Prognostic ability of VE/VCO 2 slope calculations using different exercise test time intervals in subjects with heart failure

Abstract

Background

The minute ventilation-carbon dioxide production (VE/VCO₂) slope, obtained during exercise testing, possesses prognostic value in heart failure (HF). The VE-VCO₂ relationship is generally linear thereby hypothetically producing similar slope values regardless of the exercise-test time interval used for calculation.

Design

This study assesses the ability of the VE/VCO₂ slope, calculated at different time intervals throughout a progressive exercise test, to predict 1-year cardiac-related hospitalization and mortality in subjects with HF.

Methods

Seventy-two subjects underwent symptom-limited exercise testing with ventilatory expired gas analysis. Mean age and left ventricular ejection fraction for 44 male and 28 female subjects were 51.2 years (± 13.0) and 27.0% (± 12.3) respectively. The VE/VCO₂ slope was calculated from time 0 to 25, 50, 75 and 100% of exercise time and subsequently used to create five randomly selected VE/VCO₂ slope categories.

Results

(The intraclass correlation coefficient found calculation of the VE/VCO₂ slope, when divided into quartiles, to be a reliable measure (alpha = 0.94, P<0.0001). Univariate Cox regression analysis revealed all VE/VCO₂ slope categories (25-100% and random selections) were significant predictors of cardiac-related hospitalization and mortality over a 1-year period. Multivariate Cox regression analysis revealed all VE/VCO₂ slope categories outperformed peak oxygen consumption (VO₂) in predicting hospitalization and mortality at 1 year.

Conclusions

Although the different classification schemes were not identical, these results suggest VE/VCO₂ slope maintains prognostic significance regardless of exercise-test time interval. Calculation of VE/VCO₂ slope may therefore still be valuable in subjects putting forth a sub-maximal effort while effort-dependent measures, such as peak VO₂, are not.

Keywords

ventilatory expired gas treadmill hospitalization mortality subject effort

Introduction

Symptom-limited exercise testing with ventilatory expired gas analysis provides a wealth of valuable information in the heart failure (HF) population. Peak oxygen consumption (VO₂) is presently considered the diagnostic/prognostic gold standard in HF [1]. Numerous investigations provide data to support the use of peak VO₂ in this capacity [2–5]. A potential weakness of peak VO₂ is its dependence on subject effort. Cessation of an exercise test secondary to subject request and prior to maximal physiological effort produces a falsely low peak VO₂. This obviously has negative implications for the diagnostic/prognostic value of this parameter. The risk of poor effort is of particular concern in the HF population given the potential apprehension a number of individuals may have in maximally exerting themselves during the exercise test.

Other ventilatory expired gas measures have recently emerged as potentially having diagnostic/prognostic superiority over peak VO₂. One such measure is the minute ventilation (VE)-carbon dioxide production (VCO₂) relationship commonly expressed as the VE/VCO₂ slope [3, 6–11]. Robbins et al. [6], for example, found a VE-VCO₂ relationship 44.7 at peak exercise was superior to a peak VO₂ 13.9 ml/kg per min in predicting 18-month survival in a group of patients with HF. The VE-VCO₂ relationship was also a significant predictor of survival as a continuous variable and was again superior to peak VO₂. The authors of the present study have also found VE/VCO₂ slope to be superior to peak VO₂ in predicting 1-year, cardiac-related hospitalization in a small group of patients with HF [10].

The slope of the VE-VCO₂ relationship is generally linear and should therefore remain relatively constant throughout a progressive exercise test making this measure independent of subject effort. In the event of submaximal effort, the VE/VCO₂ slope would then theoretically maintain diagnostic/prognostic significance. To date, however, no investigation has tested this hypothesis. The purpose of this investigation is therefore to assess the ability of VE/VCO₂ slope, calculated at different time intervals throughout a progressive exercise test, to predict 1-year hospitalization and mortality in a group of patients with HF.

Methods

Seventy-two consecutive subjects diagnosed with HF and undergoing a symptom-limited exercise test between 15 May 1997 and 2 May 2001, were retrospectively recruited from the HF program at the Virginia Commonwealth University Health Systems. All tests were conducted on an outpatient basis. Written informed consent was obtained from all subjects prior to testing. Virginia Commonwealth University Institutional Review Board approval was obtained for those subjects undergoing an exercise test as part of a prospective research project and not as a standard of care. The exercise tests performed as part of a subject's standard of care was done so to establish a baseline aerobic capacity. None of the tests were performed to assess worsening heart failure or arrhythmias. The exercise testing procedures were, however, identical for all subjects. Subject and pharmacological characteristics are listed in Table 1.

Exclusion criteria consisted of diagnosed pulmonary disease (per physician examination), myocardial infarction within the past 6 months, signs and/or symptoms suggestive of decompensated heart failure, and/or any orthopedic condition that would not allow the subject to ambulate on a treadmill. Inclusion criteria consisted of diagnosed compensated HF, and evidence of left ventricular dysfunction by echocardiogram or cardiac catheterization.

Table 1

Subject characteristics


Number of subjects	72 (44 male/28 female)

Age (years)	51.3±13.2 (range: 21-82)
Left ventricular ejection fraction (%)	27.0±12.3 (range: 10-55)
Heart failure etiology (ischemic/ non-ischemic)	29/43
ACE inhibitor (number of subjects)	51
Cardiac glycoside (number of subjects)	57
Diuretic (number of subjects)	60
Nitrate (number of subjects)	15
Beta-blocker (number of subjects)	39

ACE, angiotensin-converting enzyme.

Equipment calibration

Ventilatory expired gas analysis was obtained through a metabolic cart (Medgraphics CPX-D, Minneapolis, Minnesota, USA or Sensormedics Vmax29, Yorba Linda, California, USA). The oxygen and carbon dioxide sensors were calibrated using gases with known oxygen, nitrogen, and carbon dioxide concentrations prior to each test. The flow sensor was also calibrated before each test using a three-litre syringe.

Testing procedure and data collection

Symptom-limited exercise tests with ventilatory expired gas analysis were conducted through the use of a treadmill. The modified ramping protocol selected for testing consisted of approximately 2 mlO₂ kg per min increases in workload every 30 s [10, 12, 13]. Stage one began at 1.0 mile per hour (mph) and a 0% grade. Stages increased by 0.1 mph and 0.5% grade thereafter. The same conservative treadmill protocol was used to test all subjects. Monitoring consisted of continuous 12-lead electrocardiography (Quinton 4000, Seattle, Washington, USA), manual blood pressure measurements approximately every third stage (90 s), heart rate recordings every stage via the electrocardiogram, and rating of perceived exertion (Borg 15 grade scale) at each stage. Test termination criteria consisted of patient request, ventricular tachycardia, 2 mm of horizontal or down sloping ST segment depression, or a drop in systolic blood pressure of 20 mmHg during progressive exercise. A qualified exercise physiologist with physician supervision conducted each exercise test.

Breath-by-breath VCO₂ (l/min) and VE (l/min) collected pre-, during, and post-exercise, was averaged over 10-s intervals prior to analysis. The 10-s average VO₂ during the final stage of the exercise test was deemed the peak value. The VE/VCO₂ slope was determined by manually inputting the 10-s averaged VE and VCO₂ data throughout the test into spreadsheet software (Microsoft Excel; Microsoft Corp., Bellevue, Washington, USA). Four different VE/VCO₂ slope values were derived from the exercise test. The starting point for each was at the beginning of the ramping protocol. The amount of data used for the four VE/VCO₂ slope calculations were 25, 50, 75, and 100% of the exercise test. Although not an exact depiction of the VE-VCO₂ relationship given the possible non-linear response toward maximal exercise, Figure 1 provides a graphical illustration of the quartile calculations. Using the data from the 25, 50, 75, and 100% VE/VCO₂ slope calculations, a random number generator created five random categories (r1-5) for each subject. For example, in subject one, the random number generator selected the 100% (r1), 75% (r2), 75% (r3), 50% (r4), and 100% (r5) VE/VCO₂ slope calculations for the five random categories respectively.

Fig. 1

Illustration of 25, 50, 75, and 100% VE/VCO₂ slope calculations.

Subjects were tracked for hospitalization and mortality 1 year following exercise testing via medical chart review. Any hospital admission or death with a cardiac-related diagnosis within the 1-year tracking period was counted as an event for this study. Tracking continued until a subject was admitted to the hospital with a cardiac-related discharge diagnosis, died, or 1 year since the time of the exercise test had passed.

Statistical analysis

The intraclass correlation coefficient (ICC) assessed reliability amongst the 25, 50, 75, and 100% VE/VCO₂ slope categories. Univariate Cox regression analysis determined the ability of the 25, 50, 75, and 100% VE/ VCO₂ slope categories, five randomly selected VE/VCO₂ slope categories (r1-5) and peak VO₂ to predict cardiac-related hospitalization and mortality over a 1-year period. The construction of the randomly selected VE/VCO₂ slope classification schemes is described in the testing procedures and data collection section. Nine separate multivariate Cox regression analyses were conducted using each of the VE/VCO₂ slope classification schemes and peak VO₂ to predict cardiac-related hospitalization and mortality over a 1-year period. A statistical software program was used for data analysis (SPSS 10.0 for Windows; Chicago, Illinois, USA). All statistical tests with a P value < 0.05 were considered significant.

Results

The primary endpoint for each exercise test was volitional fatigue with no adverse events warranting premature test termination occurring. None of the exercise tests were terminated secondary to angina. The mean peak respiratory exchange ratio (RER) was 1.06 (± 0.11). Fifty-one of the subjects had a peak RER ≥ 1.0. Twelve of the remaining 21 subjects had a peak RER ≥ 0.95. Mean and standard deviation of peak VO₂, 25, 50, 75, and 100% VE/VCO₂ slope categories, and randomly selected VE/VCO₂ slope categories (r1-r5) are listed in Table 2. A graphic representation of mean values of the variables of interest is illustrated in Figure 2.

The ICC found reliability amongst the 25, 50, 75, and 100% VE/VCO₂ slope calculations to be highly significant (alpha = 0.94, P < 0.0001). Twenty-five of 72 (34.7%) subjects were hospitalized with a cardiac-related discharge diagnosis in the year following the exercise test. There were additionally eight (11.7%) cardiac-related deaths 1 year after exercise testing. Univariate Cox regression results are presented in Table 3. All VE/VCO₂ slope classification schemes and peak VO₂ were significant predictors of 1-year cardiac-related hospitalization and mortality.

Multivariate Cox regression results are presented in Table 4. In each of the analyses, peak VO₂ did not significantly improve prediction of 1-year cardiac-related hospitalization or mortality and was removed from the equation.

Table 2

Mean and standard deviation of pertinent ventilatory expired gas calculations∗


	Mean	Standard deviation (±)

VE/VCO₂ slope-25% (1)	34.1	9.0
VE/VCO₂ slope-50% (2)	33.1	7.4
VE/VCO₂ slope-75% (3)	33.4	7.8
VE/VCO₂ slope-100% (4)	35.2	8.7
VE/VCO₂ slope-random 1 (5)	34.0	8.3
VE/VCO₂ slope-random 2 (6)	34.2	7.9
VE/VCO₂ slope-random 3 (7)	33.7	7.4
VE/VCO₂ slope-random 4 (8)	33.7	9.0
VE/VCO₂ slope-random 5 (9)	34.6	9.2
Peak VO₂ (mlO₂/kg per min) (10)	14.3	5.0

∗

See Figure 2.

Fig. 2

Mean values of VE/VCO₂ slope categories and peak VO₂. Bar graph numbers on x-axis correspond to numbers in parenthesis in Table 2.

Table 3

Univariate Cox regression results: predictors of 1-year cardiac-related hospitalization and mortality


	Hospitalization		Mortality

	Chi-square	P value	Chi-square	P value

VE/VCO₂ slope-25%	16.9	<0.001	8.7	0.003
VE/VCO₂ slope-50%	25.6	<0.001	6.6	0.01
VE/VCO₂ slope-75%	32.2	<0.001	8.4	0.004
VE/VCO₂ slope-100%	24.8	<0.001	10.6	0.001
VE/VCO₂ slope-random 1	22.2	<0.001	8.1	0.005
VE/VCO₂ slope-random 2	18.3	<0.001	6.4	0.01
VE/VCO₂ slope-random 3	18.9	<0.001	7.5	0.006
VE/VCO₂ slope-random 4	24.6	<0.001	7.6	0.006
VE/VCO₂ slope-random 5	24.9	<0.001	7.6	0.006
Peak VO₂	8.9	0.003	4.6	0.03

Discussion

The results of the present study indicate that VE/VCO₂ slope is a reliable measure, which maintains prognostic value irrespective of the exercise test time interval used for calculation. Maintenance of prognostic value indicates that VE/VCO₂ is independent of subject effort. This may be a primary reason for the prognostic superiority of the VE/VCO₂ slope over peak VO₂ (an effort-dependent parameter) as demonstrated by several investigations [6, 8, 10].

The five random VE/VCO₂ slope categories were created to ensure that patterns within the classification schemes (25, 50, 75, and 100%) did not influence prognostic value. The random categories may more accurately reflect clinical practice as exercise effort varies among subjects. As indicated by the univariate and multivariate Cox regression analysis in Tables 3 and 4, the prognostic capability of the VE/VCO₂ slope was highly consistent.

Decreased pulmonary perfusion and cardiac function appear to be a leading mechanism behind the elevated VE/VCO₂ slope observed in some patients with HF [7, 14–16]. Additionally, the abnormal VE-VCO₂ relationship, when observed in patients with HF, is often apparent at rest and throughout exercise. Given the association between cardiopulmonary function and the VE-VCO₂ relationship during both rest and exercise, the robust predictive value of this ventilatory expired gas measure should come as no surprise.

An assumption of the present study is all subjects put forth near maximal effort represented by the 100% VE/VCO₂ slope classification scheme. Although the mean peak respiratory exchange ratio was above one (1.06), some subjects were most certainly at submaximal effort when the exercise test was terminated. While the authors recognize this assumption as a limitation, the fact that the VE/VCO₂ slope was a significant predictor of hospitalization and mortality irrespective of calculation scheme should not be dismissed.

Determining and recommending an optimal exercise test time interval for VE/VCO₂ slope calculation or cutoff value is not the intent of the present study. Quartile calculations were an arbitrary selection and should not be implemented into clinical practice. The intent was rather to simulate submaximal effort and demonstrate that the VE/VCO₂ slope maintains prognostic value, which was achieved. Presently, the VE/VCO₂ slope has primarily been calculated by either using data to the point of ventilatory threshold [8] or to peak exercise [10]. The premise for using data only to the point of ventilatory threshold for the calculation is the belief that the excessive rise in VE following an increase in anaerobic metabolism causes the VE/VCO₂ slope to become nonlinear and capturing this response adds no value. Strong scientific support for an optimal VE/VCO₂ slope expression is, however, lacking and both methods of calculation (to the point of ventilatory threshold and to maximal exercise) have produced a measure of prognostic value. The authors feel a deeper discussion of this issue is beyond the scope of this study and recommend further investigation to examine the possibility of an optimal VE/VCO₂ slope calculation method.

The choice of endpoints for the present study was based upon previous investigations and the authors view on the time period and information that potentially provides optimal clinical data. Limiting endpoints to a 1-year time period may be clinically optimal given the fluid nature of cardiac function in the HF patient. Since individuals with HF can shift from compensated to uncompensated HF (or vice versa) rather abruptly, limiting the time period a single exercise test is used for prognostic assessment seems prudent. A 1-year tracking period may strike a sufficient balance between avoiding outdated information and the economic constraints of multiple exercise tests. Additionally, most research examining the prognostic value of ventilatory expired gas parameters do not use hospitalization as an endpoint. Given that HF is the primary hospital diagnostic-related group among Medicare patients [17], analysis of measures predicting hospitalization in this population seems warranted. The ability of the VE/VCO₂ slope to effectively predict hospitalization may provide clinicians with the unique opportunity to identify high-risk patients, provide appropriate interventions, prevent non-fatal adverse events, and reduce cost of care.

Table 4

Multivariate Cox regression results: predictors of 1-year cardiac-related hospitalization


	Hospitalization		Value of adding peak VO₂ to the regression

	Chi-square	P value	Chi-square	P value

(a) Predictors of 1-year cardiac-related hospitalization
VE/VCO₂ slope-25%	16.9	<0.001	2.6	0.11
VE/VCO₂ slope-50%	25.6	<0.001	1.3	0.26
VE/VCO₂ slope-75%	32.2	<0.001	1.3	0.25
VE/VCO₂ slope-100%	24.8	<0.001	1.9	0.17
VE/VCO₂ slope-r 1	22.2	<0.001	2.4	0.12
VE/VCO₂ slope-r 2	18.3	<0.001	2.9	0.09
VE/VCO₂ slope-r 3	18.9	<0.001	0.76	0.38
VE/VCO₂ slope-r 4	24.6	<0.001	1.4	0.22
VE/VCO₂ slope-r 5	24.9	<0.001	2.0	0.16
(b) Predictors of 1-year cardiac-related mortality
VE/VCO₂ slope-25%	8.7	0.003	1.0	0.31
VE/VCO₂ slope-50%	6.6	0.01	1.6	0.20
VE/VCO₂ slope-75%	8.4	0.004	1.5	0.22
VE/VCO₂ slope-100%	10.6	0.001	1.1	0.30
VE/VCO₂ slope-r 1	8.1	0.005	1.6	0.21
VE/VCO₂ slope-r 2	6.4	0.01	1.8	0.18
VE/VCO₂ slope-r 3	7.5	0.006	0.9	0.36
VE/VCO₂ slope-r 4	7.6	0.006	1.4	0.23
VE/VCO₂ slope-r 5	7.6	0.006	1.7	0.19

Of paramount importance is the need for future investigations to clearly define the clinical role of the VE/VCO₂ slope in assessing patients with HF. With regards to exercise testing, peak VO₂ is presently the ventilatory expired gas gold standard measure when determining prognosis and guiding clinical decisions in the HF population. The dependence attainment of a true peak VO₂ has on subject effort is a limitation considering the apprehension a patient with HF may have in physically exerting himself/herself to maximal capacity. The prognostic value of the VE/VCO₂ slope, as supported by the findings of the present study, does not appear to have an effort-dependent limitation and consistently outperforms peak VO₂ in predicting outcome. If further investigations continue to demonstrate the prognostic superiority of the VE/VCO₂ slope over peak VO₂, thought should be given to revising clinical guidelines, which define interpretation of ventilatory expired gas data in the HF population. The authors are certainly not suggesting peak VO₂ should be disregarded. Analysis of this measure has consistently demonstrated prognostic value [2–4, 18] and is the cornerstone of prescribing exercise/activity guidelines [1]. However, given the building body of evidence reporting the diagnostic/prognostic value of the VE-VCO₂ relationship in the HF population [3, 5–8, 10, 19], thought should be given to optimizing the value of information obtained from exercise testing with ventilatory expired gas analysis.

In conclusion, the VE/VCO₂ slope continues to show clinical promise as a prognostic marker in HF. The fact that this measure, independent of subject effort, is reproducible at different exercise test time intervals and maintains prognostic value may help to explain the findings of this and previous investigations.

References

Gibbons

Balady

Beasley

Bricker

Duvernoy

Froelicher

ACC/AHA Guidelines for exercise testing. A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on Exercise Testing). J Am Coll Cardiol 1997; 30:260–311.

Mancini

Eisen

Kussmaul

Mull

Edmunds

Jr. Wilson

Value of peak exercise oxygen consumption for optimal timing of cardiac transplantation in ambulatory patients with heart failure. Circulation 1991; 83:778–786.

Chua

Ponikowski

Harrington

Anker

Webb-Peploe

Clark

Clinical correlates and prognostic significance of the ventilatory response to exercise in chronic heart failure. J Am Coll Cardiol 1997; 29:1585–1590.

Myers

Gullestad

Vagelos

Bellin

Ross

Cardiopulmonary exercise testing and prognosis in severe heart failure: 14 mL/kg/min revisited. Am Heart J 2000; 139:78–84.

Francis

Shamim

Davies

Piepoli

Ponikowski

Anker

Cardiopulmonary exercise testing for prognosis in chronic heart failure: continuous and independent prognostic value from VE/VCO₂ slope and peak VO₂ . Eur Heart J 2000; 21:154–161.

Robbins

Francis

Pashkow

Snader

Hoercher

Young

Ventilatory and heart rate responses to exercise: better predictors of heart failure mortality than peak oxygen consumption. Circulation 1999; 100:2411–2417.

Reindl

Wernecke

Opitz

Wensel

Konig

Dengler

Impaired ventilatory efficiency in chronic heart failure: possible role of pulmonary vasoconstriction. Am Heart J 1998; 136:778–785.

Corra

Mezzani

Bosimini

Scapellato

Imparato

Giannuzzi

Ventilatory response to exercise improves risk stratification in patients with chronic heart failure and intermediate functional capacity. Am Heart J 2002; 143:418–426.

MacGowan

Murali

Ventilatory and heart rate responses to exercise: better predictors of heart failure mortality than peak exercise oxygen consumption. Circulation 2000; 102:E182.

10.

Arena

Humphrey

Comparison of ventilatory expired gas parameters used to predict hospitalization in patients with heart failure. Am Heart J 2002; 143:427–432.

11.

Kleber

Vietzke

Wernecke

Bauer

Opitz

Wensel

Impairment of ventilatory efficiency in heart failure: prognostic impact. Circulation 2000; 101:2803–2809.

12.

McInnis

Bader

Pierce

Balady

GJ.

Comparison of cardiopulmonary responses in obese women using ramp versus step treadmill protocols. Am J Cardiol, 1999; 83:289–291, A7.

13.

Arena

Humphrey

Relationship between ventilatory expired gas and cardiac parameters during symptom-limited exercise testing in patients with heart failure. J Cardiopulm Rehabil 2001; 21:130–134.

14.

Banning

Lewis

Northridge

Elborn

Hendersen

AH.

Perfusion/ventilation mismatch during exercise in chronic heart failure: an investigation of circulatory determinants. Br Heart J 1995; 74:27–33.

15.

Lewis

Banning

Cooper

Sundar

Facey

Evans

Impaired matching of perfusion and ventilation in heart failure detected by 133xenon. Basic Res Cardiol 1996; 91 Suppl 1:45–49.

16.

Lewis

Macdougall

Willis

Henderson

AH.

The ventilatory cost of exercise compared in chronic heart failure and chronic renal anaemia. Q J Med 1992; 83:523–531.

17.

Parmley

WW.

Heart failure awareness week: February 14–21. J Am Coll Cardiol 2000; 35:534.

18.

Davies

Francis

Piepoli

Scott

Ponikowski

Coats

AJ.

Chronic heart failure in the elderly: value of cardiopulmonary exercise testing in risk stratification. Heart 2000; 83:147–151.

19.

MacGowan

Janosko

Cecchetti

Murali

Exercise-related ventilatory abnormalities and survival in congestive heart failure. Am J Cardiol 1997; 79:1264–1266.