Prolonged low-frequency force depression is underestimated when assessed with doublets compared with tetani in the dorsiflexors

Abstract

Prolonged low-frequency force depression (PLFFD) after damaging eccentric exercise may last for several days. Historically, PLFFD has been calculated from the tetanic force responses to trains of supramaximal stimuli. More recently, for methodological reasons, stimulation has been reduced to two pulses. However, it is unknown whether doublet responses provide a valid measure of PLFFD in the days after eccentric exercise. In 12 participants, doublets and tetani were elicited at 10 and 100 Hz before and after (2, 3, 5 min, 48 and 96 h) 200 eccentric maximal voluntary contractions of the dorsiflexors. Doublet and tetanic torque responses at 10 Hz were similarly depressed throughout recovery (P > 0.05; e.g., 2 min: 58.9 ± 12.8% vs. 57.1 ± 14.5% baseline; 96 h: 85.6 ± 11.04% vs. 85.1 ± 10.8% baseline). At 100 Hz, doublet torque was impaired more than tetanic torque at all time points (P < 0.05; e.g., 2 min: 70.5 ± 14.2% vs. 88.1 ± 11.7% baseline; 96 h: 83.0 ± 14.2% vs. 98.7 ± 9.5% baseline). As a result, the postfatigue reduction of the 10 Hz-to-100 Hz ratio (PLFFD) was markedly greater for tetani than for doublets (P < 0.05; e.g., 2 min: 64.3 ± 15.1% vs. 83.0 ± 5.8% baseline). In addition, the doublet ratio recovered by 48 h (99.2 ± 5.0% baseline), whereas the tetanic ratio was still impaired at 96 h (88.2 ± 9.7% baseline). Our results indicate that doublets are not a valid measure of PLFFD in the minutes and days after eccentric exercise. If study design favors the use of paired stimuli, it should be acknowledged that the true magnitude and duration of PLFFD are likely underestimated.

NEW & NOTEWORTHY Prolonged low-frequency force depression (PLFFD) will result from damaging exercise and may last for several days. After 200 eccentric maximal dorsiflexor contractions, we compared the gold-standard measure of PLFFD (calculated using trains of supramaximal stimulation) to the value obtained from an alternative technique that is becoming increasingly common (paired supramaximal stimuli). Doublets underestimated the magnitude and duration of PLFFD compared with tetani, so caution must be used when reporting PLFFD derived from paired stimuli.

INTRODUCTION

Peripheral fatigue is the reversible impairment of muscle contractile capability due to processes occurring at or below the neuromuscular junction (1). One possible manifestation of peripheral fatigue is a proportionately greater loss of force in response to low (e.g., 10 Hz)- versus high (e.g., 100 Hz)-frequency muscle stimulation. Edwards and colleagues (7) termed this phenomenon “low-frequency fatigue.” However, because this terminology is easily misinterpreted as fatigue induced by low-frequency stimulation, Allen and colleagues (1) offered “prolonged low-frequency force depression” (PLFFD) as a less ambiguous wording. This type of fatigue can be observed after any form of intense exercise [e.g., repeated all-out cycling bouts (21)], but it is most pronounced when the exercising protocol includes unaccustomed eccentric contractions (6, 12). Although PLFFD is documented using evoked muscle forces, motor neurons discharge at relatively low frequencies during low- to moderate-intensity voluntary movements, which means that PLFFD is probably responsible for the feeling of weakness that can be experienced in the days following intense exercise (1). That is, if higher than normal discharge rates or numbers of motor units are needed to achieve a submaximal force, the brain will perceive this greater descending drive (and accompanying sense of effort) as weakness (1, 15).

PLFFD can last for several days and is characterized by the absence of muscle metabolic or electrical disturbance (7, 12, 30). Although reduced myofibrillar Ca²⁺ sensitivity has been implicated (e.g., Ref. 21), PLFFD is primarily ascribed to a spatially uniform reduced Ca²⁺ release within the muscle fiber (14, 31). High stimulation frequencies coincide with the plateau portion of the force-intracellular Ca²⁺ concentration ([Ca²⁺]_i) relationship, so moderate reductions in sarcoplasmic reticulum Ca²⁺ release or myofibrillar Ca²⁺ sensitivity have little influence on muscle force compared with the sizable effect at low stimulation frequencies, which relate to the steep part of the force-[Ca²⁺]_i relationship (1).

The gold-standard measure to assess PLFFD in both animal and human studies is the ratio of low- to high-frequency force responses induced by a train of supramaximal stimuli applied to a peripheral nerve (1). However, in humans the reproducibility of the tetanic force to a train of stimuli (i.e., 0.5–1 s) can be a concern when the strong muscular contraction displaces the electrode position relative to the nerve (e.g., femoral nerve). Furthermore, supramaximal trains can be prohibitively uncomfortable (e.g., tibial nerve) and so are viewed as best suited to highly motivated subjects and unsuitable for clinical practice (18, 26). To eliminate these problems, alternative techniques for measuring PLFFD have been considered, such as submaximal stimulation via large electrodes over the muscle belly (18) or pairs, rather than trains, of supramaximal stimuli (26). Immediately and 30 min after downhill running, 0.5-s trains of submaximal stimulation over the quadriceps yielded an equivalent reduction of the 20 Hz-to-80 Hz ratio compared with 0.5-s maximal trains delivered to the femoral nerve (18). In contrast, in another study with the same exercise paradigm and testing timeline, the reduction of the 10 Hz-to-100 Hz ratio was much less in response to paired electric or magnetic stimuli compared with the ratio obtained with submaximal muscle belly stimulation (26). An underestimation of PLFFD with paired stimuli is intuitive if one considers that the continued summation of force with successive stimuli is markedly greater at a high than a low frequency, especially after fatigue (e.g., see Fig. 2 of Ref. 19). Consequently, the proportion of peak tetanic force elicited by two pulses would be much lower at 100 Hz compared with 10 Hz, resulting in a higher 10 Hz-to-100 Hz ratio with pairs than trains of stimulation.

Despite the expectation and evidence that a doublet ratio underestimates PLFFD compared with a tetanic ratio, paired stimuli have been extensively used in recent years (8, 9, 13, 16, 24, 25). Although this technique has practical appeal over trains of supramaximal stimulation, before it becomes further entrenched, it is important to characterize in more detail how well it reflects the gold standard. To do this effectively, it is necessary to consider muscle groups other than quadriceps and to measure PLFFD in the days following damaging exercise, when there may be a large disparity in force elicited by 2 versus 50–100 high-frequency pulses. Hence, the purpose of the present experiment was to compare the magnitude of PLFFD measured by pairs and trains of supramaximal stimuli in the minutes and days following 200 maximal eccentric contractions of the dorsiflexor muscles. In accordance with the physiological reasoning presented above, it was hypothesized that the underestimation of PLFFD with paired stimuli would persist for several days after damaging eccentric exercise.

METHODS

Ethical approval.

All procedures in the present study were reviewed and approved by the Clinical Research Ethical Review Board of the University of British Columbia (Application ID: H17-02716) and conformed to the standards set by the Declaration of Helsinki, except for registration in a database. Written informed consent was obtained from each participant.

Participants.

Twelve recreationally active university students (5 women, 7 men; means ± SD: age 23 ± 3 yr, height 1.73 ± 0.10 m, body mass 77 ± 14 kg) were recruited for this study.

Study design.

Participants were tested on three separate occasions at the same time of the day. Session 1 included baseline isometric neuromuscular testing, the fatiguing eccentric protocol, and acute (≤5 min) isometric neuromuscular recovery measures. In sessions 2 and 3 (48 and 96 h after the first session), the baseline isometric neuromuscular testing was repeated.

Isometric neuromuscular testing was performed on a custom-built dynamometer similar to that used by Marsh and colleagues (17) (Fig. 1). The foot of the dominant limb was placed on a footplate at 30° of plantar flexion and secured with a Velcro strap across the instep and another across the toes. A 90° angle at the knee joint was ensured, and a C-clamp was tightened over the distal portion of the thigh to avoid forward and upward movement of the thigh during dorsiflexor contractions. The torque produced during voluntary and electrically evoked isometric contractions was measured by a linear strain gauge connected to the footplate (MLP-300; Transducer Techniques, Temecula, CA). The signal was then amplified (1,000×) (CED 1902; Cambridge Electronic Design, Cambridge, UK), sampled at 1,000 Hz with a 16-bit A/D converter (CED 1401-3; Cambridge Electronic Design), and recorded with Spike2 software (version 8; Cambridge Electronic Design).

Fig. 1.

Experimental setup for isometric neuromuscular testing.

The fatigue protocol was performed on a Humac Norm multijoint dynamometer (CSMi, Stoughton, MA). Participants sat in a reclined position with 120° and 90° angles at the hip and knee joints, respectively. The foot of their dominant limb was placed on a footplate so the malleoli aligned with the rotation axis of the torque motor, and was secured with Velcro straps over the instep and toes. Adjustable shoulder straps, a seat belt across the waist, and a strap over the thigh minimized extraneous movement during dorsiflexor contractions. The range of motion of the footplate was adjusted to perform the eccentric dorsiflexor contractions from a neutral ankle position to 30° of plantar flexion.

Common fibular nerve stimulation.

To induce electrically evoked contractions, square-wave electrical stimuli (500-µs pulse width, 100–400 V) were delivered to the common fibular nerve with a computer-triggered stimulator (DS7AH; Digitimer). A bar electrode (279-930-24TP; Chalgren Enterprises, Gilroy, CA) was held in place by the operator distal and posterior to the fibular head. The optimal position of the bar electrode was marked with permanent ink to ensure reproducible placement for all visits. Stimulus intensity was set at 115% of the current required to obtain the maximal resting twitch peak torque of the dorsiflexor muscles, to ensure supramaximal stimulation throughout the protocol (10–35 mA).

Neuromuscular testing.

Data collection began with the determination of the supramaximal stimulator intensity. Stimulus intensity was raised incrementally until the peak torque of the dorsiflexor resting twitch reached a plateau. Intensity was then increased by 15%. Baseline neuromuscular testing consisted of three sets (separated by ≥90 s of recovery) of one brief (~3 s) isometric maximal voluntary contraction (MVC) and a sequence of five isometric electrically evoked contractions. During the MVC, a single stimulus was delivered to assess voluntary activation (VA). The sequence of electrically evoked contractions started ~1–2 s after relaxation from the MVC and consisted of a single twitch, doublets at 10 and 100 Hz (D10 and D100), and 1-s tetani at 10 and 100 Hz (T10 and T100). All stimuli after the MVC were separated by 1 s of rest. During each MVC, strong verbal encouragement and visual feedback were provided. To measure acute effects of the fatiguing eccentric exercise, one set of contractions was performed at 2, 3, and 5 min after the end of the fatigue protocol. To assess longer-term effects, all procedures for baseline testing were repeated 48 and 96 h after the first session.

Fatigue protocol.

In session 1, after baseline neuromuscular testing, the fatiguing task was performed. The protocol consisted of four sets (1 min of rest in between) of 50 eccentric dorsiflexor MVCs (Fig. 2A). The velocity of the footplate during the active eccentric motion was set at 60°/s, and that of the passive concentric motion was set at 30°/s. Both eccentric and concentric motions had positive and negative acceleration set to 2,000°/s². A 1-s delay separated each concentric and eccentric motion. The participant was asked to maximally contract as soon as the upward passive concentric motion finished, so that each eccentric contraction started from maximal muscle activation. Throughout the protocol, torque was continuously displayed on an oscilloscope to provide feedback to the participant.

Fig. 2.

Schematic representation of the experimental timeline (bottom) and torque responses (top) from doublets and tetani from 1 participant. A: before and after fatiguing eccentric exercise, measures of voluntary activation and peak torque were obtained from a sequence of isometric contractions that included a brief (~3 s) maximal voluntary contraction (MVC) followed by a series of electrically evoked contractions: single twitch, doublets at 10 and 100 Hz (D10 and D100), and 1-s tetani at 10 and 100 Hz (T10 and T100). The series started ~1–2 s after relaxation from the MVC, and all electrically evoked contractions were separated by 1 s of rest. At baseline as well as 48 and 96 h after the fatigue protocol, the sequence was performed 3 times, with ≥90 s of rest between each set. B: torque responses of electrically evoked contractions from 1 participant. From left to right: D10 vs. T10 before fatigue, D10 vs. T10 2 min after fatigue, D100 vs. T100 before fatigue, D100 vs. T100 2 min after fatigue. Before fatigue, the difference in peak torque between D10 and T10 was much less than the difference between D100 and T100. Thus at baseline, the torque ratio D10:100 was greater than T10:100 (see Table 1). At 2 min after fatigue, compared with baseline values, D10 and T10 declined similarly, whereas the decline of D100 was much greater than T100. Thus prolonged low-frequency force depression was underestimated when calculated using D10:D100 (see Fig. 4C).

Additional experiment.

To confirm that peak torque of D100 resided in the plateau region of the torque-frequency relationship of doublets, and to contrast the torque-frequency relationships of doublets and 1-s tetani, five additional participants (2 women, 3 men) were tested. The setup was the same as that for the isometric testing of the main study (Fig. 1). Once the stimulus intensity was determined, two series of isometric electrically evoked contractions (doublets or tetani) were elicited at the following stimulation frequencies: 30, 50, 5, 80, 10, 90, 20, 110, 40, 100, 70, and 60 Hz. A 10-s rest was provided between stimuli, and a rest of at least 3 min separated the series for doublets and tetani, which were collected in a random order.

Data analysis and statistics.

Torque data were analyzed off-line with Signal software (v. 5.08; Cambridge Electronic Design). For neuromuscular testing at baseline as well as 48 and 96 h after fatigue, all variables were calculated as the mean value from the three sets of contractions. However, at 2, 3, and 5 min after fatigue, values were measured from the lone set of contractions. For each set of voluntary and electrically evoked contractions, the peak torque of the interpolated twitch and the resting twitch were used to calculate VA of the dorsiflexors with the following equation: VA (%) = [1 − (interpolated twitch ÷ resting twitch)] × 100. Peak torque was obtained for each doublet and tetanus (D10, D100, T10, T100) and used to create the 10 Hz-to-100 Hz torque ratio from doublets and tetani (D10:100, T10:100). To calculate PLFFD, D10:100 and T10:100 at each recovery time point was expressed as a percentage of the respective value at baseline. Peak torque was also obtained for each doublet and tetanus in the additional experiment.

Statistical analysis was conducted with SPSS software (version 23; SPSS, Chicago, IL). A paired-samples t-test was conducted to compare the following measures at baseline: D10 versus T10, D100 versus T100, and D10:100 versus T10:100. All recovery data were expressed as a percentage of baseline, except for VA. One-way repeated-measures ANOVAs were conducted to compare MVC torque and VA across time (baseline, 2, 3, and 5 min, 48 and 96 h). When the output was significant, paired-samples t-tests and a Dunnett’s table were used to determine which time points were different from baseline. Two-way repeated-measures ANOVAs, with stimulation technique and time as within-subject factors, were used to compare normalized peak torques at 10 and 100 Hz as well as the normalized 10 Hz-to-100 Hz ratios. Paired-samples t-tests and a Dunnett’s table were used for post hoc investigation of main effects of stimulation technique and time. If only a main effect for time was found, doublet and tetanus data were pooled for determination of time points different from baseline. Given the sample size of five subjects, statistical tests were not performed on data collected during the additional experiment. All data are reported as means ± SD in the text and Table 1 and as means ± SE in Figs. 3, 4, and 6. The significance level was P < 0.05.

Table 1.

Baseline voluntary and electrically evoked torques, voluntary activation, and ratios of torque responses to low- and high-frequency stimulation

MVC torque, N·m	VA, %	D10, N·m	T10, N·m	D100, N·m	T100, N·m	D10:100	T10:100
43.7 ± 12.8	98.8 ± 0.9	13.3 ± 3.9	15.9 ± 4.3	17.2 ± 4.9	34.6 ± 11.1	0.78 ± 0.05	0.48 ± 0.10†

Values are means ± SD. D10, doublet peak torque at 10 Hz; D100, doublet peak torque at 100 Hz; D10:100, 10 Hz-to-100 Hz torque ratio for doublets; MVC, maximal voluntary contraction; T10, tetanus peak torque at 10 Hz; T100, tetanus peak torque at 100 Hz; T10:100, 10 Hz-to-100 Hz torque ratio for tetani; VA, voluntary activation.

^†P < 0.05 different between stimulation techniques.

Fig. 3.

Maximal voluntary torque and voluntary activation after the fatiguing eccentric protocol: mean ± SE values from 12 participants. A: maximal voluntary contraction (MVC) torque, expressed as % of baseline (BL). One-way repeated-measures ANOVA reported a main effect of time. Post hoc testing revealed that MVC torque was lower than baseline at all time points after the fatigue protocol (*P < 0.05). B: voluntary activation. One-way repeated-measures ANOVA reported a main effect of time. Post hoc testing indicated that voluntary activation was lower than baseline during the acute recovery period (*P < 0.05).

Fig. 4.

Peak torque of the 10-Hz and 100-Hz evoked doublets and tetani and prolonged low-frequency force depression: mean ± SE values from 12 participants. All values are expressed as % of baseline (BL). A: peak torque of the doublets and tetani at 10 Hz (D10 and T10 in the text, respectively). Two-way repeated-measures ANOVA reported only a main effect of time, with neither an effect of stimulation technique nor a stimulation technique × time interaction. Post hoc testing on pooled data indicated that both measures were decreased from baseline at all time points (*P < 0.05). B: peak torque of the doublets and tetani at 100 Hz (D100 and T100 in the text, respectively). Two-way repeated-measures ANOVA reported a main effect of stimulation technique and time and a stimulation technique × time interaction. Post hoc testing revealed that D100 was declined at all time points (*P < 0.05), whereas T100 was lower than baseline at 2, 3, and 5 min and 48 h (*P < 0.05). When techniques were compared, D100 was impaired more than T100 at all time points (†P < 0.05). C: prolonged low-frequency force depression (PLFFD), calculated as the 10 Hz-to-100 Hz torque ratio for doublets and tetani (D10:100 and T10:100 in the text, respectively). Two-way repeated-measures ANOVA reported a main effect of stimulation technique and time and a stimulation technique × time interaction. Post hoc testing indicated that D10:100 was lower than baseline only at 2, 3, and 5 min (*P < 0.05), whereas T10:100 was reduced at all time points (*P < 0.05). D10:100 was greater than T10:100 at all time points (†P < 0.05), indicating an underestimation of PLFFD for doublets compared with tetani.

Fig. 6.

Torque-frequency relationships for doublets and tetani: mean ± SE values from 5 participants of the evoked doublets and tetani. Although the response at 100 Hz occurs on the plateau for both forms of stimulation, it is clear from the 10 Hz vs. 100 Hz torques that 2 pulses are unsuitable to adequately differentiate low- vs. high-frequency stimulation (see also D10:100 and T10:100 values in Table 1).

Normalized values of D10 and T10, D100 and T100, as well as D10:100 and T10:100 were considered for the analysis of agreement between stimulation techniques. Within each measure, values were pooled separately for the acute (2, 3, and 5 min; n = 36) and long-term (48 and 96 h; n = 24) recovery periods. The agreement between the data acquired with the two stimulation techniques was assessed via Pearson’s correlation coefficient (r) and limits of agreement (LoA). The strength of the correlation was evaluated as follows: trivial (0–0.1), small [0.1–0.3), moderate [0.3–0.5), large [0.5–0.7), very large [0.7–0.9), and extremely large [0.9–1) (10). For the calculation of the LoA, data were checked for homogeneity of variance by correlating the test score differences (i.e., residuals) between D10 and T10, D100 and T100, as well as D10:100 and T10:100 to their mean value, following the method by Bland and Altman (5). For D10 and T10 as well as D100 and T100 no significant correlations were found (r < 0.35, P > 0.10), so residuals were used for further analysis. In contrast, significant correlations were found between residuals of D10:100 and T10:100 and their mean values for both the acute (r = −0.71, P < 0.05) and long-term (r = −0.64, P < 0.05) recovery periods. Thus, for these measures, the ratios of D10:100 to T10:100 were used for further analysis instead of the residuals (5). The residuals for D10 and T10 as well as D100 and T100 had a normal distribution, so 95% LoA were calculated as the mean absolute difference ± 1.96 times the standard deviation of the absolute differences (5). In contrast, ratios of D10:100 to T10:100 were not normally distributed, so LoA were calculated by using the median of the ratios as an estimate of the relative mean bias and the 5th and 95th percentiles as the lower and upper LoA (5).

RESULTS

Baseline values for MVC torque, VA, as well as electrically evoked torques are reported in Table 1. D10 was 16% lower than T10 [t(11) = −5.2, P < 0.05], whereas D100 was 50% lower than T100 [t(11) = −9.0, P < 0.05] (see Fig. 2B for comparison of baseline torque responses from D10 vs. T10 and D100 vs. T100, in 1 participant). As a result, D10:100 was 69% greater than T10:100 [t(11) = 12.8, P < 0.05].

The fatiguing protocol induced a significant decline in all measures. One-way repeated-measures ANOVA reported a main effect of time for MVC torque (F_1.8,19.6 = 27.6, P < 0.05: Fig. 3A) and VA (F_2.3,25.4 = 6.3, P < 0.05; Fig. 3B). Both variables were lower than baseline at 2, 3, and 5 min after the fatigue protocol (P < 0.05), and the impairment of MVC torque persisted at 96 h (P < 0.05). For D10 and T10 (Fig. 4A), two-way repeated-measures ANOVA reported only a main effect of time (F_2.6,28.5 = 87.9, P < 0.05), with neither an effect of stimulation technique (F_1,11 = 3.9, P = 0.07) nor a stimulation technique × time interaction (F_2.0,22.0 = 1.5, P = 0.24). Post hoc testing on pooled data indicated that both D10 and T10 were decreased from baseline at all recovery time points (P < 0.05). For D100 and T100 (Fig. 4B), two-way repeated-measures ANOVA reported a main effect of stimulation technique (F_1,11 = 54.5, P < 0.05) and time (F_2.1,23.2 = 22.6, P < 0.05) and a stimulation technique × time interaction (F_1.8,19.7 = 13.3, P < 0.05). Specifically, D100 was declined at all recovery time points (P < 0.05), and T100 was lower than baseline at 2, 3, and 5 min and 48 h (P < 0.05). When techniques were compared, the recovery relative to baseline was greater for T100 than D100 at all time points (P < 0.05). When D10:100 and T10:100 were compared (Fig. 4C), two-way repeated-measures ANOVA reported a main effect of stimulation technique (F_1,11 = 31.9, P < 0.05) and time (F_2.0,222.3 = 77.5, P < 0.05) and a stimulation technique × time interaction (F_2.4,26.8 = 20.6, P < 0.05). Whereas D10:100 was lower than baseline only at 2, 3, and 5 min (P < 0.05) and had recovered to baseline by 48 h, T10:100 was declined at all recovery time points (P < 0.05). When indices of PLFFD were compared, D10:100 was greater than T10:100 at all time points (P < 0.05).

Correlations between D10 and T10 were extremely large in both the acute (r = 0.92) and long-term (r = 0.95) recovery periods (P < 0.05). The correlation between D100 and T100 was large in the acute recovery period (r = 0.66; P < 0.05), whereas in the long-term recovery period the correlation was moderate (r = 0.44; P < 0.05). When D10:100 and T10:100 were considered, the correlation was large in the acute recovery period (r = 0.58; P < 0.05), whereas no significant correlation between measures was found in the long-term recovery period (r = 0.34; P = 0.11).

LoA between doublet- and tetanus-related measures are reported in Fig. 5. For D10 and T10 (Fig. 5, A and B), for both the acute and long-term recovery periods the 95% LoA did not indicate any substantial consistent difference between techniques. For D100 and T100 (Fig. 5, C and D), during both the acute and long-term recovery periods the 95% LoA reported a considerable shift toward lower values in D100 compared with T100. When agreement between D10:100 and T10:100 in the acute recovery period was examined (Fig. 5E), the LoA (expressed as ratio of the 2 measures) displayed consistently greater values (i.e., underestimation of PLFFD) for D10:100 compared with T10:100. In the long-term recovery period, LoA tended toward positive values (Fig. 5F). The difference was enough that the time frame for recovery was markedly shorter for PLFFD calculated with doublets (<48 h) compared with PLFFD calculated with tetani (>96 h) (Fig. 4C).

Fig. 5.

Ninety-five percent limits of agreements (LoA) of measures calculated with pairs and trains of stimuli. In 12 participants, data points at 2, 3, and 5 min were pooled to indicate the acute recovery period (n = 36), whereas data points at 48 and 96 h collectively represented the long-term recovery period (n = 24). In A–D, the difference between doublet and tetanic responses is plotted against their average. The solid line represents the mean difference between either D10 and T10 (A and B) or D100 and T100 (C and D), whereas the dashed lines enclosing the gray area represent upper and lower limits with 95% confidence. In E and F, the ratios D10:100 and T10:100 are plotted against their average. The solid line represents the median of the ratios, whereas the dashed lines enclosing the gray area represent the upper and lower limits based on the 5th and 95th percentiles. The dotted reference lines at 0% (A–D) or 1.0 (E and F) are included to frame the data relative to the scenario of no mean or median difference between measures. A: D10 and T10 considered in the acute recovery period. B: the same measures considered in the long-term recovery period. For both time frames, the 95% LoA did not indicate any substantial consistent difference between techniques. C: D100 and T100 considered in the acute recovery period. D: the same measures considered in the long-term recovery period. During both time frames, the 95% LoA reported a considerable shift toward lower values in D100 compared with T100. E: The ratio of D10:100 to T10:100 (both expressed as % baseline) considered in the acute recovery period. F: the same measures considered in the long-term recovery period. During acute recovery, the LoA displayed consistently greater values for D10:100 than T10:100. For long-term recovery, the LoA still displayed greater values for D10:100 than T10:100, albeit reporting lower median difference than in the acute recovery period. PLFFD, prolonged low-frequency force depression.

The torque-frequency relationships for both doublets and tetani from the additional experiment are reported in Fig. 6. Like the tetanic response at 100 Hz, the peak torque of the doublet at 100 Hz resided in the plateau of the torque-frequency relationship. In accordance with the baseline D10:100 and T10:100 data (Table 1), the disparity between the torques produced at 10 Hz versus 100 Hz was considerably smaller for doublets than tetani.

DISCUSSION

The aim of the present investigation was to examine differences in the estimation of PLFFD using doublets rather than tetani in response to supramaximal stimulation of a peripheral nerve. Our research led to two principal findings: 1) the ratio of low- to high-frequency torque using doublets compared with tetani underestimates the extent of PLFFD for at least 4 days after eccentric contractions, and 2) the discrepancy between methods is due to slower recovery of the doublet than the tetanus in response to high-frequency stimulation. In light of these findings, studies using a pair rather than a train of supramaximal stimuli should acknowledge that the magnitude and duration of PLFFD are likely to be underestimated and use caution when drawing mechanistic insights from the data.

Recovery of voluntary measures.

After the fatiguing protocol, VA was lower than baseline only in the acute recovery period, whereas the impairment of MVC torque was still present at 96 h. The impairment of isometric MVC torque following eccentric contractions is regarded as the best indirect measure of muscle damage (28). Thus the MVC torque data suggest that the eccentric protocol successfully induced muscle damage, which was not resolved 96 h after the task. Despite an acute reduction in MVC torque that was very similar to the magnitude reported in previous studies involving maximal eccentric contractions of the dorsiflexors (19, 22, 23), we observed an acute reduction in VA that was not seen in those earlier studies. In the present study, the VA value of 93.2 ± 6.0% at 2 min after fatigue was driven by two participants whose VA decreased by 12% and 18%. For the remaining acute time points (i.e., 3 and 5 min), average values of VA, although statistically decreased from baseline, are similar to those previously reported, i.e., >95% (19, 22, 23).

Doublets versus trains to measure PLFFD.

After the eccentric protocol, D10 and T10 were similarly decreased from baseline throughout recovery. In contrast, T100 was impaired less and recovered faster (within 96 h) than D100, which was still impaired at 96 h. As a consequence, the magnitude of PLFFD was lower when calculated using D10:100 compared with T10:100 for both the acute and long-term recovery periods. In addition, D10:100 suggested that PLFFD was abolished by 48 h, whereas T10:100 indicated that PLFFD persisted beyond 96 h. The underestimation of PLFFD using D10:100 compared with T10:100 almost certainly relates to the much lower release of Ca²⁺ from the sarcoplasmic reticulum after a pair versus a 1-s train of stimuli at high frequencies.

After maximal eccentric exercise such as the protocol in the present study, muscle damage ensues (11, 20). This leads to decoupling at the interface between T tubules and ryanodine receptors and, in turn, to impaired sarcoplasmic reticulum Ca²⁺ release (27). At a low frequency of stimulation (e.g., 10 Hz), a substantial reduction in Ca²⁺ release for each action potential would cause a marked impairment of peak force compared with baseline, regardless of the number of pulses delivered. In contrast, at a high frequency of stimulation (e.g., 100 Hz), a reduced Ca²⁺ release per action potential would impair greatly the peak force in response to only two pulses but not a large number of pulses. In the case of a 1-s train, the 100 pulses would presumably lead to complete saturation of myoplasmic [Ca²⁺] and a peak force much closer to its control value. The failure of two pulses to elicit an adequate high-frequency response and therefore distinguish between low- versus high-frequency stimulation is supported by the baseline ratios (Table 1) and the results of the additional experiment (Fig. 6).

In humans, the importance of the number of stimuli and the subsequent Ca²⁺ release can be observed with postactivation potentiation studies (2–4), which suggest indirectly that a pair of stimuli at 100 Hz fails to saturate myoplasmic [Ca²⁺] whereas a 0.5-s train does not. The myoplasmic [Ca²⁺] discrepancy between paired stimuli and longer trains will be even greater when excitation-contraction uncoupling is present in the days after contraction-induced muscle injury. In the specific case of PLFFD, the denominator of the low-to-high-frequency ratio is supposed to represent the torque response when myoplasmic [Ca²⁺] is saturated. Only when this requirement is satisfied can the presence of PLFFD be interpreted as a reduction of Ca²⁺ release from the sarcoplasmic reticulum or myofilament sensitivity to Ca²⁺ (14, 21, 29). Hence, in the strictest sense, the ratio D10:D100 is not appropriate for the measurement of PLFFD. This conclusion is supported by the lack of correlation between D10:100 and T10:100 (expressed relative to baseline) in the long-term recovery period.

Practical considerations.

In accordance with our hypothesis, the data indicate that a doublet ratio underestimates PLFFD compared with the gold-standard measure in the minutes and days after damaging eccentric contractions of the dorsiflexors. An obvious point to consider with respect to the generalizability of our findings is the issue of muscle group. That is, because the magnitude of PLFFD after repeated eccentric tasks is muscle specific (11, 20), the extent to which a doublet ratio will underestimate the true magnitude of PLFFD may also vary among muscle groups. However, the blunted reduction of D10:D100 compared with T10:T100 in the acute recovery period of the present study (~22% vs. 43% from baseline) is equivalent to that observed by Verges and colleagues (26) for the knee extensors. Still, with data from only two studies, there is insufficient evidence to draw a definitive conclusion, so it would be prudent to investigate additional muscle groups and also different levels of muscle damage.

As a final note, given the promising results of trains of submaximal stimuli to the muscle belly (18), that technique warrants further exploration via experiments in other muscle groups and at different levels of stimulation intensity (i.e., set to elicit different percentages of MVC force). However, even submaximal muscle belly stimulation can induce substantial discomfort, so, depending on the muscle group of interest, a short series of supramaximal stimuli to the nerve may be perceived as more tolerable. Therefore, it would be advisable to conduct a systematic study that determines the minimum number of high-frequency supramaximal stimuli required to saturate myoplasmic [Ca²⁺] and thereby produce a measure of PLFFD equivalent to the gold standard.

Conclusions.

In the minutes and days after fatiguing eccentric contractions of the dorsiflexors, the 10 Hz-to-100 Hz ratio derived from a pair of supramaximal stimuli underestimated by ~50% the magnitude of PLFFD (as measured by a 1-s train of supramaximal stimuli) and prematurely indicated its recovery by at least 48 h. The discrepancy between measures arises from the greater impairment and slower recovery of high-frequency doublets compared with high-frequency tetani. Our findings make it clear that these stimulation techniques should not be considered interchangeable. If PLFFD is a variable of primary interest, trains of supramaximal stimuli should be used to ensure accurate measurement. When paired stimuli are deemed more practical than trains, it should be acknowledged that the magnitude and the duration of PLFFD will be underestimated.

GRANTS

This work was supported by the Natural Sciences and Engineering Research Council of Canada (DG 435912-2013) and the Canada Foundation for Innovation/British Columbia Knowledge Development Fund (32260).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.R., C.D.B., and C.J.M. conceived and designed research; L.R., C.D.B., P.D.C., G.U.D., and C.J.M. performed experiments; L.R., C.D.B., P.D.C., G.U.D., and C.J.M. analyzed data; L.R., C.D.B., P.D.C., G.U.D., and C.J.M. interpreted results of experiments; L.R. and C.J.M. prepared figures; L.R. drafted manuscript; L.R., C.D.B., P.D.C., G.U.D., and C.J.M. edited and revised manuscript; L.R., C.D.B., P.D.C., G.U.D., and C.J.M. approved final version of manuscript.