Caloric restriction lifespan extension in mice

Answer

The cleanest primary figure is Di Francesco et al. 2024 (Nature): 40% caloric restriction (CR) begun at 6 months of age in female Diversity Outbred mice gave a median lifespan 36.3% greater than ad libitum (AL) feeding (CR n=182, AL n=188) 3d7ccdf7. This is adult-onset restriction maintained for life, not late-onset. The text prints no median days and no confidence interval for it. Across the corpus the printed median change under CR ranges from -22% a438336e to about +50% d0bc96d3 in standard strains, so no single number represents mouse CR.

Summary

At about 40% restriction the printed median gains are 36.3% 3d7ccdf7, 36% in B6C3F1 mice of both sexes 6bb0120f, 40% in Emory mice e57614a7, 19% from 2 months of age 30e033b3, about 50% in lymphoma-prone AKR mice from 4 weeks d0bc96d3, 28.5% in female SAMP8 mice fe580e3f, and -22% in male 107-RI mice from 6 weeks (median 729 vs 930 days) a438336e. Milder or differently delivered restriction gave smaller gains: 11% for 20% CR from 12 weeks in C57BL/6J 857e8ddc and 10.5% for evenly spread CR (875 vs 792 days) 94f9f81d. The spread therefore tracks strain, sex, onset and feeding schedule as much as restriction level, and the facts are not matched on these, so they are reported side by side and not pooled.

Lifelong and late-onset restriction are different claims. Forster et al. 2003 report 6-7 months of extra median life in male C57BL/6 and B6D2F1 mice on 40% CR from 4 months, no benefit in DBA/2, and increased mortality when restriction started at 17 months or later; no percent is printed eda5d3b5. No verified fact supplies a late-onset percent change in median lifespan.

Dose-response: Di Francesco randomised the same female population to 20% and 40% CR plus AL, all from 6 months, so a within-study graded comparison exists in the source. Only the 40% arm's number (36.3%) was extracted into the fact; the 20% value is shown only in a figure and has not been extracted 3d7ccdf7.

Evidence

Onset is stated per entry; where the abstract does not give it, it is marked not stated and is not classed as lifelong or late.

• Di Francesco et al. 2024 - 40% CR from 6 months, female Diversity Outbred, median +36.3% vs AL; adult-onset, lifelong thereafter; 20% arm also run but not extracted. 3d7ccdf7
• Sheldon et al. 1995 - B6C3F1 on 60% of AL intake, age at 50% survival +36% in both sexes, 56 per sex per diet; onset age not stated in the abstract. 6bb0120f
• Taylor et al. 1995 - Emory mouse, diet 40% below control, median lifespan extended by 40%; onset, sex and n not stated in the abstract. e57614a7
• Ikeno et al. 2005 - 40% less food from 2 months, median +19%; strain and sex not stated in the abstract. 30e033b3
• Shields et al. 1991 - AKR mice, 40% CR from 4 weeks, median about +50% (P<0.0001); sex and n not stated. d0bc96d3
• Choi and Kim 2000 - female SAMP8 on 60% of AL intake, median +28.5%, maximal +16.4%; onset not stated in the abstract. fe580e3f
• Unnikrishnan et al. 2021 - 40% DR from 6 weeks; male 107-RI median -22% (729 vs 930 days; n=45 DR, 34 AL), female 107-RI +24%, female 115-RI +50%. a438336e
• Brandon et al. 2025 - 20% CR from 12 weeks, C57BL/6J, sexes pooled, median +11%; maximum lifespans 1008 days (control) and 1179 days (CR). 857e8ddc
• Acosta-Rodriguez et al. 2022 - male C57BL/6J, CR spread over 24 hours, median 875 vs 792 days (+10.5%), n=36 and 43; 30% restriction rising to about 40% at later ages. 94f9f81d
• Smith et al. 2024 (GeroScience, Table 1 survival analysis) - male C57BL/6J, 'calorie restriction' imposed as pair-feeding at 22°C to the intake of ad libitum mice housed at 27°C (an energy imbalance, not a stated percent cut in intake): median survival 148.93 weeks versus 115.07 weeks for ad libitum controls at 27°C (n=60/group), restriction beginning at about 12 weeks. The record gives 'about +29.4% median lifespan (computed from the printed medians)'; that percentage is the submitter's own arithmetic on the two printed medians, not a figure printed in the paper. 9019670b
• Forster et al. 2003 - male mice, 40% CR from 4 months, 6-7 months longer median in C57BL/6 and B6D2F1; late-onset CR raised mortality; no percent printed. eda5d3b5
• Mitchell et al. 2019 - male C57BL/6J, 30% CR from 4 months, mean (not median) lifespan +28% (CR 132-135 weeks vs AL 104-110 weeks); median values are in supplementary tables not read; this 28% was not independently certified from the main text here. ff48cb01
• Vermeij et al. 2016 - Ercc1 progeroid mice, 30% DR from 7 weeks, median remaining lifespan from 10 to 35 weeks in males (250%) and 13 to 39 weeks in females (200%); a percent of remaining lifespan in a mutant, not a wild-type total-lifespan change. 239d5a94
• PLOS ONE 2017 (POLG mutator) - lifelong 25% CR from 2 months; male medians 473 (control, n=24) vs 482 days (CR, n=29), calculated +1.9%; female 487 (n=28) vs 460 days (n=23), calculated -5.5%; no significant survival difference in either sex; control arm not explicitly called ad libitum. e546c51b

Methodology

I used every verified fact on the task and did not pool across strains, sexes, restriction levels or feeding schedules. Facts are grouped by what the printed quantity is: a percent change in median total lifespan versus AL (Di Francesco, Sheldon, Taylor, Ikeno, Shields, Choi, Unnikrishnan, Brandon, Acosta-Rodriguez); a derived percent from printed medians (Smith, POLG); a percent of remaining lifespan in a progeroid mutant (Vermeij); a mean rather than median (Mitchell); and months without a percent (Forster). Nothing was converted beyond what the facts already print. The platform listed no measurement conflicts. The sources differ in strain, sex, onset age, restriction level and control arm, so they are not comparable measurements of one quantity and the spread between them is reported as a range.

Limitations

• Only the 40% arm of Di Francesco has an extracted numeric median change; the 20% arm value and confidence intervals sit in a figure or supplement that is not in the corpus 3d7ccdf7.
• Median days for both arms are printed only in a few facts a438336e 94f9f81d 9019670b e546c51b; confidence intervals are printed in none.
• Taylor, Sheldon and Choi leave onset age unstated, and Ikeno leaves strain and sex unstated, because only abstracts were reachable e57614a7 6bb0120f fe580e3f 30e033b3. They cannot be classed as lifelong or late-onset.
• Mitchell's 28% is a mean and its median values were not read ff48cb01; Forster prints months not percent eda5d3b5.
• Progeroid and short-lived models (Ercc1, POLG, SAMP8, AKR) answer a different question from normal ageing and are not estimates for wild-type mice 239d5a94 e546c51b.
• No late-onset percent change in median lifespan is in the corpus.