Six 30-second sprints changed a quarter of the measurable plasma proteome; 90 minutes of moderate cycling changed seven proteins
And the ones sprint raised are those linked, in 53,026 people, to lower cardiometabolic risk
Protocol
Four separate datasets, one question: does the INTENSITY of a single session change what the body puts into the blood?
1
Acute cycling. Plasma at rest, immediately after (0 h) and 3 h after, in two groups of young active healthy participants.
MIE (n=9): 90 minutes continuous cycling at 90-100% of the individually determined first lactate threshold.
SIE (n=10): 6 × 30 s all-out cycling against 0.075 kg per kg body mass, 4 min rest — about 3 minutes of total work.
Olink Explore antibody platform: 2,884 proteins detected, plus untargeted polar metabolomics.
2
After 8 weeks of training. A subset repeated the acute test after 3-4 sessions/week of their own modality (SIT n=7, MIT n=6).
3
Running. 9 trained participants, 2 h of running at 60% V̇O₂max; 2,787 proteins detected.
4
UK Biobank. 53,026 participants, 1,066 diseases, 2,920 plasma proteins, used to ask which exercise-regulated proteins track future health.
Plus tissue work: skeletal-muscle biopsies before and 3 h after; subcutaneous abdominal adipose biopsies before and 3 h after a maximal graded treadmill test (n=9, mean age 28, mean BMI 26.4); and a plasma-swap in which primary human adipocytes were incubated for 3 h with plasma drawn before or immediately after SIE or MIE.
Key Findings
| Sprint-interval (~3 min work) | Moderate 90 min | |
|---|---|---|
| Plasma proteins changed immediately after | 714 (~25% of those detected), >98% increased | 7 |
| …3 h after | ~20× fewer than at 0 h | 19 |
| Metabolites changed immediately after | 203 | 31 |
| …3 h after | 199 | 183 |
| Direct intensity × time contrast | 280 proteins differed at 0 h | 2 proteins differed at 3 h |
| Adipocytes given that plasma for 3 h | 1,128 genes up, 549 down | 14 up, 11 down |
The two intensities are not a bigger and a smaller version of one signal — the timing differs. Sprint produces a large, immediate, short-lived spike. Moderate work produces almost nothing at once and a delayed response: fatty-acid and bile-acid metabolites rise later, consistent with a shift from carbohydrate toward lipid metabolism during recovery.
Trained people still show it, and the shape persists: an immediate rise after sprint, a delayed one after moderate work. GH1, VWF and POMC remained significant after 8 weeks of training.
Running is not cycling. Moderate-intensity *running* disturbed the plasma proteome more than moderate-intensity *cycling* (111 of 2,787 proteins changed, 109 of them upward) — still far below sprint cycling. Proteins shared between sprint cycling and running included ANGPTL4, VWF, FABP4, ADAMTS1.
The health link is associative and comes from the biobank arm. Of 741 exercise-regulated proteins, 143 had a hazard or odds ratio below 1 across 14 disease chapters. Narrowing to type 2 diabetes, metabolic disorders and obesity left 33 proteins — and 32 of those 33 were regulated by sprint, at 0 h or 3 h.
Practical Application
A time-poor athlete's few hard minutes are not a lesser version of a long ride — metabolically they are a different, larger signal. Three minutes of all-out work moved a quarter of the measurable plasma proteome; 90 minutes at the first lactate threshold moved seven proteins. Use this when someone with 40 minutes asks whether it is worth training at all.
Do not read it as "sprints beat endurance". In this same paper both 8-week regimens improved maximal aerobic power and 4-km time-trial performance. The claim here is about circulating signals associated with metabolic health, not about race results.
Expect the recovery windows to differ, and say so. After sprint work the signal is immediate and largely gone by 3 h. After long moderate work it is still building at 3 h. If an athlete asks why a long ride "hits later", this is a mechanism worth naming — carefully, as a plausible explanation rather than a measured cause in them.
For an athlete already doing plenty of volume whose goal is health as much as speed, this is an argument for keeping a small amount of genuinely maximal work in the week — the protocol is 6 × 30 s all-out with 4 min recovery — provided the athlete can safely produce it. It is not an argument for making the whole week hard.
Do not name individual proteins to an athlete as if they were actionable. Nothing here is measurable in them, none of it is a target, and there is no intervention that follows from a single protein.
Limitations
The authors' own first limitation: the acute cohorts are small and skewed in composition (n=9 and n=10), and they state that further work is needed to establish whether the findings hold across the whole population. Treat every number here as a direction, not a magnitude.
The biobank arm is association, not causation. A hazard ratio below 1 says people with more of a protein had less disease; it does not say that raising it by exercising prevents disease, and the exercise cohorts and the biobank cohort are different people.
Tissue of origin is PREDICTED, not measured. The mapping uses reference expression atlases. The authors write plainly that they were unable to determine organ of origin or destination, and call it an underdeveloped area.
The authors themselves raise the strongest counter-argument: recent cell-type-selective labelling found minimal muscle-fibre-derived protein in the systemic circulation after exercise, which suggests muscle-derived proteins may act mainly *locally* inside the muscle. If that holds, part of the "interorgan" framing is overstated.
The cell-culture model has a known artefact: electrically simulating sprint work in vitro can cause cell damage and cytosolic protein leak, which mimics secretion. The authors name this as a limitation of that paradigm.
The antibody platform cannot distinguish a cleaved protein fragment from the full-length protein, so part of the "secretome" may be shed ectodomains rather than deliberately secreted signals.
The adipose work is subcutaneous abdominal tissue only and says nothing about depot-specific responses.
Two variables are confounded by design. SIE is ~3 min of work and MIE is 90 min: intensity and duration move together, and the paper cannot separate them.
No performance claim is tested acutely. Nothing here measures whether either session made anyone faster on the day.
How this entry was checked and changed
Content correction 2026-09-27
Source note (truth pass): the numbers in this entry were taken from the paper's full text when it was ingested; many of them are not in the abstract. They were not re-checked against the full text on 2026-09-27.