Share this article on:

A Submaximal Test for the Assessment of Knee Extensor Endurance Capacity


Medicine & Science in Sports & Exercise: February 2014 - Volume 46 - Issue 2 - p 398–406
doi: 10.1249/MSS.0b013e3182a59c9c
Applied Sciences

Purpose We aimed to develop an undemanding test for endurance capacity of the knee extensor muscles, which can also be applied to frail participants. We hypothesized 1) that the first objective indications for peripheral fatigue during incremental unilateral repetitive isometric knee extensor contractions could be used to assess a fatigue threshold (FT), 2) that torque at FT would depend on training status, and 3) that this torque could easily be sustained for 30 min.

Methods Five trained and five untrained participants performed 5-min bouts of 60 repetitive contractions (3-s on and 2-s off). Torque, set at 25% maximal voluntary contraction (MVC), was increased by 5% MVC in subsequent bouts. The highest torque for which rectified surface EMG remained stable during the bout was defined as the FT. On separate occasions, 30-min bouts were performed at and above the FT to assess sustainable torque. Changes in gas exchange parameters, HR, and RPE were monitored to corroborate FT.

Results At FT (RPE = 5.7 ± 1.7), torque was higher (P < 0.05) in trained (41.4% ± 5.8% MVC) than in untrained participants (30.5% ± 1.8% MVC). Sustainable torque was ∼4% higher than (P < 0.05) and significantly related to FT (r 2 = 0.79). When torque was increased by 5% MVC, significant increases in rectified surface EMG and V˙O2 were found.

Conclusions During incremental knee extensor contractions, FT could be assessed at a submaximal exercise intensity. FT was higher in trained than in untrained participants and was related to exercise sustainability. With the use of FT, changes in endurance capacity of single muscle groups can potentially also be determined in frail participants for whom exercise performed until exhaustion is unwarranted.

MOVE Research Institute Amsterdam, VU University Amsterdam, THE NETHERLANDS

Address for correspondence: Cornelis J. de Ruiter, Ph.D., MOVE Research Institute Amsterdam, FBW, VU University Amsterdam, Van der Boechorststraat 9, 1081 BT Amsterdam, The Netherlands; E-mail:

Submitted for publication May 2013.

Accepted for publication July 2013.

In many sports, the maximal exercise intensity at which power output can be sustained for a long period is an important performance-determining factor (22). Concepts like the maximal lactate steady state (12) and critical power (16) relate to such an upper limit of performance during high-intensity endurance exercise. Up until a critical limit, power output can be sustained for at least 30 min, but even a slight increase in exercise intensity will lead to early fatigue. The exact assessment of such a critical exercise intensity involves a series of 30-min exercise bouts at different power outputs (maximal lactate steady state) (12) or a series of exercise bouts sustained until failure covering a range of power outputs (critical power) (16). Recently, it has been shown that critical torque (CT), that is, the maximal sustainable torque production during repetitive muscle contractions, may be a useful parameter to evaluate the effects of training and rehabilitation interventions (4,6). The advantage of such a test is that endurance capacity of a specific muscle group can be established and directly related to the contractile properties of the muscle group being tested. This may be very useful in training studies and rehabilitation settings.

In practice, a series of tests to determine sustainable force or power often is very time consuming, and therefore, alternative tests have been developed. These alternative tests typically involve 3–5 min of all-out exercise (4,23). The critical power (or torque) determined with such all-out tests was shown to be similar to that obtained with “traditional” testing over multiple days (4,23). However, although one single all-out test does not take much time, it is highly demanding both mentally and physically. In rehabilitation, when working with frail people or following recovery from an injury, highly demanding tests may not be feasible or warranted. Recently, we showed that a critical exercise intensity (fatigue threshold [FT]) could also be determined during a single series of submaximal isometric unilateral knee extensor contractions (9). This FT had strong relations to important parameters of aerobic training status obtained during cycle ergometry (first ventilatory threshold and V˙O2max) and was also significantly related to CT determined during a 5-min all-out test (9). However, in the submaximal exercise test, the 5-min bouts of different intensities (30%–60% maximal voluntary contraction [MVC]) were applied in random order. Consequently, participants during some of the bouts exercised (well) above FT, which lead to considerable effort and to torque failure within 5 min in half of the participants. As stated above, strenuous exercise has to be avoided in many (rehabilitation) settings. Therefore, the first goal of the present study was to investigate whether an FT at a submaximal exercise intensity can objectively be determined during an incremental test of repetitive isometric knee extensions. This would allow future tests to evaluate aerobic muscle capacity and link this to (changes in) other contractile properties of individual muscle groups, without the necessity for participants to perform any unduly strenuous exercise. The second objective of the present study was to relate the FT established during the incremental test directly to peripheral fatigue and exercise sustainability for long-duration (30 min) exercise at and above FT.

We developed a progressive incremental knee extension test based on recent work of others and ourselves (4–6,9). The test consists of 5-min bouts of 3-s isometric contractions with a 2-s rest in between. On the basis of our recent study (9), we hypothesize that there will be a CT, well below the level during which torque failure occurs but above which signs of peripheral fatigue will develop. This torque will signify the threshold for peripheral fatigue (FT). We have previously demonstrated that peripheral fatigue was accompanied by increases in amplitude of the rectified surface EMG (rsEMG) and muscle deoxygenation (9). These changes are probably caused by recruitment and rate coding of additional motor units to compensate for the loss of force-producing capacity of fatigued muscle fibers (8,14) or to compensate for a diminished economy of fatigued fibers (7,13). Therefore, our second hypothesis is that repetitive torque production at FT occurs under steady-state conditions and, consequently, can be sustained for 30 min with relative ease: with moderate RPE and without clear indications of peripheral fatigue. In contrast, at intensities higher than peripheral FT, increases in EMG and oxygen uptake followed by near-maximal effort and/or torque failure are expected during 30 min of exercise.

Critical exercise intensities during whole-body activities like running, cycling, skating, rowing, etc., are often determined with or related to changes in gas exchange parameters (15,18,20). In our recent study on isolated unilateral knee extensor contractions, muscle oxygenation levels decreased above FT and dynamometer FT was related to the first ventilatory threshold and V˙O2max obtained during cycling (9). The amount of muscle mass involved in unilateral isolated knee extensor contractions only is about 2–3 kg and peak knee extensor O2 uptake during dynamic contractions was reported to be about 25% of peak pulmonary O2 uptake (17). Moreover, in the present study, knee extensor O2 consumption is probably further limited by the use of static instead of dynamic muscle contractions. Nevertheless, and this is our third hypothesis, during exercise above FT, increases in V˙E and V˙O2 are expected to accompany the expected rise in EMG. All these changes signify a non–steady-state condition during exercise above FT.

Back to Top | Article Outline



To increase heterogeneity in responses, both trained and untrained healthy participants were recruited. We did not a priori aim to split the group into two subgroups, but five of the participants (one woman; 25.4 ± 1.7 yr, 1.83 ± 0.08 m, 74.2 ± 11.1 kg) were involved in regular (4–8 h·wk−1) endurance training for cycling and/or running, while the other five were not involved in any exercise training (three women; 25.4 ± 1.2 yr, 1.75 ± 0.1 m, 73.2 ± 8.6 kg). The study was approved by the local ethics committee, and following written and verbal explanation, all participants signed informed consent forms. They visited our laboratory on three different occasions with at least a 72-h break in between and they refrained from heavy exercise 24 h before testing.

Back to Top | Article Outline

First occasion

On the first occasion, MVC and voluntary activation (VA) were determined, which was followed by an incremental unilateral knee extensor exercise test to torque failure to establish FT (see Data Analysis).

Back to Top | Article Outline


The preferred (kicking) leg was tested in a dynamometer, with knee and hip angles set at 90°. The participants were secured with straps over hips and shoulders, and a shin guard protected the lower leg, which was connected to a force transducer (KAP-E K.L.0.2; A.S.T. GmbH, Dresden, Germany). The measured forces were multiplied by the external moment arm to obtain knee extension torques. Participants started with a 2-min warming up consisting of 3-s isometric contractions with a 2-s rest, guided by audio signals and with online torque feedback on a computer monitor. These contractions were executed at a comfortable self-selected nonfatiguing intensity (18% ± 6% MVC). The warm-up was followed by three MVCs with a 2-min rest in between. If the last attempt was more than 5% higher than the previous attempts, an additional MVC was executed. The highest value was used for normalization of target torques during the incremental test, but first the participants’ ability for maximal voluntary knee extensor activation was determined. This was done using a superimposed electrical doublet (100 Hz) stimulation during the plateau phases of two additional MVCs (with a 2-min rest in between) that were both followed by doublet stimulation within 3 s after the contraction. Stimulation was applied using a computer-controlled stimulator (model DS7A; Digitimer Ltd., Welwyn Garden City, UK) via surface electrodes (80 × 130 mm; Schwamedico, Leusden, the Netherlands) placed over the proximal and distal parts of the knee extensors. Current was increased in steps of 50 mA until doublet torque leveled off and was then increased by an additional 50 mA.

Subsequently, the stimulation electrodes were removed, and after further shaving and cleansing of the skin with 70% ethanol, EMG surface electrodes (Bleu Sensor, Ambu Ølstykke, Denmark) were placed in bipolar configuration on the vastus lateralis, vastus medialis, and rectus femoris muscle. Electrodes were placed in line with muscle fiber direction with an interelectrode distance of 25 mm. A reference electrode was placed on the patella. The location of the electrodes was marked with a waterproof felt tip for precise reapplication on following test occasions. Then, participants performed one or two additional MVCs for purposes of normalization of the EMG signals to be obtained during the subsequent incremental knee extension test.

Back to Top | Article Outline

The incremental test

The incremental test was executed following 5 min of rest. It consisted of 5-min bouts of 3-s isometric contractions with a 2-s rest in between (4,6,9). The contractions were guided by an audio signal indicating the start and end of each contraction. Real-time torque feedback was displayed on a computer monitor that also showed a horizontal line indicating target torque. The test started at a torque that was 25% of the highest MVC and torque was increased with each subsequent bout in steps of 5% MVC until failure occurred. Failure occurred when participants, despite strong verbal encouragement, were unable to maintain target torque for 3 s in three subsequent contractions. Between bouts, there was a 1-min rest, which started with the participants’ RPE of the preceding bout on a scale from 0 to 10, with 10 indicating “very, very heavy” (3). During the incremental test, oxygen uptake was analyzed breath by breath with participants wearing a mask using a metabolic unit (Cosmed Quark b2®, Rome, Italy). A bidirectional digital turbine (Ø 28 mm, flow range = 0.03–20 L·s−1, accuracy = ±2%, resistance < 0.7 cm H2O·s·L−1 at 12 L·s−1, ventilation range = 0–300 L × min) allows the unit to measure flow and volume. The unit contains a temperature-controlled oxygen sensor (type = zirconia, response time < 120 ms, range = 1%–100%, accuracy = ±0.01%) and a carbon dioxide sensor (type = NDIR, response time < 120 ms, range = 0%–15%, accuracy = ±0.03%). The gas analyzers were calibrated immediately before and were verified after each test using a certified determined gas mixture (O2 concentration = 16.04% and CO2 concentration = 5.01%), the turbine was calibrated using a calibrated 3-L syringe. HR was measured using a Suunto t6 HR monitor (Suunto Oy, Vantaa, Finland).

Back to Top | Article Outline

Second and third occasions

During both occasions, participants again executed isometric knee extensor contractions (3-s on and 2-s off) at a constant target torque with a maximum duration of 30 min. The setup was the same as during the first occasion, EMG, HR, and oxygen consumption were measured, and RPE was obtained immediately after exercise.

After preparation, and similar to the first occasion, the participants performed a 2-min warm-up (3-s on and 2-s off contractions) with torque set at 20% of MVC obtained during the first occasion. This was followed by three MVCs with a 2-min rest in between. If the last attempt was more than 5% higher than the previous attempts, an additional MVC was executed. After the 5-min rest, a 30-min exercise bout was performed for which MVC obtained on the first occasion was used to set the target torque (FT + 5%, see below).

Back to Top | Article Outline

The 30-min tests

The goal of the 30-min test was to find the torque that could be maintained while EMG, V˙E, and V˙O2 would remain stable, indicating steady-state exercise conditions. Recently, we found torque at FT to be lower than CT (9). Therefore, on the second occasion, target torque was set one step (5% MVC) higher than the FT determined with the incremental test during the first occasion (see Data Analysis). Seven participants were able to sustain the exercise at FT + 5% with relative ease (RPE < 8) and without clear objective signs of peripheral fatigue (see Data Analysis). For them, torque was increased by another 5% MVC on the third occasion (FT + 10%). Two of those seven participants also completed their second 30-min test without exhaustion (RPE = 8). After a 72-h break, these two executed a third 30-min bout with target torque set at FT + 15%. In contrast, the three remaining participants developed clear signs of peripheral fatigue and had problems (or failed) completing the 30 min already during the second occasion (RPE > 8). For them, torque on the third occasion was decreased by 5% MVC (=FT). Importantly, the participants as well as the investigator who encouraged the participant during the 30-min bouts were unaware of the target torque level.

Back to Top | Article Outline

Data analysis

Torque and EMG data were analyzed using custom-written MATLAB software (version R20008b). EMG was sampled at 1000 Hz, filtered (fourth-order band-pass = 10–400 Hz). rsEMG amplitude and median frequency were calculated for each of the three superficial knee extensor muscles separately. For each muscle, rsEMG data were expressed as a percentage of the rsEMG during MVC obtained on the same day. Subsequently, the normalized rsEMG values and the median frequencies were averaged for the three muscles.

V˙E, V˙O2, V˙CO2, and RER were calculated using the Cosmed-K4b2 software. For the incremental exercise test, the first ventilatory threshold (VT1) was determined by visual inspection of the V˙O2 versus V˙CO2 curve (V slope method) (2). At VT1, the ventilatory equivalent for oxygen (V˙E/V˙O2) starts to increase while V˙E/V˙CO2 remains stable (19).

In all experimental sessions, target torque was set as a percentage of MVC obtained during the first occasion (100%). Thus, at a certain percentage target torque, participants executed the task at the same absolute torques across days. However, following data collection, all torques were expressed as a percentage of the highest MVC measured at any of the days. In this way, MVC is the best approximation of the true maximum of each participant. This is important because torques are normalized to MVC to obtain FTs that can be compared among subjects. In addition, on the first occasion, we checked to which extent participants were able to use the full force-generating capacity of their knee extensors with superimposed electrical stimulation. We planned to exclude participants with a maximal capacity for VA under 85%, but this did not happen.

VA was calculated as follows (4,9):

For all contractions of 3-s duration, the average value over the middle 2 s were used to calculate median frequency, rsEMG, and mean torque. Subsequently, for all contractions, rsEMG amplitude was divided by torque to adjust rsEMG amplitude for small fluctuations in torque. These EMG/torque ratios will be presented. In addition, torque time integral (TTI) was calculated for each contraction to obtain a measure of total torque production, which was used for normalization of V˙O2 to torque production (V˙O2/TTI).

For all parameters obtained during the incremental test, mean values were calculated per 30-s exercise, resulting in 10 data points for each 5-min bout. These data points were subsequently normalized to the mean value obtained during the lowest intensity bout (25% MVC) for each participant, which was set at 100%. This was done to facilitate comparing the relative changes of the different parameters within and among participants.

When during the 5-min bouts parameter values changed, this always occurred in an approximate linear fashion (Figs. 1 and 2). Therefore, changes within bouts were quantified by taking the slopes of linear regression lines fitted through the 10 data points. The highest target torque, at which slope of EMG/torque was smaller than 2%·min−1, was defined as the FT.





During the first 2–4 min of all 30-min exercise bouts, HR and ventilation parameters often increased relatively fast, signifying kinetic adjustments to the required exercise load (e.g., Fig. 2). Therefore, the first 4 min of exercise was non–steady state, and therefore, these data were not included for quantification of any (fatigue-related) changes. Moreover, when changes occurred in the remaining part of the bouts, these were often nonlinear. To accommodate for these variations, averaged values of the last 2 min of exercise (without torque failure, these were minutes 29 and 30) were expressed as a percentage of the averaged values of the fifth and sixth minutes and were subsequently divided by time (usually 30 − 6 min = 24 min) to obtain averaged changes as percentages per minute.

Back to Top | Article Outline


Data are presented as mean ± SD and 95% confidence intervals (CIs) are provided where appropriate. Paired t-tests or repeated-measures ANOVA (SPSS Version 20.0; IBM, New York, NY) with Bonferroni post hoc comparisons was used when applicable, and effect sizes (partial η 2) are reported. Pearson correlation coefficient was calculated to establish significance of linear relations. Mann–Whitney U test was used to test for differences between subgroups (trained vs untrained and male vs female). Wilcoxon signed rank test for related samples was used to compare RPE at the end of the two 30-min exercise bouts. Statistical significance was accepted at P < 0.05.

Back to Top | Article Outline


Incremental testing

Maximal VA was 96.0% ± 3.5% (range = 89.8%–99.1%). MVC on the first day was 245 ± 57 N·m, which was not significantly different (P = 0.19, η 2 = 0.18) from MVC on the second (257 ± 60 N·m) and third days (257 ± 60 N·m). For each participant, MVC was subsequently taken as the highest value measured at any of the three test days, which was 264 ± 64 N·m on average, with similar values (P = 0.84) for trained (257 ± 47 N·m) and untrained (272 ± 82 N·m) participants.

All data regarding the incremental test are presented as a percentage of the average values obtained during the lowest-intensity bout (25% MVC). At 25% MVC, the measured values for the respective parameters were as follows: rsEMG = 20.8% ± 5.8% MVC, median frequency of the EMG = 61.6 ± 6.0 Hz, V˙E = 13.8 ± 2.8 L·min−1, V˙O2 = 563 ± 138 mL·min−1, V˙CO2 = 464 ± 105 mL·min−1, HR = 78 ± 7 bpm, RER = 0.83 ± 0.06, V˙E/V˙O2 = 25.0 ± 3.7, and V˙E/V˙CO2 = 30.2 ± 4.2.

In Figure 1, EMG/torque data obtained during the incremental test were first realigned according to the individually determined FT and were subsequently averaged. At FT, normalized EMG/torque (averaged over the last 2 min of the bout: 106.1% ± 8.3%, n = 10) was significantly different from 100% (95% CI = 100.1%–112.0%) and from exercise at FT − 5% (97.9% ± 6.7%, P = 0.02). However, a clear increase (to 134.7% ± 17.4%) was only found at FT + 5% (95% CI = 122.3%–147.2%). Moreover, and in accordance with our definition of FT, during 5 min of exercise at FT − 5% and at FT, EMG/torque did not increase for any of the participants. The slopes (%·min−1) of the regression lines at FT − 5% (−1.7 to 0.7) and at FT (−0.6 to 1.4) were not significantly different from zero. However, when target torque was increased by another 5% MVC to FT + 5%, in all participants, normalized EMG/torque consistently increased (Fig. 1) during the 5 min, with an average slope of 6.1%·min−1 ± 3.2%·min−1 (3.8%·min−1–8.4%·min−1) of the regression lines (r 2 = 0.68± 0.15).

FT was reached at 36.0% ± 7.0% of the highest MVC. RPE at FT was 5.7 ± 1.7 (range = 3–8), indicating that participants did not yet rate the exercise as being extremely heavy. Failure within the 5 min of a bout occurred, on average, at 50.7% ± 7.3% MVC, which was 14.7% ± 4.2% MVC above FT. In line with the expectations, RPE at failure was 9.5 ± 0.7. There was a significant linear relation between FT and torque at failure (r 2 = 0.68). FT was similar (P = 0.83) between men (n = 6, 36.1% ± 6.0% MVC) and women (n = 4, 35.8% ± 9.4% MVC) but was significantly higher (P = 0.008) in trained participants (n = 5, 41.4% ± 5.8% MVC) than in untrained participants (n = 5, 30.5% ± 1.8% MVC).

Changes in the other parameters are illustrated by the data of participant no. 9 (untrained female) in Figure 2. Around FT, median frequency of the EMG started to decline faster, while RER, V˙O2/TTI, and HR, in particular, showed faster increases above FT (Fig. 2).

Back to Top | Article Outline

Median frequency of the EMG

At FT, the normalized median frequency of the EMG (99.1% ± 3.4% over the last 2 min of the bout) was similar (P = 0.67) to the median frequency at FT − 5%. At FT + 5%, the median frequency was significantly lower (96.8% ± 3.8%, P = 0.02) than at FT − 5%, and it tended to be lower compared to FT (P = 0.07). However, within the bouts, the decline of median EMG was limited (in one participant, it was even absent) and the slopes of the regression lines were not significantly (P = 0.55) different between FT − 5% (1.44 ± 0.93, r 2 = 0.46 ± 0.31), FT (−1.6 ± 0.7%·min−1, r 2 = 0.47 ± 0.19), and FT + 5% (−1.94 ± 1.36, r 2 = 0.67 ± 0.23).

Back to Top | Article Outline


As would be expected, HR significantly increased with exercise intensity, also around FT where it increased (P < 0.001, η 2 = 0.83) from 109.6% ± 6.3% at FT − 5% to 128.0% ± 10.2% at FT + 5%. However, during the 5-min bouts, the slopes of the regression lines significantly changed (P = 0.007, η 2 = 0.50). Post hoc testing revealed that at FT − 5%, HR slightly (P < 0.05) increased (0.27%·min−1–1.89%·min−1) over 5 min, and the average slope of the regression line was 1.08%·min−1 ± 1.13%·min−1 (r 2 = 0.41 ± 0.29). This increase was not different (P = 0.76) from that at FT (1.56%·min−1 ± 1.02%·min−1, r 2 = 0.45 ± 0.15), but at FT + 5%, the slope of the regression line (r 2 = 0.56 ± 0.27) doubled (P = 0.015) to 3.04%·min−1 ± 1.76%·min−1.

Back to Top | Article Outline


Similar to the HR, RER increased (P = 0.002, η 2 = 0.58) from 104.3% ± 3.6% at FT − 5%, to 109.0% ± 8.8% at FT, and to 114.2% ± 4.4% at FT + 5%. The slopes of the regression lines tended to increase around FT (P = 0.06, η 2 = 0.30).

Back to Top | Article Outline


Similar to EMG/torque, V˙O2/TTI also significantly (P = 0.04, η 2 = 0.35) increased around FT, from 90.2% ± 10.5% at FT − 5%, to 91.7% ± 8.4% at FT, and to 99.9% ± 15.6% at FT + 5%. Post hoc testing showed that V˙O2/TTI at FT was not different (P = 0.52) from that at FT − 5%, whereas the value obtained at FT + 5% was higher than at FT (P = 0.02) and tended to be higher compared to FT − 5% (P = 0.06). Moreover, the slopes of the regression lines during the 5-min bouts tended to increase around FT (P = 0.06, η 2 = 0.38).

Back to Top | Article Outline

V˙E/V˙CO2 and V˙E/V˙O2

V˙E/V˙CO2 significantly decreased (P = 0.04, η 2 = 0.33) around FT. Only in two participants that clear increases in V˙E/V˙CO2 were seen, and these occurred at intensities near torque failure. V˙E/V˙O2 did not significantly increase around FT (P = 0.28, η 2 = 0.13), although for eight of the participants, V˙E/V˙O2 increased at some intensity above FT, but the magnitude of this increase varied among participants. The slopes of the regression lines were not significantly (P = 0.31, η 2 = 0.12) different around FT. VT1 could be determined for nine participants and was, on average, found at a torque of 41.4% ± 6.2% MVC, which was significantly higher (P = 0.01) than torque at FT for those nine participants (36.8% ± 7.0% MVC). Both parameters were significantly related (r 2 = 0.60, P = 0.01).

Back to Top | Article Outline

30-min exercise bouts

We expected that rsEMG amplitude and gas exchange parameters would no longer stabilize during the 30-min bouts when peripheral fatigue developed. In general, our findings were in line with those expectations. A typical example is presented in Figure 3. During exercise at the highest intensity, peripheral fatigue developed in all participants (Table 1). Seven completed that 30-min bout with considerable effort (RPE > 8), while three, at some point (13–20 min), failed to deliver the required torque. Changes in most parameters (except for RER and V˙E/V˙O2) were significantly greater for exercise at the higher compared to the (5% MVC) lower torque (Table 1). At the lower of the two intensities, all participants completed the 30-min bout with relative ease as indicated by the moderate RPE values: 6.7 ± 0.9 versus 9.1 ± 0.9 at the higher torque level (P = 0.005). EMG/torque (−0.009%·min−1 to 0.46%·min−1) and V˙O2/TTI (−0.20%·min−1 to 0.55%·min−1) did not significantly increase over time at the lowest of the two intensities, but at the higher intensity, increases were considerable (Table 1). The latter was also found for V˙E, V˙CO2, and HR, although these three parameters already showed small (<0.91%·min−1) but significant increases when participants exercised at the lower of the two torques. A representative example is shown in Figure 3.





During 30-min repetitive knee extensor contractions, objective signs of fatigue appeared when participants exercised at a torque of 44.3% ± 7.0% MVC. The highest exercise intensity without clear increases in EMG, HR, and gas exchanges parameters was defined the 30-min threshold torque. The 30-min threshold torque was 39.7% ± 8.0% MVC, which was almost 4% higher (P < 0.05) than FT established during the incremental test (36.0% ± 7.0% MVC) but was comparable to the torque at which VT1 was found: 41.4% ± 6.2% MVC. Similar to the FT established during the incremental test, the trained participants produced significantly higher (P = 0.007) torques at the 30-min threshold torque (46.1% ± 5.7% MVC) than the untrained participants did (33.4% ± 2.9% MVC). Moreover, the 30-min threshold torque was significantly related to FT (r 2 = 0.79).

Back to Top | Article Outline


The present study confirms recent findings (9) by showing that, based on objective parameters, an FT can be established during submaximal repetitive isometric knee extensor contractions. It elaborates on these previous results (9) by showing 1) that determination of FT is possible solely based on changes in EMG amplitude, 2) that an incremental test can be used to establish FT, and 3) that FT is related to exercise sustainability. Using an incremental test allows establishing FT while avoiding severe fatigue and the necessity of maximal or near-maximal effort for participants. This is of importance when contractile and physiological muscle properties of fragile people have to be investigated; in other situations, it can also be an advantage to obtain a valid indication of aerobic capacity of muscle groups that is basically independent of the participants’ motivation. Although in the present study the participants still exercised until torque failure, FT was already found when RPE was still moderate and well (about 15% of MVC torque) below the intensity at which torque failure occurred. In addition, the submaximal character of the presented method of establishing FT was confirmed during continuous exercise bouts of long (30 min) duration; disproportional increases in V˙E, V˙O2, and EMG and high (>8) RPE only occurred at torques well above (about 8% MVC) FT.

FT was established based on changes in surface EMG amplitude, since consistent FT-related changes in rsEMG were recently found with a similar exercise protocol (9). The increases in rsEMG amplitude and muscle deoxygenation that occurred above FT in that study and the increases in EMG and V˙O2 found in the current study during constant torque production above FT are assumed to be signs of non–steady-state exercise conditions. Signs of non–steady-state metabolism, such as increased blood lactate and V˙O2 (21), increased [Pi], and decline in pH (15), in the exercising muscles have been found by others, when a critical exercise intensity was surpassed. These metabolic changes in the muscle have been related to reduced force production of the muscle fibers: peripheral fatigue (1). It seems plausible that, when torque has to be maintained, there is a certain torque threshold above which a fatigue-induced loss of force-producing capacity of the active muscle fibers has to be compensated for. According to the size principle, this would occur by increasing the firing rate of the already-active motor units and recruitment of new muscle fibers that belong to larger and more fatigable motor units. Consequently, in addition to the increase in rsEMG, the increases in muscle deoxygenation in our previous study (9) and in V˙O2 in the present study probably also reflect increased neuromuscular activation, possibly in combination with a reduction of the efficiency of the already-recruited muscle fibers (7,13). Although care has to be taken with the interpretation of changes in surface EMG (11), when increases in rsEMG amplitude are found, this is usually related to increased motor unit activity (8,14). Also in the present study, torque and rsEMG amplitude increased in a near-linear manner, with each 5% MVC increase in torque: the ratio between rsEMG amplitude and torque (EMG/torque) remained constant at torques below FT. However, above FT, this ratio increased, not only during the 5-min exercise bouts but also with each 5% step increase in target torque, a progressive increase in EMG/torque was found (Fig. 1). These findings are in line with a fatigue-induced increase in motor unit activity relative to torque production during repetitive contractions above FT.

A lack of increase in rsEMG amplitude, on the other hand, does not always signify the absence of increased motor unit activity. Particularly with longer-lasting contractions and during fatigue, neuromuscular activation may increase without clear increases in rsEMG (8), for instance, due to increased signal cancelation. Indeed, in three of the present participants, a continuous increase in rsEMG was not found during the 30-min exercise at the highest intensity, despite clear increases in V˙E, V˙CO2, V˙O2, (near) maximal RPE, and even torque failure, which all indicated a considerable increase in neural drive. Thus, rsEMG amplitude appeared a less reliable indicator for peripheral fatigue in some subjects during 30 min of repetitive contractions. Interpretation of changes in gas exchange parameters also had some limitations, but conversely to the changes in EMG, the changes in gas exchange parameters during the 30-min exercise bouts seemed somewhat more consistent than those during the incremental test. This may be related to the specific, rather static, nature of the exercise. The 3-s isometric contractions considerably influenced breathing patterns, particularly at the highest intensities during the incremental test, where inhaling is restricted to the 2-s rest in between contractions. This may have contributed to the variation in the measurements that we encountered during the incremental test.

Notwithstanding some of the above considerations, the presented results indicate that FT established during the incremental test signifies a meaningful border, above which the first signs of non–steady-state exercise conditions, such as increases in rsEMG, HR, and V˙O2, appear. In the present study, FT was significantly related to VT1, to torque at which failure occurred during the incremental test, as well as to the 30-min threshold torque. Moreover, and as would be expected, torque at FT (%MVC) was significantly higher in participants involved in regular endurance training compared to the participants who did not train. In this respect, the current findings fully concur with our previous study in which torque at FT was found to be significantly related to V˙O2max (r 2 = 0.68) and the first (r 2 = 0.45) and second ventilatory thresholds (r 2 = 0.63) obtained during cycle ergometry (9).

Torque at FT in the present study (36.7% ± 7.0% MVC) was comparable to torque at FT obtained with similar exercise bouts that were presented in a nonincremental way (40.0% ± 8.1% MVC) in our previous study (9). In that recent study, CT was found at 53.1% ± 10.0% MVC. The present results also strongly indicate that FT occurred at a considerably lower exercise intensity than CT. Although the present study did not directly establish CT, it seems reasonable to state that the 30-min threshold torque (at 39.7% ± 8.0% MVC) in the present study is a fair approximation of CT. Both FT (at 36.7% MVC) and 30-min threshold torque (at 39.7% MVC) in the present study are higher than CT recently obtained with repetitive exercise to failure at different torque levels (34% ± 2% MVC) by Burnley et al. (6) who investigated the same type of repetitive (3-s on and 2-s off) isometric knee extensor contractions. They did not report the training status of their participants, but the 30-min threshold torque of the present untrained participants was 33.4%, which is similar to the CT they reported.

The participants of Burnley et al. (6) had a time-to-task failure of 17.6 ± 2.2 min when torque level was increased by a mere 3.4% MVC above CT. Therefore, if the 30-min threshold torque would be similar to CT, our participants were expected to fail well within 30 min while exercising 5% MVC above the 30-min threshold torque. Indeed, three of our participants failed after about 15 min. However, the others completed exercise at the 30-min threshold torque +5% (at 44. 0% ± 9.4% MVC, n = 7), albeit with considerable effort and signs of peripheral fatigue. These findings suggest that not only torque at FT but also even the slightly higher torque at the 30-min threshold torque for some of our subjects still was below CT. These findings are in line with the recent study of Burnley et al., who showed that peripheral fatigue already developed at exercise intensities below CT, as indicated by a near 20% decrease in electrically evoked doublet torque during the first 30 min of 60-min repetitive knee extensor contractions at torques 3%–6% MVC below CT (their Fig. 3A). Noteworthy, their results indicated that it was the rate of change in markers of fatigue rather than the presence or absence of fatigue that demarcated CT (6).

The development of peripheral fatigue during exercise at and just above CT makes establishing maximal exercise duration at torques near CT dependent on volitional drive and the participants’ motivation. The advantage of working with a more “conservative” FT is that it can be established without straining the subjects too much, which is especially relevant when frail people are tested. We do recognize that, to measure maximal sustainable output and exercise performance, a direct assessment of CT may be more meaningful than obtaining FT. However, FT and CT are related (10). Moreover, not only CT (by definition) but also FT is related to sustainable exercise capacity (present results), V˙O2max, and ventilatory thresholds established during cycle ergometry (10) and during repetitive knee extensor contractions (present study). In addition, in the present study, torque at FT of participants who were involved in endurance training was considerably higher than that of their inactive but healthy counterparts. However, it has to be noted that the latter finding was based on preliminary data of five participants only in each group. Consequently, it needs to be confirmed in future studies that FT is closely related to training status. Clearly, frail participants have to be included to verify the proposed feasibility of the current test in these people.

To make application of the current test cost-effective in a clinical setting, total exercise duration has to be reduced, for instance, by decreasing bout duration. For the present study, we used 5-min exercise bouts in order not to change too many variables at the same time, which allowed us to make direct comparisons with our recent study (9). The present results (e.g., Figs. 1 and 2) indicate that, when steady state was attained, this occurred within 2 min. Thus, it may be feasible to reduce bout duration by about 50%.

In conclusion, early signs of peripheral fatigue leading to increases of rsEMG amplitude can be used to establish an FT during an incremental test of unilateral isometric knee extensor contractions. This threshold is related to sustainable exercise intensity and training status. The disproportional increase in EMG at torques above FT were substantiated by increases in HR and V˙O2 during exercise above FT. The submaximal nature of the test makes it a promising alternative for tests involving maximal effort, which is not always warranted, particularly with frail or injured participants.

The authors did not receive external funding for this study. The authors have no professional relationships with companies or manufacturers.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.

Back to Top | Article Outline


1. Allen DG, Lamb GD, Westerblad H. Skeletal muscle fatigue: cellular mechanisms. Physiol Rev. 2008; 88 (1): 287–332.
2. Beaver WL, Wasserman K, Whipp BJ. A new method for detecting anaerobic threshold by gas exchange. J Appl Physiol. 1986; 60 (6): 2020–7.
3. Borg GA. Psychophysical bases of perceived exertion. Med Sci Sports Exerc. 1982; 14 (5): 377–81.
4. Burnley M. Estimation of critical torque using intermittent isometric maximal voluntary contractions of the quadriceps in humans. J Appl Physiol. 2009; 106 (3): 975–83.
5. Burnley M, Vanhatalo A, Fulford J, Jones AM. Similar metabolic perturbations during all-out and constant force exhaustive exercise in humans: a (31)P magnetic resonance spectroscopy study. Exp Physiol. 2010; 95 (7): 798–807.
6. Burnley M, Vanhatalo A, Jones AM. Distinct profiles of neuromuscular fatigue during muscle contractions below and above the critical torque in humans. J Appl Physiol. 2012; 113 (2): 215–23.
7. Cannon DT, White AC, Andriano MF, Kolkhorst FW, Rossiter HB. Skeletal muscle fatigue precedes the slow component of oxygen uptake kinetics during exercise in humans. J Physiol. 2011; 589 (Pt 3): 727–39.
8. de Ruiter CJ, Elzinga MJH, Verdijk PWL, van Mechelen W, de Haan A. Voluntary drive-dependent changes in vastus lateralis motor unit firing rates during a sustained isometric contraction at 50% of maximum knee extension force. Pflügers Arch. 2004; 447: 436–44.
9. de Ruiter CJ, Maas EA, Wesseling MG, de Haan A. Knee extensor fatigue threshold is related to whole-body V˙O2max. Med Sci Sports Exerc. 2012; 44 (7): 1366–74.
10. de Ruiter CJ, May AM, van Engelen BG, Wevers RA, Steenbergen-Spanjers GC, de Haan A. Muscle function during repetitive moderate-intensity muscle contractions in myoadenylate deaminase-deficient Dutch subjects. Clin Sci. 2002; 102 (5): 531–9.
11. Farina D, Merletti R, Enoka RM. The extraction of neural strategies from the surface EMG. J Appl Physiol. 2004; 96 (4): 1486–95.
12. Hauser T, Bartsch D, Baumgartel L, Schulz H. Reliability of maximal lactate-steady-state. Int J Sports Med. 2013; 34 (3): 196–9.
13. Hepple RT, Howlett RA, Kindig CA, Stary CM, Hogan MC. The O2 cost of the tension-time integral in isolated single myocytes during fatigue. Am J Physiol Regul Integr Comp Physiol. 2010; 298 (4): R983–8.
14. Hunter SK, Duchateau J, Enoka RM. Muscle fatigue and the mechanisms of task failure. Exerc Sport Sci Rev. 2004; 32 (2): 44–9.
15. Jones AM, Vanhatalo A, Burnley M, Morton RH, Poole DC. Critical power: implications for determination of V˙O2max and exercise tolerance. Med Sci Sports Exerc. 2010; 42 (10): 1876–90.
16. Jones AM, Wilkerson DP, DiMenna F, Fulford J, Poole DC. Muscle metabolic responses to exercise above and below the “critical power” assessed using 31P-MRS. Am J Physiol Regul Integr Comp Physiol. 2008; 294 (2): R585–93.
17. Krustrup P, Hellsten Y, Bangsbo J. Intense interval training enhances human skeletal muscle oxygen uptake in the initial phase of dynamic exercise at high but not at low intensities. J Physiol. 2004; 559 (Pt 1): 335–45.
18. Lorenzo S, Minson CT, Babb TG, Halliwill JR. Lactate threshold predicting time-trial performance: impact of heat and acclimation. J Appl Physiol. 2011; 111 (1): 221–7.
19. Lucia A, Hoyos J, Perez M, Chicharro JL. Heart rate and performance parameters in elite cyclists: a longitudinal study. Med Sci Sports Exerc. 2000; 32 (10): 1777–82.
20. Pires F, Noakes TD, Lima-Silva AE, et al. Cardiopulmonary, blood metabolite and rating of perceived exertion responses to constant exercises performed at different intensities until exhaustion. Br J Sports Med. 2011; 45 (14): 1119–25.
21. Poole DC, Ward SA, Gardner GW, Whipp BJ. Metabolic and respiratory profile of the upper limit for prolonged exercise in man. Ergonomics. 1988; 31 (9): 1265–79.
22. Smith JC, Dangelmaier BS, Hill DW. Critical power is related to cycling time trial performance. Int J Sports Med. 1999; 20 (6): 374–8.
23. Vanhatalo A, Doust JH, Burnley M. Determination of critical power using a 3-min all-out cycling test. Med Sci Sports Exerc. 2007; 39 (3): 548–55.


© 2014 American College of Sports Medicine