Creating and Evaluating a Prediction Equation for VO_2peak in Individuals with Early Stage, Never Medicated Parkinson’s Disease

Abstract

Background:

Parkinson’s disease (PD) is a neurodegenerative nervous system condition causing motor and nonmotor symptoms. Endurance training is commonly prescribed in people with PD for possible slowing of disease progression. Since people with PD exhibit lower cardiorespiratory fitness, it is important to understand peak aerobic capacity (VO_2peak) in people with PD. VO_2peak prediction equations may be used when cardiopulmonary exercise testing (CPET) is unavailable; however, exercise-based PD-specific prediction equations are lacking. The purpose of the study was to develop a PD-specific VO_2peak prediction equation and to compare this equation to published VO_2peak prediction equations.

Methods:

This study included N = 127 never-medicated individuals with PD, aged 40–80 yr, Hoehn and Yahr stages 1–2, within 5 yr of diagnosis, and exercising ≤3 d/wk, who completed a treadmill CPET. Linear regression analyses were performed to generate the VO_2peak equation from a validation subsample, which was applied to a cross-validation subsample. The equation was compared with two published equations for healthy adults.

Results:

The PD-specific VO_2peak equation was: VO_2peak (mL/kg/min) = 12.466 + 0.149 × (treadmill speed [m/min]) + 85.7 × (treadmill grade [%, as a decimal]) – 2.383 × (sex [0 = male, 1 = female]) – 0.135 × (age [years]). There was no difference between estimated and measured VO_2peak in the cross-validation subsample. Our equation successfully predicted VO_2peak in early PD, whereas VO_2peak was over- and underestimated in people with PD by the American College of Sports Medicine and Foster equations, respectively.

Conclusions:

Clinicians can estimate VO_2peak in individuals with PD to identify those for whom endurance exercise training should be a major health priority, develop an exercise prescription, and assess changes in VO_2peak over time.

Parkinson’s disease (PD) is an age-related neurodegenerative nervous system condition marked by motor and nonmotor signs and symptoms. Motor signs include tremor, bradykinesia, rigidity, postural instability, and balance deficits (1). In addition to these presentations, several nonmotor symptoms can present, such as anxiety, depression, constipation, hallucination, sleep disturbance, cognitive decline, and autonomic dysfunction (1). People with PD have a statistically significant reduction in cardiorespiratory fitness (CRF) as measured by peak aerobic capacity (VO_2peak) as compared with healthy adults (2–4). This may result in an increased risk for cardiovascular disease in PD compared with their healthy peers (5).

Although there is currently no cure for PD, several dopaminergic drugs and nondopamine targeting drugs exist, which aim to control the motor symptoms present with PD. Oftentimes, these drugs involve an extensive regimen, leading to several side effects. Physical exercise, in the form of both resistance and endurance exercise, is an additional important treatment option (6). Endurance exercise, or sustained movements undertaken to improve cardiovascular health and overall fitness (7), has been shown to increase VO_2peak and functional capacity, and decrease Unified Parkinson’s Disease Rating Scale (UPDRS) scores in people with PD (8). This exercise-induced reduction in UPDRS scores in PD represents the disease-modifying potential of exercise in this population (9–11). Regarding this type of training, endurance exercise can be prescribed based on the percentage of measured VO_2peak (7). CRF, as measured by VO_2peak, can serve as a vital sign (12) and quantify improvement over time following endurance exercise training.

VO_2peak is commonly assessed via cardiopulmonary exercise testing (CPET), the gold standard for assessing CRF. This commonly involves treadmill testing with a standardized protocol by which speed and/or incline are altered to capture the maximal amount of oxygen that the participant can utilize during exercise via direct measurement of expired respiratory gases. While this assessment technique results in the most accurate measure of VO_2peak, time, energy, resources, and cost often limit the feasibility of a full CPET, and the cost-effectiveness has not yet been studied (13). An alternative option for estimating CRF is using a VO_2peak prediction equation. The American College of Sports Medicine (ACSM) equation serves as a commonly utilized VO_2peak prediction equation, calculating predicted oxygen consumption on a treadmill based on steady-state speed and grade (7). This equation has been demonstrated to overestimate VO_2peak during CPET, resulting in the need to refine such an equation to better represent the metabolic cost of treadmill exercise during CPET, irrespective of test protocol (14). Although these work rate-based VO_2peak prediction equations have been developed using data from healthy individuals (7,14), an equation that estimates VO_2peak in individuals with de novo PD has yet to exist.

However, for several reasons, there is a need for a specific equation, which estimates the VO_2peak in individuals with PD. Oxygen consumption has been shown to be higher in people with PD during submaximal exercise (15). This difference in metabolic efficiency suggests that responses to endurance exercise are different in people with PD, potentially due to rigidity or bradykinesia associated with this condition (15). These patients may also demonstrate altered gait kinematics. The alteration in gait kinematics generally includes a slow gait with a short step shuffling pattern and forward stooped posture accompanied by asymmetrical arm swinging (16,17). These patterns, along with increased knee and ankle coactivation (18), directly increase energy expenditure and decrease gait efficiency (16,19). These differences may result in inaccurate predictions of VO_2peak in people with PD when employing traditional VO_2peak prediction equations, which were developed from healthy adults. Further, the financial and resource costs associated with CPET may be a challenge for widespread implementation of this assessment technique in the exercise prescription framework in PD. Therefore, the purposes of the study were (1) to develop a PD-specific VO_2peak prediction equation and (2) to compare the performance of this PD-specific VO_2peak prediction equation to published VO_2peak equations.

METHODS

Study Design

The Study in Parkinson’s Disease of Exercise (SPARX) was a randomized controlled trial in participants with de novo PD, in which participants were tested at outpatient and community-based facilities. The current study used retrospective analyses of baseline SPARX data. The Institutional Review Board for each study site (University of Colorado, University of Illinois at Chicago, Northwestern University, University of Pittsburgh, and Rush University Medical Center) approved the study protocol. Participant informed consent was obtained via written forms before any study activities.

Study Participants

Participants with de novo, unmedicated PD were enrolled from 2012 to 2015 from Chicago, IL, Pittsburgh, PA, and Denver, CO. Inclusion criteria required that participants were aged 40–80 yr, Hoehn and Yahr (H&Y) stages 1–2, and not participating in endurance exercise more than 3 days a week. Exclusion criteria for the study were uncontrolled cardiovascular disease, abnormal CPET results, failure to meet the UK Parkinson’s Disease Society Brain Break criteria for idiopathic PD (20), and usage of dopaminergic medication for the management of PD.

Health History and Clinical Assessment

Demographic and clinical data were collected via self-reported questionnaires. A baseline health history and medical records review were conducted before participation by trained clinicians. Objective measurements were obtained via the UPDRS and Movement Disorders Society UPDRS.

Cardiopulmonary Exercise Testing

Treadmill testing was conducted using the modified Gardner Protocol (21). Supine and standing blood pressure measurements were obtained before initiation of exercise testing. These measurements were taken after 5 min of quiet rest. The treadmill was kept at a constant speed throughout the test for any given participant; however, there were interindividual differences in treadmill speed based on participant walking ability. The grade was increased by 2% every 2 min until volitional fatigue was reached. That is, each participant was encouraged to exercise as long as they were able, and all tests were terminated when the participant could no longer continue. Heart rate was measured using electrocardiography during the test. VO_2peak was measured via expired gas analysis (ParvoMedics TrueOne 2400, Salt Lake City, UT, for Chicago and Denver; AEI Technologies, MOXUS, Bastrop, TX, for Pittsburgh)

Following completion of the CPET, maximal effort was evaluated via a respiratory exchange ratio ≥1.05 (22), heart rate response, rating of perceived exertion, and occasional investigation of a plateau in exercising VO₂. VO_2peak was identified as the average of the two highest consecutive 30-s values (23), which were reported automatically throughout the test by the metabolic cart’s internal software and derived from the carts’ mixing chambers.

Comparison VO_2peak Prediction Equations

Comparison equations for prediction of VO_2peak include the ACSM equation (7) and the Foster equation (14). The ACSM equation is VO_2peak = 3.5 + (0.1 × speed) + (1.8 × speed × grade). The Foster equation is VO_2peak = 0.694 × (ACSM’s predicted) + 3.33. These equations use speed in meters per minute and grade as a percentage expressed in decimal format. The output of both equations is VO_2peak in mL/kg/min.

Statistical Analysis

Participant demographic and clinical characteristics were summarized for our participants, including mean (SD) and frequency (percentage). Multiple linear regression analyses were performed to predict VO_2peak in a validation subsample comprised of a randomly selected half of our study sample. Randomization into the validation subsample was accomplished using a computer-based random number generator. The highest grade of the treadmill achieved when the CPET was stopped was used in the regression analyses. The VO_2peak prediction equation was then applied to a cross-validation subsample, which consisted of the other half of the SPARX sample. The directly measured VO_2peak and the predicted VO_2peak in the cross-validation subsample were compared using a paired samples t-test. We further analyzed the performance of the prediction equation using a Pearson correlation between measured and predicted VO_2peak in the cross-validation subsample. A Bland-Altman plot was generated to depict the difference between measured and predicted VO_2peak values. The PD-specific prediction equation was then compared with general VO_2peak prediction equations previously established in healthy cohorts (7,14). All statistical analyses were performed in Statistical Package for the Social Sciences (Version 29, Chicago, IL). Significance for all analyses was set at P <0.05.

RESULTS

Our study included 127 individuals (57.5% male) with a mean age of 63.6 yr, and study participants were enrolled on average 0.7 ± 0.9 yr after diagnosis. During CPET, mean maximal treadmill speed and grade were 3.0 ± 0.6 mph and 10.8 ± 3.5%, respectively. Participants recorded a mean VO_2peak of 23.8 ± 6.0 mL/kg/min and achieved 94.7 ± 10.4% of their age-predicted maximal heart rate. Demographic, clinical, and CPET characteristics of study participants are included in Table 1.

TABLE 1.

Demographic, clinical, and CPET characteristics of participants, data are mean (SD).

	Total Sample (n = 127)	Validation Subsample (n = 64)	Cross-Validation Subsample (n = 63)
Sex, N (%)
Male	73 (57.5%)	37 (57.8%)	36 (57.1%)
Female	54 (42.5%)	27 (42.2%)	27 (42.9%)
Age (years)	63.6 (9.3)	62.8 (9.9)	64.3 (8.6)
BMI (kg/m2)	26.7 (4.1)	26.8 (4.3)	26.7 (4.0)
H&Y, N (%)
Stage 1	33 (26.0%)	17 (26.6%)	16 (25.4%)
Stage 2	94 (74.0%)	47 (73.4%)	47 (74.6%)
Years since diagnosis	0.7 (0.9)	0.8 (0.9)	0.6 (0.9)
UPDRS MBM	0.9 (1.2)	0.9 (1.1)	0.8 (1.3)
UPDRS ADL	5.9 (3.5)	6.6 (3.8)	5.2 (3.1)
UPDRS motor	16.7 (7.0)	16.3 (6.9)	17.1 (7.0)
UPDRS total	23.5 (9.0)	23.8 (9.5)	23.1 (8.5)
MDS total	34.0 (11.8)	34.0 (11.9)	34.0 (11.8)
Max speed (mph)	3.0 (0.6)	3.0 (0.6)	3.0 (0.6)
Max grade (%)	10.8 (3.5)	11.2 (3.5)	10.3 (3.5)
Max HR (bpm)	148.3 (19.4)	149.8 (20.7)	146.8 (18.1)
VO2peak (mL/kg/min)	23.8 (6.0)	24.4 (6.1)	23.1 (5.8)
APMHR (%)	94.7 (10.4)	95.1 (10.5)	94.3 (10.4)
Peak respiratory exchange ratio	1.06 (0.07)	1.05 (0.08)	1.06 (0.07)

APMHR, baseline percentage of age-predicted maximum heart rate; BMI, body mass index; H&Y, Hoehn and Yahr scale; HR, heart rate; MDS Total, Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale Total Score; UPDRS ADL, Unified Parkinson’s Disease Rating Scale—Activity of Daily Living; UPDRS MBM, Unified Parkinson’s Disease Rating Scale—Motor Behavior Module; UPDRS Motor, Unified Parkinson’s Disease Rating Scale—Motor Examination; UPDRS Total, Unified Parkinson’s Disease Rating Scale—Total Score.

Participants were randomly assigned to either the validation or cross-validation group. Multiple regression analysis of the validation group (n = 64) was performed and showed that sex, treadmill (TM) speed, grade, and participant age were significant predictors of VO_2peak. Table 2 presents the nonstandardized regression equation coefficients (b), standardized coefficients (β), and increase in variance explained by the four predictor variables that were included in the regression model.

TABLE 2.

VO_2peak regression equation coefficients for the PD-specific equation.

Variables Included	Nonstandardized Coefficient (b)	Standardized Coefficients (β)	New Variance Explained (R2 Change)	P-Value
Constant	12.466	–	–	0.022
Max speed (m/min)	0.149	0.417	0.354	<0.001
Max grade (%, as a decimal)	85.7	0.485	0.253	<0.001
Sex (0 = male, 1 = female)	−2.383	−0.194	0.035	0.015
Age (years)	−0.135	−0.218	0.031	0.012

The regression model successfully predicted VO_2peak in the validation group (R = 0.820, R² = 0.672, P < 0.001) with the standard error of the estimate (SEE) of 3.62 mL/kg/min. The regression equation we developed for the prediction of PD-specific VO_2peak was as follows:

$${\displaystyle \begin{array}{c}\mathrm{V}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}(\mathrm{m}\mathrm{L}/\mathrm{k}\mathrm{g}/\mathrm{m}\mathrm{i}\mathrm{n})=12.466+0.149\times \left(\mathrm{T}\mathrm{M}\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}[\mathrm{m}/\mathrm{m}\mathrm{i}\mathrm{n}]\right)\\ +85.7\times \left(\mathrm{T}\mathrm{M}\mathrm{g}\mathrm{r}\mathrm{a}\mathrm{d}\mathrm{e}[\%,\mathrm{a}\mathrm{s}\mathrm{a}\mathrm{d}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{m}\mathrm{a}\mathrm{l}]\right)\\ \text{–}2.383\times \left(\mathrm{s}\mathrm{e}\mathrm{x}[0=\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e},1=\mathrm{f}\mathrm{e}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e}]\right)\text{–}0.135\times \left(\mathrm{a}\mathrm{g}\mathrm{e}\left[\mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}\mathrm{s}\right]\right)\end{array}}$$

The regression equation was also developed using max treadmill speed in miles per hour and max treadmill grade as a percentage, providing an equation that may more closely reflect treadmill settings directly available in the United States of America. This regression equation was as follows:

$${\displaystyle \begin{array}{c}{\displaystyle }\mathrm{V}{\mathrm{O}}_{2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}(\mathrm{m}\mathrm{L}/\mathrm{k}\mathrm{g}/\mathrm{m}\mathrm{i}\mathrm{n})=12.466+3.984\times \left(\mathrm{T}\mathrm{M}\mathrm{s}\mathrm{p}\mathrm{e}\mathrm{e}\mathrm{d}\left[\mathrm{m}\mathrm{p}\mathrm{h}\right]\right)\\ {\displaystyle }+0.857\times \left(\mathrm{T}\mathrm{M}\mathrm{g}\mathrm{r}\mathrm{a}\mathrm{d}\mathrm{e}\left[\%\right]\right)\text{–}2.383\times \left(\mathrm{s}\mathrm{e}\mathrm{x}[0=\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e},1=\mathrm{f}\mathrm{e}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e}]\right)\\ {\displaystyle }\text{–}0.135\times \left(\mathrm{a}\mathrm{g}\mathrm{e}\left[\mathrm{y}\mathrm{e}\mathrm{a}\mathrm{r}\mathrm{s}\right]\right)\end{array}}$$

Upon application of the nonstandardized regression equation to the cross-validation cohort (n = 63), there was no difference between mean measured and mean predicted VO_2peak (23.1 ± 5.8 versus 23.7 ± 4.4 mL/kg/min, respectively, P = 0.249). There was a significant correlation in the cross-validation group between measured and predicted VO_2peak (P < 0.001, R = 0.710). Figure 1 shows the positive and significant association between measured and predicted VO_2peak in the cross-validation group based on the PD-specific equation.

FIGURE 1.

Association between measured and predicted VO_2peak in the cross-validation group.

The Bland-Altman plot comparing the measured and predicted VO_2peak in the cross-validation group is shown in Figure 2. There was minimal bias (−0.599 mL/kg/min), with the upper and lower limits of agreement being 7.58 and −8.78 mL/kg/min, respectively.

FIGURE 2.

Bland-Altman plot representing average deviations of measured VO_2peak from equation-predicted VO_2peak data in the cross-validation group. Dashed lines represent upper and lower limits of agreement at the 95% confidence interval.

Comparison of the PD-specific VO_2peak prediction equation with previously published VO_2peak prediction equations in healthy adult cohorts is shown in Figure 3. The mean measured VO_2peak in the cross-validation cohort was 23.1 ± 5.8 mL/kg/min, which is represented by the dotted line in the figure. Mean predicted VO_2peak in the cross-validation cohort according to the ACSM equation was 26.4 ± 6.9, which was significantly higher than measured VO_2peak (P < 0.001). Conversely, the mean predicted VO_2peak in the cross-validation cohort according to the Foster equation was 21.6 ± 4.8, which was significantly lower than the measured VO_2peak (P = 0.007).

FIGURE 3.

Mean predicted VO_2peak in the cross-validation cohort according to the PD, ACSM, and Foster equations. The dotted line represents the mean measured VO_2peak in the cross-validation cohort (23.1 mL/kg/min). *Significantly different than the mean measured VO_2peak in the cross-validation cohort.

DISCUSSION

The primary findings of this study demonstrate that the PD-specific VO_2peak equation-predicted VO_2peak in people with de novo PD with good accuracy. This equation used TM speed, TM grade, sex, and age to predict VO_2peak. Using these four variables, the prediction equation was able to account for 67.2% of the variance in VO_2peak and had an SEE of 3.62 mL/kg/min. The secondary findings of this study support that our PD-specific VO_2peak prediction equation outperformed previously published prediction equations, which were developed using middle-aged healthy adults (7,14) in being able to predict VO_2peak in those with early PD. These models used a combination of TM speed and grade and resulted in a mean overestimation of measured VO_2peak by 3.24 mL/kg/min for the ACSM equation and a mean underestimation of measured VO_2peak by 1.49 mL/kg/min for the Foster equation when compared with directly measured VO_2peak. In contrast, our PD-specific equation overestimated measured VO_2peak by an average of 0.60 mL/kg/min in the cross-validation sample. This suggests that the present PD-specific prediction equation may be more appropriate for this clinical population, as data from our cross-validation sample suggest a consistently lower mean estimate (in mL/kg/min) when compared with these alternative equations. This results in increased confidence in results for the clinicians relying on this prediction equation when full CPET may be unavailable. Importantly, the current prediction equation’s SEE of 3.62 mL/kg/min represents a value very similar to the 3.5 mL/kg/min (1 MET) threshold that has been associated with significant improvements in survival from estimates of VO_2peak based on treadmill speed and grade (24). Additionally, it is important to note that this equation’s SEE is lower than a collection of nonexercising VO_2peak prediction equations developed using data from healthy adults, which ranged from 4.1 to 6.2 mL/kg/min (25). While this equation represents an improvement in VO_2peak prediction for those with PD over generic equations developed in healthy populations, additional refinement is required before such a prediction equation in PD can be considered for evaluation of mortality risk. Rather, this equation serves as a valuable resource for evaluating baseline CRF and changes in VO_2peak following exercise training based on treadmill speed and grade achieved at maximal exercise.

Our PD-specific prediction equation aligned closely with expectations. Factors such as TM speed and TM grade have positive coefficients, confirming that a higher TM speed or grade results in a higher VO_2peak and building upon previously published equations (7,14). Our equation further adds the additional independent variables of sex and age, which are both known to impact VO_2peak (26). In our equation, female sex is associated with lower VO_2peak, confirming known sex-based differences likely resulting from differences in lean tissue (27), cardiac output (28), and hemoglobin concentration (29). Finally, age has a negative coefficient, meaning that as age increases, VO_2peak tends to decrease. This finding aligns with current research on age and VO_2peak (30,31) and is associated with well-documented reductions in lean tissue (32), physical activity participation (33), and cardiac output (34) with age. The standardized coefficients in this PD-specific equation demonstrate that sex has the strongest influence on predicted VO_2peak, followed by TM speed, TM grade, and age.

ACSM’s VO_2peak prediction equation has been shown to over-predict measured VO_2peak in healthy individuals (35), and our study is the first to show this is also true in a sample of recently diagnosed people with PD. The Foster equation by contrast systematically underestimated VO_2peak in our sample of people with recently diagnosed PD. There are PD-specific characteristics that may help to explain why the Foster equation, which is also based on treadmill speed and grade, is less accurate for people with PD than is desirable. Specifically, studies in individuals with PD have reported increased metabolic costs with gait in people with PD, especially at higher walking speeds (36–38), which are likely utilized during CPET to elicit a maximal effort. Biomechanical factors in the gait of people with PD may also factor into increases in metabolic requirements for a given TM speed and grade, including a short step shuffling pattern and asymmetrical arm swing (16,17), or increased knee and ankle coactivation (18). Individuals with PD may also experience fatigability in the performance of exercise, likely resulting in part from impairments related to the disease (39,40). While these PD-related factors may help to explain such differences in the performance of VO_2peak prediction equations in this population, there is a general lack of research regarding systematic differences in CRF and submaximal energy expenditure in people with PD compared with healthy controls. Careful analysis of a recent systematic review investigating potential differences in CRF between people with PD and healthy individuals (41) reveals that data do suggest a reduction in VO_2peak (2–4). The few studies that may suggest no difference in this important health metric either still demonstrate a trend in this direction (15,42) or present data from healthy controls likely not reflective of a generally healthy adult population (43). Additional research is required to more thoroughly identify potential differences and better inform future VO_2peak prediction equations in this population.

Additional VO_2peak prediction equations have been developed using participant behavior and self-reported participation in physical activity. One such VO_2peak prediction equation in people with PD used factors like age, gender, waist size, body mass index, self-reported International Physical Activity Questionnaire, and Short Physical Performance Battery (44). Participants in that study were further along in their PD diagnoses as the research included participants within H&Y stages 1–3 and were in the on-phase of their dopaminergic medication. Studies with larger sample sizes are required to validate this equation and better understand the impact that these additional clinical and self-reported metrics have on VO_2peak. A recent study by Bonde-Jensen and colleagues suggests that CRF may be associated with motor symptoms and other important clinical measures (45), presenting an important future avenue of inquiry to better refine such VO_2peak prediction equations. Importantly, such clinical measures were not associated with VO_2peak in our study. As PD progresses, one may observe changes in gait, such as shorter step length, slower walking speeds, and development of freezing of gait (46), which may further alter the metabolic demand associated with any given TM-based work rate. Because of this, VO_2peak prediction equations that are not only PD-specific but are also disease stage-specific may be an important area of future investigation.

Strengths of our analysis include the objectivity of the variables included in our equation, as well as the homogeneity of the sample group. This homogeneity limited potential confounding variables such as medication usage or underlying cardiovascular comorbidities that may have limited one’s ability to exercise and increased the applicability of this equation to the population of individuals with PD reflective of our eligibility criteria. There are no variables within our prediction equation that require a person with PD to complete a questionnaire or subjectively report physical activity habits. In addition, the gas exchange information available during CPET ensures that the participants are truly reaching maximal exertion instead of by self-report, which may over- or underestimate effort during exercise testing. All 127 participants were classified as H&Y stages 1–2, with 57.5% male participants, an average age of 63.6 ± 9.3 yr, and no participants were taking any dopaminergic medications. These demographics are consistent with individuals with early PD and increase confidence in the application of this prediction equation in this participant population (47).

Our study has several limitations. First, while study participants were encouraged not to use the handrail for support during CPET, not all participants followed this guidance, resulting in some tests likely lasting longer (and therefore resulting in a higher grade) than the participant would achieve solely based on aerobic capacity. Additionally, as this is a retrospective analysis of previously collected data, we did not have the ability to analyze submaximal, stage-by-stage CPET data. Future investigations may develop improved prediction equations by only utilizing data from completed CPET stages. Further, the sample size of the current study limited the number of independent variables that could be entered into the multiple linear regression analysis. While our PD-specific equation still outperformed previously published metabolic prediction equations in this population, future studies with larger sample sizes should be able to further improve the predictive capability of similar equations by including additional variables such as body fat percentage (48), or other variables that have been associated with VO_2peak, as well as PD-specific clinical characteristics including time since diagnosis and fatigue scales. Furthermore, future studies in a more heterogeneous sample including PD participants on dopaminergic therapies or those with higher H&Y stages will continue to provide data to empower clinicians working with a more diverse participant population to utilize VO_2peak in their baseline and follow-up assessments, especially as individuals in more advanced stages of PD are likely still able to provide a maximal effort during exercise testing (49). Additionally, future studies may benefit from investigating submaximal exercise data to further understand predictors of metabolic demand.

Clinicians can use this prediction equation to identify individuals who would benefit from endurance exercise training to increase CRF, and track changes in VO_2peak over time. For instance, predicting VO_2peak at baseline may reveal that an individual’s fitness level is well below established reference values compared with age- and sex-matched peers (30), providing an opportunity for development of a targeted endurance exercise prescription to improve this metric over time. Alternatively, a predicted VO_2peak value in alignment with higher age- and sex-matched percentiles may lead to an exercise program targeting maintenance rather than improvement of fitness. VO_2peak is an important marker of health status, and decreases in VO_2peak are linked with loss of independence (50,51) and poor prognostic outcomes (52,53) over time. Further, increased CRF is associated with improvements in neural network connectivity (54) and increased volume of the thalamic and pallidum volume in the brain in individuals with PD (55), providing additional support for the importance of endurance exercise training and evaluating VO_2peak in this population. Since the data used to create the current prediction equation were derived from the SPARX study, parallels can be drawn between the mean change in VO_2peak of 1.9 ± 2.9 mL/kg/min with high-intensity exercise and predicted improvements for patients in a clinical setting who follow endurance exercise guidelines (56). Such improvements have been demonstrated with a variety of exercise prescriptions and modalities (57). These exercise recommendations are especially important when considering the expected reduction in physical activity over time in this population (58).

Another example of the use of fitness-related equations related to PD can be found in a recent paper by Rosenfeldt and colleagues (59). Advancements in the understanding of age-related maximal heart rate decline in PD, including refinement of PD-specific maximal heart rate prediction equations, will continue to facilitate more effective and individualized exercise prescriptions, maximizing benefit and maintaining function across the healthspan.

CONCLUSIONS

The PD-specific VO_2peak prediction equation successfully predicted VO_2peak in participants with early PD. In contrast, VO_2peak was over- and underestimated in people with PD by the ACSM and Foster equations, respectively. This underscores the importance of utilizing a PD-specific regression equation when predicting VO_2peak in individuals with PD. Using the prediction equation as a part of patient education can provide clinicians with an effective, quantifiable approach to modifying exercise behaviors to promote improved VO_2peak and subsequently reduce the risk of mortality.

The research reported in this publication was supported by the National Institute of Neurological Disorders and Stroke and the National Institutes of Health under Award Numbers U01NS113851. Research is also supported by the National Institutes of Health’s National Center for Advancing Translational Sciences, Grant Number UL1TR001422. It is also supported by a generous philanthropic gift in honor of Howard Gilbert and a gift from the JCS Family Foundation. No conflicts of interest were disclosed. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine.