Can young microbiota transfer age-related traits back towards youth?

The useful question is which material, transferred into which recipient under which preparation, changes which outcome. This article addresses donor material and transferable phenotypes; centenarian associations, probiotic trials and fecal microbiota transplantation (FMT) are different exposures. s1 s2

01What can a transplant comparison identify?

Boehme's 2021 study compared young- and aged-donor material in aged male mice, with young recipients receiving young material as a reference. The main behavioral cohort contained 16, 16 and 21 mice. Transfer began four weeks before testing and continued during testing. Retrieved transfer methods do not describe antibiotic conditioning; protocols from other studies cannot be imported. s2 s3

Young material came from 40 donors and aged material from 83, including animals from the aged recipient cohort. Material was pooled by age, and donor sources differed. More donors broaden a pool without producing independent age-treatment replications. The design mainly identifies differences between these materials in these recipients; age cannot be fully separated from pool, source and collection batch. More recipients cannot solve the single-pool limitation. s3

Allocation balanced aged recipient body weight, and reporting describes randomization and blinding. Public behavioral tables lack cage identifiers and global mouse IDs. Our row-based estimates assume independent recipients and do not control shared housing. Without an untreated aged arm, superiority of young over aged material cannot separate benefit from young material from harm from aged material. Metabolites and viruses also travel with fecal material; transfer alone does not identify living bacteria as the mediator. s3

02“Not significantly different from young” does not establish recovery

In the water-maze probe, aged old-donor recipients were slower than young references, while aged young-donor recipients were not significantly different from references. Those statements do not establish a difference between aged arms or equivalence to youth. We directly compared aged young-pool minus aged old-pool recipients after excluding source-table asterisks. Intervals below are pointwise Welch intervals assuming recipient independence, without cage adjustment. s2 s4

Figure 1. The open-arm contrast is clearer under author-flag exclusions. Including all values gives +13.77 s, interval −5.84 to 33.37, without statistical significance. This assesses exclusion sensitivity; it does not establish that the authors' ROUT rule was erroneous. Source exclusions leave 13 aged old-pool recipients, versus 14 in the final caption; this discrepancy remains unresolved. s4

Open-arm time also exceeds the young reference: excluded-data means are 6.13 s in young references and 4.89 and 21.20 s in the aged arms. Anxiety-like behavior, exploration and risk preference can change without memory recovery. Increased NOR interaction alongside an unchanged recognition index likewise separates exploration from recognition. The author-posted proof's index formula and source-table scale remain unreconciled, so we compare only the source-reported index and do not test against a chance level. s2 s3 s4

03Normalized learning signals also matter

The final extended-data caption reports improvements in selected normalized water-maze measures, including path length on days three to five and distance to platform on day five. These are different estimands from the native-unit family. Each mouse's first day is set to 100; values are percentages of baseline, rather than percentage changes after subtracting 100. s4

After reading this positive evidence, we added four normalized AUCs and 16 day-two-to-five contrasts as a separate exploratory family. Normalized path AUC differed by −61.26 percentage-days, pointwise interval −106.03 to −16.50, unadjusted P=0.0088; Holm P in the added family was 0.168. The authors used different three-group and day-specific comparisons. Our wider family does not overturn their reported findings. It shows a learning-related signal whose strength depends on scale and comparison scope. s4

Figure 2. Intervals remain pointwise. Initial path means were 1006.45 cm in aged old-pool recipients, 1189.09 cm in aged young-pool recipients and 1218.48 cm in young references. A smaller denominator can affect later ratios. We verified path and velocity normalization against daily native values, but did not obtain native daily distance-to-platform or escape-latency tables. Initial behavior and regression to the mean remain alternatives, without an identified numerical bias estimate. s4

The early experiment supports selected phenotype changes, without independently establishing broad cognitive restoration, whole-body rejuvenation or longer life. Nonsignificance does not establish zero effect.

04A newer mechanism study locates a specific pathway

Cox's 2026 study used co-housing, germ-free transfer and bacterial colonization to investigate memory changes. Under the tested conditions, P. goldsteinii and secreted medium-chain fatty acids, including 3-hydroxyoctanoic acid (3-HOA), were candidates. The proposed direction is enhanced local inflammation through peripheral myeloid GPR84, weakening vagal afferent and hippocampal novelty responses and affecting memory; metabolites need not enter the brain directly. Metabolite, receptor, marrow and neural interventions tested this chain. s5

Figure 3. PLX affects central and peripheral cells, so it cannot alone identify peripheral immunity. Clodronate, CCR2 experiments and mixed marrow chimeras provide further localization; neuronal Il1r1 deletion and selective neural manipulation connect immune and afferent signals. Exact cell locations and the full brain route remain unresolved. s5 s6

We separately audited seven source-data contrasts. The 2026 NOR index is novel-object exploration fraction minus 0.5 and cannot be pooled with the 2021 index. Mice had to accumulate 30 seconds of object exploration within ten minutes or were excluded. Estimates concern analyzed animals; complete exclusion denominators per experiment are unavailable, limiting generalization to all treated animals. Old versus young FMT in young germ-free recipients differed by −0.131 (pointwise 95% interval −0.217 to −0.044). The four-arm clodronate-by-3-HOA difference of differences was +0.153 (0.095 to 0.211), showing a change in the 3-HOA effect under intervention. All seven passed our within-family Holm threshold, but they are experiments within one paper, not seven independent study replications. Intervals assume independent recipients. s5 s6

Figure 4. The first two contrasts change exposure; others test pathway interventions. Mixed marrow, neuronal conditional deletion and aged-drug comparisons are conditional group differences, not complete factorial interactions. Individual sex labels are unavailable for mixed-sex experiments, preventing sex adjustment. Counts, definitions and all estimates remain in the package. s6

Memory changes did not require matching changes in neurogenesis, frailty or permeability. A phage altered bacterial function and improved performance without lowering P. goldsteinii abundance; killing a harmful bacterium is an unsupported rewrite. This pathway need not generalize across strains, diets or hosts, and has not been established as a driver of human cognitive ageing. s5

05The recipient changes the meaning of donor age

Kundu's 2019 abstract reports increased hippocampal neurogenesis and intestinal growth after old-donor material was transferred to young germ-free recipients. Prolongevity signaling describes molecular pathways, not measured lifespan extension. We read the abstract only and do not fill in missing design details. This opposing endpoint cautions against treating old material as universally harmful or germ-free recipients as ordinary aged hosts. s7

Parker's independent 2022 experiment combined antibiotic conditioning, donor material and bedding transfer. Brain, retinal and intestinal inflammatory measures changed, while short-term NOR and Y-maze outcomes did not show improvement in aged recipients of young material. Microbiome statistics accounted for cages; that does not mean every behavioral analysis did. Inflammatory markers are not memory substitutes, and Iba1 staining cannot alone separate cell number from marker expression. s8

Sommer's 2025 lifelong intervention used antibiotic conditioning and young or age-matched transfer every eight weeks. Rotarod performance favored young material; grip strength, glucose tolerance and FITC permeability did not reach significance, while circulating LPS was lower. The lifespan endpoint did not reach significance (log-rank P=0.067), with observation to approximately 120 weeks. Neither equal lifespan nor established extension follows. Early deaths from gavage lesions occurred in both arms; late measurements among survivors also require attention to selection and procedural burden. s9

These studies are not one interchangeable young-microbiota treatment: recipient age, conventional or germ-free status, sex, preparation, source, duration and outcomes differ. A separate Korean 2022 microbiome dataset cannot supply missing behavioral mouse IDs or cages for Boehme's experiment. s10

06Human benefits require their own evidence

The load-bearing evidence obtained here does not establish young-donor FMT benefits for memory or healthspan in ordinary older adults. Disease-specific trials, probiotics and registry records cannot substitute for that population and comparison. The FDA's listed VOWST indication is prevention of recurrent C. difficile infection in adults after antibacterial treatment for recurrent infection; this indication cannot be extended to ageing. s11

Human studies have started. In the 2026-10-07 registry snapshot, NCT06649981 tests young, active-donor material for muscle, cognitive and metabolic function and is active but no longer recruiting. NCT05598112 is recruiting, with reduced frailty at 96 weeks as its primary outcome. Newly registered NCT07826793 has not begun recruitment and specifies eight-week DunedinPACE as its primary outcome. None has posted registry results; that does not prove absence of other publications. A short-term clock change would not substitute for healthspan. s13

Healthy donors do not imply risk-free material. The FDA's March 12, 2020 alert described six EPEC/STEC infections after FMT, with four hospitalizations and two additional deaths. The next-day update found one death-associated product negative and no longer suspected STEC transmission from it. The other product was positive, but STEC's contribution to death remained unknown. The two deaths cannot both be attributed to FMT. Transmission despite screening still belongs in benefit–harm assessment; this article provides no donor-selection, conditioning or home-transplant procedure. s12

Multiple independent donor pools, documented cage and allocation units, untreated and procedural aged controls, prespecified functional outcomes and long-term adverse events would most improve inference. Human research must distinguish patient symptom improvement from prevention of ordinary ageing and establish whether the mechanism operates in people.

07Calculation and reading boundaries

The package retains all results in the original 20-outcome family, added normalized 20-outcome family and seven mechanistic contrasts, with cell provenance. Repeated tests are not independent animals. Author exclusions and all-value sensitivity coexist; no new outlier search was performed. Holm adjustment covers each listed family, not every molecular or exploratory analysis. Unreconciled counts, index versions, cages and native-unit gaps remain. For 2021, methods were read in an author-posted proof and checked against final reporting/captions; the final main PDF was not obtained. We do not claim full-text review or complete replication.

Download code, derived inputs and results. Original figures and complete workbooks remain private; article figures are newly drawn. Transfer can change selected phenotypes; specific mechanisms, human benefits and longer life each require separate evidence.