REF-1132 · Pastorio et al., 2026 · Scandinavian Journal of Medicine & Science in Sports · Paper · PubMed · Free full text
Sex differences in durability following heavy intensity cycling exercise in trained athletes
After 90 min at 110% of GET on water only, power at the first threshold fell 16% in men and 2% in women, while peak power, VO2peak and neuromuscular function fell alike
Pastorio et al. asked whether trained women keep their thresholds better than trained men during a long ride. Sixteen men and sixteen women, recruited above sex-specific VO2peak floors (55 and 48 mL/kg/min), rode 90 min at 110% of the power at their gas exchange threshold (GET, the first threshold) with water only, then repeated the ramp test from their fresh visit.
Power at GET fell 16 ± 15% in men and 2 ± 13% in women (sex × time P = 0.013; the women's fall itself was not significant, P = 0.460). Power at the respiratory compensation point (RCP, the second threshold) fell 13 ± 10% vs 6 ± 9% (interaction P = 0.049). Peak power (−10%) and VO2peak (−7%) fell by the same amount in both sexes.
Neuromuscular fatigue (MVIC −16%, twitch forces, voluntary activation) showed no sex difference. During the ride men's heart rate, %VO2peak and energy cost per kg rose more and their blood lactate was higher. In women RER and carbohydrate oxidation fell from 45-60 min, but fat oxidation itself did not differ by sex.
The men did 25% more work per kg (17.56 vs 14.08 kJ/kg) and sat slightly higher in the heavy domain (21% vs 17% of the GET-RCP gap, P = 0.07).
The authors conclude that women have greater durability of submaximal thresholds; our inference: the work and intensity mismatch means the sex gap may be partly a dose gap (the authors concede only that the work mismatch could have influenced the magnitude of fatigue).
Protocol
Participants: 16 women (27 ± 4 y, 56.8 kg, VO2peak 51.2 ± 3.1 mL/kg/min, Pmax 4.7 W/kg, 11 ± 3 h/week) and 16 men (29 ± 6 y, 75.9 kg, VO2peak 58.3 ± 6.8, Pmax 5.4 W/kg, 9 ± 4 h/week); training volume, frequency and age not different. Competitive cyclists and triathletes, used to rides of 2 h or more, a race in the last 3 months.
Inclusion floors: VO2peak at least 48 mL/kg/min for women and 55 for men, which the authors call "equitable performance levels" by the De Pauw (men) and Decroix (women) classification guidelines; both groups averaged performance level 3 ± 1 (P = 0.265). Two recruits were excluded for missing the floor. The relative VO2peak gap left by the floors is 12% (58.3 vs 51.2), close to the ~14% gap among 580 competitive cyclists in REF-1017 (our comparison).
Women: 11 naturally menstruating, tested in the early follicular phase (bleeding onset to day 5); 1 combined monophasic pill user, tested in the withdrawal week; 1 vaginal-ring user, tested in the ring-free week; 2 hormonal IUD (Mirena) and 1 progestin implant (Nexplanon) users, tested at any time. No hormone measurement is reported.The authors describe all the women as tested in a low endogenous hormonal state; for the two hormonal-IUD users, tested at any time, that rests on the authors' assumption of a constant hormonal environment, not on timing or measurement.
Sites: Northumbria (n = 11), Newcastle (n = 12) and Pavia (n = 9). Same protocol and most equipment; the load cell and EMG system differed between sites. One team did all the analysis. No cross-site check with the same participants.
Visit 1: familiarisation with the neuromuscular tests, then a step-ramp test: 6 min at a moderate step (1.5 × body mass in W for men, 1.3 × for women), then a ramp of 25 W/min (men) or 20 W/min (women) to failure. GET and RCP were set by two investigators from gas exchange and corrected for VO2 mean response time.
Visit 2 (at least 72 h later, mean 85 ± 7 h, same time of day): neuromuscular test, 5 min at 75% of GET power, then 90 min at 110% of GET power at 80-90 rpm (men 247 ± 48 W, women 149 ± 23 W). Neuromuscular test again within 60 s of stopping, then the same step-ramp test.
Feeding: habitual diet, a 24-h record before visit 1 copied before visit 2, a "high carbohydrate meal" (amount not given) at least 2 h before each visit; no alcohol or vigorous exercise for 48 h, no supplements for 24 h, no caffeine for 6 h. During the 90 min: water ad libitum, no food or calorie-containing drinks. Body mass was not measured after the ride.
Measured during the ride: heart rate, NIRS (vastus lateralis) and EMG throughout; gas exchange for the first 15 min, then 5 min in every 15 (substrate oxidation and energy expenditure from gas exchange); RPE (6-20) and earlobe lactate every 15 min.
Measured before and after: MVIC, voluntary activation, potentiated 100 Hz and 10 Hz doublets, single twitch and Mmax of the knee extensors; ramp-test GET and RCP power and VO2, Pmax, VO2peak.
Three men stopped early (60, 78 and 86 min) and one woman (60 min); all four did the post test and were kept in the analysis.
Key Results
| Outcome | Men | Women | Statistics (full text) |
|---|---|---|---|
| Fresh GET | 226 ± 42 W; 66% ± 4% VO2peak | 135 ± 21 W; 63% ± 5% VO2peak | W P < 0.001; %VO2peak P = 0.053 |
| Fresh RCP | 331 ± 25 W; 89% ± 3% VO2peak | 222 ± 35 W; 89% ± 5% VO2peak | W P < 0.001; %VO2peak P = 0.607 |
| Ride intensity | 247 ± 48 W; 21% ± 7% of the GET-RCP gap | 149 ± 23 W; 17% ± 6% of the gap | gap P = 0.07; VO2 at 4-6 min 112.7% of VO2 at GET in both (P = 0.982); both ~81% HRpeak, ~73% VO2peak at the start |
| Work in the 90 min | 17.56 ± 2.82 kJ/kg | 14.08 ± 2.25 kJ/kg | P < 0.001 |
| Power at GET, post vs pre | −16% ± 15% (P = 0.001) | −2% ± 13% (P = 0.460) | sex × time F1,30 = 6.93, P = 0.013, ηp² = 0.19 |
| Power at RCP, post vs pre | −13% ± 10% | −6% ± 9% | sex × time F1,30 = 4.19, P = 0.049, ηp² = 0.12 |
| VO2 at GET / at RCP | fell (time P = 0.019) / no change | same | no sex interaction (P = 0.096 / P ≥ 0.169) |
| Pmax | −10% ± 8% (pooled) | same | time P < 0.001; no sex effect or interaction (P ≥ 0.072; abstract P ≥ 0.073) |
| VO2peak | −7% ± 7% (pooled) | same | time P < 0.001; no sex effect or interaction |
| VEpeak | −18% ± 16% | −9% ± 10% | interaction P = 0.049 |
| MVIC, Db100, Db10:100, Qtw,pot, VA | −16%, −15%, −22%, −27%, −6% (pooled) | same | all time P < 0.001; no sex interaction (P ≥ 0.096) |
| Mmax | −12% ± 21% (P = 0.082) | +19% ± 35% (P = 0.003) | interaction P = 0.001 |
| Heart rate in the ride | rose more | rose less | interaction P < 0.006; men higher at 60 (P = 0.042) and 90 min (P = 0.032) |
| VO2 as % of fresh VO2peak | rose more, to 82% | to 76% | interaction P = 0.008, from 15 min on (P ≤ 0.021); vs post-ride VO2peak 90% ± 9% vs 83% ± 9% |
| Energy expenditure per kg | rose more | rose less | interaction P = 0.006 |
| Gross efficiency | fell over time | same | no sex effect or interaction (P ≥ 0.741) |
| RER | fell less | fell more | interaction P = 0.001; at 60 (P = 0.015) and 90 min (P = 0.016) |
| Carbohydrate oxidation | not changed (P ≥ 0.874) | fell from 45 min (P ≤ 0.008) | interaction P = 0.003 |
| Fat oxidation | rose over time | same | no sex effect or interaction (P ≥ 0.741) |
| Blood lactate | higher | lower | sex effect P = 0.030 (at 30, 75, 90 min); no upward drift in either group |
| Muscle O2 extraction (HHbMb) | +49% | +13% | interaction P = 0.005; TSI fell more in men (P < 0.001) |
| RPE | rose over time | same | no sex effect or interaction (P ≥ 0.741) |
Is 110% GET the same relative intensity for both sexes? By the authors' anchors, yes: VO2 was 112.7% of the VO2 at GET in both, and both started at about 81% HRpeak and 73% VO2peak. Measured against the upper edge of the heavy domain it was not quite equal.The men sat 21% of the way from GET to RCP and the women 17% (P = 0.07), and the men's GET was at a slightly higher fraction of VO2peak (66% vs 63%, P = 0.053). Our calculation from the group means: the ride was ~75% of RCP power for the men vs ~67% for the women, and ~60% vs ~55% of Pmax, because the women's GET sits lower relative to their RCP (61% vs 68%).The men therefore rode nearer their second threshold, did 25% more work per kg, and three of them stopped early against one woman. Every one of these differences points the same way and each is modest or borderline. The confound is unresolved, not disproved.
Spread: the SDs are almost as large as the GET difference. Our calculation: d ≈ 1.0 for GET and ≈ 0.7 for RCP. Under a normal approximation, in about one random man-woman pair in four the man held GET better, and roughly a quarter of the women lost more than 10% while about a quarter of the men lost less than 5%. Figure 2 plots each rider; the individual values are not tabulated.
Statistics: the GET result is a proper sex × time interaction (P = 0.013, ηp² = 0.19). The RCP interaction sits at P = 0.049, among some 20 outcome ANOVAs with no correction across outcomes. Several F tests carry denominator degrees of freedom (1,58-1,59) that do not fit a 32-person repeated-measures ANOVA; the paper does not explain this.
Mechanisms
Measured: the metabolic rate (VO2) at GET and at RCP shifted similarly in both sexes, while the POWER at those thresholds fell more in men. In men VO2 drifted more (4.30 vs 2.51 mL/kg/min; 82% vs 76% of fresh VO2peak by the end), and so did energy expenditure per kg, heart rate and muscle O2 extraction (HHbMb +49% vs +13%); blood lactate was higher in men throughout but drifted upward in neither group.
Measured, and against the efficiency story: gross efficiency fell over time with no sex difference (P ≥ 0.741). The energy-expenditure interaction was tested on raw J/s/kg values, not percentage change, and the men worked at more W/kg (our calculation: ~3.3 vs ~2.6 W/kg), so part of their larger rise may simply reflect their higher work rate.
Authors' interpretation: men lost more efficiency during the ride (a larger VO2 slow component, attributed to more type II fibre recruitment), so the same metabolic threshold arrived at a lower power. The type II fibre share was not measured.
Measured: RER fell more in women from 60 min, and carbohydrate oxidation fell in women from 45 min but not in men; fat oxidation rose over time with no sex difference.
Authors' interpretation: women spared glycogen by relying more on fat, which protected their thresholds. Glycogen was not measured (no biopsies). The authors themselves attribute the men's maintained carbohydrate use partly to their rising relative demand and to the extra work they did. The substrate gap may therefore be a consequence of the harder ride as much as a cause.
Measured: neuromuscular fatigue (MVIC −16%, voluntary activation −6%, twitch forces −15 to −27%) did not differ by sex, contrary to the authors' hypothesis and to earlier cycling studies. Authors' interpretation: the sex difference in durability is not explained by neuromuscular fatigue resistance.
Inference, not tested: some or all of the sex difference could come from the dose mismatch (more kJ/kg, a ride nearer RCP), not from sex. A work-matched or RCP-anchored comparison would settle it; the authors list work-matching as future research.
Practical Application
Are women more durable? One lab study: after 90 min of unfuelled heavy riding, women's GET power fell 2 ± 13% vs 16 ± 15% in men, but peak power fell alike, the men did 25% more work per kg, and individuals overlapped widely. That supports "women's thresholds slid less in this protocol", which is narrower than "women are more durable".The 16 women and 16 men were matched by sex-specific fitness floors. Second-threshold power fell 6% vs 13%; VO2peak and neuromuscular function fell equally. The men rode slightly nearer their second threshold (21% vs 17% of the GET-RCP gap, P = 0.07), though at the same 110% of GET, no race or time-trial performance was measured, and no carbohydrate was given in the ride.Coach each athlete from her or his own measured fade, not from sex.
Do not raise a woman's late-ride targets on the strength of this paper. The female group mean barely moved at GET, but the spread was ±13%, so an individual woman can lose as much as a typical man. The base's rule stands: late-session targets come down from fresh zones (SYNTH-32) until her own repeat test or decoupling says otherwise.
For pacing a long event, the authors suggest women may "drift" into a harder domain later than men. That is their inference; this study measured no pacing and no field data.
Cycle phase: the women were tested in the early follicular phase or the pill-free or ring-free week, or on a hormonal IUD or implant at any time — what the authors call a low-hormone state, with no hormones measured. Whether durability changes across the cycle is untested; the base's default on phase is SYNTH-37.
Limitations
One laboratory protocol: 90 min, heavy domain, water only. No time trial, race or field outcome; "durability" here means ramp-test threshold power after the ride.
The groups were matched on duration and on percentage of GET, not on work: men did 25% more work per kg (P < 0.001), rode slightly nearer RCP (21% vs 17% of the gap, P = 0.07), and three men stopped early against one woman. The authors acknowledge the work mismatch could have influenced the magnitude of fatigue.
Large within-group spread (±13-15% at GET); the women's GET change was not significant on its own (P = 0.460). The RCP interaction is borderline (P = 0.049) with no correction across about 20 outcomes.
Men were fitter in absolute and relative terms (VO2peak 58.3 vs 51.2). The sex-specific floors (55 vs 48) and equal performance levels make the groups comparable under the De Pauw and Decroix guidelines the authors cite, but that is a matching convention, not identical training status. Training intensity distribution and history were not collected.
Women: mixed hormonal status (11 natural cycles, 5 hormonal contraceptive users), tested in what the authors call a low-hormone state, without hormone verification (the two hormonal-IUD users at any time); results may not hold in the luteal phase.
Three sites with different load cells and EMG systems; the site-by-sex split is not reported.
No glycogen, fibre-type or core-temperature measurements; the substrate and fibre explanations are inferred. Post-ride body mass was not measured.
Neuromuscular tests started within 60 s of stopping, which may allow some recovery.
Funding: "The authors have nothing to report"; no conflicts declared.
How this entry was checked and changed
Why this level
n=32 (16 men, 16 women), sized a priori for the GET interaction; sex cannot be randomised, work per kg was not matched (men +25%), the men rode slightly nearer their second threshold, one unfuelled laboratory protocol; read from the full text
Full text (checked 2026-10-13)
Sixteen males (58 ± 6 mL·kg−1·min−1) and 16 females (51 ± 3); inclusion floors 48 mL kg−1 min−1 for females and 55 for males; Performance Level 3 ± 1 vs. 3 ± 1, p = 0.265 (full text).
Women: n = 11 naturally menstruating (early follicular phase, onset of bleeding until day 5), n = 1 combined monophasic pill, n = 2 Mirena coil, n = 1 Nexplanon implant, n = 1 vaginal ring (full text).
Sites: Northumbria (n = 11), Newcastle (n = 12), Pavia (n = 9) (full text).
During the 90 min: water ad libitum, no food or calorie-containing drinks; before each visit a high carbohydrate meal, arriving ≥ 2 h post prandial (full text).
Ride 247 ± 48 vs. 149 ± 23 W, at 21% ± 7% and 17% ± 6% of the GET-RCP difference (p = 0.07); VO2 112.7% of VO2GET in both (p = 0.982) (full text).
Fresh GET 66% ± 4% vs 63% ± 5% VO2peak (Table 1 p 0.053); RCP 89% ± 3% vs 89% ± 5% (Table 1 p 0.607) (full text).
PO at GET −16% ± 15% (p = 0.001) vs −2% ± 13% (p = 0.460), interaction F 1,30 = 6.93, p = 0.013, ηp2 = 0.19 (full text).
PO at RCP −13% ± 10% vs −6% ± 9%, interaction F 1,30 = 4.19, p = 0.049, ηp2 = 0.12 (full text).
Pmax −10% ± 8% and VO2peak −7% ± 7%, no sex effect or interaction (p ≥ 0.072) (full text).
MVIC −16% ± 12%, Qtw·pot −27% ± 13%, VA −6% ± 6%, no time × sex interaction (p ≥ 0.096) (full text).
Total work 17.56 ± 2.82 vs 14.08 ± 2.25 kJ kg−1, p < 0.001 (full text).
End of ride 82% vs 76% of VO2peakPRE; 90% ± 9% vs 83% ± 9% of VO2peakPOST; HHbMb +49% vs +13% (full text).
Gross efficiency and fat oxidation: time effect, no main effect of sex or time × sex interaction (p ≥ 0.741); carbohydrate oxidation interaction p = 0.003 (full text).
Funding: "The authors have nothing to report" (full text).