REF-428 · de Pablos, Valenzuela, Martínez-Cava et al., 2026 · Scandinavian Journal of Medicine & Science in Sports, 2026; · Paper
The Role of Pedaling Intensity During Torque Training in Cyclists: A Three-Arm Randomized Controlled Trial
High-load vs low-load torque training vs control
Participants
n=46 well-trained male cyclists, VO2max 62.7 ± 5.8 mL·kg⁻¹·min⁻¹, randomly allocated to three groups.
Protocol (10 weeks)
Control (n=15): kept habitual cycling training
HIGH-Load (n=16): habitual training + two weekly torque sessions of all-out pedaling cycles against 70% of maximum dynamic force (MDF)
LOW-Load (n=15): habitual training + two weekly torque sessions of 4-min intervals against 35% of MDF
Outcomes: maximum aerobic power (MAP), ventilatory thresholds, efficiency, time to exhaustion at the respiratory compensation point (TTE), body composition, quadriceps morphology (cross-sectional area, CSA)
Number of pedal cycles, gradient, cadence and recovery per session are not in the abstract
Key Findings
Within-group: only HIGH-Load improved Wingate peak and mean power (p < 0.001), endurance indicators (MAP and ventilatory thresholds, p < 0.05) and quadriceps CSA (p = 0.030)
HIGH-Load vs control: larger improvements in Wingate peak and mean power and in quadriceps CSA (all p < 0.01)
LOW-Load: no significant within-group changes; vs control, a between-group difference only for Wingate mean power (p = 0.008)
HIGH-Load vs LOW-Load: no significant differences, despite moderate effect sizes for some outcomes (Cohen's d > 0.5 for MAP, Wingate performance and TTE)
Authors' summary: HIGH-Load torque training seems to improve different performance indicators and muscle morphology in well-trained cyclists, but its superiority over other torque training should be confirmed in larger studies
Percentage changes per group and the efficiency result are not in the abstract
Mechanism
Inference, not tested: all-out pedal cycles against 70% MDF demand far more force per stroke than 4-min intervals at 35% MDF; the quadriceps CSA gain and the Wingate gains in HIGH-Load only are consistent with a neuromuscular/structural stimulus delivered on the bike. Efficiency was an outcome but the abstract does not report its change.
Practical Application
In 46 well-trained male cyclists (VO2max 62.7) over 10 weeks, two weekly torque sessions of all-out pedal cycles against 70% of maximum dynamic force (HIGH-Load) improved Wingate peak and mean power, maximal aerobic power, ventilatory thresholds and quadriceps cross-sectional area within the group, and beat control on Wingate power and CSA (p < 0.01); 4-min intervals against 35% MDF (LOW-Load) produced no significant within-group change and differed from control only on Wingate mean power (p = 0.008).Head-to-head, HIGH- and LOW-Load were NOT significantly different (d > 0.5 in favour of HIGH-Load for MAP, Wingate and TTE) — all-out heavy-torque work is the version with evidence of adaptation, but its superiority over moderate low-cadence intervals is not proven.
Do not tell an athlete that moderate low-cadence intervals "do nothing": LOW-Load beat control on Wingate mean power, and the HIGH vs LOW comparison was non-significant at this sample size.
Not from this paper; the base's own practice: run HIGH-Load-style sessions as all-out seated efforts in a big gear from a slow or stationary start with full recovery between reps, two sessions a week as in the trial, in a base or strength block rather than inside a concentrated VO2max block; keep them away from athletes with active knee or lower-back problems (the trial reports no injury data).Off-bike heavy strength as the primary tool: REF-147 Rønnestad 2017, REF-149 Rønnestad & Mujika 2014; on-bike resistance training: REF-158 Pallarés 2025.
Limitations
Male only; well-trained (VO2max 62.7), not elite
15/16/15 per arm — HIGH vs LOW non-significant despite d > 0.5: underpowered for the head-to-head, as the authors state
10 weeks; no follow-up, no injury data
Session details (number of pedal cycles, gradient, cadence, recovery) not in the abstract
Quadriceps CSA by imaging; no biopsy
Single research group; independent replication pending
How this entry was checked and changed
Why this level
regraded 2026-09-27 from A: RCT with n=46 (15/16/15 per arm)
Content correction 2026-09-27
Per-group percentage tables (MAP +4.9/+1.0/+1.3%, VT2 +3.2%, Wingate +5.8/+2.9%, CSA +3.5/+0.7/-3.0%) -> deleted; the abstract reports p-values only.
"HTRQ 5 × 7 stationary starts, 6% gradient, ~40 rpm, 10 s, 4 min recovery; MTRQ 5 × 4 min at 50 rpm, 85-90% FTP, 2 min recovery" -> reduced to the abstract's description (all-out pedaling cycles at 70% MDF vs 4-min intervals at 35% MDF, two sessions/week).
"MTRQ neither effective nor harmful, neutral vs control" -> LOW-Load differed from control on Wingate mean power (p = 0.008); "definitive head-to-head" -> HIGH vs LOW not significantly different (d > 0.5 for MAP, Wingate, TTE).
"Cycling efficiency unchanged in all groups" -> deleted (not reported in the abstract). HTRQ template, decision matrix and when/when-not lists -> cut to a short "base's own practice" line. Level A -> A-; Tier and Sport lines added.
2026-10-01: cross-reference repointed from a merged framework to the one that replaced it (SYNTH-01→31, 05→34, 09→70, 12/15→32, 19→26).