REF-1746 · Guisado-Cuadrado et al., 2026 · Scandinavian Journal of Medicine & Science in Sports, · Paper · PubMed
Two to fourteen days of deliberately low energy availability in athletes (mostly 15 kcal/kg fat-free mass a day, against an adequate-energy control) raised a bone-resorption marker in 4 of 8 interventions and lowered a bone-formation marker in 5 of 8, lowered leptin in 5 of 8 and testosterone in 4 of 8, while estradiol, T3 and IGF-1 mostly did not move
A systematic review without pooling of 13 small experiments (145 participants) measuring blood markers only, no bone density, performance or health outcome, in which low carbohydrate often came with the deficit
Main caveat: Vote counting of significant results with no effect sizes: a marker that "did not change" in a study of 6-14 people may simply have been underpowered.
Guisado-Cuadrado and colleagues (Universidad Politécnica de Madrid) collected experiments in which athletes were put into low energy availability on purpose for more than a day, with energy availability fixed per kilogram of fat-free mass and compared against a control condition with adequate energy, and asked which blood markers change in the short term.
Thirteen studies with 145 participants qualified; the deficits lasted 2 to 14 days, and in most the low condition was set at 15 kcal/kg FFM/day. Because no marker had 10 or more comparisons, the authors did not pool and instead counted, for each marker, how many interventions reported a significant pre-post change.
Bone-turnover markers moved in about half: the resorption marker βCTX-1 rose in 4 of 8 interventions and the formation marker P1NP fell in 5 of 8. Leptin fell in 5 of 8 and testosterone in 4 of 8, while IGF-1 (1 of 8 lower), T3 (2 of 8 lower, both in women), estradiol (0 of 6), calcium-related, inflammatory and iron markers showed no consistent change.
The authors read these as early, probably reversible adaptations, and warn that many of the diets were also low in carbohydrate, which may have caused or added to the changes.
Participants
13 studies, 145 participants: 9 studies of men and 5 of women in Table 1 (one study tested both sexes as separate groups), 79 men and 66 women by our sum; group sizes 6 to 14.
Training status by the studies' labels: elite racewalkers (two studies with the same size, age and VO2max, which may be one cohort; our reading), well-trained long-distance runners, trained, exercising, active, moderately or highly active, recreationally active. Sedentary people and clinical groups (eating disorders, obesity, chronic disease) were excluded.
Women: menstrual phase at testing early follicular (one study), early to mid follicular (two), early and late follicular (one), not reported (one); contraceptive use is not reported in the tables.
Protocol
Search: PubMed, Web of Science and Scopus; keyword blocks for energy availability, outcome, population, minus reviews and animal studies; full strings in the supporting information (not read). English, peer-reviewed only.
Inclusion: athletes of any sport classifiable as McKay Tier 1-5; an induced low-energy condition of more than 24 h stated as 30 kcal/kg FFM/day or below; a within-person pre-post comparison and a control condition with adequate energy; randomised or non-randomised, crossover or parallel. Where a crossover had several low conditions, the lowest was taken.
Low-energy conditions (Table 3): energy availability 15 to 25 kcal/kg FFM/day, most at 15; diets 2 to 14 days; exercise during the deficit ranged from none to running at 65-70% of VO2max or VO2peak, racewalking, ergometer cycling at 60% of VO2peak, or resistance plus endurance work.Several diets gave under 5 g/kg/day of carbohydrate (values marked in Table 3 from 2.5 to 4.8 g/kg), while the authors state protein met at least 1.2 g/kg/day (Table 3 nonetheless lists one protein value of 0.9 g/kg).
Outcomes: bone (βCTX-1, P1NP, osteocalcin, sclerostin), calcium metabolism (calcium, magnesium, phosphate, PTH), energy regulation (leptin, cortisol, ghrelin, IGF-1), inflammation (IL-6, CRP), iron status, sex hormones and thyroid.
Quality: NIH quality assessment tool for before-after studies with no control group, chosen because the outcomes were within-person changes; two reviewers; no study excluded for its score.
Funding and interests: the authors report no funding and no conflicts of interest.
Key Results
| Marker | Interventions with a significant change after low energy availability |
|---|---|
| βCTX-1 (bone resorption) | raised in 4 of 8 |
| P1NP (bone formation) | lowered in 5 of 8 |
| Osteocalcin | lowered in 1 of 3 |
| Sclerostin | no change in 3 |
| Calcium, magnesium, phosphate (5 each), PTH (3) | no change in any |
| Leptin | lowered in 5 of 8 |
| Cortisol | raised in 1 of 3 |
| IGF-1 | lowered in 1 of 8 (the Discussion says unchanged in all eight) |
| Ghrelin, IL-6, CRP | no change (1 each) |
| Ferritin; serum iron | raised in 1 of 2; raised in 1 of 2 (the authors note the rise also occurred in the control condition) |
| Hepcidin (2), haemoglobin, haematocrit, red cells, transferrin, saturation (1 each), reticulocytes (2) | no change |
| Estradiol | no change in 6 |
| FSH, LH, progesterone | no change (1 each) |
| SHBG | raised in 1 of 1 |
| Testosterone | lowered in 4 of 8, over 4 to 14 days |
| T3 | lowered in 2 of 8, both in women |
| TSH | no change (1) in the Results text; Table 1 marks it lowered |
Quality scores 6 to 8 out of 10, mean 7.6 ± 0.7; the item-level picture is in a figure and supplement not in the text read.
No magnitudes, effect sizes or intervals are given for any marker; the counts are of significant results only.
Between-study observations by the authors (single studies, not tested): in one study bone markers changed in women but not men; in another βCTX-1 rose when the deficit came from eating less but not when it came from extra exercise; jumping exercise blunted the bone-marker change in a third.
Practical Application
A few days of under-fuelling already shows in blood markers of bone turnover, leptin and testosterone: in controlled experiments of 2-14 days at about 15 kcal/kg fat-free mass, bone resorption rose and bone formation fell in about half the interventions, leptin fell in 5 of 8 and testosterone in 4 of 8. These are markers, not measured harm: no study measured bone density, injury or performance, and the authors judge the changes likely reversible with refeeding.
A normal estradiol, T3 or IGF-1 after a short deficit does not rule out low energy availability. These mostly did not move within two weeks. This agrees with SYNTH-54, which already says a normal short-term testosterone does not rule it out (REF-878, one of the 13 studies here), and with its rule to screen by the athlete's signs rather than a number; nothing in this review contradicts the guide.
Part of the effect may be low carbohydrate, not low energy. Several of these deficits also put carbohydrate under 5 g/kg/day, and the review cannot separate the two. Inference, not tested here: in any deliberate deficit keep carbohydrate around key sessions, as SYNTH-54 already says.
Inference, not tested here: weight-bearing or impact work may blunt the bone-marker response to a short deficit (one study with jumping, one with exercise-induced deficit); that is a reason to keep lifting and impact in a cut, not a licence to under-eat.
Suspected energy deficiency goes to a sports physician or sports dietitian. This review is not a diagnostic tool and these markers have no tested cut-offs; the assistant does not order, read or act on such blood panels, does not diagnose REDs, and says so.
Limitations
Vote counting of significant results with no effect sizes: a marker that "did not change" in a study of 6-14 people may simply have been underpowered.
The quality tool was built for uncontrolled before-after studies and was applied to controlled designs; no certainty grading; item-level scores were not read.
Short deficits only; chronic or repeated low energy availability, bone density, menstrual function and performance were not studied.
Low carbohydrate travelled with low energy in several diets, so the two cannot be separated (the authors' own caution).
Small, mixed samples; men and women in separate studies; menstrual phase controlled in some studies and not reported in one; hormonal contraception not reported.
Internal inconsistencies: IGF-1 is "lower in 1 of 8" in the Results and "unchanged across all eight" in the Discussion; the text counts 8 interventions for leptin, IGF-1, T3 and testosterone where Table 1 lists 7 (our count); citation numbers in the text and tables do not always match the studies named; TSH is unchanged in the Results but marked lowered in Table 1; protein is said to meet 1.2 g/kg/day while Table 3 lists one diet at 0.9 g/kg.
Two included studies appear to share one cohort of elite racewalkers (our reading), which would double-count it.
No funding; no conflicts declared.
How this entry was checked and changed
Why this level
a systematic review of controlled experiments whose quality appraisal we read in the full text, A- on design at best because the 13 studies are small and the synthesis counts significant results without pooling; one step down to B+ because every outcome is a blood marker, not bone density, performance or health (rule 3)