key: cord-0938851-zxh7wr9k
authors: Frazão, Murillo; Eugênio Silva, Paulo; de Assis Pereira Cacau, Lucas; Rocha Petrucci, Tullio; Cometki Assis, Mariela; da Cruz Santos, Amilton; do Socorro Brasileiro-Santos, Maria
title: EMG breakpoints for detecting anaerobic threshold and respiratory compensation point in recovered COVID-19 patients
date: 2021-06-08
journal: J Electromyogr Kinesiol
DOI: 10.1016/j.jelekin.2021.102567
sha: 5d26db7d74fac2242b22e2cc498c71cae4ed3cea
doc_id: 938851
cord_uid: zxh7wr9k

INTRODUCTION: A huge number of COVID-19 patients should be referred to rehabilitation programmes. Individualizing the exercise intensity by metabolic response provide good physiological results. The aim of this study was to investigate the validity of EMG as a non-invasive determinant of the anaerobic threshold and respiratory compensation point, for more precise exercise intensity prescription. METHODS: An observational cross-sectional study with 66 recovered COVID-19 patients was carried out. The patients underwent a cardiopulmonary exercise test with simultaneous assessment of muscle electromyography in vastus lateralis. EMG breakpoints were analyzed during the ramp-up protocol. The first and second EMG breakpoints were used for anaerobic threshold and respiratory compensation point determination RESULTS: EMG and gas exchange analysis presented strong correlation in anaerobic threshold (r = 0.97, p < 0.0001) and respiratory compensation point detection (r = 0.99, p < 0.0001) detection. Bland-Altman analysis demonstrated a bias = -4.7 watts (SD = 6.2 watts, limits of agreement = -16.9 to 7.6) for anaerobic threshold detection in EMG compared to gas exchange analysis. In respiratory compensation point detection, Bland-Altman analysis demonstrated a bias = -2.1 watts (SD = 4.5 watts, limits of agreement = -10.9 to 6.6) in EMG compared to gas exchange analysis. EMG demonstrated a small effect size compared to gas exchange analysis in oxygen uptake and power output at anaerobic threshold and respiratory compensation point detection. CONCLUSIONS: EMG analysis detects anaerobic threshold and respiratory compensation point without clinical significant difference than gas exchange analysis (gold standard method) in recovered COVID-19 patients.

Patients who have successfully recovered from the acute COVID-19 pneumonia will require health support to define and quantify the consequences of the disease. The follow-up is currently the new challenge as it was in the beginning for intensive care units. Indeed, it is not clear if COVID-19 will leave permanent lung and/or physical damage, and if so, to what extent. Persisting limitations in respiratory function and gas exchange will likely be more pronounced in the subgroup of severe patients (Xiaoneng Mo, Wenhua Jian, Zhuquan Su, Mu Chen, Hui Peng, Ping Peng, Chunliang Lei, Ruchong Chen, Nanshan Zhong, 2020) . In addition, as in non-COVID-19 related acute respiratory distress syndrome, we can anticipate a high incidence of intensive care unit acquired weakness that is associated with poor short-as well as long-term outcomes (Polastri et al., 2020) .

Considering the expected high burden of respiratory, physical and psychological impairment following the acute phase of COVID-19, a huge number of patients should be early referred to a rehabilitation program (Polastri et al., 2020) . Multinational task force recommends early rehabilitation for patients affected by severe . The pulmonary rehabilitation model may suit as a framework, particularly in a subset of patients with long term respiratory consequences (Spruit et al., 2020) .

Individualizing the exercise intensity by exercise testing evaluation provides good physiological results. The best choice of exercise intensity might also play a fundamental role in the rehabilitation program adherence (Vandoni et al., 2016) . Moreover, the use of relative terms such as maximum oxygen uptake percentage or heart rate has been substantially criticized (Meyer et al., 1999) .

Due to most physiological responses to exercise being intensity dependent, reliance on these parameters alone without considering the anaerobic threshold and respiratory compensation point is not sufficient. The threshold-based exercise intensity method (zone between anaerobic threshold and respiratory compensation point) provides great improvements in cardiorespiratory fitness (Wolpern et al., 2015) . Due to physical limitations in recovered severe COVID-19 patients, a threshold-based intensity program could be a reliable and efficient method for exercise prescription in this population (Huang et al., 2016) .

Cardiopulmonary exercise test (CPET) is the gold standard method for anaerobic threshold and respiratory compensation point detection. Gas exchange analysis has been used for decades as a precise tool for metabolic behavior determination. However, CPET is a high cost method, not available in a large number of rehabilitation centers. Surface electromyography (EMG) has been used for anaerobic threshold and respiratory compensation point detection in the last decades in healthy and chronic disease populations (Hug et al., 2003 ) (Lucía et al., 1997) (Tikkanen et al., 2012) . EMG is a low cost, portable and noninvasive method which can be incorporated in rehabilitation programs.

In view of the above, the aim of this study was to investigate the validity of EMG as a non-invasive determinant of the metabolic response to incremental exercise. We studied the relationship between EMG activity and the gold standard method of anaerobic threshold and respiratory compensation point detection.

This is an observational cross-sectional study. Patients underwent a cardiopulmonary exercise test with simultaneous assessment of muscle electromyography in a single-day evaluation. This study was approved by the local research ethics committee and was registered in the Brazilian Clinical Trial Registration Platform (Number: RBR-6xqcr4).

A group of COVID-19 patients who were referred for functional evaluation by CPET at the Exercise Physiology Laboratory of the Federal University of Paraíba from July 4 th to August 14 th were considered eligible for this study.

COVID-19 diagnosis was established by clinical symptoms (fever, fatigue, muscle soreness, cough, dyspnea, etc.) associated with a positive laboratory test (nasal swab or serology) and/or chest tomography (ground-glass opacity).

Patients were classified as mild (major clinical symptoms without dyspnea or respiratory failure) or severe (major clinical symptoms with dyspnea or respiratory failure), as postulated by Tian (Tian et al., 2020) . Patients who met the following inclusion criteria were enrolled: recovered (less than 30 days) from mild to severe COVID-19. Exclusion criteria were based on comorbidity confounding factors.

Thus, patients with critical COVID-19 (i.e. who had required intubation and mechanical ventilation) and those with previous cardiac, pulmonary, neurological, hematological or muscular diseases were excluded.

The technical procedures for CPET followed the American Thoracic Society/American College of Chest Physicians guidelines for cycle ergometer testing (ATS/ACCP, 2003). The CPET was performed on a CG-04 cycle ergometer (INBRAMED, Porto Alegre, Brazil) . Each subject performed a rampup protocol, starting with a warm-up by unloaded pedaling for 2 minutes, followed by an individually-selected workload increment to achieve maximum effort within 8 to 12 min. Subjects were instructed to keep a cadence of 60 rotations per minute and were strongly encouraged by verbal stimuli to achieve maximum effort. The VO2000 (MedGraphics, St. Paul, Minnesota, USA) was used for gas exchange analysis, and it was calibrated according to the manufacturer's instructions. Data were filtered (mean of 7 points) to avoid noise and analyzed by 10s-averages.

Resting spirometry was conducted before the CPET, in which forced expiratory volume in one second (FEV 1 ) was measured (ASMA-1, Vitalograph, United Kingdom) to calculate maximum voluntary ventilation (MVV = FEV 1 x 35).

Next, the following variables were considered for analyses: power output, peak oxygen uptake (VO 2 ), percentage of predicted VO 2 (Hansen et al., 1984) , respiratory exchange ratio at maximal effort (RER), oxygen pulse at maximal effort (O 2 Pulse), peak ventilation (VE) and breathing reserve used during maximal effort (BR = VE/MVV). Ventilatory equivalents for oxygen (VE/VO 2 ) and carbonic gas exchange (VE/VCO 2 ) were used for anaerobic threshold and respiratory compensation point analysis (Gas Exchange analysis). The anaerobic threshold was determined by the first inflection in the VE/VO 2 curve (Wasserman et al., 1994) . The respiratory compensation point was determined by the second inflection in the VE/VO 2 curve ( Figure 1 ).

Neuromuscular activity during CPET was analyzed by EMG ( Figure 1) using a signal acquisition module with a 12-bit resolution A/D converter (EMG800C, EMG System, São José dos Campos, Brazil). The sampling frequency was adjusted to 1000 Hz, frequency band to 20-500 Hz and gain to 1000 times. Bipolar Ag/AgCl self-adhesive surface electrodes were used and placed 20 mm apart (center to center) on the right vastus lateralis (2/3 of the way from the anterior superior iliac spine to the lateral side of the patella), according to Surface Electromyography for the Non-Invasive Assessment of Muscle recommendations (Hermens et al., 1999) . A reference electrode was placed on the ulna. The subject's skin was shaved, abraded and cleaned with alcohol prior to electrode placement.

Root mean square (RMS) values were used for analysis. EMG breakpoints were analyzed during the ramp-up protocol. A visual method based on previous reports by Lucía et al. was employed (Lucía et al., 1999) . The increased EMG amplitude reflects the recruitment of additional motor units (Hug, 2009 ). Based on this, the first EMG breakpoint was assumed to be type IIa fiber activation, and the second EMG breakpoint was assumed to be type IIb fiber activation (Henneman's principle) (Henneman and Somjen, 1965) . The first and second EMG breakpoints were used for anaerobic threshold and respiratory compensation point determination (EMG analysis) (figure 1).

The raw signal and exercise test time were exported and merged into a non-commercial software called Shengo ® (INBRAFIC, Brasília, Brazil). The root mean square voltage was then computed at every five seconds throughout whole test. The signal analysis area was automatically determined by the Shengo ® algorithm using a trigger signal acquired from a device adjusted in the cycle ergometer. The window analysis was 500 milliseconds positioned from the signal peak median. Next, the signal was smoothed by seven-point means.

We used the Shengo ® algorithm which models RMS response to exercise using multisegment linear regression to establish objective criteria to determine breakpoints in the EMG power output response. With this method, a single linear regression was initially fitted to all data points. A brute force method was then used to fit two lines to the data points. The program calculated regression lines for all possible divisions of the data into two contiguous groups, and the pair of lines yielding the least pooled residual sum of squares was chosen as representing the best fit. Thereafter, the program attempts to fit a third line to the data in order to detect another breakpoint in the EMG data. The third middle segment was obtained by methodically adding points on the left side of the two- 

Data normality was verified using the Shapiro-Wilk test. The effect size was calculated by the T-test (difference between two dependent means) and post hoc analysis. The input parameters were total sample size, means and standard deviations, and error probability α = 0.05. The effect size points were small = 0.2, medium = 0.5 and large = 0.8 (Cohen, 1992) . The effect size was used to determine clinically significant differences (medium or large effect size was assumed as clinically significant). The relationship between values was tested using the Pearson's or Spearman's correlation tests according to Gaussian distribution. Agreement between variables was analyzed by the Bland-Altman test. A statistical significance value of p ≤ 0.05 was set for all analyses. GraphPad Prism 7.0 and GPower 3.0.10 software programs were used. According to data normality distribution, the data are presented as means ± standard deviations or as medians and interquartile ranges and percentages.

A total of 66 patients were enrolled, however 18 were excluded (comorbidities: asthma = 9, heart failure = 3, critical COVID-19 = 3, COPD = 2 and fibromyalgia = 1). Anthropometric characteristics, main COVID-19 symptoms and main drug therapy are presented in Table 1 . Exercise characteristics are presented in Table 2 .

The EMG analysis demonstrated a small effect size compared to the gas exchange analysis in oxygen uptake at the anaerobic threshold ( The Bland-Altman analysis demonstrated a bias = -4.7 watts (SD = 6.2 watts, limits of agreement = -16.9 to 7.6) for anaerobic threshold detection in the EMG analysis compared to the gas exchange analysis. The stratified data showed a bias = -4.2 watts (SD = 6.5 watts, limits of agreement = -16.9 to 8.5) for mild and a bias = -5.5 watts (SD = 5.9 watts, limits of agreement = -17.0 to 6.1) for severe patients (Figure 4) . The Bland-Altman analysis demonstrated a bias = -2.1 watts (SD = 4.5 watts, limits of agreement = -10.9 to 6.6) in the EMG analysis compared to the gas exchange analysis in the respiratory compensation point detection. The stratified data showed a bias = -1.9 watts (SD = 4.8 watts, limits of agreement = -11.4 to 7.6) for mild and a bias = -2.4 watts (SD = 3.8 watts, limits of agreement = -9.8 to 5.0) for severe patients (Figure 4 ).

The main finding of this study was that: 1) EMG analysis detects the anaerobic threshold and respiratory compensation point without clinical significant difference than gas exchange analysis; 2) EMG analysis presented good correlation and agreement in anaerobic threshold and respiratory compensation point detection compared to gas exchange analysis.

Houtz and Fischer (Houtz and Fischer, 1959) were the first to record surface electromyograms during pedaling in 1959, and since then numerous investigators have reported EMG analyses of pedaling for different purposes (Hug et al., 2004) (Duc et al., 2008) (Sarre et al., 2003) . Some studies used visual methods to identify breakpoints in myoelectric signal response during incremental exercise ramp protocol. Hug (Hug et al., 2006) postulated that EMG activity of the vastus lateralis muscle presents a non-linear increase during incremental rampup cycling. The results of the Hug study suggest that determining EMG breakpoints can be used as a reliable method for studying neuromuscular fatigue during cycling exercise in this specific muscle, and better than in other leg muscles. Bearden and Moffatt (Bearden and Moffatt, 2001 ) demonstrated that the two breakpoints of the RMS/power output ratio coincided with the first and second ventilatory equivalent inflection points, as well as in the present study.

Glass (Glass et al., 1998) analyzed rectus femoris and vastus lateralis electromyography during CPET, finding that VO 2 at EMG breakpoints was not significantly different from the gas exchange method (similar to our data).

Our results are in overall agreement with those of previous studies. Lucía (Lucía et al., 1997) evaluated vastus lateralis electromyography during CPET in cardiac transplant patients. They showed that EMG analysis presented strong correlation (r = 0.89, p < 0.05) with gas exchange analysis in detecting the anaerobic threshold. There was no evidence of a significant difference between methods when expressed as oxygen uptake (again, similar to our data). The data in Lucía's study also demonstrated that the first EMG breaking point occurred around 60% of maximum oxygen uptake, similar to our results. Unfortunately, they did not measure the second EMG breaking point. Zamunér ) evaluated vastus lateralis electromyography during CPET in sedentary middle-aged men. As in the present study, their data did not show significant differences in power output at the anaerobic threshold in the EMG and gas exchange methods.

Quantifying the breakpoints can be achieved by dynamically analyzing these variables, as their disproportion increases are relative to the cardiorespiratory adjustments necessary to supply the growing metabolic demand from increased motor unit recruitment. The EMG breakpoints may occur as a result of a change in the motor unit recruitment pattern from predominantly slow-twitch motor units to fast-twitch motor units, which could contribute to the accumulation of circulating lactate during exercise (Lucía et al., 1999) (Viitasalo et al., 1985) . Lactic acid is produced during fast-twitch motor unit activation (glycolytic pathway utilization). The H + produced in cells as lactate must be immediately buffered upon its formation. Since HCO 3is a volatile buffer, the resulting H 2 CO 3 does not remain in the cell, but leaves upon its formation as CO 2 , thereby removing H + from the intracellular environment (Wasserman and Whipp, 1975) . The ventilation normally increases at a rate required to remove CO 2 added to the capillary blood by metabolism while minimizing the increase in arterial H + concentration.

Vastus lateralis activity analysis is representative in prescribing thresholds for whole body activities. Jürimäe et al. (2007) demonstrated that the EMG breakpoint happens at similar workload in different lower extremity muscles (vastus lateralis, vastus medialis, biceps femoris and gastrocnemiuslateralis).

An EMG breakpoint analysis in lower extremity muscles can be done independently of the effort method or ergometer used. Tikkanen et al. (2012) demonstrated that the EMG analysis had similar a breakpoint pattern during treadmill running analysis as in bicycling. The timing of the EMG breakpoints (thresholds) was similar to ventilatory threshold and onset of blood lactate accumulation.

The power output bias in the Bland-Altman analysis of anaerobic threshold and respiratory compensation point detection in the present study was similar to the findings by Lucía (Lucía et al., 1999) . This was probably due to "metabolic delay". EMG analyses fiber membrane depolarization, so it can detect muscle fiber activation in real time. Gas exchange measurements analyses the metabolic response to muscle fiber activation. VO 2 responds with linear first-order dynamics for power outputs with a time constant approximately equal to 25 to 35 seconds and a "delay" of 15 to 20 seconds (Whipp, 1987 

The EMG analysis detected the anaerobic threshold and respiratory compensation point without a clinically significant difference compared to the gas exchange analysis (gold standard method) in recovered COVID-19 patients. The 

ATS/ACCP Statement on Cardiopulmonary Exercise Testing

Identification and agreement of first turn point by mathematical analysis applied to heart rate , carbon dioxide output and electromyography

Leg electromyography and the VO2-power relationship during bicycle ergometry

A power prime

EMG Threshold Determination in Eight Lower Limb Muscles During Cycling Exercise : A Pilot Study

Occurrence of electromyographic and ventilatory thresholds in professional road cyclists

Heterogeneity of muscle recruitment pattern during pedaling in professional road cyclists : a magnetic resonance imaging and electromyography study

Electromyographic analysis of pedaling : A review

Influence of cycling cadence on neuromuscular activity of the knee extensors in humans

Identifying the integrated electromyographic threshold using different muscles during incremental cycling exercise

Functional significance of cell size in spinal motoneurons

European Recommendations for Surface ElectroMyoGraphy

Dose -response relationship of cardiorespiratory fitness adaptation to controlled endurance training in sedentary older adults

Predicted values for clinical exercise testing

Electromyographic activity related t o aerobic and anaerobic threshold in ergometer bicycling

Aerobic -anaerobic transition intensity measured via EMG signals in athletes with di V erent physical activity patterns

Determination of the anaerobic threshold by gas exchange biochemical considerations, methodology and physiological effects

Electromyographic Response to Exercise in Cardiac Transplant Patients * A New Method for Anaerobic Threshold

Analysis of the aerobic-anaerobic transition in elite cyclists during incremental exercise with the use of electromyography

Early View Original article COVID-19 : Interim Guidance on Rehabilitation in the Hospital and Post-Hospital Phase from a

Is determination of exercise intensities as percentages of VO2max or HRmax adequate

COVID-19 and pulmonary rehabilitation : preparing for phase three

Muscular activity during uphill cycling : Effect of slope , posture , hand grip position and constrained bicycle lateral sways

An analysis of muscle action and joint excursion during exercise on a stationary bicycle

Desenvolvimento e validação de um sistema para a determinação do limiar ventilatório e do ponto de compensação respiratória, por meio da eletromiografia de superfície

Characteristics of COVID-19 infection in Beijing

Ventilatory threshold during incremental running can be estimated using EMG shorts

Psychophysiological Responses to Group Exercise Training Sessions : Does Exercise Intensity Matter?

Excercise physiology in health and disease

Dynamics of pulmonary gas Exchange

Is a thresholdbased model a superior method to the relative percent concept for establishing individual exercise intensity ? a randomized controlled trial

Abnormal pulmonary function in COVID-19

Identification and agreement of first turn point by mathematical analysis applied to heart rate , carbon dioxide output and electromyography