to natural log transform CPETleg. For the ease of interpretation, the linear regression results were presented on the original scale, with the consequence that the resulting PIs were asymmetric and non-constant. For simplicity, the predictive ability of CPETarm was evaluated against the predefined acceptable widths at the CPETleg sample mean. Arm and leg measurements were compared using paired t-tests. The assumption of normality for the mean of the paired differences was assessed using the Q–Q plots. Scatter plots of CPETleg vs CPETarm were constructed and boundaries representing CPETarm as a percentage of CPETleg were added. The purpose of these plots was to establish whether a simple relative relationship rule such as ‘CPETarm is typically x–y% of CPETleg’ could be advised. Finally, nonparametric receiver operating characteristic (ROC) curves were used to determine whether CPETarm could discriminate
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Methods
Loughney et al.
BJA
Oxygen uptake in arm or leg ergometry
between high- and low-risk patients and hence become a useful perioperative risk stratification tool. Risk groups (high/ low) were established by dichotomizing CPETleg at 10.2 ml ˙ 2 at uˆL and 15 ml kg21 min21 for Vo ˙ 2 at peak kg21 min21 for Vo based on published cut-off points (≥cut-off—low risk and ,cut-off—high risk).6 All results were considered as statistically significant at the 5% level. All statistical analyses were carried out using Stata version 12.1 (StataCorp., 2011. Stata Statistical Software: Release 12, College Station, TX, USA).
Results
obtained from CPETleg (Fig. 5). The area under the ROC (AUROC) ˙ 2 at uˆL in the patient group was 0.84 [95% confidence for Vo ˙ 2 at peak, the AUROC was interval (CI): 0.66, 1.0], while for Vo 0.76 (95% CI: 0.54, 0.99).
Discussion This study clearly showed that it is not possible to predict ˙ 2 obtained by CPETleg from CPETarm, in patients or healthy Vo ˙ 2 at uˆL and peak were unacceptvolunteers. The PIs for both Vo ably wide. We also found that CPETarm has a discrete ability to ˙ 2 obtained by CPETleg at predict CPETleg when dichotomizing Vo the cut-off points described by Hartley and colleagues;6 however, it must be emphasized that in the proposed model, ˙ 2 at uˆL cut-off was not used in isolation but as part of a the Vo more complex risk prediction model. Furthermore, we must also emphasize that although this might be deemed useful in perioperative risk stratification (especially when extremes of ˙ 2 are obtained by CPETarm), we are unable to justify the use Vo of CPET by arm ergometry in this setting without further robust studies. Finally, when exploring the proportionality rela˙ 2 obtained from CPETarm and CPETleg, we tionship between Vo ˙ 2 values of CPETarm as a profound very high variability in the Vo portion of CPETleg, such that any linear relationship would appear to be overly simplistic and would not accurately reflect the relative relationship between the two measurements. It is not uncommon, in our practice, to encounter patients who are unsuitable for CPET on a cycle ergometer due to co-morbidity affecting the lower limbs. In these patients, arm ergometry is, intuitively, a potential alternative which, however, has not yet been clinically validated. We decided to test our hypothesis in two different groups of subjects (patients and healthy volunteers) because we suspected that elderly vascular patients could show a higher degree of variability between the two tests due to subclinical co-morbidity (e.g. mild joint arthritis in shoulder/hip/knee or COPD). In general, ˙ 2 measured by the we found a strong correlation between Vo two tests, at uˆL and at peak exercise, with CPETarm being systematically lower than CPETleg. Although our findings in patients were mirrored by those in volunteers, it was notable ˙ 2 at uˆL during CPETarm was never meathat, in all volunteers, Vo sured at more than three-quarters of the value recorded during CPETleg, while in patients, higher proportional values were seen
59
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
We prospectively screened 26 and recruited 23 (19 men, four women) vascular surgery patients. Three patients dropped out due to withdrawal of consent, aneurysmal rupture, and expedited elective surgery. Twenty patients were included with the median (IQR) age of 74 (70 –77) yr and with body mass index of 27 (25– 29) kg m22. Baseline pulmonary function tests showed a generalized mild –moderate obstructive pattern [according to GOLD spirometric criteria for chronic obstructive pulmonary disease (COPD) severity] with a median (IQR) FEV1 of 93% (75.5 –105.3), FVC 110% (97.8 –128.5), and FEV1/FVC ratio of 63.5 (59 –69) of their predicted values. Seventeen out of 20 patients had a median FEV1/FVC ratio of 61 (59 –69); of which, six had an FEV1 of ,80%. Patients were undergoing preoperative evaluation before elective AAA repair (n¼16), elective thoraco-abdominal aortic aneurysm repair (n¼1), combined AAA repair and cancer resection (n¼2), and elective aorto-mesenteric bypass (n¼1). Twenty healthy volunteers (10 men and 10 women) were also recruited, median (IQR) age of 31 (24–42) yr and median (IQR) body mass index of 26 (24–28) kg m22. The order of the CPET was not randomized. Arm and leg CPET was not carried out on the same visit and not more than 14 days apart [median 7 (range 6–14) days]. ˙ 2 and workload For both patients and healthy volunteers, Vo values were lower during CPETarm compared with CPETleg (P,0.0001; Table 1). One patient experienced chest pain with electrocardiographic changes during CPETleg, but no signs or symptoms were reported during CPETarm, while another experienced significant oxygen desaturation (to 85%) during CPETleg, but not CPETarm. In both groups, there was a strong correlation between ˙ 2 measured by the two tests, at uˆL (patient group—r¼0.72, Vo P¼0.0003; healthy volunteers—r¼0.64, P¼0.0024) and at peak exercise (patient group—r¼0.75, P¼0.0001; healthy ˙ 2 CPETarm being sysvolunteers—r¼0.77, P¼0.0001), with Vo ˙ 2 CPETleg. After fitting a simple tematically lower than Vo ˙ 2 at uˆL of 11.1 linear regression model, at the mean CPETleg Vo ml kg21 min21 for the patient group, the PI’s width was 8.0 ˙ 2 at peak, the total width of (4.0) ml kg21 min21 (Fig. 1A). For Vo the PI was 14.0 ml kg21 min21 at the mean of 16.6 ml kg21 min21; approximately +7.0 ml kg21 min21 (Fig. 1B). In the ˙ 2 at uˆL of 21.7 ml kg21 healthy volunteer group, at the mean Vo 21 min , the total PI width was 19.9 (10.0) ml kg21 min21. For ˙ 2 at peak, the total width of the PI was 25.3 ml kg21 min21 Vo
at the mean of 35.8 ml kg21 min21; approximately +12.5 ml kg21 min21 (Fig. 2A and B). The scatter plots in Figures 3 and 4 depict the relationship ˙ 2 at uˆL and Vo ˙ 2 at peak for CPETarm and CPETleg between Vo with overlaid percentage boundaries for CPETarm as a propor˙ 2 at uˆL values ranged tion of CPETleg. In the patient group, Vo ˙ 2 at peak values from 55% to 108% (median 79%) andVo ranged from 54% to 105% (median 80%). In the healthy volun˙ 2 at uˆL , the range was 37–77% (median 56%) and teer group, Vo ˙ 2 at peak, the range was 41 –79% (median 64%). We used for Vo ROC curve analysis to test the ability of CPETarm to discriminate ˙ 2 at uˆL between high- and low-risk patients by dichotomizing Vo ˙ 2 at peak at 15 ml kg21 min21 as at 10.2 ml kg21 min21 and Vo
BJA
Loughney et al.
˙ 2 ) and workload during CPET at estimated lactate threshold (uˆL ) and peak exercise (Vo ˙ 2 peak). Data presented as Table 1 Oxygen uptake (Vo median (IQR) Patients
P-value
Arm ˙ 2 at uˆL (ml kg21 min21) Vo ˙ 2 at peak (ml kg21 min21) Vo
Leg
Healthy
P-value
Arm
Leg
8.9 (6.9 –9.6)
10.9 (8.9 –12.8)
,0.00005
11.5 (10.7 –13.6)
20.8 (18.2 –27.9)
,0.00005
12.1 (11.1 – 14.7)
16.2 (12.5 –20.6)
,0.00005
20.6 (19.9 –24.2)
34.5 (29.4 –40.7)
,0.00005
Workload at uˆL (W)
42.0 (32 – 52)
50.0 (46 – 66)
0.0001
58.0 (50 –68)
121 (100 – 155))
,0.00005
Workload at peak (W)
73.0 (52 – 89)
92.0 (69 – 117)
0.0001
101 (86 –137)
214 (188 – 256)
,0.00005
· ^ Vo2 at qL leg (ml kg–1 min–1)
25
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
A
Patient Linear fit Prediction interval Line of equality
20
15 -1
10
5 4
Vo2 at peak leg (ml kg–1 min–1)
B
40
6
10 8 · ^ Vo2 at qL arm (ml/kg–1min–1)
12
14
Patient Linear fit Prediction interval Line of equality
35 30 25 20 15
·
10 5 8
10
12 14 16 · Vo2 at peak arm (ml kg–1 min–1)
18
20
Fig 1 Linear regression modelling of the relationship between CPETarm and CPETleg for the vascular surgery patient group. The curvature in the linear fits and PIs is a result of fitting a linear regression model to the log-transformed CPETleg. The limits of the 95% PIs give an indication of our uncertainty in predicting CPETleg using CPETarm.
60
BJA
Oxygen uptake in arm or leg ergometry
A
Healthy Linear fit Prediction interval Line of equality
50
· ^ Vo2 at qL leg (ml kg–1 min–1)
45 40 35 30 25 20 15 10
4
B 80
Vo2 at peak leg (ml kg–1 min–1)
8
10 12 14 · ^ –1 Vo2 at qL arm (ml/kg min–1)
16
18
20
Healthy Linear fit Prediction interval Line of equality
70
·
6
60 50 40 30 20 10 8
10
12
14
16 18 20 22 24 26 28 · Vo2 at peak arm (ml kg–1 min–1)
30
32
34
Fig 2 Linear regression modelling of the relationship between CPETarm and CPETleg for the healthy volunteer group. The curvature in the linear fits and PIs is a result of fitting a linear regression model to the log-transformed CPETleg. The limits of the 95% PIs give an indication of our uncertainty in predicting CPETleg using CPETarm.
in half of the cases, with one even showing marginally higher ˙ 2 during CPETarm. Vo ˙ 2 obtained from CPETarm to When evaluating the ability of Vo ˙ 2 obtained from CPETleg using the width of the estipredict Vo mated PIs at the CPETleg sample mean, we have shown that ˙ 2 at uˆL and peak were prohibitively in both groups PIs for Vo wide. This coupled with a lack of proportionality relationship indicates that CPETarm should not be used to determine ˙ 2 obtained from CPETleg. Vo Although CPETarm is unable to substitute CPETleg in perioperative risk stratification, it might still be a useful adjunct in risk assessment. Estimated AUROCs indicated that CPETarm may be useful in discriminating between low- and high-risk
patients (since they both exceed 0.75); however, the CIs are wide, suggesting that a larger study is needed to confirm our findings. CPETarm may be useful in identifying patients at low risk of developing postoperative complication when patients ˙ 2 values. Furthermore, it also has the ability to attain high Vo detect subclinical co-morbidity (ECG changes, cardiac or ventilatory limitation) which might only become apparent when the cardiorespiratory system is stressed, albeit to a lesser degree than by using CPETleg. Cycle ergometry has been used in perioperative risk stratification since the late 1990s. Findings by Older and colleagues20 in elderly patients undergoing major intra-abdominal surgery ˙ 2 at uˆL ,11.0 ml kg21 min21 suggest that preoperative Vo
61
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
5
BJA
Loughney et al.
· ^ Vo2 at qL leg (ml kg–1 min–1)
A
40–50%
20
15
50–60%
60–70%
70–80%
90–100%
100%
5
B
30
Vo2 at peak leg (ml kg–1 min–1)
0
25
5
10 · ^ Vo2 at qL arm (ml/kg–1min–1)
15
20
40–50% 50–60% 60–70% 70–80% 80–90% 90–100%
100%
10 5 Line of equality 0 0
5
10
15 20 · Vo2 at peak arm (ml kg–1 min–1)
25
30
˙ 2 at uˆL (A) and Vo ˙ 2 at peak (B). Fig 3 Percentage plots for the vascular surgery patient group Vo
was associated with increased cardiovascular mortality. In a later study and review,21 22 Older investigated the effects of ˙ 2 at uˆL ,11 ml kg21 min21 and triaging patients having a Vo concludes that using a small number of important variables obtained from CPETleg, we can accurately predict future response to perioperative stress. This was then replicated in various observational studies,1 – 6 23 which conclude that variables derived from CPETleg can be used in accurate risk prediction modelling with reasonable accuracy. To date, CPETusing arm ergometry in a perioperative setting has mainly focused on the ability of detecting ECG changes in patients with coronary artery disease.24 25 Arm exercise testing has also been utilized to determine physical fitness of vascular amputee patients before commencement of rehabilitation programmes.26 Traditionally, it has been suggested that if a person is healthy and has not undergone specific upper
62
˙ 2 for arm cycling will be 50– extremity training, the peak Vo 9 13 27 70% of that for leg cycling. We have now demonstrated that the relationship between the two tests is too weak for ˙ 2 values during arm such a rule. In healthy subjects, lower Vo exercise are expected due to a smaller muscle mass of the arms compared with the legs and the lack of experience with rhythmic arm exercise which can result in fatiguing sooner.13 However, more recently, the interaction between limb blood flow and cardiac output (which limits the maximum rate in a given muscle group), the amount of work done, and its ATP cost, together with the ancillary work done by other muscles ˙ 2 measured at the mouth, also which contribute to the Vo need to be considered when accounting for arm-to-leg differences.28 29 This difference is also apparent in COPD patients, where other reasons are suggested for the lower ˙ 2 values.30 Interestingly, baseline pulmonary function Vo
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Line of equality
0
·
80–90%
BJA
Oxygen uptake in arm or leg ergometry
A
40–50%
35
50–60%
60–70%
70–80%
80–90% 90–100%
· ^ Vo2 at qL leg (ml kg–1 min–1)
30 100%
15 10 5
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
Line of equality 0 0
Vo2 at peak leg (ml kg–1 min–1)
B
·
5
10
15 20 · ^ –1 Vo2 at qL arm (ml/kg min–1) 40–50%
60
50–60%
25
60–70%
30
70–80%
35
80–90% 90–100%
50 100% 20 10 Line of equality 0 0
5
10
15
20
25 30 35 40 · –1 –1 Vo2 at peak arm (ml kg min )
45
50
55
60
˙ 2 at uˆL (A) and Vo ˙ 2 at peak (B). Fig 4 Percentage plots for the healthy volunteer group Vo
tests in our vascular patient group presented with a generalized obstructive pattern. Strengths of this study include the homogeneous nature of the study population, the blinded reporting of objectively measured CPET variables, and the prospective nature of the study. An inherent limitation is that we used the width of PIs as a measure of the predictive ability of CPETarm since this width is partly determined by the sample size (i.e. a larger sample size would result in a reduced width). However, the width of a PI does not approach zero as the sample size increases. Using our results, the PI widths were re-estimated for a large (infinite) sample size (results not shown), but the reduction in PI widths was only modest and still remained prohibitively wide. Another limitation of our study is that perhaps CPETleg
could be more accurately predicted if other variables, alongside CPETarm, were considered as part of a multivariable model; however, this was beyond the scope of this project. In conclusion, although a relationship exists between ˙ 2 values for CPETarm and CPETleg, this was insufficient for Vo ˙ 2 obtained from CPETleg to be accurately predicted using Vo CPETarm alone. Similarly, any relative or absolute relationship rule between the two tests would be overly simplistic and not encouraged. While these results do not necessarily preclude ˙ 2 values, subclinical ECG changes, the use of high or low Vo or the discovery of cardiac or ventilatory limitation in the perioperative risk stratification of patients with lower limb dysfunction, further evaluation of CPETarm as a perioperative risk assessment tool is necessary.
63
BJA
Loughney et al.
A 1.00
Sensitivity
0.75
0.50
0.25
0.00
0.25
0.50
0.75
1.00
0.75
1.00
1–Specificity Area under ROC curve = 0.8385
B
1.00
Sensitivity
0.75
0.50
0.25
0.00 0.00
0.25
0.50 1–Specificity
Area under ROC curve = 0.7677 ˙ 2 at peak (B) (cut-off Vo ˙ 2 at uˆL 10.2 ml kg21 min21 and Vo ˙ 2 at peak of 15 ml kg21 min21). ˙ 2 at uˆL (A) and Vo Fig 5 ROC for CPETarm Vo
Authors’ contributions
Declaration of interest
L.L., M.W., S.P., D.L., E.C., S.J., and F.T. contributed to the design of the study and interpretation of these data. L.L., M.W., S.P., and E.C. contributed to data collection. D.L., F.T., and S.J. provided clinical support and helped with data interpretation within their specialist area. D.L. performed all statistical analysis. D.L., F.T., M.W., and S.J. produced the figures and tables. L.L., M.W., and S.P. wrote the first draft. L.L., M.W., S.P., D.L., E.C., S.J., and F.T. contributed to manuscript revisions. L.L., M.W., S.P., D.L., E.C., S.J., and F.T. provided critical revisions and produced the final manuscript. All authors have approved the final version.
None declared.
64
Funding Vascular Research Fund at University Hospitals Aintree, Liverpool, UK.
References 1 Smith TB, Stonell C, Purkayastha S, et al. Cardiopulmonary exercise testing as a risk assessment method in non cardio-pulmonary surgery: a systematic review. Anaesthesia 2009; 64: 883– 93
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
0.00
BJA
Oxygen uptake in arm or leg ergometry
17
18
19
20
21
22
23
24
25 26
27
28
29
30
Committee to Revise the 2002 Guideline). J Am Coll Cardiol 2007; 50: 159– 241 Palange P, Ward SA, Carlsen KH, et al. Recommendations on the use of exercise testing in clinical practice. Eur Respir J 2007; 29: 185– 209 Wasserman K, Hansen JE, Sue DY, et al. Principles of Exercise Testing and Interpretation: Pathophysiology and Clinical Applications, 4th Edn. Lippincott: Williams & Wilkins, 2005; 33 –89 Sue DY, Wasserman K, Moricca RB, et al. Metabolic acidosis during exercise in patients with chronic obstructive pulmonary disease: use of the V-slope method for anaerobic threshold determination. Chest 1988; 94: 931– 8 Older P, Smith R, Courtney P, et al. Preoperative evaluation of cardiac failure and ischemia in elderly patients by cardiopulmonary exercise testing . Chest 1993; 104: 701–4 Older P, Hall A, Hader R. Cardiopulmonary exercise testing as a screening test for perioperative management of major surgery in the elderly. Chest 1999; 116: 355–62 Stringer W, Casaburi R, Older P. Cardiopulmonary exercise testing: does it improve perioperative care and outcome? Curr Opin Anaesthesiol 2012; 25: 178– 84 Colson M, Baglin J, Bolsin S, et al. Cardiopulmonary exercise testing predicts 5 yr survival after major surgery. Br J Anaesth 2012; 109: 735– 41 Hanson P, Pease M, Berkoff H, et al. Arm exercise testing for coronary artery disease in patients with peripheral vascular disease. Clin Cardiol 1988; 11: 70 –4 Ilias NA, Xian H, Inman C, et al. Arm exercise testing predicts clinical outcome. Am Heart J 2009; 157: 69 –76 Priebe M, Davidoff G, Lampman RM. Exercise testing and training in patients with peripheral vascular disease and lower extremity amputation. West J Med 1991; 154: 598– 601 Casaburi R, Barstow TJ, Robinson T, et al. Dynamic and steady-state ventilatory and gas exchange responses to arm exercise. Med Sci Sports Exerc 1992; 24: 1365– 74 Boushel R, Gnaiger E, Calbet JAL, et al. Muscle mitochondrial capacity exceeds maximal oxygen delivery in humans. Mitochondrion 2011; 11: 303–7 Helge JW. Arm and leg substrate utilization and muscle adaptation after prolonged low-intensity training. Acta Physiol (Oxford, England) 2010; 199: 519–28 Carter R. Peak physiologic responses to arm and leg ergometry in male and female patients with airflow obstruction. Chest 2003; 124: 511– 8
Handling editor: H. C. Hemmings
65
Downloaded from http://bja.oxfordjournals.org/ at Northern Kentucky University on August 11, 2015
2 West M, Jack S, Grocott MPW. Perioperative cardiopulmonary exercise testing in the elderly. Best Pract Res Clin Anaesthesiol 2011; 25: 427– 37 3 Snowden CP, Prentis JM, Anderson HL, et al. Submaximal cardiopulmonary exercise testing predicts complications and hospital length of stay in patients undergoing major elective surgery. Ann Surg 2010; 251: 535– 41 4 Wilson RJT, Davies S, Yates D, et al. Impaired functional capacity is associated with all-cause mortality after major elective intraabdominal surgery. Br J Anaesth 2010; 105: 297– 303 5 Hennis PJ, Meale PM, Grocott MPW. Cardiopulmonary exercise testing for the evaluation of perioperative risk in noncardiopulmonary surgery. Postgrad Med J 2011; 87: 550–7 6 Hartley RA, Pichel AC, Grant SW, et al. Preoperative cardiopulmonary exercise testing and risk of early mortality following abdominal aortic aneurysm repair. Br J Surg 2012; 99: 1539–46 7 Schrieks IC, Barnes MJ, Hodges LD. Comparison study of treadmill versus arm ergometry. Clin Physiol Funct Imaging 2011; 31: 326–31 8 Lyons S, Richardson M, Bishop P, et al. Excess post-exercise oxygen consumption in untrained men following exercise of equal energy expenditure: comparisons of upper and lower body exercise. Diabetes Obes Metab 2007; 9: 889–94 9 Bar-Or O, Zwiren LD. Maximal oxygen consumption test during arm exercise reliability and validity. J Appl Physiol 1975; 38: 424– 6 10 Franklin B. Exercise testing, training and arm ergometry. J Sports Med 1985; 2: 100– 19 11 Martin T, Zebalos R, Weisman I. Gas exchange during maximal upper extremity exercise. Chest 1991; 99: 420–5 12 Reybrouck T, Heigenhauser GF, Faulkner JA, et al. Limitations to maximum oxygen uptake in arms, leg, and combined arm–leg ergometry. J Appl Physiol 1975; 38: 774–9 13 Davis J, Vodak P, Wilmore J, et al. Anaerobic threshold and maximal aerobic power for three modes of exercise. J Appl Physiol 1976; 41: 544– 50 14 Orr JL, Williamson P, Anderson W, et al. Cardiopulmonary exercise testing: arm crank vs cycle ergometry. Anaesthesia 2013; 68: 497– 501 15 Schneider DA, Wing AN, Morris NR. Oxygen uptake and heart rate kinetics during heavy exercise: a comparison between arm cranking and leg cycling. Eur J Appl Physiol 2002; 88: 100– 6 16 Fleisher L, Beckman JA, Brown KA, et al. ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing
