Eccentric-Only Versus Concentric-Only Isokinetic Strength Training Effects on Maximal Voluntary Eccentric, Concentric and Isometric Contraction Strength: A Systematic Review and Meta-analysis

Abstract

Background

Conflicting results have been reported regarding the effects of resistance exercise training with eccentric (lengthening muscle) versus concentric (shortening muscle) contractions on changes in maximal voluntary contraction (MVC) strength assessed by different contraction modes.

Objective

The main objective of this systematic review with meta-analyses was to compare the effectiveness of maximal isokinetic eccentric-only and concentric-only strength training for changes in maximal voluntary eccentric (MVC_ECC), concentric (MVC_CON), and isometric contraction (MVC_ISO) strength in healthy adults.

Methods

We conducted a systematic search in PubMed, SPORTDiscus, and Google Scholar from February to March 2024 for studies that met the following criteria: (1) randomized controlled trials; (2) inclusion of eccentric-only and concentric-only strength training groups; (3) use of an isokinetic dynamometer for training and testing; (4) reporting changes over time in MVC_CON and MVC_ECC; and (5) using healthy adult participants. The certainty of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation approach. A multilevel random-effects model meta‑analyses with robust variance estimation were performed in Rstudio software using metafor and clubSandwich packages. Moreover, sensitivity analysis was performed, excluding the highly influential studies. The potential moderating role of sex, training status and age of the participants, muscles, velocity in training and testing, initial MVC_ECC, MVC_CON, and MVC_ECC/MVC_CON ratio, and training-related variables such as number of repetitions per set, number of sets, rest period between sets, number of sessions per week, and duration of the training protocol were also assessed.

Results

Twenty-seven studies matched with the criteria, and overall 162 study results were identified and included in the meta-analyses. Greater effects on MVC_ECC were found after eccentric-only than concentric-only training (Hedges’ g: 1.51; 27 vs. 10%; p < 0.001). However, no differences were evident between the training modalities for changes in MVC_CON (Hedges’ g: –0.10; 13% vs. 14%, p = 0.726) and MVC_ISO (Hedges’ g: –0.04; 18 vs. 17%; p = 0.923). The subgroup analyses showed smaller effect of eccentric-only than concentric-only training on MVC_CON when eccentric-only training was performed at higher velocities than the velocities of MVC_CON testing (Hedges’ g: –0.99; p = 0.010). Meta-regressions showed that the longer the training period, the greater the superior effect of eccentric-only over concentric-only training on MVC_ECC.

Conclusions

Eccentric-only strength training is more effective for improving MVC_ECC, but both concentric-only and eccentric-only training provide similar effects on improving MVC_CON and MVC_ISO. Further studies are necessary to investigate the mechanisms underpinning the superior effect of eccentric-only training.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40798-025-00887-w.

Key Points

• Eccentric-only strength training is more effective than concentric-only strength training for increasing maximal voluntary eccentric contraction strength (27% vs 10%).

• The superiority of eccentric-only training for improving maximal voluntary eccentric contraction strength becomes more prominent with longer training periods, but the effect of faster velocity maximal eccentric-only training on slower maximal voluntary concentric contraction strength is limited.

• Eccentric-only and concentric-only strength training similarly increase maximal voluntary concentric and isometric contraction strength, and thus eccentric-only training appears to produce more versatile effects than concentric-only training.

• Maximal eccentric-only training methods should be actively promoted among strength and conditioning practices to effectively enhance eccentric strength and the movements dependent on it.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40798-025-00887-w.

Introduction

Muscle strength is crucial not only for athletic performance [1] but also for health and quality of life [2]. Resistance training is the most effective way to increase muscle strength, and its optimal protocols to increase muscle strength more effectively have been investigated [3]. Since neuromuscular and functional changes induced by resistance training differ in contraction modes [4], it has been a topic of interest [5]. In resistance exercises, eccentric (lengthening muscle), concentric (shortening muscle), and isometric (static) contractions are used, and all of them increase muscle strength [6]. However, it is not necessarily clear how the contraction modes used in a resistance training affect strength gains in different contraction modes after the training [7].

Nuzzo et al. [8] reported in their review article that maximal voluntary eccentric contraction strength (MVC_ECC) is 41% greater than maximal voluntary concentric contraction strength (MVC_CON) (95% credible interval: 38–44%), with a high effect of movement velocity and a low effect of age and sex on the results, along with substantial heterogeneity between joint actions (32–161%). It should be noted that approximately 98% of the comparisons were derived from the studies in which isokinetic contractions were used. Nevertheless, these findings suggest that eccentric contractions are more advantageous than concentric contractions for improving muscle strength. Eccentric contractions enable a greater mechanical stimulus to be imposed on muscles, leading to more pronounced peripheral and central adaptations that contribute to strength gains [9]. Resistance exercises involving eccentric contractions also appear to trigger mechanotransduction pathways greater [10], stimulating muscle protein synthesis and promoting greater activation of satellite cells, which also results in a greater increase in the number of sarcomeres in parallel and especially in series [11, 12]. Moreover, during eccentric contractions, passive muscular tension is generated by lengthening the extracellular matrix and titin [13]. The combined tension from contractile and noncontractile elements strengthens not only the muscle [13] but also noncontractile elements including tendons [14]. It appears that most of the strength gains in shorter than eight weeks of eccentric training are more attributable to increased neural drive [15], possibly due to the more robust downregulation of peripheral inhibitory pathways and higher activity of the central nervous system [4, 16] indicated by an increased motor unit discharge rate [17].

Due to their potent effects and the advent of new technologies [18, 19], eccentric resistance exercises including those consisting of eccentric-only contractions (concentric contractions are performed without load or with a minimal load) are becoming increasingly prevalent in strength training, rehabilitation, and injury prevention programs [20]. However, considerable methodological variations exist in the studies comparing the effects of eccentric versus concentric strength training, which makes conclusions regarding the eccentric versus concentric strength training effects challenging [5, 11]. The variability stems from the specificity of strength measurements such as one-repetition maximum and maximal voluntary contraction force/torque; training modalities (e.g., isokinetic, isoinertial, and isoweight [20]), muscles trained (e.g., upper body, lower body), and training-related variables including number of repetitions per set, sets per session, frequency of training, duration of the training protocols, and exercise tempo or velocity [7, 21, 22]. Participant characteristics such as age [23], training status, and initial strength [24], also contribute to this variability. Given that eccentric contractions have higher force generation capability and fatigue tolerance [25], optimal loading during the training should consider these characteristics.

Previous meta-analysis studies have shown favourable effects of eccentric over concentric resistance exercise training on muscle hypertrophy [5, 26]. Moreover, the meta-analysis study by Roig et al. [5] identified favourable effects of eccentric-only strength training on MVC_ECC, but did not find significant differences between the two in changes in MVC_CON and maximal voluntary isometric strength (MVC_ISO). They concluded that the better effects of eccentric-only over concentric-only training were produced by maximal but not submaximal contractions performed in the training [5]. They also stated that eccentric-only strength training adaptations were more velocity-specific. However, the authors included only two effect sizes in the subgroup analysis of velocity specific adaptations [5]. As the magnitude of MVC_ECC to MVC_CON ratio is known to be influenced by factors such as age, muscle groups assessed, testing velocity, and sex [8], and while different training modalities (e.g., isokinetic dynamometry, free weights) were used for training and testing in various studies included in the review, these could have affected the overall heterogeneity of the results. These should be considered to compare eccentric-only and concentric-only resistance training for their effects on muscle strength.

Eccentric training has been proposed to improve muscle mechanical function to a greater extent than other contraction modes [5, 9, 11, 12, 27]; however, the quantification of this effect remains unknown. To investigate the effects of the muscle contraction mode on changes in muscle strength, we thought that the most reliable way would be to perform meta-analyses of existing studies which compared maximal eccentric-only and concentric-only strength training using an isokinetic dynamometer for changes in MVC_ECC, MVC_CON, and MVC_ISO strength. Therefore, the main aim of the present systematic review and meta-analyses was to examine the hypothesis that eccentric-only strength training would provide superior effects on not only MVC_ECC but also MVC_CON and MVC_ISO strength when compared with concentric-only strength training. Additionally, changes in MVC_CON, MVC_ECC, and MVC_ISO strength were compared within eccentric-only and concentric-only maximum isokinetic contraction training groups to evaluate the transferability of the strength training on strength measures in different contraction modes. We hypothesized that eccentric-only strength training would produce greater increases in not only MVC_ECC strength but also MVC_CON and MVC_ISO strength when compared with concentric-only strength training, and thus eccentric-only strength training would produce greater transferable strength effect on different contraction modes.

Methods

Study Design

This systematic review followed the “Preferred Reporting Items for Systematic Reviews and Meta-Analyses” guidelines [28]. A review protocol was not pre-registered for this review.

Search Strategy

To identify all potentially relevant data from the experimental studies, initial systematic literature searches were conducted between February and March in 2024, with no limitations based on publication date. Searches included the following databases: MEDLINE/PubMed, SPORTDiscus, and Google Scholar. Moreover, we used “snowballing” strategies (i.e., reference screening from most relevant studies and citation tracking using a scholarly publication discovery tool supported by artificial intelligence, Research Rabbit [29]) as described by Greenhalgh and Peacock [30]. Electronic databases were searched using the combination of the following search terms: (eccentric or lengthening) AND (shortening or concentric) AND (exercise or training) AND (strength or torque or force or isokinetic or isometric or muscle).

Data Extraction

Articles were screened following a three-stage process: (1) duplicates of articles identified across numerous search databases were removed; (2) article title and abstracts were screened for suitability. Where a definitive decision could not be made at this stage, studies were taken forward for a full study review; and (3) full articles were screened according to the inclusion and exclusion criteria. All of these were done by one of the authors (DS).

Eligibility Criteria

Studies were considered to be eligible for inclusion according to the PICOS criteria (Participants, Intervention, Comparator, Outcome, and Study design). The articles that used healthy (i.e., the absence of injury or illness) adult (i.e., 18–64 years) human participants were chosen. Only randomized controlled exercise intervention studies including both eccentric-only (intervention) and concentric-only strength training protocols (comparator) were included. Training and testing were required to be performed on an isokinetic dynamometer and involved exclusively eccentric or concentric contractions in the training protocols. The primary outcomes were percentage changes (mean and standard deviation) in MVC_CON and MVC_CON before and after training. The secondary outcome was percentage change in MVC_ISO, if reported. Only studies with randomized design and published in peer-reviewed journals were qualified to be included. If the aforementioned criteria were not fulfilled, or if the strength training protocol was not defined appropriately, or if strength measures were taken outside the training joint range of motion area [31, 32], they were excluded. Studies from which we could not extract enough information to calculate the effect size and include them in the qualitative data synthesis were also excluded. Data were extracted by one investigator (DS). Consensus or arbitration by a second investigator (KN) was used to settle any disputes [28].

If a study did not report percent changes and standard deviation of the change within both training groups (for example only means and standard deviations for pre- and post-intervention within each group were reported or only raw change with SD was reported), we calculated the percent change and SD of the percent change relative to the baseline raw value within each group. Finally, for papers in which data were presented in figures, muscle strength values were estimated using a graph digitizer (WebPlotDigitizer, https://apps.automeris.io/wpd/).

In accordance with the recommendation for meta-analysis [33], in case of incomplete data, we calculated missing change score SDs (SDchange) from SD values at baseline (SDpre) and postintervention (SDpost) using the following formula: √((SDpre^2/N) + (SDpost^2/N)); where N was the number of participants. In one study only change over time in raw values was reported [34] for training groups and a control group (which did not train). In this case, percent improvements of the eccentric training and concentric training groups were calculated relative to control group results using the formula ((eccentric or concentric–control group)/control group)*100.

In addition to main outcomes, the extracted data from papers included study type (randomized controlled trial or cross design randomized controlled trial); sample size (for eccentric and concentric training groups), sex of the participants (male, female, mixed sample); age of the participants (years), training status of the participants (untrained, moderately trained, trained); muscles trained and tested; testing and training isokinetic velocity (°/s); joint angle of strength assessment (only in the case of MVC_ISO; in degrees [°]); number of repetitions per set; number of sets per session; rest period between sets [seconds], training frequency (number of sessions per week); duration of the training (in weeks) and participants’ initial MVC_ECC and MVC_CON [force or torque, not normalized to body mass]. Additionally, initial eccentric to concentric strength ratio was calculated from the initial MVC_ECC and MVC_CON data.

Methodological Quality Assessment

The Physiotherapy Evidence Database scale (PEDro), TIDieR (Template for Intervention Description and Replication) and, finally, Grading of Recommendations Assessment, Development and Evaluation (GRADE) checklists were used to assess the risk of bias, completeness of intervention descriptions and quality of evidence, respectively. Using the PEDro scale [35], the listed studies’ methodological quality was evaluated by the investigator (DS). The PEDro scale consists of 11 items designed to assess methodological quality. Studies were categorized as at low risk (≥ 6 points), moderate risk (4–5 points), and high risk (≤ 3 points) of bias. The intervention descriptions were evaluated for completeness using the TIDieR checklist [36]. Finally, the GRADE approach was used to evaluate the overall quality of the evidence [37, 38]. Quality assessment was performed separately for each meta-analysis (Table S3 of the Electronic Supplementary Material [ESM] 1). High quality of evidence was initially assumed and then downgraded based on the following criteria: a) risk of bias: downgraded by one level if the median PEDro score was indicative of moderate risk (4 or 5 points) or two levels if score was indicative of high risk (< 4 points), b) inconsistency: downgraded by one level if the Cochrane Q test for heterogeneity was significant (p < 0.05) or total I² exceeded 50%; c) indirectness: in all cases considered at low risk, because the PICOS criteria were ensured; d) imprecision [39]: downgraded by one level if the 95% confidence interval crossed the threshold for a small effect size [− 0.2 to 0.2] when comparing training modes, or [− 5 to + 5%] for within training mode strength gains; both of which we have considered to be practically important strength changes over time, and d) publication bias: downgraded by one level if Egger’s test, assessing the asymmetry in the funnel plot, was significant (p < 0.10). The level of certainty was considered as high (considerable confidence exists that the true effect is similar to the estimated effect), moderate (true effect is probably close to the estimated effect), low (true effect might be markedly different from the estimated effect), or very low (true effect is probably markedly different from the estimated effect).

Statistical Analyses

While some studies provided multiple outcomes, multilevel random-effects model meta‑analyses with robust variance estimation were performed to control for dependent effect sizes in a meta-regression models [40, 41]. Following the main objective of our research, MVC_ECC, MVC_CON and MVC_ISO changes following the eccentric-only and concentric-only training protocols were compared in separate meta-analyses. Standardized mean differences as Hedges’ g (effect size corrected for small sample sizes [33]) with 95% CIs between eccentric-only training and concentric-only training protocols were calculated for individual studies and to test for overall effect. The following categories were used to categorize the size of the effects: trivial (< 0.20), small (0.21–0.60), moderate (0.61–1.20), large (1.21–2.00), very large (2.01–4.00), and extremely large (> 4.00) [42]. Forest plots were displayed to graphically represent effects of each individual study, the magnitude of overall effect and its direction (favouring eccentric or concentric training).

To aid interpretation of the findings, separate multilevel random-effects model meta‑analyses of single means with robust variance estimation were used to synthesize the percent (%) outcomes from individual studies for improvements of MVC_ECC, MVC_CON and MVC_ISO after eccentric-only and concentric-only training protocols. Percent improvements of MVC_ECC, MVC_CON and MVC_ISO were additionally compared within the training mode by pairwise meta‑analyses (improvement of MVC_ECC vs. MVC_CON vs. MVC_ISO, separately for eccentric and concentric training, respectively) to observe the effect of mode-specificity of the training.

Publication bias was evaluated with observing asymmetry of the funnel plots by calculating the Egger’s statistics. A substantial publication bias was considered to be present when the p-value was less than 0.10. Moreover, heterogeneity was investigated using the Cochrane Q-test (Chi² statistics), σ² test (sigma²; variance normalized to effect sizes) and the I² (relative measure of heterogeneity among studies; %). Values of 25, 50, and 75% for I² signified low, moderate, and high statistical heterogeneity [43]. The heterogeneity was additionally partitioned across two levels (i.e., Level 2: within-study and Level 3: between-study heterogeneity).

For all meta-analytic models, influential case diagnostics were performed to identify studies that had a large influence on the overall effect size. Cook’s distance, which combines information about both the leverage and outliers’ impact on the analysis was calculated to identify if a particular study effect size had a potential effect on the estimated coefficients. Individual cases were red flagged if Cook’s distance’s values exceeded more than three times their respective mean. Robustness of each meta-analysis model was checked with sensitivity analysis, excluding red-flagged studies from the analysis.

To explain the variation of the effects, subgroups analyses were performed for categorical variables: sex [male, female, mixed]; muscle trained [upper body, lower body]; training status of the participants [untrained, moderately trained, highly trained] and training to testing isokinetic velocity [with four categories: training at lower velocity than the testing, training at higher velocity than the testing, training at the same velocity as the testing and velocity spectrum pyramidal ordering concept training] (please see the notes in Table 1 for further explanation). Moreover, to explain the variation of the effects by continuous moderator variables, random-effects meta-regressions were performed. These moderators were separated into participant-related (age of the participants, initial MVC_CON, initial MVC_ECC and initial MVC_ECC/MVC_CON ratio) and into training-related moderators (number of repetitions per set, number of sets, number of training sessions per week and duration of the resistance training protocol in weeks). For training-related moderators, multiple meta-regression was performed to distinguish the effect of each training-related variable on the effect size while holding other training-related variables constant. The estimated proportional reduction in the total variance was computed using the variance accounted for, a pseudo R² value (i.e., the amount of heterogeneity accounted for by the moderator(s)).

Table 1.

Summary of the studies with quality evaluation and reporting completeness

Study	Measures	Sample characteristics	Muscle groups	Training protocol (Sets × repetitions and training velocity, rest, frequency, duration)	Risk of biasa	TIDieRb
Akınoğlu et al. [45]	MVCCON and MVCECC at 30, 60, 90 120, 150 and 180°/s	mean age = 32	Knee extensors, knee flexors	c3 × 10 at 60°/s and 3 × 15 at 180°/s,	Low	6
		Untrained		NR, 3, 6
		Sex = MF
		ECC training n = 14
		CON training n = 14
Barak et al. [46]	MVCCON and MVCECC at 30, 90°/s	mean age = 24	Knee extensors	4 × 10 at 30°/s or 90°/s,	low	6
	MVCISO at 45° knee flexion	Moderately trained		1, 3, 6
		Sex = F
		Velocity 30°/s:
		ECC training n = 14
		CON training n = 13
		Velocity 90°/s:
		ECC training n = 13
		CON training n = 14
Benford et al. [47]	MVCCON and MVCECC at 30°/s	mean age = 23	Knee extensors	4 × 8 at 30°/s,	low	9
	MVCISO at 30 and 90° knee flexion	Moderately trained		1, 2, 5
		Sex = M
		ECC training n = 8
		CON training n = 8
Blazevich et al. [48]	MVCCON and MVCECC at 30°/s	mean age = 24	Knee extensors	4–6 × 6 at 30°/s,	Low	6
		Moderately trained		1, 3, 10
		Sex = MF
		ECC training n = 12
		CON training n = 12
Cadore et al. [49]	MVCCON and MVCECC at 60°/s	mean age = 23	Knee extensors	2–5 × 8–12 at 60°/s,	Low	9
	MVCISO at 60° knee flexion	Moderately trained		NR, 2, 6
		Sex = MF
		ECC training n = 11
		CON training n = 11
Duncan et al. [50]	MVCCON and MVCECC at 60, 120, 180°/s	mean age = 24	Knee extensors	1 × 10 at 120°/s,	Moderate	6
		Status NR		/, 3, 6
		Sex = M
		ECC training n = 16
		CON training n = 14
Ellenbecker et al. [51]	MVCCON and MVCECC at 60, 180, 210°/s	mean age = NR	Shoulder internal rotators,	c2 × 10 at 60°/s, 2 × 10 at 180°/s and 2 × 10 at 210°/s,	Moderate	6
		Trained	shoulder external rotators	NR, 2, 6
		Sex = MF
		ECC training n = 11
		CON training n = 11
Farthing and Chilibeck [52]	MVCCON and MVCECC collapsed between 30 and 180°/s	mean age = 20	Elbow flexors	2–6 × 8 at 30°/s or 180°/s,	Low	8d
		Untrained		1, 3, 8
		Sex = MF
		ECC training n = 13
		CON training n = 13
Higbie et al. [17]	MVCCON and MVCECC at 60°/s	mean age = 20	Knee extensors	3 × 10 at 60°/s,	Low	7
		Moderately trained		3, 3, 10
		Sex = F
		ECC training n = 19
		CON training n = 16
Hilliard-Robertson et al. [53]	MVCCON and MVCECC at 60°/s	mean age = 33	Knee extensors	4 × 10 at 60°/s,	Low	6
		Moderately trained		NR, 3, 5
		Sex = MF
		ECC training n = 11
		CON training n = 11
Hortobágyi et al. [54]	MVCCON and MVCECC at 60°/s	mean age = 22	Knee extensors	4–6 × 8–12 at 60°/s,	low	7
	MVCISO at 45° knee flexion	Untrained		1, 3, 12
		Sex = M
		ECC training n = 7
		CON training n = 8
Hortobágyi et al. [55]	MVCCON and MVCECC at 60°/s	mean age = 22	Knee extensors	4–6 × 8–12 at 60°/s,	Moderate	6
	MVCISO at 45° knee flexion	Moderately trained		1, 3, 12
		Sex = MF
		ECC training n = 12
		CON training n = 12
Kim et al. [56]	MVCCON and MVCECC at 60°/s	mean age = 28	Shoulder abductors	4–6 × 6–8 at 60°/s,	Low	7
	MVCISO at 60° shoulder abduction	Status NR		1, 3, 8
		Sex = MF
		ECC training n = 7
		CON training n = 6
Paschalis et al. [57]	MVCCON and MVCECC at 60°/s	mean age = 21	Knee extensors	5 × 15 at 60°/s,	Low	7
	MVCISO at 90° knee flexion	Moderately trained		2, 1, 8
		Sex = F
		ECC training n = 10
		CON training n = 10
Ruas et al. [58]	MVCCON and MVCECC at 60°/s	mean age = 23	Knee extensors,	e1-6 × 10 at 60°/s, 90°/s, 120°/s°/s, 150, 180°/s or 210°/s,	Low	7
	MVCISO at 60° knee flexion	(2 ECC groups and 1 CON group for knee flexors; 2 CON groups and 1 ECC group for knee extensors)	knee flexors	NR, 2, 6
		Untrained
		Sex = M
		ECC training all n = 10
		CON training all n = 10
Ruas et al. [59]	MVCCON and MVCECC at 60°/s	mean age = 23	Knee extensors,	e1-6 × 10 at 60°/s, 90°/s, 120°/s°/s, 150, 180°/s or 210°/s,	Low	6
		(2 ECC groups and 1 CON group for knee flexors; 2 CON groups and 1 ECC group for knee extensors)	knee flexors	NR, 2, 6
		Untrained
		Sex = M
		ECC training all n = 10
		CON training all n = 10
Miller et al. [60]	MVCCON and MVCECC at 60°/s	mean age = 38	Knee extensors, knee flexors	1–5 × 6 at 60°/s,	Low	8
		Moderately trained		1, 3, 20
		Sex = F
		ECC training n = 17
		CON training n = 21
Miller et al. [61]	MVCCON and MVCECC at 60°/s	mean age = 20	Collapsed elbow extensors and flexors	1–5 × 6 at 60°/s,	Low	7
		Status NR		NR, 3, 20
		Sex = F
		ECC training n = 32
		CON training n = 22
Moore et al. [62]	MVCCON and MVCECC at 45 and 300°/s	mean age = 22	Elbow flexors	NR × NR at 45°/s,	Low	8
	MVCISO at 60° elbow flexion	Moderately trained		NR, 2, 9
		Sex = M
		ECC training n = 9
		CON training n = 9
Nickols-Richardson et al. [63]	MVCCON and MVCECC at 60°/s	mean age = 20	Collapsed knee extensors, flexors and	1–5 × 6 at 60°/s,	Low	8
		Moderately trained	Collapsed elbow extensors and flexors	1, 3, 20
		Sex = F
		ECC training n = 33
		CON training n = 37
Sato et al. [64]	MVCCON and MVCECC at 30 and 180°/s	mean age = 22	Elbow flexors	1 × 1 at 30°/s,	Low	8
	MVCISO at 20, 55 and 90° elbow flexion	Moderately trained		/, 5, 4
		Sex = MF
		ECC training n = 13
		CON training n = 13
Seger et al. [65]	MVCCON and MVCECC at 30, 90 and 270°/s	mean age = 25	Knee extensors	4 × 10 at 90°/s,	Low	6d
	MVCISO at 60° knee flexion	Moderately trained		2, 3, 10
		Sex = M
		ECC training n = 10
		CON training n = 10
Seger and Thorstensson [66]	MVCCON and MVCECC at 30, 90, 270°/s	mean age = 25	Knee extensors	4 × 10 at 90°/s,	Low	6
		Moderately trained		2, 3, 10
		Sex = M
		ECC training n = 5
		CON training n = 5
Sharma et al. [67]	MVCCON and MVCECC at 60°/s	mean age = 26	Knee flexors	3 × 10 at 60°/s,	Low	10
		Moderately trained		1, 2, 6
		Sex = M
		ECC training n = 15
		CON training n = 15
Takayanagi et al. [68]	MVCCON and MVCECC at 60, 120, 180, 240°/s	mean age = 20	Knee extensors,	5–8 × 10 at 180°/s,	Low	6
		Moderately trained	knee flexors	2, 3, 6
		Sex = MF
		ECC training n = 10
		CON training n = 10
Timmins et al. [69]	MVCCON and MVCECC at 60 and 180°/s	mean age = 22	Knee flexors	c2-3 × 6–8 at 60 and 2–3 × 6–8 at 90°/s,	Low	9d
		Moderately trained		0.5, 2–3, 6
		Sex = M
		ECC training n = 14
		CON training n = 14
Tomberlin et al. [70]	MVCCON and MVCECC at 100°/s	mean age = 27	Knee extensors	3 × 10 at 100°/s,	Low	7
		Status NR		3, 3, 6
		Sex = MF
		ECC training n = 21
		CON training n = 19

CON Concentric, ECC Eccentric, F Female, ISO Isometric, M Male, MF mixed sex sample, MVC Maximal Voluntary Contraction, NR Not Reported, collapsed indicates the averaged MVC values for two muscle groups as an outcome, rest time period between sets in minutes, frequency number of training sessions per week, duration of training period duration in weeks

^aThe Physiotherapy Evidence Database scale results, assessing risk of bias (grade on the scale to 11)

^bTemplate for Intervention Description and Replication completeness (grade on the scale to 11)

^{c, e}Refers to training conditions where sets within one training session were performed at different velocities (^c), or the velocity of training contractions changed from low to high or high to low throughout the training period (^e); both approaches are referred to as the velocity spectrum concept

^dCrossover study design where each participant was randomly assigned to a sequence of eccentric and concentric trainings (including a washout periods)

Meta-analyses were performed in the RStudio: Integrated Development Environment for R (v4.3.3.; Posit team [2024], Boston, MA; http://www.posit.co/, accessed in April 2024). The robust variance estimation method was implemented using the clubSandwich package. Sampling variance–covariance matrix was prepared with estimating 0.6 degree of correlation between sample variances of different outcomes within a study. The matrix was then included into the metafor package. Moreover, confidence intervals of robust meta-analyses were obtained by adjusting for small samples. The cut-off for statistical significance was set at p < 0.05 [42].

Results

Search Results

The initial search yielded 6,614 studies, and we observed that using different search terms to narrow them down could overlook some of the of the most relevant research. Therefore, we adopted a more conservative approach. Identified records were filtered using keywords in titles and abstracts through the systematic review software Rayyan [44] (https://www.rayyan.ai/, accessed in March 2024). Additionally, reference screening was performed on the most relevant studies, and citation tracking was conducted using Research Rabbit software [29] (https://www.researchrabbit.ai/, accessed in March 2024), which yielded an additional 50 potentially useful studies. Our systematic review and meta-analysis ultimately included 27 studies. Of these, 11 studies reported changes in all three strength measures (MVC_ECC, MVC_CON, and MVC_ISO) following both concentric-only and eccentric-only training. However, 16 studies did not report changes in MVC_ISO. In total, we gathered 162 study results (expressed as changes in percent units). This enabled us to calculate 71 standardized effect sizes for differences in improvements of MVC_ECC and MVC_CON between both training groups (71 results in each group) and 20 standardized effect sizes for differences in improvements of MVC_ISO between training groups (20 results in each group). To aid interpretation of the findings, results of multiple studies within each training group and contraction mode were summarized and compared in separate meta-analyses. The stages of the search and study selection process are presented in Fig. 1.

Fig. 1.

Literature search flow chart. n number of studies

Study Characteristics

Individual study characteristics are presented in Table 1, and the raw data included in the analyses are provided in ESM 2. Summary of study characteristics reporting MVC_CON and MVC_CON by categorical subgroup variables are presented in Table 2. The summary of study characteristics by categorical subgroup variables for the studies reporting MVC_CON, MVC_CON and additionally MVC_ISO are presented in Table 3. Additionally, in studies reporting MVC_CON and MVC_ECC (Table 2), the number of repetitions per set ranged from 1 to 15 (mode = 10; mean ± SD = 9.6 ± 2.4), the number of sets ranged from 1 to 7 (mode = 3; mean ± SD = 4.7 ± 1.6), rest periods between sets ranged from 0.5 to 3 min (mode = 1; mean ± SD = 1.4 ± 0.7), the number of training sessions per week ranged from 1 to 5 (mode = 3; mean ± SD = 2.8 ± 0.6) and training protocol duration ranged from 4 to 20 weeks (mode = 6; mean ± SD = 7.8 ± 3.8). The mean age of the participants was 25 years (SD = 4.7; range 20–38). The number of participants was 364 for the eccentric training group and 354 for the concentric training group. When summarizing multiple results from the same studies, the totals were 904 and 899, respectively.

Table 2.

Summarized study characteristics comparing changes in maximal voluntary concentric (MVC_CON) and eccentric contraction (MVC_ECC) strength between eccentric and concentric training protocols

Subgroup variable	Category	n studies (outcomesa)	outcomesa (%)
Sex	Only men	9 (24)	34
	Only women	7 (11)	15
	Mixed sample	11 (36)	51
Status	Untrained	5 (23)	32
	Moderately	18 (36)	52
	Trained	1 (6)	8
	Not reported	4 (6)	8
Muscle lower body	All lower body	20 (56)	79
	Knee extensors	17 (37)	66
	Knee flexors	3 (18)	32
	Knee flexors and extensors	1 (1)	2
Muscle upper body	All upper body	7 (15)	21
	Elbow flexors only	3 (6)	40
	Elbow flexors and extensors	2 (2)	13
	Shoulder rotators	1 (6)	40
	Shoulder abductors	1 (1)	7
Isokinetic training to testing velocity (°/s)	Low to highb	8 (10)	14
	High to lowc	7 (10)	14
	The samed	22 (25)	35
	Velocity spectrume	4 (26)	37

^aNumber of comparisons between eccentric-only and concentric-only training results

^bTraining at lower velocity than testing

^cTraining at higher velocity than testing

^dTraining and testing at same isokinetic velocity

^eVelocity spectrum pyramidal ordering concept training

Table 3.

Summarized study characteristics comparing changes in maximal voluntary concentric (MVC_CON), eccentric (MVC_ECC) and isometric (MVC_ECC) strength between eccentric and concentric training protocols

Subgroup variable	Category	n Studies (outcomesa)	Outcomesa (%)
Sex	Only men	5 (9)	45
	Only women	2 (5)	25
	Mixed sample	4 (6)	30
Status	Untrained	2 (5)	25
	Moderately	8 (14)	70
	Not reported	1 (1)	5
Muscle	All lower body	8 (15)	75
lower body	Knee extensors	8 (13)	87
	Knee flexors	1 (2)	13
Muscle	All upper body	3 (5)	25
upper body	Elbow flexors	2 (4)	80
	Shoulder abductors	1 (1)	20
Isokinetic training	30	3 (7)	35
velocity (°/s)	45	1 (1)	5
	60	5 (5)	25
	90	2 (3)	15
	Velocity spectrumb	1 (4)	20

^aNumber of comparisons between eccentric-only and concentric-only training results

^bVelocity spectrum pyramidal ordering concept training

In the studies where MVC_ISO was reported (Table 3), the number of repetitions per set ranged from 1 to 15 (mode = 10; mean ± SD = 9.0 ± 3.1), the number of sets ranged from 1 to 6 (mode = 3; mean ± SD = 3.7 ± 1.2), rest periods between sets ranged from 0.5 to 3 min (mode = 1; mean ± SD = 1.18 ± 0.4), the number of training sessions per week ranged from 1 to 5 (mode = 3; mean ± SD = 2.9 ± 0.9) and training protocol duration ranged from 4 to 12 weeks (mode = 6; mean ± SD = 6.4 ± 2.0). The mean age of the participants was 23.3 years (SD = 1.3; range 21–28). The number of participants was 111 for the eccentric training group and 110 for the concentric training group. When summarizing multiple results from the same studies, the totals were 215 and 215, respectively.

Quality of Evidence and Completeness of Reporting

PEDro scale values and completeness of reporting of the controlled randomized study items are presented for each particular study in Table 1. PEDro scores ranged from 5 to 9 (mode = 6; mean ± SD = 6.3 ± 1.0) indicating low to moderate risk of bias (Table S1 of the ESM 1). As shown in Fig. 2, all studies reported the execution of the testing and training procedures, but only 19% reported who performed the training or testing protocol, 44% clarified where the training and testing was performed, 15% clarified if sample size was calculated, and 37% of the studies reported information regarding the dropout of the participants. None of the studies reported tailoring and/or modifications of training protocols (Table S2 of the ESM 1).

Fig. 2.

Percentage of studies achieving each Template for Intervention Description and Replication (TIDieR) checklist item The checklist key components: name a brief name to easily identify the intervention, why the rationale or theory behind the study, what (materials) detailed descriptions of the materials and procedures, what (procedures) detail descriptions of the intervention procedures, how modes of training delivery, where the setting of the intervention, when and how much the timing and quantity of the intervention delivery, how well (actual) reporting the intervention's actual fidelity, who descriptions of the qualifications and background of the intervention providers, how well (planned) planning the intervention's fidelity, tailoring considerations for the tailoring of the intervention to individual needs, modifications any modifications made during the study

Altogether, due to considerable heterogeneity (Q test results) and publication bias (Egger’s statistics; further reported in the text following the results of the particular meta-analysis and summarized in Table S3 of the ESM 1), quality of evidence was downgraded from high quality to low quality according to GRADE approach [38] for MVC_ECC results. Additionally, due to imprecision (confidence interval was crossed by a small effect size), MVC_CON and MVC_ISO results were downgraded to very low quality.

Meta-analyses Results

Effect of Different Resistance Training Modes on Eccentric Strength Gain (MVCECC)

A meta-analysis of 27 studies with 71 comparisons showed statistically significantly beneficial effects of eccentric-only in comparison to concentric-only strength training for improvement of MVC_ECC (Hedges’ g = 2.03, 95% CI 0.74–3.32; p < 0.01; very large effect) (Figure S1 of the ESM 1). A small sample adjustment to the robust meta-analysis results expanded the confidence intervals of Hedges’ g to 0.69–3.39. Five individual effect sizes from two studies were identified as highly influential. However, the overall results were robust to their exclusion from the model as the interpretation of the model did not change. Overall, Hedges’ g decreased to 1.51 (95% CI 0.59–2.42; p < 0.01; large effect), with the small sample adjustment for the 95% CI being 0.55–2.47 (Fig. 3).

Fig. 3.

Meta-analysis results with forest plot for changes in maximal voluntary eccentric contraction (MVC_ECC) strength following eccentric-only versus and concentric-only training. Study characteristics in brackets: muscle trained and tested [KE knee extensors, KF knee flexors, EF elbow flexors, EE elbow extensors, SER shoulder external rotarors, SIR shoulder internal rotators, SABD shoulder abductors], sex [M male, F female, M + F mixed sample], participants training status [UN untrained, MO moderately trained, TR highly strength trained, NR not reported], velocity [training velocity – testing velocity, VS velocity spectrum pyramidal ordering concept training]. ^a and ^b present outcomes from two eccentric isokinetic training groups from the same study

Egger’s test results indicated publication bias for the MVC_ECC (p < 0.01) meta-analysis indicating smaller studies showing higher benefits in favour of eccentric or concentric training, respectively. Moreover, statistically significant overall heterogeneity among the studies was found (Q = 647.9, df = 65; p < 0.01; I² = 96.5%). Within study effect size variability (Level 2) was low to moderate (29%), while between study variability (Level 3) was moderate to high (68%).

Effect of Resistance Training Modes on Concentric Strength Gain (MVCCON)

A meta-analysis of 27 studies with 71 comparisons did not show statistically significantly different benefits of eccentric-only and concentric-only strength training for improvement of MVC_CON (Hedges’ g = –0.71, 95% CI –1.65 to 0.22; p = 0.13; small effect) (Figure S2 of the ESM 1). A small sample adjustment to the robust meta-analysis results expanded the confidence intervals of Hedges’ g to –1.69 to 0.27. Five individual effect sizes from three studies were identified as highly influential. However, the overall results were robust to their exclusion from the model as the interpretation of the model did not change. Overall, Hedges’ g decreased to trivial, i.e. –0.10 (95% CI –0.69 to 0.48) with small sample adjustment 95% CI ranging from –0.72 to 0.51 (Fig. 4).

Fig. 4.

Meta-analysis results with forest plot for changes in maximal voluntary concentric contraction (MVC_CON) strength following eccentric-only versus concentric-only training. Study characteristics in brackets: muscle trained and tested [KE knee extensors, KF knee flexors, EF elbow flexors, EE elbow extensors, SER shoulder external rotarors, SIR shoulder internal rotators, SABD shoulder abductors], sex [M male, F female, M + F mixed sample], participants training status [UN untrained, MO moderately trained, TR highly strength trained, NR not reported], velocity [training velocity – testing velocity, VS velocity spectrum pyramidal ordering concept training]. ^a and ^b present outcomes from two eccentric isokinetic training groups from the same study

Egger’s test results indicated publication bias for the MVC_CON (p < 0.01) meta-analysis indicating smaller studies showing higher benefits in favour of eccentric or concentric training, respectively. Moreover, statistically significant overall heterogeneity among the studies was found (Q = 358.4, df = 65; p < 0.01; I² = 92%). Within study effect size variability (Level 2) was low (21%), while between study variability (Level 3) was moderate to high (71%).

Effect of Resistance Training Modes on Isometric Strength Gain (MVCISO)

A meta-analysis of 11 studies with 20 comparisons did not show statistically significantly different effects between eccentric and concentric resistance training protocols for improvement of MVC_ISO (Hedges’ g = –0.31, 95% CI –2.40 to 1.75; p = 0.77; small effect) (Figure S3 of the ESM 1). A small sample adjustment to the robust meta-analysis results expanded the confidence interval (Hedges’ g = –0.31, 95% CI –2.65 to 2.02). Two individual effect sizes from two studies were identified as highly influential. After removing them from the analysis, the overall effect size changed sign and narrowed the confidence intervals from negative (favouring concentring training) to positive trivial (Hedges’ g: 0.04 with 95% CI –0.82 to 0.90), favouring eccentric training (Fig. 5). Confidence intervals (95%) for robust meta-analysis with small samples adjustment ranged from –0.96 to 1.05.

Fig. 5.

Meta-analysis results with forest plot for changes in maximal voluntary isometric contraction (MVC_ISO) strength following eccentric-only versus concentric-only training. Study characteristics in brackets: muscle trained and tested [KE knee extensors, KF knee flexors, EF elbow flexors, EE elbow extensors, SER shoulder external rotarors, SIR shoulder internal rotators, SABD shoulder abductors], sex [M male, F female, M + F mixed sample], participants training status [UN untrained, MO moderately trained, NR not reported]), velocity-angle [training velocity – joint angle of strength assessment, VS velocity spectrum pyramidal ordering concept training]. ^a and ^b present outcomes from two eccentric isokinetic training groups from the same study

Egger’s test results indicated publication bias for the MVC_ISO (p < 0.01) meta-analysis indicating smaller studies showing higher benefits in favour of eccentric or concentric training, respectively. Moreover, statistically significant overall heterogeneity among the studies was found (Q = 120.7, df = 25; p < 0.01; I² = 89%). Within study effect size variability (Level 2) was low (11%), while between study variability (Level 3) was moderate to high (71%).

Analyses of Moderators

Despite high statistical heterogeneity among the studies comparing the magnitude of increase of MVC_ECC between eccentric and concentric training, no statistically significant differences were found within subgroups of sex (Chi² = 3.2; p = 0.20; pseudo R² = 2.7%), muscle (Chi² = 0.59; p = 0.44; pseudo R² = 6%), and training status (Chi² = 0.89; p = 0.83; pseudo R² = 18%). Effect of eccentric-only training was more superior when testing was performed at the same velocity as training in comparison to when testing was performed at the lower velocity as the training (difference of 1.53 in Hedges’ g, p < 0.05), nevertheless training to testing isokinetic velocity subgroups together could not explain the variability of the effect (Chi² = 7.2; p = 0.07; pseudo R² = 1%). Participant-related continuous moderators analyses showed no effect of age (Chi² = 3.0; p = 0.08; pseudo R² = 12.7%), initial MVC_ECC (Chi² = 0.16; p = 0.69, pseudo R² = 10%), initial MVC_CON (Chi² = 0.55; p = 0.46, pseudo R² = 12%) and MVC_ECC/MVC_CON ratio (Chi² = 2.05; p = 0.15; pseudo R² = 12%). Moreover, training related variables included in multiple meta-regression (number of repetitions, sets, frequency and duration of training), could not statistically significantly explain the variability (Chi² = 8.2; p = 0.08; pseudo R² = 19.5%). Within training-related factors, only duration of the training protocol had shown a statistically significant influence on the effect size (β = 0.25 [95% CI 0.05–0.45]; SE = 0.10; z = 2.4; p < 0.05). Thus, the longer the training protocol, the more beneficial the eccentric training was over concentric training for improving MVC_ECC when controlling for the rest of training-related factors (Table S4 of the ESM 1). As rest periods between sets were reported for only 35 out of 71 outcomes, and 90% of these indicated durations of one or two minutes, they were not included in the multiple meta-regression analysis of moderators.

Comparing the magnitude of increase of MVC_CON between eccentric and concentric training, no statistically significant differences were found between subgroups of sex (Chi² = 3.90; p = 0.14; pseudo R² = 4.4%), muscle (Chi² = 0.00; p = 0.98; pseudo R² = 10%) and training status (Chi² = 1.89; p = 0.60; pseudo R² = 21%). Statistically significant differences were found within subgroups of training to testing isokinetic velocity (Chi² = 8.4; p < 0.05; pseudo R² = 49%). A statistically significant lower effect of eccentric training compared to concentric training on the improvement of MVC_CON was observed when the eccentric training was performed at higher velocities than those used in the MVC_CON testing in comparison to the effect of the same training and testing velocity (Hedges’ g = –0.99 [95% CI from –1.75 to –0.23]; p = 0.01). Participant-related continuous moderators showed no effect of age (Chi² = 0.68; p = 0.41; pseudo R² = 13%), initial MVC_ECC (Chi² = 0.08; p = 0.77; pseudo R² = 14.5%), initial MVC_CON (Chi² = 0.19; p = 0.66; pseudo R² < 19.7%) and MVC_ECC/MVC_CON ratio (Chi² = 1.84; p = 0.17; pseudo R² = 5.2%). Moreover, training related variables (number of repetitions, sets, frequency and duration of training), could not statistically significantly explain the variability (Chi² = 5.7; p = 0.22; pseudo R² = 40.7%) (Table S5 of the ESM 1). Rest periods between sets were once again excluded from the multiple meta-regression analysis of moderators.

Among the studies comparing the magnitude of increase of MVC_ISO between eccentric and concentric training, no statistically significant differences were found within subgroups of sex (Chi² = 0.02; p = 0.99; pseudo R² = 47%), muscle (Chi² = 0.20; p = 0.66; pseudo R² = 18%) and training status (Chi² = 2.85; p = 0.24; pseudo R² = 9%). Participant-related continuous moderators showed no effect of age (Chi² = 0.3; p = 0.60; pseudo R² = 14.7%), initial MVC_ECC (Chi² = 0.10; p = 0.75, pseudo R² = 3.6%), initial MVC_CON (Chi² = 0.5; p = 0.47, pseudo R² = 29%) and MVC_ECC/MVC_CON ratio (Chi² = 0.17; p = 0.68; pseudo R² = 31%). Moreover, training related variables (number of repetitions, sets, frequency and duration of training), could not statistically significantly explain the variability (Chi² = 0.65; p = 0.96; pseudo R² = 14%) (Table S6 of the ESM 1). As rest periods between sets were reported for only 11 out of 21 outcomes, and 9 of these indicated a duration of one minute, they were not included in the multiple meta-regression analysis of moderators.

Changes in Strength Within Training Group

As shown in Fig. 6, results from individual studies indicate that eccentric-only training resulted in an improvement of MVC_ECC by 27.3% (95% CI 19.4–35.2%; p < 0.05; robust 95% CI 18.2–36.4%), MVC_CON by 12.6% (95% CI 8.6–16.7%; p < 0.05; robust 95% CI 8.0–17.3%), and MVC_ISO by 18.2% (95% CI 11.7–24.7%; p < 0.05; robust 95% CI 10.2–26.2%) (Table S7 of the ESM 1).

Fig. 6.

Percent improvement with 95% robust confidence intervals of maximal voluntary eccentric (MVC_ECC), concentric (MVC_CON) and isometric contraction (MVC_ISO) strength after eccentric training and concentric training. * significant (p < 0.05) difference between eccentric training and concentric training. # significantly (p < 0.05) different from MVC_ECC of respective training. † significantly (p < 0.05) different from MVC_CON of respective training

Improvements after concentric-only training were observed as follows: MVC_ECC increased by 10.2% (95% CI 8.2–12.3%; p < 0.05; robust 95% CI 7.2–13.2%), MVC_CON by 13.8% (95% CI 10.0–17.5%; p < 0.05; robust 95% CI 9.7–17.9%), and MVC_ISO by 16.9% (95% CI 9.5–24.2%; p < 0.05; robust 95% CI 7.8–25.9%) (Table S7 of the ESM 1).

No influential individual results were identified, and Egger’s test results showed no publication bias, with p-values ranging from 0.08 for MVC_CON after concentric training to 0.797 for MVC_ISO after eccentric training. Statistically significant heterogeneity was confirmed by Q-test (p < 0.05) in all cases, with I² values exceeding 98% in all cases. Specifically, level 2 I² varied from 0% for MVC_ISO after concentric training to 27% for MVC_CON after concentric training, while level 3 I² ranged from 72% for MVC_CON after concentric training to 99% for MVC_ISO after concentric training (Table S7 of the ESM 1).

Differences in Strength Changes Within Training Group and Between Groups

Pairwise comparisons revealed differences in strength improvement among testing contraction modes following eccentric-only training: MVC_ECC versus MVC_CON showed an 11.5% difference (95% CI 11.2–11.7; robust 95% CI –1.9 to 24.8; p < 0.05), illustrated in Fig. 6. The difference between MVC_ECC and MVC_ISO was 8.6% (95% CI 8.0–9.3; robust 95% CI –44.4 to 61.7; p < 0.05), and between MVC_CON and MVC_ISO was 3.8% (95% CI 3.3–4.3; robust 95% CI –5.7 to 13.2; p < 0.05) (Table S8 of the ESM 1).

For concentric-only training, the respective differences were also notable: MVC_ECC versus MVC_CON at 1.7% (95% CI 1.5–1.9; robust 95% CI 4.1–7.5; p < 0.05), MVC_ECC versus MVC_ISO at 6.2% (95% CI 5.7–6.7; robust 95% CI 1.4–11.0; p < 0.05), and MVC_CON versus MVC_ISO at 2.1% (95% CI 1.7–2.6; robust 95% CI –7.0 to 11.2; p < 0.05) (Table S8 of the ESM 1).

No influential individual results sizes were detected. Egger’s test revealed publication bias (p < 0.10) in all comparisons except for MVC_ECC versus MVC_ISO after concentric training (p = 0.44), and statistically significant heterogeneity was confirmed in all cases with Q-test’s Chi² statistics (p < 0.05). I² values exceeded 90% for all comparisons. Specifically, level 2 (within-study) I² ranged from 38.6% for the MVC_CON and MVC_ISO comparison for concentric training to 66.4% for the MVC_ECC and MVC_ISO comparison post concentric training. Level 3 (between-study) I² ranged from 32.3% for the MVC_ECC vs MVC_ISO comparison post concentric training to 62% for the MVC_ECC vs MVC_CON comparison post eccentric training (Table S8 of the ESM 1).

Discussion

The main finding of the present review was that both eccentric-only and concentric-only isokinetic strength training increased MVC_ECC, MVC_CON, and MVC_ISO, but eccentric-only strength training was more effective than concentric-only training for increasing MVC_ECC, while increases in MVC_CON and MVC_ISO did not significantly differ between the two training types. The results were in line with our hypothesis. The findings also indicate that the longer the training period, the greater the improvement in MVC_ECC with eccentric-only training compared to concentric-only training. Additionally, higher-velocity eccentric-only training had a smaller effect on slower MVC_CON strength gains than higher-velocity concentric-only training.

Specificity of Strength Gain

While both eccentric-only and concentric-only strength training were effective for increasing MVC_ECC, eccentric-only training led to a greater increase (27%) than concentric-only training (10%) as shown in Fig. 6. As depicted in Fig. 3, 29 out of 71 individual effect sizes (41%) favoured eccentric-only over concentric-only training, and only 4 effect sizes (6%) favoured concentric-only training. Both eccentric-only and concentric-only strength training increased MVC_CON similarly (13% and 14%) (Fig. 6), and 13 out of 71 effect sizes (18%) favoured eccentric-only training, while 11 effect sizes (15%) showed a better effect of concentric-only training on MVC_CON (Fig. 4). Additionally, both eccentric-only and concentric-only training increased MVC_ISO similarity (18% and 17%, respectively) (Fig. 5), and 5 out of 20 effect sizes (25%) favoured eccentric-only training and 3 effect sizes (15%) favoured concentric-only training. These meta-analysis results were in line with some original studies investigating the specificity of contraction mode strength gains [34, 54, 71], suggesting that eccentric strength training is superior for improving MVC_ECC [7] and neither concentric nor eccentric training is superior to the other in improving MVC_ISO. Although Morrissey et al. [7] reported that concentric training is preferable for increasing MVC_CON, this was not supported by the present study showing that maximal eccentric-only isokinetic training was as effective as maximal concentric-only isokinetic strength training for MVC_CON strength gains. Kataoka et al. [47] have stated in their review article that the effects of eccentric loading in training are transferable to concentric strength, although the mechanistic reasons why eccentric-biased training carries over to maximal concentric strength remains to be elucidated.

The increases in muscle strength result from a combination of neural and muscle adaptations, and the former includes improved muscle activation through greater muscle motor unit recruitment, higher discharge rate, and motor unit synchronization [11]. Muscle adaptations that contribute to strength gains include increases in muscle cross-sectional area, changes in muscle architecture, and greater musculotendinous stiffness [11]. These factors could contribute to the strength increases after both eccentric-only and concentric-only training. However, it is possible that some unique adaptations depicted below could explain the superiority of maximal eccentric-only to concentric-only training on MVC_ECC and its favourable transfer to other contraction modes [72]. Strength gains in the initial several weeks of training are primarily driven by neural adaptations [4, 6]. High mechanical forces in eccentric-only training induce unique neural adaptations that are more pronounced [15, 54] than those observed in concentric-only training, where the forces are lower [8]. Eccentric-only resistance training may also provide greater excitability in the motor cortex, corticospinal pathways, and neuromuscular junction than concentric-only resistance training [73]. High mechanical forces during eccentric-only resistance training lead to a reduction in protective inhibitory peripheral afferent sensory mechanisms, which limit force generation in the muscle–tendon unit [74]. Downregulation of spinal inhibition, presumably guided by Renshaw cell activity [75], is regulated by central descending pathways and has been observed in previous studies [4, 10, 15, 76–78]. This is evidenced by the proportionally greater improvement of motor unit discharge rate after eccentric-only than concentric-only training [54]. Downregulation of spinal inhibition is also evident after maximal concentric training [74]; however, this may be more significant after maximal eccentric training, which plays a more critical role in increasing MVC_ECC due to the more pronounced protective mechanisms during the initial stages of exercise, attributable to higher mechanical demands [9, 79]. This could also be due to lower voluntary activation during eccentric contractions, commonly found in resistance training studies involving participants naive to eccentric training [80, 81].

The greater increases in MVC_ECC following eccentric-only than concentric-only training may be attributed to adaptations in both contractile muscle structures and noncontractile components such as muscle connective tissue and tendons [12]. It has been documented that mechanical tension, exercise-induced muscle damage, and metabolic stress mediate the hypertrophic signalling response to training [82]. Eccentric-only training may provide higher mechanical stress to induce greater protein signalling cascades and acute inflammatory responses to muscle damage, which are thought to upregulate protein synthesis more effectively when compared with concentric-only training [82, 83]. In contrast to concentric contractions, EMG activity does not change with increasing muscle–tendon force in eccentric contractions [84], indicating that noncontractile elements significantly contribute to force production [9, 85]. Noncontractile proteins such as titin, connective tissue surrounding muscle fibers and fascicles (i.e., extracellular matrix) are stretched during the lengthening of the muscle–tendon unit [86]. This stretching induces adaptations in the connective tissue structures over time, increasing their ability to resist tensile forces and improve sarcomere integrity during eccentric contraction [87]. The role of titin in force production during eccentric, but less in concentric or isometric, contraction has been documented [87–91]. Titin molecules differ between fiber types, being larger and stiffer in type II than in type I [88]. While they also act as mediators for hypertrophic signalling [87], this may explain the beneficial effect of eccentric training on type II fibers [85] and muscle hypertrophy [92, 93].

The time course of muscle–tendon morphological adaptations is longer than that of neural adaptations [76, 94–97], and thus the beneficial effects of eccentric-only training over concentric-only training on the structural components of the muscle–tendon unit, may take longer to be observed. This speculation aligns with the results of our meta-regression, which showed that the superiority of eccentric-only training for MVC_ECC improvement was more pronounced in the studies with longer training durations [55, 61, 63, 65, 66] (please also see Table S4 in ESM 1). Previous studies have also suggested that a longer recovery time after eccentric than concentric training due to muscle damage may be related to the longer time taken for neuromuscular adaptations to be observed after eccentric training [11, 98].

In contrast, the present study showed no significant difference between eccentric-only and concentric-only strength training for changes in MVC_CON and MVC_ISO (Fig. 6). It could be speculated that the neural and morphological adaptations mentioned above are not specific to eccentric-only training. It may be that the adaptations are also induced by concentric-only training, which induced similar increases in MVC_CON and MVC_ISO increases after training (Fig. 6). Since neural and morphological adaptations for strength gains could be greater after eccentric-only than concentric-only training as discussed above, it seems reasonable to assume that eccentric-only training could also increase MVC_CON and MVC_ISO greater than concentric-only training, but this was not found. An extra excitatory descending drive to compensate for spinal inhibition (recurrent inhibition and Ib afferent inhibition) may increase through the activation of different cortical areas in eccentric versus concentric contractions [99, 100]. Nevertheless, the factors contributing to the modulation of voluntary activation at spinal and supraspinal levels remain unknown [101]. It has been reported that hypertrophic responses to eccentric versus concentric contractions are achieved through different adaptations in muscle architecture [12]. Eccentric training results in a significantly greater increase in fascicle length, while concentric training promotes greater changes in pennation angle, likely reflecting the differential addition of sarcomeres either in series or in parallel, respectively [12, 102]. Increased muscle cross-sectional area due to eccentric training has been associated with increased fascicle length rather than changes in pennation angle [47, 103, 104]. It could be speculated that the addition of more contractile material in parallel enhances MVC force production across all contraction modes. However, eccentric-only training leads to a greater addition of contractile material in series [12, 105], which is associated with increased muscle force production at longer muscle lengths, a broader plateau in the muscle's length-tension relationship and also improves muscle shortening velocity and peak isotonic power [106]. It has also been documented that an increase in serial sarcomere number may allow individual sarcomeres to shorten more slowly and remain closer to their optimal length for force production during contraction [106]. These adaptations could contribute to better improvement of peak muscle force production specific to eccentric contractions; however, further studies in humans are needed to confirm this assumption [106].

It has been shown that an increase in MVC_CON is specific to the velocity used in concentric training [7]. The results of subgroup analyses in the present study indicate that higher-velocity concentric-only training was more effective for improving slower MVC_CON when compared to higher-velocity isokinetic eccentric-only training (Table S5 in ESM 1). This finding contrasts with the finding of a previous study showing that eccentric training with contractions lasting for 2–6 s increased concentric one-repetition maximum similarly [107]. Previous studies indicated that fast eccentric training increased muscle thickness [93] and IIb fiber composition [85]. This could theoretically benefit increases in MVC_CON and MVC_ISO, regardless of the movement velocity. Conversely, by increasing the velocity of eccentric contractions, the muscle–tendon force-generating capacity may rely more on noncontractile structures [9, 84, 85, 108]. This could lead to more favourable adaptations in noncontractile than contractile elements, which might explain why faster eccentric-only training does not necessary increase MVC_CON and MVC_ISO greater than concentric-only training. Furthermore, while fast eccentric-only resistance training has been shown to be more effective than slow eccentric-only training for improving isometric rate of force development (possibly due to increases in muscle fascicle length [109]), it would be interesting for future research to investigate whether eccentric-only training enhances muscle power and rate of force development more effectively than concentric-only training and what determines the specificity or non-specificity of adaptations to distinct contraction modes.

Strengths and Limitations

This review with meta-analyses was conducted rigorously, with a priori specifications for all criteria. A major strength of our analyses was the high number of included studies and outcomes and comparisons, which enhances the robustness of the results. We also adhered to the recommendations by Kadlec et al. [110] to further improve the result quality. Additionally, performing sensitivity analyses provided us with insights into the influence of outliers and smaller studies on the overall effects.

However, the findings of this review should be interpreted with some limitations, which in turn offer useful guidance for future research. The quality of the study results was very-low to low due to high between-study inconsistency, wide confidence intervals and publication bias. Although we attempted to explore the causes of the heterogeneity by performing subgroup analyses and meta-regressions, we could only explain variability with one moderator for each of MVC_ECC and MVC_CON improvements. Moreover, a small number of studies were eligible for inclusion in some subgroups (for example, only one study included trained participants) (Tables 2 and 3; Tables S4-6 in ESM 1), and continuous moderator values were homogeneous across the studies (for example, age ranged from 20 to 38), similar to problems identified in previous studies in the field [5]. It is also important to note that some of the subgroups analysed included fewer than the suggested eight effect sizes [111] which increases the risk of overfitting. Moreover, moderator variables were not reported in all studies (Tables S4–S6 in ESM 1). As stated in the results, the rest periods were not included in the moderator analysis because they were reported in only 50% of the studies. More studies are needed in the future to better capture the variability of results, and researchers should be encouraged to report training-related variables more precisely. This is particularly true for MVC_ISO, where no moderator could explain the variability of the effects from our study.

As shown in Fig. 6, concentric-only training increased MVC_ISO by 17%, followed by MVC_CON at 14%, and MVC_ECC at 10%. The comparison between training mode specific (i.e., MVC_CON) and non-specific (i.e., MVC_ISO) gains should be interpreted with caution, as a lower number of studies were eligible to be included in the MVC_ISO analysis than in the MVC_CON and MVC_ECC. Readers should also be cautious about generalising the findings of this review. Due to the characteristics of the studies included in the present review, the results are most generalizable to single-joint exercises and maximal eccentric- or concentric-only training using an isokinetic dynamometer. Therefore, it is not necessarily the case that the single-joint, mode-specific strength increases would also improve multi-joint, contraction-type-specific movements involved in daily activities or sports [112]. Furthermore, isokinetic dynamometers are expensive and not commonly available in typical fitness facilities, restricting their use primarily to clinical or research settings. Moreover, a previous review [113] showed that specific strength gains were even more pronounced when eccentric training was performed with an isotonic (constant external load) modality in comparison to an isokinetic one, with a tendency toward higher and earlier hypertrophy gains after isokinetic training and greater agonist muscle activation gains after isotonic eccentric training. It could also be speculated that despite the high efficiency in modality-specific strength gains, muscle activation strategies when performing MVC at a constant velocity over the exercise range of movement do not mimic gravity-based resistance muscle activation in real-life situations [114], thereby limiting strength transfer to functional movements, such as the countermovement jump [115]. Nevertheless, future studies are required to quantify these speculations, as the effects of isokinetic and isotonic strength training have not yet been compared in standardized settings [113]. Our results are also most generalizable to healthy, middle-aged, moderately trained individuals naive to eccentric training; only one study included strength-trained athletes.

Practical Implications and Future Research

Both eccentric and concentric muscle contractions are included in the majority of human movements. Therefore, it is crucial to consider the specific mechanisms associated with each type of contraction independently and in combination. For the maintenance or improvement of health and quality of life related to muscle strength and mass [2], eccentric contractions should be emphasized during resistance exercise training due to more versatile strength improvements by eccentric training. Moreover, MVC_ECC predisposes individuals to better stretch–shortening cycle performance [27], particularly evident in changes of direction [116] and jumping performance [117, 118]. Eccentric training has proven superior to concentric training for improving the stiffness of elastic elements [11] and the recoil of elastic energy [27]. Thus, the present study results support the use of maximal eccentric-only training as the preferred method to increase MVC_ECC and consequently enhance performance, while the potential of concentric-only training for improving MVC_ECC is limited.

The present study results indicate that the transfer of strength gain from fast maximal eccentric-only training to slow concentric strength is limited; therefore, fast eccentric contractions may not be ideal for improving slow MVC_CON. Additionally, as the superiority of eccentric-only to concentric-only training for improving MVC_ECC increases over time, strength and conditioning coaches should plan a longer training period to maximize neuromuscular adaptations and performance enhancements using eccentric-only training [11, 98]. To obtain credible insights into the distinct underlying mechanisms or adaptations between training types, particularly morphological changes that require extended periods of consistent training to manifest [94], testing in research settings should be conducted at later stages following the completion of the training protocols. Future research should also involve more trained participants and extend over longer training time periods to ensure more comprehensive results.

Furthermore, given the higher within-set fatigue tolerance and possible muscle damage specific to eccentric contractions, the dose–response relationships of eccentric training protocols for maximizing strength gains and minimizing the overtraining effect [11, 85, 98] should be carefully considered in future research. This aspect was overlooked in the studies included in our review, with none reporting the conduct of familiarization protocols despite the maximal training intensity expected to cause severe muscle damage [119, 120] which can result in suppressed force production and suboptimal training intensity [121, 122], potentially limiting strength gains in the early stages of training.

Eccentric exercise-mode specificity of MVC_ECC gain [5, 7] was confirmed by the results of the present study. Nevertheless, further randomized controlled studies are necessary to elucidate the mechanisms underlying the variability of the effects and to assess the applicability of the results to diverse populations, including young and elderly individuals, across various training modalities (i.e., isotonic, isokinetic, and isoinertial [72]), and in the context of more complex, multi-joint human movements.

Conclusions

The present review with meta-analysis demonstrates the superiority of isokinetic eccentric-only training over concentric-only training in improving MVC_ECC. Furthermore, the results indicate a higher transfer effect between training modes with eccentric-only training, while its effects on MVC_CON and MVC_ISO were similar to those of concentric-only training. It is important to note that the effect of eccentric-only training on MVC_CON is greater than the effect of concentric-only training on MVC_ECC, and the magnitude of increase in MVC_ISO is greater for eccentric-only than concentric-only training. This suggests that eccentric-only training is more versatile than concentric-only training. It seems that despite the unique neural control and muscle-force generating mechanisms involved, eccentric-only strength training leads to adaptations over time that better predispose the muscles' overall force-generating capacities compared to concentric-only strength training. Therefore, the use of eccentric training should be actively promoted in strength and conditioning practice.

Author contributions

DS performed the analyses, visualized the data, and wrote the first draft of the manuscript. DS and KN contributed equally to the conception and design of the study, interpretation of the data, drafting and critical revision of the manuscript. Both authors read and approved the final manuscript.

Funding

The Slovenian Research Agency's program 'Kinesiology of Monostructural, Polystructural, and Conventional Sports (P5-0147)' provided partial salary support for DS. Additionally, DS received financial support for a visiting professorship at the Centre for Human Performance, School of Medical and Health Sciences, Edith Cowan University in Australia for three months, through the '[RSF] Internal Call for Co-Financing Mobility of Assistants, Assistants with Doctorates, and Higher Education Teachers (Educational Staff) at Higher Education Institutions Abroad for 2023–2024 (B.II.3)' from the University of Ljubljana, Slovenia. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Availability of data and material

All data generated or analysed during this study are included in the article and its Supplementary files.

Declarations

Competing interest

The authors declare that they have no competing interest relevant to the content of this review.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.