The Prognostic Value of Exercise Testing in Patients with Cystic Fibrosis

Study Population and Procedures

One hundred nine patients with cystic fibrosis, 7 to 35 years of age, underwent pulmonary-function and exercise testing in the late 1970s and were followed for eight years to determine the factors associated with subsequent mortality. Pulmonary-function testing was performed with standard spirometric techniques.⁸ Progressive exercise tests were performed on an electronically braked cycle ergometer according to the Godfrey protocol.⁹ The work rate was set at 0 watts (i.e., no added resistance) for the first minute and was increased by 10, 15, or 20 watts (depending on the patient's height) each minute thereafter, until the patient was unable to continue. The patient was encouraged to make a maximal effort.

Oxygen consumption (V̇_O2), work rate (Ẇ), carbon dioxide production (V̇_CO2), and end-tidal carbon dioxide tension (PCO₂) were measured throughout the test with a computerized system,¹⁰ and peak values were obtained during the last minute of exercise. Peak oxygen consumption and work capacity were expressed as percentages of the normal predicted values.¹¹ The test results were considered to be valid or near maximal if the respiratory exchange ratio (calculated as V̇_CO2/V̇_O2) exceeded 1.10 at the peak exercise level.¹² Oxyhemoglobin saturation (Sa_O2) was monitored throughout the exercise period with a Hewlett–Packard ear oximeter. The patient's electrocardiogram was also monitored continuously throughout the test. In addition to records of age and sex, measurements obtained at base line for each patient included the body-mass index, calculated as the weight in kilograms divided by the square of the height in meters, and the forced expiratory volume in one second (FEV₁), expressed as a percentage of the predicted value. Colonization of the respiratory tract by P. cepacia was documented throughout the eight-year period. The study was approved by the institutional review board of the University Hospitals of Cleveland, and written informed consent was obtained from the participants (or from a parent or legal guardian if the patient was under 18 years of age).

Statistical Analysis

Survival rates were calculated with standard life-table methods.¹³ Cox proportional-hazards regression models were used to determine the crude and adjusted relative risks of mortality.¹⁴ Colonization of the respiratory tract by P. cepacia was entered into the models as a time-dependent measure — i.e., the time from the date of the baseline measurement to the date P. cepacia was first isolated from a culture of the respiratory tract. Two sets of Cox regression analyses were performed. In one set, continuous explanatory variables, such as FEV₁, were entered into the model as continuous variables. In the second set, variables were divided into high, middle, and low categories and entered into the model as categorical variables. The former analysis made more complete use of the data, but the latter analysis yielded similar results and permitted the presentation of the risk of mortality associated with, for example, having an FEV₁ in the "worst" or lowest category as compared with the "best" or highest category.

Cox regression analyses were performed with a subsample of 58 patients for whom there was no evidence of colonization of the respiratory tract by P. cepacia. The values were divided at the median (because the sample was too small for them to be divided into thirds) in order to define categorical variables that were subsequently entered into the model.

Chi-square analysis for gradient proportions¹⁵ was used to determine whether there were significant linear trends in the proportions of patients who died during the eight-year follow-up according to age, body-mass index, FEV₁, end-tidal PCO₂, and oxygen consumption at peak exercise (V̇_O2peak). The chi-square test was used to determine the association of sex, Sa_O2, and presence of P. cepacia with mortality.

A two-tailed P value below 0.05 was considered to indicate statistical significance.

Figure 1. Figure 1. Survival among 109 Patients with Cystic Fibrosis, According to Fitness Level.

See the Results section for definitions of the fitness levels.

The mean age of the 109 patients at base line was 17 years (range, 7 to 35). Forty-seven percent were female. The mean FEV₁ was 59 percent of the predicted value (range, 24 to 95 percent). The mean V̇_O2peak was 35.0 ml per kilogram of body weight per minute (range, 13.6 to 59.9), or 70 percent of the predicted value (range, 21 to 131 percent). Only one patient was lost to follow-up. Data for this patient were included in the survival analyses and were censored during the first year. The survival rate for the total group at eight years was 56 percent. Figure 1 shows survival rates for the eight-year period according to fitness level. The patients in the highest fitness group (V̇_O2peak, ≥82 percent of predicted) had a survival rate of 83 percent, as compared with survival rates of 51 percent and 28 percent for the middle (V̇_O2peak, 59 to 81 percent of predicted) and the lowest (V̇_O2peak, ≤58 percent of predicted) fitness groups, respectively.

Table 1. Table 1. Crude Mortality Rates Eight Years after Pulmonary-Function and Exercise Testing in 108 Patients with Cystic Fibrosis, According to Base-Line Characteristics.

Table 1 shows the proportion of patients who died during the eight-year follow-up period, according to base-line characteristics. Older patients had significantly higher mortality than younger patients. Female patients were more likely to die than male patients, and smaller patients (body-mass index, <16.0) were more likely to die than larger patients (body-mass index, ≥18.6), although these differences were not statistically significant.

There was a significant inverse relation between FEV₁ at base line and the mortality rate. Seventy percent of patients with FEV₁ ≤50 percent of the predicted value died, as compared with only 26 percent of those with FEV₁ ≥65 percent of the predicted value. End-tidal PCO₂ at peak exercise was directly related to higher mortality. Sa_O2 at peak exercise was inversely related to higher mortality. Sixty-one percent of the patients who had colonization by P. cepacia at any time during the eight-year period died, as compared with only 29 percent of those who never had evidence of P. cepacia. V̇_O2peak (expressed as the percentage of predicted) was also significantly related to the survival rate, with a higher proportion of deaths in the lowest fitness category than in the highest category.

The relation between fitness and colonization of the respiratory tract by P. cepacia is also shown in Table 1. Among the patients with the lowest aerobic fitness (V̇_O2peak, ≤58 percent of predicted), 65 percent of those with P. cepacia died, as compared with 69 percent of those without P. cepacia. In the middle and the highest fitness groups, a higher proportion of the patients who had P. cepacia died; however, the difference between the groups with and without P. cepacia was significant only in the middle fitness group.

Table 2. Table 2. Prognostic Factors for Eight-Year Mortality in 108 Patients with Cystic Fibrosis.*

Univariate and multivariate Cox proportional-hazards regression analyses were performed to estimate crude and adjusted relative risks of mortality corresponding to various prognostic factors (Table 2). Univariate analysis indicated that lower pulmonary function (as measured by FEV₁), lower aerobic fitness (as measured by V̇_O2peak), higher end-tidal PCO₂ and lower Sa_O2 at peak exercise, and colonization by P. cepacia were associated with a significantly higher risk of dying. The increased risk of death associated with older age, female sex, and smaller body size was not statistically significant. When the significant univariate factors were entered into the multivariate model, only aerobic fitness and P. cepacia were significant independent correlates of mortality. Patients in the lowest category for aerobic fitness were more than three times as likely to die as patients in the highest fitness group. Patients with P. cepacia had a fivefold greater risk of dying than patients without P. cepacia. Patients with an end-tidal PCO₂ >41 mm Hg at peak exercise were more than twice as likely to die as patients with an end-tidal PCO₂ ≤36 mm Hg, although this difference was not statistically significant. Patients with severe lung disease (FEV₁, ≤50 percent of predicted) were not at significantly higher risk of dying than patients with mild lung disease (FEV₁, ≥65 percent of predicted) when fitness, presence of P. cepacia, and end-tidal PCO₂ at peak exercise were taken into account.

Multivariate Cox regression analysis was repeated with peak work capacity (Ẇ_peak, expressed as the percentage of predicted) used in place of V̇_O2peak. This was done because the measurement of oxygen consumption may not be readily available, and at times it may be difficult to obtain a valid result. Patients with the lowest work capacity (defined by a Ẇ_peak ≤73 percent of predicted) were three times more likely to die than patients in the highest work-capacity group (Ẇ_peak, >92 percent of predicted), after adjustment for FEV₁, presence of P. cepacia, and end-tidal PCO₂.

Similar analyses were carried out in the subgroup of 58 patients who did not have colonization of the respiratory tract by P. cepacia. Seventeen of the 58 patients died during the eight-year period. When the data for this group were divided at the median to create categorical variables, Cox multivariate regression analysis revealed that only lower aerobic fitness was associated with a significantly increased risk of mortality when FEV₁ and age were included in the model. Patients with lower aerobic fitness (defined by a V̇_O2peak <68 percent of predicted) were 5.8 times (95 percent confidence interval, 1.3 to 26.3; P<0.023) more likely to die than patients with higher aerobic fitness (defined by a V̇_O2peak ≥68 percent of predicted). Worse pulmonary function (defined by an FEV₁ ≤50 percent of predicted, as compared with an FEV₁ >50 percent of predicted) was associated with a risk of dying that was 3.5 times higher (95 percent confidence interval, 0.9 to 14.2; P<0.075), but this difference was not statistically significant.

The results of this investigation show a significant association between aerobic fitness (expressed as either V̇_O2peak or Ẇ_peak) and survival in a large population of patients with cystic fibrosis. This finding persisted after adjustment for age, sex, colonization of the respiratory tract by P. cepacia, body-mass index, pulmonary function, and end-tidal PCO₂ at peak exercise. Only one previous study has suggested an association between lower levels of physical fitness and mortality in patients with cystic fibrosis.⁷ Zinman et al.⁷ reported significantly lower initial work tolerance (Ẇ_peak) in 8 patients who died during a 26-month trial of nocturnal oxygen therapy in very ill patients, as compared with 20 patients who survived for the duration of the trial.

Previous studies have found lower body weight,¹⁶ worse pulmonary function (as measured by FEV₁),¹⁷ and colonization by P. cepacia⁴ to be significant prognostic indicators. Although we found these factors to be associated with increased mortality when they were examined individually, after adjustment for aerobic fitness (in a multivariate regression model) only P. cepacia colonization continued to be a significant independent correlate of mortality. The study design did not permit us to determine whether there was a significant interaction between aerobic fitness and P. cepacia in influencing mortality. Furthermore, in patients whose respiratory tracts were not colonized by P. cepacia, aerobic fitness was the only independent correlate of mortality that was statistically significant.

Studies of healthy populations have shown aerobic fitness to be associated with reduced mortality from all causes, as well as with reductions in cardiovascular disease and cancer mortality.¹⁸ Aerobic fitness is determined by both genetic and environmental factors. In the general population, 30 percent of the variation in aerobic fitness between people may be explained by a genetic component,¹⁹ whereas the remaining variation may be explained by environmental factors. We and others have shown that aerobic fitness in patients with cystic fibrosis can be improved by aerobic exercise training.^20,21 In addition to improved aerobic fitness, patients also had increased cardiovascular efficiency and ventilatory-muscle endurance after aerobic exercise training. Improvements in ventilatory-muscle endurance may help delay muscle fatigue and may potentially delay the development of respiratory failure.^{22 23 24} Since the deaths of most patients with cystic fibrosis may be attributed to respiratory failure,²⁵ aerobic exercise training may improve the prognosis by delaying this condition.

It is also possible that aerobic fitness is simply a marker for less severe systemic illness. When deaths during the first two years were excluded from the survival analysis, however, the eight-year survival rate increased by only 5 percent (from 28 percent to 33 percent) in the lowest fitness group. Even if fitness does reflect more severe illness, it is still strongly correlated with survival and is thus a powerful marker, suggesting that exercise testing is a useful tool. If, on the other hand, better aerobic fitness itself imparts a better prognosis, this would be important to know, since it would encourage the use of exercise training as a therapeutic measure.

In summary, the results of this study indicate a notable association between aerobic fitness and survival at eight years among patients with cystic fibrosis. This association persisted when we controlled for confounding factors such as age, sex, nutritional status, pulmonary function, colonization of the respiratory tract by P. cepacia, and end-tidal PCO₂ at peak exercise. Although these results do not allow us to determine which patient will die in an eight-year period, we can say which patient has a higher risk of dying on the basis of measurements of peak oxygen uptake or peak work capacity. The results suggest that exercise testing may be a valuable tool for determining prognosis in patients with cystic fibrosis. Further research is warranted to clarify the potential role of exercise training in influencing mortality among patients with cystic fibrosis.