# HSEF v5.3 review: Trained Immunity

> Sources: 63 primary reports; claims: 69; verified: 69; unverified: 0. Verification is bounded attribution. Access: 8 partial-body and 55 complete abstract/status examinations. Literature cutoff: 2026-10-07; source examination UTC date: 2026-10-08. No complete-paper audit or blinded calibration claimed.

> **Adversarial review:** Generated by Sol 6.1; adversarially reviewed by Claude Opus 5.5 (Anthropic) on 2026-10-14. 5 corrections applied (see ADVERSARIAL REVIEW AND APPROVAL). Status: **Approved as revised.**

## AT A GLANCE

**Trained innate responsiveness is causally demonstrated, and conditional cell-intrinsic persistence is supported; a universal autonomous histone–metabolic memory mechanism and universal clinical protection remain unestablished.**

| Level | Confidence | Causal role | Momentum | Stability | Divergence |
|---|---|---|---|---|---|
| Ε — Context-specific causality established | Moderate | reinforcing / stabilizing factor | Mechanistic only | Stable | Contested |

**Confidence basis:** Moderate reflects evidence density and reproducibility: the heterologous-response phenomenon is replicated across many independent groups, models and assay classes, whereas the purified-transfer persistence result that sets the letter rests on a small number of laboratories with heterogeneous conditioning and purification controls, and the clinical arm is split between replicated adult nulls and single-setting neonatal benefits. Reviewer access depth is a limitation of this document, not a component of the confidence band.

**Deciding experiment:** After a defined nonreplicating inducer has cleared by sensitive labeled-ligand and molecular assays, isolate barcoded LT-HSCs and self-renewing primary macrophages. Randomize delayed cell-specific removal of original receptor/niche signaling, candidate chromatin-maintenance machinery and matched rescue. Transfer into ligand-free recipients with matched conditioning and engraftment; follow clones, baseline competence and heterologous challenge function over months.

**Belief versus evidence:** Preregistered primary-subclaim aggregation uses SC3 and returns Ε, context-dependent causality, with Moderate confidence. Purified post-resolution and long-term donor-cell transfer support conditional intrinsic persistence without continued original tissue exposure. Complete ligand exclusion and independence from all ongoing signaling are not demonstrated; normal homeostatic maintenance is compatible with cell-intrinsic memory. The strict universal signal/antigen-free extension remains Γ. Weakest-link sensitivity is Γ because histone-carrier specificity and human clinical mediation remain unresolved; scope-matched primary sensitivity is Ε. The final letter does not certify the full conjunctive mechanism or universal clinical protection.

**Best-supported rival:** Dose- and context-dependent innate set-point. Apply delayed original-receptor/niche subtraction after measured ligand clearance in barcoded purified donor cells; require preserved baseline competence and separate clone/lineage effects from functional memory.

## COMPETING MECHANISTIC MODELS

**Rank 1: Dose- and context-dependent innate set-point (Ε).** Stimulus, tissue and lineage determine a persistent responsive state that can protect, tolerate or cause harm. Distinct stimulus programs, AM transfer harm despite whole-animal benefit, and opposite microglial/clinical effects fit conditional memory. A set-point description alone does not identify its molecular carrier.

**Rank 2: Cell-intrinsic memory with unresolved carrier (Ε).** Cells/progenitors preserve altered responsiveness after removal from the original exposure context. Purified LT-HSC transfer after resolution and autoimmune LT-HSC transfer without recipient autoimmune activity sustain altered descendants. Ligand quantification, normal niche dependence and selective histone inheritance remain unresolved.

**Rank 3: Progenitor selection and differentiation bias (Ε).** Some durable outputs arise from altered HSC composition, lineage propensity or selected clones. β-glucan and periodontitis transfers produce myeloid bias, with matched chimerism. Stringent autoimmune LT-HSC transfer yields functional memory without overall lineage skew; bias cannot explain every case.

**Rank 4: Maladaptive inflammatory amplification (Ε).** Past disease exposure can transmit a susceptibility state that amplifies later pathology. Autoimmune, periodontal, diabetic, stroke and heart-failure transfer models alter subsequent disease or inflammatory response. Not a universal explanation for protective or resolving macrophage states.

**Rank 5: Persistent antigen, metabolites or external signaling (Γ).** Some long-lived effects reflect continuing inducer, microbiome or niche inputs. Live BCG dissemination, Western-diet backgrounds and constitutive mevalonate excess keep this alternative live. Purified cells transferred after resolution and prolonged recipients reduce its sufficiency as a blanket explanation; not directly excluded in every model.

The tournament does not alter the nominal level. Compatible pathways and flexible alternatives are not automatically proved by explaining observations.

## FULL REVIEW

### Registered claim and scope

Innate immune cells and their hematopoietic progenitors acquire intrinsic, long-lasting, autonomously maintained epigenetic and metabolic memory that enhances responses to later heterologous challenges.

Use the index SC3 autonomy gate, with mandatory alternative aggregation. Clinical heterologous protection is scored separately and cannot establish signal-independent memory. Recheck the index assertion about maintenance evidence rather than adopting its implied result.

Selection: public-interest rank 30 from the editorial ranking compiled 2026-09-23; categorical prior verdicts were visible. Private fresh review; earlier documents are preserved.

### Evidence and causal interpretation

The phenomenon is stronger than its single-mechanism interpretation. Candida/β-glucan pretreatment in lymphocyte-deficient mice and human BCG vaccination produce heterologous responses, with monocyte/NOD2/dectin signaling contributions. This establishes functional innate plasticity within tested systems, not universally protective or antigen-free memory (PMID22901542, PMID22988082).

Metabolic induction has genuine causal support: HIF-1α, glycolysis, fumarate/KDM5 and mevalonate pathways affect training. Most manipulations intervene during induction or perturb broad enzymes. Constitutive mevalonate excess in genetic disease is continuing stimulation, not proof of autonomous memory. H3K18 lactylation is a recent candidate, but inhibitor timing/viability and mark-specific causal necessity were not examined from its body (PMID25258083, PMID27926861, PMID27866838, PMID29328908, PMID40318634).

Original Mitroulis sections show sorted LT-HSC transfer at day 28 followed by recipient myeloid bias at 12 weeks and heterologous LPS/chemotherapy responses. Its glycolysis, IL1, statin and GM-CSF interventions occur on days 0–1, separating induction evidence from untested delayed maintenance. De Laval reports acute-LPS persistent HSC accessibility and C/EBPβ dependence, but exact deletion timing and two linked corrections remain unexamined. BCG marrow access in Kaufmann keeps persistent antigen plausible (PMID29328910, PMID32169166, PMID29328912).

Original Li Results impose 21 days of periodontitis followed by 14 days of resolution before transfer. Purified LT-HSCs with support marrow yield responses after 12–16 weeks, with matched chimerism and effects in both CD41 strata. IL1R deletion precedes the original inflammation; it does not test delayed maintenance. The authors acknowledge incomplete molecular explanation and unknown duration. This supports conditional donor-cell persistence away from the original periodontal tissue, not measured total ligand absence or literal mark copying (PMID35483374).

An independent original autoimmune transfer uses stringently purified LT-HSCs and an 18-week endpoint. Recipients lack spontaneous autoimmune activity and preserve overall lineage proportions, yet derived macrophages show enhanced killing/cytokines. Their glycolysis, respiration, ATP and metabolic-gene accessibility decline. This challenges both a composition-only account and universal increased-metabolism memory, while preserving conditioning/selection limits (PMID39413777).

Original Cirovic Methods/Results compare 15 BCG recipients with five controls at day 90. Marrow M. bovis tests are reported negative with data not shown; a skin lesion remains, so this is not total body antigen exclusion. HSPC/monocyte programs and HNF1-associated network/genotype results are not HNF1 knockout or human lineage tracing. Adult cell frequencies do not significantly shift. Later single-cell marrow and neutrophil reports strengthen concordance but may share platforms and do not isolate inheritance (PMID32544459, PMID39153479, PMID33207187).

Original Sun experiments use immortalized macrophages, responder sorting and 14 divisions after washout. Early NFκB activation and most active marks fade; retained/new H3K4me1 and late interferon programs accompany later Salmonella control. TF maintenance is inferred from expression/motif coupling, without a delayed TF deletion. Continued endogenous TF regulation is not identical to continued original external stimulus. A separate epidermal stem-cell comparator actually preserves recall after delayed STAT3 loss, but does not prove myeloid memory (PMID36999024, PMID34320411).

Original alveolar-macrophage sections show IFNAR dependence during induction and no effect of induction-stage methylation/acetylation inhibitors on trained IL6. Trained AM transfer worsens pneumococcal infection despite protection from whole-animal LPS exposure. Virus-experienced AM memory reportedly requires T-cell help for priming rather than maintenance; malaria monocyte induction and CD40-TRAF6 experiments demonstrate adaptive crosstalk in other settings. These timing distinctions prevent a blanket conclusion of either universal autonomy or universal ongoing T-cell dependence (PMID35856089, PMID30433869, PMID35642634, PMID39178113).

Western-diet and diabetic marrow transfers, severe-COVID signatures, stroke and heart-failure models identify durable susceptibility states with substantial niche/selection alternatives. Microglial schedules induce training or tolerance with opposite pathology effects. Enhanced procoagulant responses can be harmful, whereas CAPS macrophages reduce NLRP3/IL1 after β-glucan and KLF4/MERTK AM states favor resolution. Higher cytokines are not a universal functional definition (PMID29328911, PMID34255973, PMID37597510, PMID38787963, PMID39043180, PMID29643512, PMID40053582, PMID32716363, PMID37615937).

Primary human macrophages in the 2026 report retain stimulus-specific responses at six days, with some memories lost by day 11; TF licensing is associative. Its 2025 preprint from SciSpace is the same authors/design and is not counted again. Live BCG microbiome/dissemination and pathogen-specific influenza/pneumococcal versus SARS-CoV-2 effects retain external-input and scope alternatives (PMID42320478, PMID36456739, PMID35235831, PMID39383035).

Randomized BCG–yellow-fever vaccine challenge reduces viremia without identifying ordinary disease prevention or exclusive innate mediation. A small malaria challenge produces earlier/more severe symptoms. ACTIVATE interim infection results and the BLOW2 NICU neonatal mortality benefit are clinically relevant in their settings; BLOW2 combines BCG with OPV and changes a secondary infection-mortality endpoint during recruitment. Guinea-Bissau trials have favorable secondary/pooled neonatal outcomes despite nonsignificant standalone primary contrasts, while Danish neonatal overall infections are null (PMID29324233, PMID30787276, PMID32941801, PMID40983468, PMID21673035, PMID29579158, PMID27429204).

Original BRACE Methods/Results show no six-month symptomatic/severe COVID benefit among 3988 stage-2 randomized healthcare workers. Its severe definition mainly includes incapacitation for three days, recruitment stops early, and follow-up censors COVID vaccination. Twelve-month follow-up reports more symptomatic COVID; the 6828 combined-stage secondary analysis is the same platform with a different denominator. Dutch, Danish and BCG-PRIME randomizations independently find no COVID benefit. Clinical nulls restrict universal protection without disproving conditional cellular persistence (PMID37099341, PMID39127450, PMID38774675, PMID36976004, PMID37774494, PMID36736662).

Consensus broad output assigns Moderate support to universal SARS-CoV-2 protection despite its own BRACE null; that claim is rejected. Targeted retrieval omits four requested originals and substitutes an unrelated alveolar WNT paper for Sun. Its de Laval summary does not validate maintenance-stage deletion. Eight original partial-body examinations supply the decisive timing, purification and endpoint checks. SciSpace leads are deduplicated, and two Scite attempts yield abstracts with denied body access. Retrieval failure is not a research gap or full-text validation.

## HYPOTHESIS STATE EVALUATION FRAMEWORK (HSEF v5.3)

<!-- consensus-consistency-2026-10-09 -->
### Consensus provenance reconciliation — 9 October 2026

Case-specific Consensus discovery records were located and linked to this final edition. This supplements the review's provenance; it leaves the original assessed claim, grades and source counts intact. Provider summaries and full-text badges do not establish examination of original primary passages by this reviewer. Source access and incorporation retain the limits stated in the original methods and any dated research update.

Recorded evidence artifacts:

- `deliverables/hsef_v53_popularity_ranks_26_30_2026-10-07/trained-immunity/consensus/broad-browser-snapshot.txt`
- `deliverables/hsef_v53_popularity_ranks_26_30_2026-10-07/trained-immunity/consensus/query-ledger.json`
- `deliverables/hsef_v53_popularity_ranks_26_30_2026-10-07/trained-immunity/consensus/reconciliation.md`
- `deliverables/hsef_v53_popularity_ranks_26_30_2026-10-07/trained-immunity/consensus/targeted-browser-snapshot.txt`

The located artifacts do not state a reliable source-comparison count. No count is inferred.

No earlier adversarial approval exists for any edition of this review; this provenance supplement is not an independent or blinded adjudication. No score or confidence change is made by this reconciliation.
<!-- end-consensus-consistency -->


### Subclaim matrix and aggregation

| Subclaim | Scope | Level | Decision |
|---|---|---|---|
| SC1 A primary stimulus changes innate responses to heterologous secondary challenge. | defined dose, timing and cell populations | Ζ | Multiple distinct causal manipulations establish altered heterologous innate responses, including lymphocyte-deficient protection, primary human vaccination/challenge and transferred cells. Responses may be protective, harmful or tolerant; enhanced cytokines are not universal benefit. Independent assay/context classes support Ζ for the phenomenon within defined settings. |
| SC2 Histone and metabolic rewiring encode the altered state. | specified chromatin/metabolic nodes | Γ | Induction-stage metabolic nodes causally contribute in defined systems, but histone co-occurrence, broad p300/LDH inhibition and TF deletion do not isolate a specific inheritable chromatin carrier. The conjunctive histone-encoding claim remains Γ under rubric 03; metabolic contribution alone is stronger. A purified autoimmune HSC transfer yields enhanced function despite reduced metabolism, excluding universal increased glycolysis as the carrier. |
| SC3 Memory persists cell-intrinsically without retained antigen or continued upstream signaling. | post-washout mature cells and progenitor compartments | Ε | Purified LT-HSC transfer after resolved periodontitis and independent stringent autoimmune LT-HSC transfer into recipients without spontaneous autoimmune activity support donor-cell intrinsic persistence away from the original tissue input for months. This crosses the context-specific Ε boundary. It does not establish universal absence of retained molecular ligands, complete independence from recipient niches or normal intracellular maintenance, or permanent mark copying. Continued homeostatic TF activity alone does not refute original-inducer independence. Stronger signal/antigen exclusion remains an unmet scope expansion. |
| SC4 Progenitor memory transmits through division to circulating progeny. | lineage-defined HSPCs and descendants | Ε | Purified donor HSCs yield altered descendants and challenge responses after 12–18 weeks, with chimerism and lineage controls. The autoimmune experiment preserves overall lineage proportions, showing that differentiation bias is not the sole explanation. Irradiated niches, clonal selection and no direct mark copying limit the broader mechanistic inference to Ε. Human HSPC/monocyte concordance does not constitute human lineage tracing. |
| SC5 Training produces clinically meaningful heterologous protection in humans. | prespecified BCG/vaccine contexts and incidence endpoints | Γ | Clinical heterologous benefit occurs in defined vaccination contexts, but the examined trials do not selectively establish trained innate memory as the mediator. Combined BCG–OPV neonatal benefit, interim elderly infection benefit and attenuated yellow-fever viremia contrasts cannot be merged with adult COVID nulls into universal protection. Clinical efficacy in its tested setting can be Ε; trained-memory-specific clinical mediation is Γ. |

Preregistered aggregation: primary-subclaim. Preregistered primary-subclaim aggregation uses SC3 and returns Ε, context-dependent causality, with Moderate confidence. Purified post-resolution and long-term donor-cell transfer support conditional intrinsic persistence without continued original tissue exposure. Complete ligand exclusion and independence from all ongoing signaling are not demonstrated; normal homeostatic maintenance is compatible with cell-intrinsic memory. The strict universal signal/antigen-free extension remains Γ. Weakest-link sensitivity is Γ because histone-carrier specificity and human clinical mediation remain unresolved; scope-matched primary sensitivity is Ε. The final letter does not certify the full conjunctive mechanism or universal clinical protection. Sensitivities: weakest-link Γ; primary-subclaim Ε; scope-matched Ε.

### Decisive-perturbation inventory and causal tally

[1] BCG-vaccinated volunteers show heterologous monocyte cytokine responses for three months; NOD2-dependent experiments and SCID Candida survival support innate contribution. Human cytokines do not demonstrate clinical protection or elimination of all retained BCG/signaling; histone inhibitor specificity and adaptive cytokine contributions remain limits.

[2] Candida/β-glucan pretreatment protects lymphocyte-deficient mice in a monocyte-dependent manner and dectin-1/Raf-1 signaling contributes. Induction necessity and stable marks do not isolate antigen-free maintenance or direct mark copying.

[3] Akt/mTOR/HIF-1α perturbations block training; myeloid HIF-1α loss abolishes trained sepsis protection. Predominantly induction-stage metabolic necessity is not mark-specific maintenance; broad metabolic/viability effects require control.

[5] BCG training requires glycolytic rewiring in pharmacological/genetic experiments, with associated histone changes. Most tests target induction or broad metabolic enzymes, not delayed locus-specific histone maintenance.

[6] Glutaminolysis/fumarate and mevalonate-related metabolism contribute to β-glucan induction; fumarate inhibits KDM5 and induces similar marks. Metabolite sufficiency and broad inhibitors do not demonstrate exclusive mark-based inheritance or autonomous maintenance.

[7] Mevalonate/IGF1R/mTOR perturbations regulate training; mevalonate-kinase-deficient patients have constitutive trained phenotypes. Constitutive metabolite excess is ongoing upstream stimulation, not antigen-independent memory; pleiotropy limits a selective epigenetic instrument.

[8] Sorted LT-HSCs taken 28 days after β-glucan transmit myeloid bias after 12 weeks in recipients with matched chimerism; heterologous LPS and chemotherapy responses improve. Transfer lineage bias is not direct mark copying; IL-1/glycolysis/statin/GM-CSF interventions occur on days 0–1 and test induction, not delayed maintenance.

[9] BCG-educated HSCs generate protective macrophages; chimeric/parabiotic and transfer designs support sustainable lineage training. BCG bone-marrow access and potential persistence constrain full ligand independence; exact original transfer/clearance details inaccessible here; Mtb challenge is mycobacterial, not a distant heterologous pathogen.

[10] Acute LPS produces persistent HSC enhancer accessibility and improved Pseudomonas defense; C/EBPβ deletion erases long-term marks/expression. Exact deletion timing and viability/transfer controls are not examined; deletion of a transcription factor is not selective proof of mark copying; two corrections linked.

[11] After 21-day periodontitis and 14-day resolution, purified LT-HSCs transmit myeloid bias and increased arthritis/LPS responses after 12–16 weeks; both CD41 strata transfer effects. IL-1R deletion precedes training; no delayed maintenance-stage IL-1 subtraction, ligand quantification or exclusive epigenetic mechanism; irradiated niche and finite duration limit scope.

[12] Western-diet withdrawal normalizes serum inflammation but retains myeloid responses; NLRP3 loss prevents diet-induced reprogramming. Ldlr deficiency and remaining lipid/niche signals may persist; constitutive knockout tests induction and associated inflammation rather than late maintenance.

[13] Diabetic donor marrow produces macrophage changes in physiological glucose and increases atherosclerosis in normoglycemic Ldlr-null recipients. Whole marrow and recipient Western diet allow selection/niche feedback; RUNX1 inhibition is broad; human signatures are associative.

[14] 15 BCG recipients versus five placebo controls show day-90 HSPC/monocyte remodeling and increased heterologous IL1β; marrow M. bovis testing is reported negative. Assay data/sensitivity not shown, a skin lesion remains, and no human lineage tracing or causal HNF1 deletion; no significant adult cell-frequency shift despite transcriptional bias.

[15] Single-cell human marrow before/90 days after BCG/placebo shows HSPC expression/accessibility changes correlated with paired PBMC IL1β. Correlation across compartments does not demonstrate literal mark inheritance or no persistent ligand; indexed preprint is the same study, not replication.

[16] BCG alters neutrophil antimicrobial function and H3K4me3 profiles for at least three months. Repeated circulating cells cannot prove persistence in the same mature cell or autonomous progenitor mechanism; no clinical incidence endpoint.

[17] Respiratory-virus-experienced alveolar macrophages retain antibacterial memory; T-cell help is reported required for priming rather than maintenance. Original detailed depletion/antigen controls unavailable; innate maintenance after adaptive induction differs from completely innate induction; normal niche requirements remain.

[18] LPS/IFN-β train cultured alveolar macrophages after rest; macrophage IFNAR loss blocks induction, while transferred trained AMs worsen pneumococcal infection despite whole-animal LPS protection. Methylation/acetylation inhibitors during induction do not alter IL6 training here; six-day context and transfer niche differ from clinical vaccination; pooling technical replicates limits sample interpretation.

[19] Mtb IFN-I/iron signaling impairs myelopoiesis and trained defense, contrasting BCG/β-glucan IFN-II/IL1 programs. Chronic pathogenic infection and progenitor death are distinct from autonomous beneficial memory; ongoing infection is a confounder for maintenance.

[20] β-glucan partly reverses LPS tolerance in vitro and restores cytokines in monocytes from experimental endotoxemia. Chromatin recovery accompanies reactivation; it does not prove the marks exclusively encode memory or clinical sepsis rescue.

[21] IRG1/itaconate/SDH processes distinguish tolerance from β-glucan training, including human genetic variants. Induction-stage metabolic control does not establish autonomous maintenance or a universal glycolytic phenotype.

[22] Immortalized macrophages retain changing molecular/function states after washout and 14 divisions; late Salmonella control parallels newly induced interferon programs. TF necessity is inferred from expression/motif coupling, not directly tested by delayed TF loss; initial responder sorting and immortalization limit generalization; mark persistence does not prove mark copying.

[23] H3K18 lactylation persists after washout and at 90 days after BCG; LDH/p300 inhibition attenuates training. Exact inhibitor timing and viability controls unexamined; metabolic and p300 effects are broad, and correlation with H3K18la does not isolate that mark as the memory carrier.

[24] Primary human macrophages retain stimulus-specific responses at six days, with stronger recall to similar challenges; some memories are lost by day 11. Finite and stimulus-dependent persistence contradicts a uniform durable state; TF licensing/chromatin are associated rather than direct maintenance subtraction; same authors/design as 2025 preprint counted once.

[25] Four inducers yield distinct monocyte programs/subpopulations, replicated with in vivo signatures. Single-cell heterogeneity and patient signatures do not isolate self-maintenance or clinical protection.

[26] In epidermal stem cells, delayed STAT3 loss after resolution preserves recall; post-induction AFOS leaves memory domains accessible with homeostatic TF support. Nonimmune tissue is a mechanistic comparator, not proof in myeloid cells; homeostatic maintenance factors are incompletely dissected.

[27] Peripheral inflammatory schedules induce six-month microglial training or tolerance with opposite AD/stroke pathology effects. Schedules and tissue context determine response; disease-transgene and persistent niche changes constrain intrinsic antigen-free maintenance.

[28] HDM inflammation produces TNF/2-HG/PGE2-related macrophage memory; TNF blockade or myeloid Tnf loss prevents imprinting. Allergen and disease milieu can continue; prevention does not establish delayed signal-independent maintenance.

[29] Stringently purified autoimmune-exposed LT-HSCs yield trained macrophages 18 weeks after transfer without spontaneous recipient autoimmune activity or lineage skewing. Transferred progeny have reduced metabolism/accessibility despite increased killing/cytokines; irradiation and selection remain; not direct mark-copying or universal metabolic necessity.

[31] Heart-failure donor marrow transmits cardiac dysfunction and altered susceptibility, with HSC TGFβ changes. Whole marrow/niche and disease carryover can coexist; TGFβ perturbation is not locus-specific epigenetic inheritance.

[32] Post-stroke myeloid changes transmit cardiac dysfunction; IL1β neutralization/CCR2/5 blockade prevent remote cardiac injury. Intervention timing and whole-cell milieu do not isolate late autonomous memory; patient concordance is observational.

[33] β-glucan/heme training increases myeloid procoagulant responses and four-week splenic activity via sphingomyelinase/tissue-factor decryption. Enhanced responsiveness can be harmful; systemic thrombotic clinical outcomes and autonomous maintenance are not demonstrated.

[34] Four-hour IL1β exposure of human marrow progenitors produces monocyte offspring with enhanced heterologous cytokines and metabolism. Culture expansion/differentiation and normal growth signaling remain; no in vivo transplant or delayed maintenance blockade.

[35] T-cell CD40-TRAF6 regulates myeloid functional/metabolic/chromatin training; combined blockade promotes graft acceptance. Adaptive crosstalk and combined CTLA4-Ig effects prevent attributing outcome solely to cell-autonomous innate memory.

[36] Malaria-trained purified monocytes fail without lymphocytes; transwell CD4/IFNγ restores heterologous cytokine responses. This tests induction crosstalk, not proof that every established memory requires ongoing lymphocyte signals.

[37] Subcutaneous BCG induces AM memory via gut microbiome/metabolite changes and mycobacterial dissemination, independently of circulating monocytes. Distal persistent upstream changes provide an alternative to direct cell-intrinsic induction; linked correction unexamined.

[38] BCG protects against influenza but not SARS-CoV-2, with different tissue tropism and cytokine responses. Pathogen-specific animal benefit cannot establish universal human heterologous protection or purely innate mediation.

[39] Acute LPS generates persistent resident-AM phenotypes and pneumococcal protection with minimal SARS-CoV-2 protection. Whole-lung response and stable population size do not isolate same-cell autonomous memory or universal protection.

[40] Influenza-experienced AMs exert long-lasting tissue-specific antitumor effects with IFNγ/NK induction dependence. Human tumor association is observational; induction dependence is not continued maintenance dependence.

[41] Lineage tracing and AM transfer identify a KLF4/MERTK efferocytic trained subset that limits lung injury. Subset expansion/differentiation and tissue niche contribute; anti-inflammatory resolution differs from simply higher cytokine secretion.

[42] β-glucan protects against Mtb and expands myeloid progenitors; IL1R deficiency abolishes protection. Constitutive receptor deficiency and early stimulus do not distinguish induction/maintenance; human ex vivo response is not clinical efficacy.

[43] STAT1 CMC patient cells have Candida-specific training defects while STAT3 HIES cells train normally. Genetic pleiotropy and pathogen specificity prevent a universal innate-memory instrument or proof of antigen-free maintenance.

[44] β-glucan-experienced human macrophages reduce NLRP3/caspase1/IL1β responses, including CAPS patient cells. Cytokine attenuation can coexist with memory; ex vivo anti-inflammatory effects do not establish clinical disease trajectory.

[45] IL37 reverses training-associated metabolic/mark changes and reduces Candida protection in models. Broad anti-inflammatory effects and treatment timing do not isolate a specific self-maintaining histone carrier.

[46] β-glucan-trained differentiated macrophage transfer protects Pseudomonas; sustained metabolic/chemokine state is independent of Dectin1/TLR2 here. Constitutive chemokine secretion differs from a fully resting state; ligand clearance and feedback remain unresolved; receptor dependence differs by model.

[47] BCG outer-membrane vesicles induce TLR2-dependent progenitor expansion/metabolism and sepsis protection. Vesicle cargo/duration and experimental sepsis do not establish long-term autonomous maintenance or human safety/benefit.

[48] β-glucan nine weeks before Listeria increases antimicrobial myeloid responses and survival. Elapsed time alone does not exclude retained particulate ligand or ongoing niche stimulation; multiple population/metabolic changes coexist.

[49] IRF4 controls metabolic uptake and memory expansion of virus-specific NK cells. Antigen-specific NK memory is a comparator, not demonstration of nonspecific trained myeloid memory; generic survival effects constrain mechanism.

[50] Randomized human BCG/malaria challenge alters activation and yields earlier/more severe symptoms; parasitemia inversely correlates with activation. 20 volunteers and correlational mediation do not establish clinical protection; enhanced activation can accompany adverse effects.

[51] Randomized BCG attenuated-yellow-fever challenge reduces viremia with IL1β-associated monocyte changes. Vaccine-strain viremia is not ordinary clinical disease; correlated cytokines do not prove innate mediation or autonomous persistence.

[52] ACTIVATE interim report of 198 elderly patients finds longer time to first infection and fewer infections with BCG. Small interim analysis/probable viral diagnoses and clinical heterogeneity limit generalization; no selective trained-memory mediation test.

[53] BLOW2 randomizes 5420 NICU newborns under 2000 g to early BCG plus OPV and reduces neonatal all-cause mortality. Combined vaccines cannot isolate BCG or trained innate mechanism; infection-death secondary outcome added during trial; setting/timing-specific clinical benefit.

[54] Randomized low-birthweight infant BCG substudy shows increased heterologous cytokines four weeks later. Same Guinea-Bissau trial platform and immune substudy do not establish cytokine mediation of mortality or independent clinical replication.

[55] Danish neonatal randomized trial finds no overall benefit in parent-reported infections through 13 months. Maternal-BCG early subgroup is not the overall endpoint; low-mortality context may differ from NICU/LBW settings.

[56] 1221 Danish healthcare workers show no overall BCG benefit for absenteeism, COVID or hospitalization. Secondary revaccination analyses diverge and have wide intervals; clinical null does not disprove all cellular memory.

[57] BRACE stage-2 randomized 3988 healthcare workers shows no lower six-month symptomatic/severe COVID risk. Early recruitment stop and vaccine-related censoring limit precision; severe chiefly means incapacitated three days, not hospital disease; no verified selective innate engagement in this report.

[58] BRACE 12-month follow-up shows increased symptomatic COVID and no severe/breakthrough benefit. Same trial as six-month report, not independent replication; severity definition and evolving vaccination/exposure limit universal conclusions.

[59] Dutch healthcare-worker randomization with serologic detection shows no infection/duration/severity benefit despite early antibody effects. Small hospitalized counts and clinical context constrain severe-disease estimates; antibody enhancement is not clinical or innate-memory mediation.

[60] BCG-PRIME randomizes 6112 older comorbid adults and finds no COVID or clinically relevant respiratory benefit. Six-month regimen-specific null does not eliminate every clinical context or demonstrate selective mechanistic engagement.

[61] BRACE combined-stage secondary analysis reports modestly more febrile/respiratory illness without severe-disease benefit. 6828 combined stages differs from 3988 stage-2 primary denominator; same trial, not independent cohort.

[62] Third Guinea-Bissau early-BCG trial has lower infection-related neonatal mortality and favorable pooled neonatal estimate. Its standalone primary neonatal all-cause contrast is nonsignificant; polio-campaign censoring and pooled trials must be separated; mechanism unisolated.

[63] Early BCG in 2320 low-birthweight infants has a favorable four-week secondary mortality contrast. Primary infant mortality is nonsignificant; neonatal secondary effect and subgroup weights do not turn the primary endpoint positive or establish autonomous training.

Tally: 61 discriminating contrasts. Null and negative contrasts are discriminators, not positive threshold crossings. Reports and replications are not independent causal lines. Null RCTs and harmful transfer results are counted as discriminating contrasts, not positive causal crossings. BRACE primary, follow-up and combined-stage secondary reports are one trial platform, with different denominators. Nonimmune epidermal memory and antigen-specific NK memory are explicit comparators, not myeloid or clinical proof.

### Cap audit, precedence and Β/Γ/Δ boundary

Rubric 05 governs the immune phenomenon, conditional persistence and clinical scope. Rubric 03 separately governs the conjunctive SC2 histone-encoding mechanism and its literal co-occurrence cap. Cell-transfer evidence is not re-capped merely because another leg is associative. Conversely, transfer of function does not by itself identify a histone carrier. The primary aggregation was registered before retrieval; Γ weakest-link sensitivity is retained beside Ε rather than hidden.

Gamma hard-rule audit: Histone marks and chromatin accessibility alone do not establish a memory carrier. Induction-stage blockade is not delayed maintenance necessity. Whole-body clinical benefit does not identify the trained-innate mediator. Purified transfer with functional challenge crosses a different, context-specific boundary than isolated ex vivo cytokines.

SC2 cap: “Histone mark co-occurrence alone” Induction-stage metabolic nodes causally contribute in defined systems, but histone co-occurrence, broad p300/LDH inhibition and TF deletion do not isolate a specific inheritable chromatin carrier. The conjunctive histone-encoding claim remains Γ under rubric 03; metabolic contribution alone is stronger. A purified autoimmune HSC transfer yields enhanced function despite reduced metabolism, excluding universal increased glycolysis as the carrier.

SC5 cap: “Correlational human data plus in vitro mechanism” Clinical heterologous benefit occurs in defined vaccination contexts, but the examined trials do not selectively establish trained innate memory as the mediator. Combined BCG–OPV neonatal benefit, interim elderly infection benefit and attenuated yellow-fever viremia contrasts cannot be merged with adult COVID nulls into universal protection. Clinical efficacy in its tested setting can be Ε; trained-memory-specific clinical mediation is Γ.

SC1 field-valid definition: “
- Cell-type-specific conditional knockout or depletion
- Pathway-specific blockade (not broad immunosuppression) changing disease course
- Human targeted therapy with mechanistic alignment showing disease-modifying effect” Positive full-scope result: yes. Multiple distinct causal manipulations establish altered heterologous innate responses, including lymphocyte-deficient protection, primary human vaccination/challenge and transferred cells. Responses may be protective, harmful or tolerant; enhanced cytokines are not universal benefit. Independent assay/context classes support Ζ for the phenomenon within defined settings.

SC2 field-valid definition: “
- Endogenous degron of a condensate component with matched rescue, AND evidence that the specific condensate state (not generic protein loss) drives the phenotype
- Locus-specific disruption of looping with locus-specific effect
- Combinatorial histone-mark perturbation showing direct output dependence
- Critically: specificity against cell-cycle collapse, global transcriptional collapse, and replication stress must be demonstrated” Positive full-scope result: no. Induction-stage metabolic nodes causally contribute in defined systems, but histone co-occurrence, broad p300/LDH inhibition and TF deletion do not isolate a specific inheritable chromatin carrier. The conjunctive histone-encoding claim remains Γ under rubric 03; metabolic contribution alone is stronger. A purified autoimmune HSC transfer yields enhanced function despite reduced metabolism, excluding universal increased glycolysis as the carrier.

SC3 field-valid definition: “
- Cell-type-specific conditional knockout or depletion
- Pathway-specific blockade (not broad immunosuppression) changing disease course
- Human targeted therapy with mechanistic alignment showing disease-modifying effect” Positive full-scope result: yes. Purified LT-HSC transfer after resolved periodontitis and independent stringent autoimmune LT-HSC transfer into recipients without spontaneous autoimmune activity support donor-cell intrinsic persistence away from the original tissue input for months. This crosses the context-specific Ε boundary. It does not establish universal absence of retained molecular ligands, complete independence from recipient niches or normal intracellular maintenance, or permanent mark copying. Continued homeostatic TF activity alone does not refute original-inducer independence. Stronger signal/antigen exclusion remains an unmet scope expansion.

SC4 field-valid definition: “
- Cell-type-specific conditional knockout or depletion
- Pathway-specific blockade (not broad immunosuppression) changing disease course
- Human targeted therapy with mechanistic alignment showing disease-modifying effect” Positive full-scope result: yes. Purified donor HSCs yield altered descendants and challenge responses after 12–18 weeks, with chimerism and lineage controls. The autoimmune experiment preserves overall lineage proportions, showing that differentiation bias is not the sole explanation. Irradiated niches, clonal selection and no direct mark copying limit the broader mechanistic inference to Ε. Human HSPC/monocyte concordance does not constitute human lineage tracing.

SC5 field-valid definition: “
- Cell-type-specific conditional knockout or depletion
- Pathway-specific blockade (not broad immunosuppression) changing disease course
- Human targeted therapy with mechanistic alignment showing disease-modifying effect” Positive full-scope result: no. Clinical heterologous benefit occurs in defined vaccination contexts, but the examined trials do not selectively establish trained innate memory as the mediator. Combined BCG–OPV neonatal benefit, interim elderly infection benefit and attenuated yellow-fever viremia contrasts cannot be merged with adult COVID nulls into universal protection. Clinical efficacy in its tested setting can be Ε; trained-memory-specific clinical mediation is Γ.

### Upward threshold ladder

SC1 Δ: “Requires pathway- or cell-type-resolved causal perturbation affecting disease or phenotype.” Satisfied: yes. 

SC1 Ε: “Requires context-specific causality with explicit scope restriction.” Satisfied: yes. 

SC1 Ζ: “Requires multiple convergent causal manipulations and reduced plausibility of downstream-only interpretations.” Satisfied: yes. 

SC1 Η: “Requires robust upstream causal evidence across systems and interventions.” Satisfied: no. Independent staged interventions with clonal/lineage controls predicting functional response across the registered cell and stimulus classes, including tolerance and adverse-response boundaries.

SC2 Δ: “Requires at least one endogenous perturbation showing the proposed feature contributes causally to the claimed output.” Satisfied: no. Delayed endogenous, locus/mark-specific subtraction and matched rescue that remove recall while preserving viability, differentiation and baseline inducibility; separate mark storage from TF/metabolic maintenance.

SC3 Δ: “Requires pathway- or cell-type-resolved causal perturbation affecting disease or phenotype.” Satisfied: yes. 

SC3 Ε: “Requires context-specific causality with explicit scope restriction.” Satisfied: yes. 

SC3 Ζ: “Requires multiple convergent causal manipulations and reduced plausibility of downstream-only interpretations.” Satisfied: no. Purified-cell transfer with sensitive original-ligand quantification, delayed removal of original upstream receptor/niche input, matched resting-state controls and preserved viability; reproduce across mature cells and progenitors.

SC4 Δ: “Requires pathway- or cell-type-resolved causal perturbation affecting disease or phenotype.” Satisfied: yes. 

SC4 Ε: “Requires context-specific causality with explicit scope restriction.” Satisfied: yes. 

SC4 Ζ: “Requires multiple convergent causal manipulations and reduced plausibility of downstream-only interpretations.” Satisfied: no. Barcoded single-HSC serial transfer with matched engraftment and clone composition, recipient ligand surveillance and mark-specific functional rescue across generations.

SC5 Δ: “Requires pathway- or cell-type-resolved causal perturbation affecting disease or phenotype.” Satisfied: no. A prespecified human heterologous clinical endpoint with randomized intervention and a selective trained-memory mediator contrast, distinguishing adaptive immunity, combined vaccines and baseline exposure.

Upper ladder: Upper levels remain live. Human vaccination and clinical randomized outcomes, receptor/natural genetic defects, dose/context contrasts and orthogonal purified-cell transfers were considered. They support the response phenomenon and conditional persistence, but do not jointly identify an exclusive histone carrier, eliminate retained inputs across all compartments, or selectively mediate clinical benefit. Ethical impossibility of universal human receptor subtraction is not itself a ceiling; endogenous staged animal/cell and human natural experiments can advance the claim.

Independence: SC1 Ζ rests on different assay/context classes, not publication count. Purified transfer, lymphocyte-deficient functional challenge and randomized human exposure exclude distinct alternatives. β-glucan/BCG laboratory platforms and human marrow cohorts may share methods/participants; irradiated transfer recipients share niche confounding. Repeated BRACE reports are one randomization. No single comparison establishes literal epigenetic inheritance.

Heterologous response without adaptive lymphocytes [2, 1]: Lymphocyte-deficient survival and monocyte-pathway perturbations exclude adaptive memory as the sole explanation in those models; not every human clinical mediator.

Original-context removal and purified long-term progenitor transfer [11, 29]: Different donor disease models, purification schemes and stable chimerism reduce ongoing original-tissue input and bulk composition explanations. Shared recipient conditioning remains a failure mode.

Human intervention and functional challenge [51, 50]: Randomization and heterologous human challenge differ from mouse transplantation and isolated cytokine assays; benefit and harm bound scope rather than proving identical molecular mediation.

### Falsification and pitfalls

Universal heterologous protection fails. BRACE, BCG-PRIME and independent healthcare-worker trials are null for clinical COVID benefit; animal influenza/SARS-CoV-2 responses also differ. Additional downgrade: False.

Universal enhanced metabolism fails. Purified autoimmune LT-HSC progeny retain enhanced killing/cytokines despite lower metabolic flux and ATP. Additional downgrade: False.

Purely autonomous induction fails in some contexts. Malaria monocyte induction needs T-cell IFNγ and CD40-TRAF6 crosstalk can determine training. This does not test all later maintenance. Additional downgrade: False.

Permanent uniform persistence is unsupported. Primary human macrophage responses are stimulus specific and some fade by day 11. Months of HSC transfer memory remain finite. Additional downgrade: False.

Induction treated as maintenance. Day 0–1 metabolic/IL1 interventions, pre-training IL1R deletion and constitutive IFNAR/Cebpb deletion do not automatically prove delayed necessity.

TF association treated as demonstrated TF necessity. Sun expression/motif coupling is not a delayed TF-loss experiment; self-renewing immortalized responder-sorted cells differ from normal primary cells.

Homeostatic maintenance confused with persistent original stimulus. A memory state can require normal endogenous regulation. Resetting every TF/metabolite to naive state would erase the state being tested.

Clinical endpoint and mechanism conflated. BCG+OPV neonatal mortality benefit does not isolate BCG or trained innate mediation; viremia is not clinical disease.

Provider gap statement treated as exhaustive absence. Consensus No Papers outputs and substituted papers cannot establish scientific absence.

### Adversarial steelman: above, then below

The strongest Ζ case combines resolved-periodontitis purified HSC transfer, independent autoimmune LT-HSC transfer without recipient inflammation, acute-LPS HSC memory and virus-experienced AM maintenance without continuing T-cell help. These causal lines make a blanket downstream-only or persistent-tissue explanation implausible for all settings.

The strongest Γ case demands strict direct proof that no original ligand survives, and that no continuing upstream niche feedback maintains the state; most interventions test induction, and transferred recipients are conditioned. Mark and metabolic associations do not identify a self-copying carrier.

Reject a blanket Γ for conditional persistence because purified donor-cell transfer after source-context removal and long-term functional challenge are genuine causal evidence. Reject broad Ζ for SC3 because sensitive ligand exclusion, delayed maintenance-selective intervention and unconditioned cross-compartment replication remain incomplete. Retain Ε with a Γ strict universal extension and Γ weakest-link sensitivity. These qualifications are not post hoc changes to the registered claim.

### Final score, causal direction and M/S/D

Preregistered primary-subclaim aggregation uses SC3 and returns Ε, context-dependent causality, with Moderate confidence. Purified post-resolution and long-term donor-cell transfer support conditional intrinsic persistence without continued original tissue exposure. Complete ligand exclusion and independence from all ongoing signaling are not demonstrated; normal homeostatic maintenance is compatible with cell-intrinsic memory. The strict universal signal/antigen-free extension remains Γ. Weakest-link sensitivity is Γ because histone-carrier specificity and human clinical mediation remain unresolved; scope-matched primary sensitivity is Ε. The final letter does not certify the full conjunctive mechanism or universal clinical protection.

Prior exposure establishes a later response state; that state modulates or amplifies subsequent immunity and disease in defined settings. It may protect, worsen injury or favor resolution. It is not established as a universal initiating cause of disease or universal clinical protection.

Momentum: Mechanistic only. The named 2025 H3K18-lactylation study and 2026 primary-macrophage stimulus-specific study refine candidate carriers and temporal boundaries; no presently recruiting decisive clinical mediation trial is asserted.

Stability: Stable. No verified pending decisive result is identified as likely to move SC3 imminently; current carrier and clinical-mediation limits remain open.

Divergence: Contested; no single implied-level arithmetic. These dimensions do not change level or confidence.

## PROPOSED DECISIVE EXPERIMENTS

After a defined nonreplicating inducer has cleared by sensitive labeled-ligand and molecular assays, isolate barcoded LT-HSCs and self-renewing primary macrophages. Randomize delayed cell-specific removal of original receptor/niche signaling, candidate chromatin-maintenance machinery and matched rescue. Transfer into ligand-free recipients with matched conditioning and engraftment; follow clones, baseline competence and heterologous challenge function over months.

Readout: Single-clone descendants, preserved viability/baseline inducibility, ligand burden, mark/TF/metabolic trajectories and challenge killing or disease phenotype. Controls: Naive transferred cells; ligand-spiked assay sensitivity; induction-only versus delayed blockade; clone and lineage matching; inactive degron/rescue; nonconditioned or matched-niche comparison; adaptive-cell controls. Supporting: Recall persists after verified original-input removal, disappears selectively after delayed candidate-carrier subtraction and returns with matched rescue without baseline impairment. Falsifying: Recall tracks detectable original ligand/niche input, disappears with input removal rather than carrier-specific subtraction, or apparent loss is explained by clone selection or general cell dysfunction.

## RESEARCH TIMELINE

| Year | Milestone | Primary source | Interpretation |
|---|---|---|---|
| 2011 | Early BCG in low-birthweight infants | [PMID21673035](https://pubmed.ncbi.nlm.nih.gov/21673035/) | Primary infant mortality null; secondary neonatal signal. |
| 2012 | BCG heterologous monocyte response | [PMID22988082](https://pubmed.ncbi.nlm.nih.gov/22988082/) | Three-month response is not global ligand clearance. |
| 2012 | Candida/β-glucan innate functional training | [PMID22901542](https://pubmed.ncbi.nlm.nih.gov/22901542/) | Lymphocyte-deficient protection. |
| 2014 | HIF-1α metabolic induction | [PMID25258083](https://pubmed.ncbi.nlm.nih.gov/25258083/) | Myeloid pathway necessity; no indexed correction record. |
| 2016 | Fumarate/KDM5 metabolic–chromatin link | [PMID27866838](https://pubmed.ncbi.nlm.nih.gov/27866838/) | Broad metabolic perturbation, not literal mark inheritance. |
| 2018 | Mevalonate pathway | [PMID29328908](https://pubmed.ncbi.nlm.nih.gov/29328908/) | Constitutive metabolite excess versus post-exposure memory. |
| 2018 | Purified β-glucan HSC transfer | [PMID29328910](https://pubmed.ncbi.nlm.nih.gov/29328910/) | Long-term lineage bias; induction blockers early. |
| 2018 | BCG HSC education | [PMID29328912](https://pubmed.ncbi.nlm.nih.gov/29328912/) | Potential marrow antigen persistence. |
| 2018 | Human yellow-fever vaccine challenge | [PMID29324233](https://pubmed.ncbi.nlm.nih.gov/29324233/) | Viremia benefit, clinical/mediator limits. |
| 2018 | Long-lived microglial training/tolerance | [PMID29643512](https://pubmed.ncbi.nlm.nih.gov/29643512/) | Opposite disease effects by exposure schedule. |
| 2018 | Virus-experienced AM memory | [PMID30433869](https://pubmed.ncbi.nlm.nih.gov/30433869/) | Reported T-cell priming versus maintenance distinction. |
| 2020 | Acute-LPS HSC enhancer memory | [PMID32169166](https://pubmed.ncbi.nlm.nih.gov/32169166/) | C/EBPβ result; exact timing/correction details unexamined. |
| 2020 | Human marrow after BCG | [PMID32544459](https://pubmed.ncbi.nlm.nih.gov/32544459/) | Day-90 signatures; no direct lineage tracing. |
| 2020 | ACTIVATE clinical infections | [PMID32941801](https://pubmed.ncbi.nlm.nih.gov/32941801/) | Interim context-specific benefit. |
| 2021 | Diabetic donor marrow transfer | [PMID34255973](https://pubmed.ncbi.nlm.nih.gov/34255973/) | Recipient disease amplification with niche limits. |
| 2021 | Epidermal stem-cell memory comparator | [PMID34320411](https://pubmed.ncbi.nlm.nih.gov/34320411/) | Delayed STAT3 loss preserves recall outside myeloid scope. |
| 2022 | Resolved-periodontitis LT-HSC transfer | [PMID35483374](https://pubmed.ncbi.nlm.nih.gov/35483374/) | Purified long-term intrinsic susceptibility. |
| 2022 | AM transfer benefit/harm divergence | [PMID35856089](https://pubmed.ncbi.nlm.nih.gov/35856089/) | Whole exposure and transferred cells differ. |
| 2023 | Memory through immortalized-cell divisions | [PMID36999024](https://pubmed.ncbi.nlm.nih.gov/36999024/) | TF necessity inferred, not experimentally subtracted. |
| 2023 | BRACE COVID primary null | [PMID37099341](https://pubmed.ncbi.nlm.nih.gov/37099341/) | Six-month trial; severe definition includes incapacitation. |
| 2023 | BCG-PRIME COVID null | [PMID36736662](https://pubmed.ncbi.nlm.nih.gov/36736662/) | 6112 older comorbid adults. |
| 2024 | Purified autoimmune LT-HSC transfer | [PMID39413777](https://pubmed.ncbi.nlm.nih.gov/39413777/) | Enhanced function without increased metabolism or lineage skew. |
| 2024 | BRACE 12-month follow-up | [PMID39127450](https://pubmed.ncbi.nlm.nih.gov/39127450/) | Same randomized cohort; no protective clinical result. |
| 2025 | BLOW2 neonatal BCG+OPV trial | [PMID40983468](https://pubmed.ncbi.nlm.nih.gov/40983468/) | Setting-specific mortality benefit; combined intervention. |
| 2025 | H3K18 lactylation candidate | [PMID40318634](https://pubmed.ncbi.nlm.nih.gov/40318634/) | Broad inhibitors do not isolate mark-specific storage. |
| 2026 | Stimulus-specific primary macrophage memory | [PMID42320478](https://pubmed.ncbi.nlm.nih.gov/42320478/) | Finite persistence; preprint counted once with published report. |

## KNOWLEDGE GAPS

Sensitive original-ligand quantification and delayed source-input removal are missing across the main mature-cell and progenitor contexts; elapsed time or washout alone is insufficient.

Separate induction from maintenance with delayed, cell-specific intervention and preserved baseline competence; most examined metabolic/receptor knockouts intervene early.

Identify specific chromatin marks/readers as causal carriers with endogenous, locus-relevant subtraction and rescue, rather than broad p300/LDH effects or histone correlation.

Disentangle clone selection, lineage propensity and literal information transmission using barcoded single-HSC serial transfer and matched engraftment.

Test whether conditional memory persists in an unconditioned physiological niche; irradiation and support marrow can reset selection and feedback.

Explain enhanced function with reduced metabolism in autoimmune-derived progeny; a single glycolytic-memory signature cannot cover every state.

Map induction stimulus, cell class, dose and duration boundaries, including fading primary-macrophage memory, tolerance, resolution and harmful transfer phenotypes.

Identify selective innate-memory mediation of human clinical benefit separately from adaptive responses, combined vaccines, exposure and setting; preserve adult clinical nulls.

Resolve the linked correction details for de Laval (PMID 32386557, PMID 36608675) and the parenteral-BCG gut-lung report (author correction PMID 37161093) before publication-grade claims about exact assays or maintenance timing; no correction record is indexed for the HIF-1α report.

Resolve human marrow cohort overlap and obtain complete Methods, supplements and raw-data/image checks for the major purified-transfer and clinical claims.

## METHODS AND CONSENSUS CONTRIBUTION

Consensus was used through the signed-in browser for discovery, counterevidence, research-gap and timeline suggestions, and attempted checks of specific indexed full-text passages. Returned Methods/Results passages were reconciled with original-paper passages where available; failed retrievals, source substitutions and unverified aspects are recorded in the case reconciliation. Successful passage checks are distinguished from provider summaries. Provider screening counts, citation counts and meters were not used as primary counts or HSEF scores. This source log records access depth; it does not claim complete-paper or supplement review. The monthly API allowance was conserved: zero Consensus API calls in this batch.

## REVIEWER NOTES

63 primary reports examined: eight selected partial-body and 55 complete abstract/status examinations. No whole-paper or supplement audit. SC2 is explicitly routed to rubric 03; other subclaims use rubric 05. Primary SC3 Ε is conditional; strict universal autonomy and weakest-link sensitivity are Γ. Two Consensus Pro queries, zero API/Deep; one SciSpace discovery query and two Scite denied-body attempts. Corrections are status links, not counted studies. Prior ranking/index guidance is visible, and carried controls are not fresh blinded calibration.

Search leads, reviews, protocols and correction notices do not pad the primary count. Downloaded XML does not earn examination credit. Linked notices remain attached; unexamined correction details limit inference. This is a coordinator source/claim consistency assessment, not independent blinded calibration.

## ADVERSARIAL REVIEW AND APPROVAL

**Authoring model:** Sol 6.1. **Adversarial reviewer:** Claude Opus 5.5 (Anthropic). **Date:** 2026-10-14.

**Scope of checks.** PMIDs verified 63/63 against PubMed `get_article_metadata`: title, first author, journal and DOI
matched for every reference; 0 unresolvable, 0 wrong-record substitutions. Retraction / expression-of-concern screen
run on all 63 records: none found. Three linked corrections were confirmed to exist and to attach to the right parents
(PMID 32386557 and PMID 36608675 to de Laval [10]; PMID 37161093, an Author Correction, to Jeyanathan [37]); a fourth
claimed correction, asserted for the HIF-1α report [3] with no identifier, has no erratum record indexed in PubMed and
the claim was withdrawn. 13 load-bearing claims spot-checked against primary abstracts (BRACE n=3988 and its
incapacitated-for-three-days severity definition; BRACE 12-month increased symptomatic COVID; BRACE combined-stage
n=6828; BCG-PRIME n=6112; Danish healthcare workers n=1221; Dutch healthcare-worker serology; ACTIVATE n=198 interim;
BLOW2 n=5420 <2000 g with the infection-mortality secondary added partway through the trial; controlled human malaria
n=20 with earlier and more severe symptoms; yellow-fever viremia with IL-1β correlation; Aaby 2011 n=2320 with a null
primary and a favourable four-week secondary; Biering-Sørensen 2017 nonsignificant neonatal primary with a 43%
infectious-death reduction; Mills 2024 decoupling of function from glycolysis) — all matched. Scoring logic re-derived
against `hsef-hypothesis-review-v5/SKILL.md` and `references/hsef-framework.md`. Internal consistency checked across
review.md, scorecard.json and validation-report.json.

| ID | Severity | Location | Original → corrected | Evidence |
|---|---|---|---|---|
| TI-1 | Major | Consensus provenance reconciliation | "The earlier adversarial approval refers to its original edition" → "No earlier adversarial approval exists for any edition of this review" | No edition of this review carries an `ADVERSARIAL REVIEW AND APPROVAL` section or an approval line; the 2026-10-06 adversarial pass covered 30 other slugs, not this one. The spliced sentence asserted a provenance that does not exist. |
| TI-2 | Minor | Evidence card [3] status line | "ErratumIn: PMIDnot supplied." → explicit correction-screen token with `status_checked_on` | Two defects in one string: the renderer emitted the placeholder "PMIDnot supplied", and PubMed holds no erratum, corrigendum or retraction record for PMID 25258083. A status flag may not assert a correction it cannot identify. |
| TI-3 | Minor | Decisive-perturbation item [3], card [3], RESEARCH TIMELINE 2014 row | "linked author correction noted" / "linked correction" → removed / "no indexed correction record" | Same evidence as TI-2; the dependent prose repeated the unverifiable correction claim in three places. |
| TI-4 | Minor | KNOWLEDGE GAPS, correction-resolution item | "linked correction details for de Laval, HIF-1α and microbiome/BCG reports" → the two de Laval errata and the gut-lung Author Correction named by PMID, with the HIF-1α claim withdrawn | The gap statement grouped a nonexistent correction with two real ones; naming the identifiers makes the residual limitation checkable. |
| TI-5 | Major | AT A GLANCE (new **Confidence basis** paragraph); `final_score.confidence_rationale` and its two mirrored copies | "density, heterogeneous controls and incomplete access set the confidence" → rationale grounded in evidence density and reproducibility, with reviewer access depth explicitly excluded | Lead policy, and `hsef-framework.md` § Confidence band: the band reflects evidence density and reproducibility. How much of a paper the reviewer could read is neither. Applying the lead's two-step test, the band was then re-derived from what remains and **Moderate stands**: the review documents substantive evidence-quality concerns that are valid determinants — irradiated recipients and conditioning effects, few laboratories behind the purified-transfer result, induction-stage interventions standing in for maintenance tests, and a clinical arm split between replicated adult nulls and single-setting neonatal benefits. |

**Editorial fix applied without a finding ID:** "Constitututive" → "Constitutive" (RESEARCH TIMELINE, 2018 mevalonate row).

**Scoring logic — checked and upheld.** Hard caps are applied per subclaim before thresholds. The declared
`primary-subclaim` rule is applied as written: SC3 is named the primary subclaim and the overall letter is taken from
it (Ε), not from the highest subclaim (SC1 Ζ). The reported alternatives are arithmetically correct — weakest-link Γ
(SC2 and SC5 are the floor), scope-matched Ε. Cap 1 (in vitro alone cannot exceed Γ) and the rubric-03 histone
co-occurrence cap are correctly the binding constraints on SC2; the review correctly declines to re-cap the
cell-transfer evidence because another leg is associative, which is what the cap-precedence rule requires. The Ζ on
SC1 carries a written independence basis naming the distinct systematic errors each line excludes, as a Ζ or above
must. Momentum "Mechanistic only" is correct and was not inflated: the named artifacts (the 2025 histone-lactylation
report and the 2026 primary-macrophage report) are completed publications, which under lead policy can never satisfy
"Decisive work underway". Divergence is recorded as **Contested**, which the framework permits without implied-level
arithmetic where the field is genuinely split; the adult clinical nulls and the neonatal-mortality literature make
that the right value here.

**Rejected findings (checked, no change).**
1. *"Best-supported rival" in AT A GLANCE naming the dose-and-context set-point model.* Checked against the score
   card: that entry is `rank_in_tournament: 1` with `is_nominal: false`, and the review's own nominal (cell-intrinsic
   memory with unresolved carrier, M1) ranks 2. The renderer defect seen elsewhere — printing the review's own
   nominal as its best-supported rival — did not occur here. No change.
2. *Three references whose print year differs by one from the PubMed e-pub year* (Domínguez-Andrés, Cell Metab
   2019/2018; Lechner, JACI 2022/2021; Jensen, J Infect Dis 2015/2014). Rejected per the standing convention.
3. *Gaps at [4] and [30] in the decisive-perturbation inventory.* The inventory is numbered against the citation
   ledger, and those two ledger rows carry no discriminating contrast. 61 numbered items, tally reports 61. Consistent.
4. *The inline `PMID22901542` token form in the prose, against `PMID 22901542` in REFERENCES.* A systematic house
   token used identically in the `Study:` ledger field throughout the corpus, not a dropped space inside a word. Not
   rewritten.
5. *`v5.claim_form` "phenomenon" warning in the validator report.* Rejected per the standing convention: the
   registered claim states acquisition and persistence **plus** the enhanced-response consequence.
6. *Scope-matched aggregation.* Not the declared rule here, and the headline is not raised above the scope-matched
   result in any case, so the misread the framework warns about cannot arise.

**Residual limitations.** Access depth is unchanged by this pass: eight partial-body and 55 abstract/status
examinations, no complete-paper or supplement audit. The claims that the Cirovic cohort comprised 15 BCG recipients
against five controls, that the Mills transfer endpoint was 18 weeks, and that recipient lineage proportions were
preserved, rest on the original reviewer's partial-body reading and could not be re-verified from abstracts; they are
not contradicted by any record examined. The two de Laval errata and the gut-lung Author Correction remain unexamined
in substance.

**Approval:** Following adversarial review and the corrections listed above, the revised document is approved. — Claude Opus 5.5 (Anthropic), 2026-10-14.

## REFERENCES

1. Kleinnijenhuis J, Quintin J, Preijers F et al. Bacille Calmette-Guerin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes. Proceedings of the National Academy of Sciences of the United States of America. 2012. [PMID 22988082](https://pubmed.ncbi.nlm.nih.gov/22988082/). DOI: 10.1073/pnas.1202870109. Access: abstract_metadata. Study: PMID22988082.

2. Quintin J, Saeed S, Martens JHA et al. Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes. Cell host & microbe. 2012. [PMID 22901542](https://pubmed.ncbi.nlm.nih.gov/22901542/). DOI: 10.1016/j.chom.2012.06.006. Access: abstract_metadata. Study: PMID22901542.

3. Cheng SC, Quintin J, Cramer RA et al. mTOR- and HIF-1α-mediated aerobic glycolysis as metabolic basis for trained immunity. Science (New York, N.Y.). 2014. [PMID 25258083](https://pubmed.ncbi.nlm.nih.gov/25258083/). DOI: 10.1126/science.1250684. Access: abstract_metadata. Study: PMID25258083.

4. Saeed S, Quintin J, Kerstens HH et al. Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity. Science (New York, N.Y.). 2014. [PMID 25258085](https://pubmed.ncbi.nlm.nih.gov/25258085/). DOI: 10.1126/science.1251086. Access: abstract_metadata. Study: PMID25258085.

5. Arts RJW, Carvalho A, La Rocca C et al. Immunometabolic Pathways in BCG-Induced Trained Immunity. Cell reports. 2016. [PMID 27926861](https://pubmed.ncbi.nlm.nih.gov/27926861/). DOI: 10.1016/j.celrep.2016.11.011. Access: abstract_metadata. Study: PMID27926861.

6. Arts RJ, Novakovic B, Ter Horst R et al. Glutaminolysis and Fumarate Accumulation Integrate Immunometabolic and Epigenetic Programs in Trained Immunity. Cell metabolism. 2016. [PMID 27866838](https://pubmed.ncbi.nlm.nih.gov/27866838/). DOI: 10.1016/j.cmet.2016.10.008. Access: abstract_metadata. Study: PMID27866838.

7. Bekkering S, Arts RJW, Novakovic B et al. Metabolic Induction of Trained Immunity through the Mevalonate Pathway. Cell. 2018. [PMID 29328908](https://pubmed.ncbi.nlm.nih.gov/29328908/). DOI: 10.1016/j.cell.2017.11.025. Access: abstract_metadata. Study: PMID29328908.

8. Mitroulis I, Ruppova K, Wang B et al. Modulation of Myelopoiesis Progenitors Is an Integral Component of Trained Immunity. Cell. 2018. [PMID 29328910](https://pubmed.ncbi.nlm.nih.gov/29328910/). DOI: 10.1016/j.cell.2017.11.034. Access: full_text_partial. Study: PMID29328910.

9. Kaufmann E, Sanz J, Dunn JL et al. BCG Educates Hematopoietic Stem Cells to Generate Protective Innate Immunity against Tuberculosis. Cell. 2018. [PMID 29328912](https://pubmed.ncbi.nlm.nih.gov/29328912/). DOI: 10.1016/j.cell.2017.12.031. Access: abstract_metadata. Study: PMID29328912.

10. de Laval B, Maurizio J, Kandalla PK et al. C/EBPβ-Dependent Epigenetic Memory Induces Trained Immunity in Hematopoietic Stem Cells. Cell stem cell. 2020. [PMID 32169166](https://pubmed.ncbi.nlm.nih.gov/32169166/). DOI: 10.1016/j.stem.2020.01.017. Access: abstract_metadata. Study: PMID32169166.

11. Li X, Wang H, Yu X et al. Maladaptive innate immune training of myelopoiesis links inflammatory comorbidities. Cell. 2022. [PMID 35483374](https://pubmed.ncbi.nlm.nih.gov/35483374/). DOI: 10.1016/j.cell.2022.03.043. Access: full_text_partial. Study: PMID35483374.

12. Christ A, Günther P, Lauterbach MAR et al. Western Diet Triggers NLRP3-Dependent Innate Immune Reprogramming. Cell. 2018. [PMID 29328911](https://pubmed.ncbi.nlm.nih.gov/29328911/). DOI: 10.1016/j.cell.2017.12.013. Access: abstract_metadata. Study: PMID29328911.

13. Edgar L, Akbar N, Braithwaite AT et al. Hyperglycemia Induces Trained Immunity in Macrophages and Their Precursors and Promotes Atherosclerosis. Circulation. 2021. [PMID 34255973](https://pubmed.ncbi.nlm.nih.gov/34255973/). DOI: 10.1161/CIRCULATIONAHA.120.046464. Access: abstract_metadata. Study: PMID34255973.

14. Cirovic B, de Bree LCJ, Groh L et al. BCG Vaccination in Humans Elicits Trained Immunity via the Hematopoietic Progenitor Compartment. Cell host & microbe. 2020. [PMID 32544459](https://pubmed.ncbi.nlm.nih.gov/32544459/). DOI: 10.1016/j.chom.2020.05.014. Access: full_text_partial. Study: human-BCG-marrow-platform-overlap-not-resolved.

15. Sun SJ, Aguirre-Gamboa R, de Bree LCJ et al. BCG vaccination alters the epigenetic landscape of progenitor cells in human bone marrow to influence innate immune responses. Immunity. 2024. [PMID 39153479](https://pubmed.ncbi.nlm.nih.gov/39153479/). DOI: 10.1016/j.immuni.2024.07.021. Access: abstract_metadata. Study: human-BCG-marrow-platform-overlap-not-resolved.

16. Moorlag SJCFM, Rodriguez-Rosales YA, Gillard J et al. BCG Vaccination Induces Long-Term Functional Reprogramming of Human Neutrophils. Cell reports. 2020. [PMID 33207187](https://pubmed.ncbi.nlm.nih.gov/33207187/). DOI: 10.1016/j.celrep.2020.108387. Access: abstract_metadata. Study: PMID33207187.

17. Yao Y, Jeyanathan M, Haddadi S et al. Induction of Autonomous Memory Alveolar Macrophages Requires T Cell Help and Is Critical to Trained Immunity. Cell. 2018. [PMID 30433869](https://pubmed.ncbi.nlm.nih.gov/30433869/). DOI: 10.1016/j.cell.2018.09.042. Access: abstract_metadata. Study: PMID30433869.

18. Zahalka S, Starkl P, Watzenboeck ML et al. Trained immunity of alveolar macrophages requires metabolic rewiring and type 1 interferon signaling. Mucosal immunology. 2022. [PMID 35856089](https://pubmed.ncbi.nlm.nih.gov/35856089/). DOI: 10.1038/s41385-022-00528-5. Access: full_text_partial. Study: PMID35856089.

19. Khan N, Downey J, Sanz J et al. M. tuberculosis Reprograms Hematopoietic Stem Cells to Limit Myelopoiesis and Impair Trained Immunity. Cell. 2020. [PMID 33125891](https://pubmed.ncbi.nlm.nih.gov/33125891/). DOI: 10.1016/j.cell.2020.09.062. Access: abstract_metadata. Study: PMID33125891.

20. Novakovic B, Habibi E, Wang SY et al. β-Glucan Reverses the Epigenetic State of LPS-Induced Immunological Tolerance. Cell. 2016. [PMID 27863248](https://pubmed.ncbi.nlm.nih.gov/27863248/). DOI: 10.1016/j.cell.2016.09.034. Access: abstract_metadata. Study: PMID27863248.

21. Domínguez-Andrés J, Novakovic B, Li Y et al. The Itaconate Pathway Is a Central Regulatory Node Linking Innate Immune Tolerance and Trained Immunity. Cell metabolism. 2019. [PMID 30293776](https://pubmed.ncbi.nlm.nih.gov/30293776/). DOI: 10.1016/j.cmet.2018.09.003. Access: abstract_metadata. Study: PMID30293776.

22. Sun S, Aguirre-Gamboa R, Barreiro LB et al. Transmission of stimulus-induced epigenetic changes through cell division is coupled to continuous transcription factor activity. Frontiers in immunology. 2023. [PMID 36999024](https://pubmed.ncbi.nlm.nih.gov/36999024/). DOI: 10.3389/fimmu.2023.1129577. Access: full_text_partial. Study: PMID36999024.

23. Ziogas A, Novakovic B, Ventriglia L et al. Long-term histone lactylation connects metabolic and epigenetic rewiring in innate immune memory. Cell. 2025. [PMID 40318634](https://pubmed.ncbi.nlm.nih.gov/40318634/). DOI: 10.1016/j.cell.2025.03.048. Access: abstract_metadata. Study: PMID40318634.

24. O'Farrell A, Niu Z, Li J et al. Human macrophages encode stimulus-specific information of prior exposures through trained immunity. Cell systems. 2026. [PMID 42320478](https://pubmed.ncbi.nlm.nih.gov/42320478/). DOI: 10.1016/j.cels.2026.101647. Access: abstract_metadata. Study: PMID42320478.

25. Zhang B, Moorlag SJ, Dominguez-Andres J et al. Single-cell RNA sequencing reveals induction of distinct trained-immunity programs in human monocytes. The Journal of clinical investigation. 2022. [PMID 35133977](https://pubmed.ncbi.nlm.nih.gov/35133977/). DOI: 10.1172/JCI147719. Access: abstract_metadata. Study: PMID35133977.

26. Larsen SB, Cowley CJ, Sajjath SM et al. Establishment, maintenance, and recall of inflammatory memory. Cell stem cell. 2021. [PMID 34320411](https://pubmed.ncbi.nlm.nih.gov/34320411/). DOI: 10.1016/j.stem.2021.07.001. Access: full_text_partial. Study: PMID34320411.

27. Wendeln AC, Degenhardt K, Kaurani L et al. Innate immune memory in the brain shapes neurological disease hallmarks. Nature. 2018. [PMID 29643512](https://pubmed.ncbi.nlm.nih.gov/29643512/). DOI: 10.1038/s41586-018-0023-4. Access: abstract_metadata. Study: PMID29643512.

28. Lechner A, Henkel FDR, Hartung F et al. Macrophages acquire a TNF-dependent inflammatory memory in allergic asthma. The Journal of allergy and clinical immunology. 2022. [PMID 34974067](https://pubmed.ncbi.nlm.nih.gov/34974067/). DOI: 10.1016/j.jaci.2021.11.026. Access: abstract_metadata. Study: PMID34974067.

29. Mills TS, Kain B, Burchill MA et al. A distinct metabolic and epigenetic state drives trained immunity in HSC-derived macrophages from autoimmune mice. Cell stem cell. 2024. [PMID 39413777](https://pubmed.ncbi.nlm.nih.gov/39413777/). DOI: 10.1016/j.stem.2024.09.010. Access: full_text_partial. Study: PMID39413777.

30. Cheong JG, Ravishankar A, Sharma S et al. Epigenetic memory of coronavirus infection in innate immune cells and their progenitors. Cell. 2023. [PMID 37597510](https://pubmed.ncbi.nlm.nih.gov/37597510/). DOI: 10.1016/j.cell.2023.07.019. Access: abstract_metadata. Study: PMID37597510.

31. Nakayama Y, Fujiu K, Oshima T et al. Heart failure promotes multimorbidity through innate immune memory. Science immunology. 2024. [PMID 38787963](https://pubmed.ncbi.nlm.nih.gov/38787963/). DOI: 10.1126/sciimmunol.ade3814. Access: abstract_metadata. Study: PMID38787963.

32. Simats A, Zhang S, Messerer D et al. Innate immune memory after brain injury drives inflammatory cardiac dysfunction. Cell. 2024. [PMID 39043180](https://pubmed.ncbi.nlm.nih.gov/39043180/). DOI: 10.1016/j.cell.2024.06.028. Access: abstract_metadata. Study: PMID39043180.

33. Rehill AM, McCluskey S, Ledwith AE et al. Trained immunity causes myeloid cell hypercoagulability. Science advances. 2025. [PMID 40053582](https://pubmed.ncbi.nlm.nih.gov/40053582/). DOI: 10.1126/sciadv.ads0105. Access: abstract_metadata. Study: PMID40053582.

34. Flores-Gomez D, Hobo W, van Ens D et al. Interleukin-1β induces trained innate immunity in human hematopoietic progenitor cells in vitro. Stem cell reports. 2024. [PMID 39515317](https://pubmed.ncbi.nlm.nih.gov/39515317/). DOI: 10.1016/j.stemcr.2024.09.004. Access: abstract_metadata. Study: PMID39515317.

35. Jacobs MME, Maas RJF, Jonkman I et al. Trained immunity is regulated by T cell-induced CD40-TRAF6 signaling. Cell reports. 2024. [PMID 39178113](https://pubmed.ncbi.nlm.nih.gov/39178113/). DOI: 10.1016/j.celrep.2024.114664. Access: abstract_metadata. Study: PMID39178113.

36. Crabtree JN, Caffrey DR, de Souza Silva L et al. Lymphocyte crosstalk is required for monocyte-intrinsic trained immunity to Plasmodium falciparum. The Journal of clinical investigation. 2022. [PMID 35642634](https://pubmed.ncbi.nlm.nih.gov/35642634/). DOI: 10.1172/JCI139298. Access: abstract_metadata. Study: PMID35642634.

37. Jeyanathan M, Vaseghi-Shanjani M, Afkhami S et al. Parenteral BCG vaccine induces lung-resident memory macrophages and trained immunity via the gut-lung axis. Nature immunology. 2022. [PMID 36456739](https://pubmed.ncbi.nlm.nih.gov/36456739/). DOI: 10.1038/s41590-022-01354-4. Access: abstract_metadata. Study: PMID36456739.

38. Kaufmann E, Khan N, Tran KA et al. BCG vaccination provides protection against IAV but not SARS-CoV-2. Cell reports. 2022. [PMID 35235831](https://pubmed.ncbi.nlm.nih.gov/35235831/). DOI: 10.1016/j.celrep.2022.110502. Access: abstract_metadata. Study: PMID35235831.

39. Kang A, Ye G, Afkhami S et al. LPS-induced lung tissue-resident trained innate immunity provides differential protection against pneumococci and SARS-CoV-2. Cell reports. 2024. [PMID 39383035](https://pubmed.ncbi.nlm.nih.gov/39383035/). DOI: 10.1016/j.celrep.2024.114849. Access: abstract_metadata. Study: PMID39383035.

40. Wang T, Zhang J, Wang Y et al. Influenza-trained mucosal-resident alveolar macrophages confer long-term antitumor immunity in the lungs. Nature immunology. 2023. [PMID 36807642](https://pubmed.ncbi.nlm.nih.gov/36807642/). DOI: 10.1038/s41590-023-01428-x. Access: abstract_metadata. Study: PMID36807642.

41. Chakraborty S, Singh A, Wang L et al. Trained immunity of alveolar macrophages enhances injury resolution via KLF4-MERTK-mediated efferocytosis. The Journal of experimental medicine. 2023. [PMID 37615937](https://pubmed.ncbi.nlm.nih.gov/37615937/). DOI: 10.1084/jem.20221388. Access: abstract_metadata. Study: PMID37615937.

42. Moorlag SJCFM, Khan N, Novakovic B et al. β-Glucan Induces Protective Trained Immunity against Mycobacterium tuberculosis Infection: A Key Role for IL-1. Cell reports. 2020. [PMID 32433977](https://pubmed.ncbi.nlm.nih.gov/32433977/). DOI: 10.1016/j.celrep.2020.107634. Access: abstract_metadata. Study: PMID32433977.

43. Ifrim DC, Quintin J, Meerstein-Kessel L et al. Defective trained immunity in patients with STAT-1-dependent chronic mucocutaneaous candidiasis. Clinical and experimental immunology. 2015. [PMID 25880788](https://pubmed.ncbi.nlm.nih.gov/25880788/). DOI: 10.1111/cei.12642. Access: abstract_metadata. Study: PMID25880788.

44. Camilli G, Bohm M, Piffer AC et al. β-Glucan-induced reprogramming of human macrophages inhibits NLRP3 inflammasome activation in cryopyrinopathies. The Journal of clinical investigation. 2020. [PMID 32716363](https://pubmed.ncbi.nlm.nih.gov/32716363/). DOI: 10.1172/JCI134778. Access: abstract_metadata. Study: PMID32716363.

45. Cavalli G, Tengesdal IW, Gresnigt M et al. The anti-inflammatory cytokine interleukin-37 is an inhibitor of trained immunity. Cell reports. 2021. [PMID 33826894](https://pubmed.ncbi.nlm.nih.gov/33826894/). DOI: 10.1016/j.celrep.2021.108955. Access: abstract_metadata. Study: PMID33826894.

46. Stothers CL, Burelbach KR, Owen AM et al. β-Glucan Induces Distinct and Protective Innate Immune Memory in Differentiated Macrophages. Journal of immunology (Baltimore, Md. : 1950). 2021. [PMID 34740960](https://pubmed.ncbi.nlm.nih.gov/34740960/). DOI: 10.4049/jimmunol.2100107. Access: abstract_metadata. Study: PMID34740960.

47. Gong Y, Hao W, Xu L et al. BCG-Derived Outer Membrane Vesicles Induce TLR2-Dependent Trained Immunity to Protect Against Polymicrobial Sepsis. Advanced science (Weinheim, Baden-Wurttemberg, Germany). 2025. [PMID 40552375](https://pubmed.ncbi.nlm.nih.gov/40552375/). DOI: 10.1002/advs.202504101. Access: abstract_metadata. Study: PMID40552375.

48. Théroude C, Reverte M, Heinonen T et al. Trained Immunity Confers Prolonged Protection From Listeriosis. Frontiers in immunology. 2021. [PMID 34603295](https://pubmed.ncbi.nlm.nih.gov/34603295/). DOI: 10.3389/fimmu.2021.723393. Access: abstract_metadata. Study: PMID34603295.

49. Santosa EK, Kim H, Rückert T et al. Control of nutrient uptake by IRF4 orchestrates innate immune memory. Nature immunology. 2023. [PMID 37697097](https://pubmed.ncbi.nlm.nih.gov/37697097/). DOI: 10.1038/s41590-023-01620-z. Access: abstract_metadata. Study: PMID37697097.

50. Walk J, de Bree LCJ, Graumans W et al. Outcomes of controlled human malaria infection after BCG vaccination. Nature communications. 2019. [PMID 30787276](https://pubmed.ncbi.nlm.nih.gov/30787276/). DOI: 10.1038/s41467-019-08659-3. Access: abstract_metadata. Study: PMID30787276.

51. Arts RJW, Moorlag SJCFM, Novakovic B et al. BCG Vaccination Protects against Experimental Viral Infection in Humans through the Induction of Cytokines Associated with Trained Immunity. Cell host & microbe. 2018. [PMID 29324233](https://pubmed.ncbi.nlm.nih.gov/29324233/). DOI: 10.1016/j.chom.2017.12.010. Access: abstract_metadata. Study: PMID29324233.

52. Giamarellos-Bourboulis EJ, Tsilika M, Moorlag S et al. Activate: Randomized Clinical Trial of BCG Vaccination against Infection in the Elderly. Cell. 2020. [PMID 32941801](https://pubmed.ncbi.nlm.nih.gov/32941801/). DOI: 10.1016/j.cell.2020.08.051. Access: abstract_metadata. Study: PMID32941801.

53. Adhisivam B, Kamalarathnam C, Bhat BV et al. Effect of BCG Danish and oral polio vaccine on neonatal mortality in newborn babies weighing less than 2000 g in India: multicentre open label randomised controlled trial (BLOW2). BMJ (Clinical research ed.). 2025. [PMID 40983468](https://pubmed.ncbi.nlm.nih.gov/40983468/). DOI: 10.1136/bmj-2025-084745. Access: abstract_metadata. Study: PMID40983468.

54. Jensen KJ, Larsen N, Biering-Sørensen S et al. Heterologous immunological effects of early BCG vaccination in low-birth-weight infants in Guinea-Bissau: a randomized-controlled trial. The Journal of infectious diseases. 2015. [PMID 25210141](https://pubmed.ncbi.nlm.nih.gov/25210141/). DOI: 10.1093/infdis/jiu508. Access: abstract_metadata. Study: Guinea-Bissau-LBW-BCG-trial-platform.

55. Kjærgaard J, Birk NM, Nissen TN et al. Nonspecific effect of BCG vaccination at birth on early childhood infections: a randomized, clinical multicenter trial. Pediatric research. 2016. [PMID 27429204](https://pubmed.ncbi.nlm.nih.gov/27429204/). DOI: 10.1038/pr.2016.142. Access: abstract_metadata. Study: PMID27429204.

56. Madsen AMR, Schaltz-Buchholzer F, Nielsen S et al. Using BCG Vaccine to Enhance Nonspecific Protection of Health Care Workers During the COVID-19 Pandemic: A Randomized Controlled Trial. The Journal of infectious diseases. 2024. [PMID 37774494](https://pubmed.ncbi.nlm.nih.gov/37774494/). DOI: 10.1093/infdis/jiad422. Access: abstract_metadata. Study: PMID37774494.

57. Pittet LF, Messina NL, Orsini F et al. Randomized Trial of BCG Vaccine to Protect against Covid-19 in Health Care Workers. The New England journal of medicine. 2023. [PMID 37099341](https://pubmed.ncbi.nlm.nih.gov/37099341/). DOI: 10.1056/NEJMoa2212616. Access: full_text_partial. Study: BRACE-stage2-and-combined-stage-platform.

58. Messina NL, Pittet LF, McDonald E et al. BCG vaccination of healthcare workers for protection against COVID-19: 12-month outcomes from an international randomised controlled trial. The Journal of infection. 2024. [PMID 39127450](https://pubmed.ncbi.nlm.nih.gov/39127450/). DOI: 10.1016/j.jinf.2024.106245. Access: abstract_metadata. Study: BRACE-stage2-and-combined-stage-platform.

59. Claus J, Ten Doesschate T, Gumbs C et al. BCG Vaccination of Health Care Workers Does Not Reduce SARS-CoV-2 Infections nor Infection Severity or Duration: a Randomized Placebo-Controlled Trial. mBio. 2023. [PMID 36976004](https://pubmed.ncbi.nlm.nih.gov/36976004/). DOI: 10.1128/mbio.00356-23. Access: abstract_metadata. Study: PMID36976004.

60. Koekenbier EL, Fohse K, van de Maat JS et al. Bacillus Calmette-Guérin vaccine for prevention of COVID-19 and other respiratory tract infections in older adults with comorbidities: a randomized controlled trial. Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases. 2023. [PMID 36736662](https://pubmed.ncbi.nlm.nih.gov/36736662/). DOI: 10.1016/j.cmi.2023.01.019. Access: abstract_metadata. Study: PMID36736662.

61. Pittet LF, Messina NL, McDonald E et al. Bacille Calmette-Guérin vaccination to prevent febrile and respiratory illness in adults (BRACE): secondary outcomes of a randomised controlled phase 3 trial. EClinicalMedicine. 2024. [PMID 38774675](https://pubmed.ncbi.nlm.nih.gov/38774675/). DOI: 10.1016/j.eclinm.2024.102616. Access: abstract_metadata. Study: BRACE-stage2-and-combined-stage-platform.

62. Biering-Sørensen S, Aaby P, Lund N et al. Early BCG-Denmark and Neonatal Mortality Among Infants Weighing <2500 g: A Randomized Controlled Trial. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America. 2017. [PMID 29579158](https://pubmed.ncbi.nlm.nih.gov/29579158/). DOI: 10.1093/cid/cix525. Access: abstract_metadata. Study: PMID29579158.

63. Aaby P, Roth A, Ravn H et al. Randomized trial of BCG vaccination at birth to low-birth-weight children: beneficial nonspecific effects in the neonatal period? The Journal of infectious diseases. 2011. [PMID 21673035](https://pubmed.ncbi.nlm.nih.gov/21673035/). DOI: 10.1093/infdis/jir240. Access: abstract_metadata. Study: Guinea-Bissau-LBW-BCG-trial-platform.

## Source-specific evidence cards

### [1] Bacille Calmette-Guerin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes.

**Shows:** BCG-vaccinated volunteers show heterologous monocyte cytokine responses for three months; NOD2-dependent experiments and SCID Candida survival support innate contribution.

**Does not show:** Human cytokines do not demonstrate clinical protection or elimination of all retained BCG/signaling; histone inhibitor specificity and adaptive cytokine contributions remain limits.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC5.

### [2] Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes.

**Shows:** Candida/β-glucan pretreatment protects lymphocyte-deficient mice in a monocyte-dependent manner and dectin-1/Raf-1 signaling contributes.

**Does not show:** Induction necessity and stable marks do not isolate antigen-free maintenance or direct mark copying.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [3] mTOR- and HIF-1α-mediated aerobic glycolysis as metabolic basis for trained immunity.

**Shows:** Akt/mTOR/HIF-1α perturbations block training; myeloid HIF-1α loss abolishes trained sepsis protection.

**Does not show:** Predominantly induction-stage metabolic necessity is not mark-specific maintenance; broad metabolic/viability effects require control.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. Correction screen: no erratum, corrigendum or retraction record is indexed in PubMed for this report; status_checked_on 2026-10-14. Subclaims: SC1, SC2.

### [4] Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity.

**Shows:** Primary monocyte-derived trained and tolerant macrophages acquire different transcriptomic/enhancer programs.

**Does not show:** Differentiation and histone co-occurrence do not show the specific marks cause or autonomously store memory.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [5] Immunometabolic Pathways in BCG-Induced Trained Immunity.

**Shows:** BCG training requires glycolytic rewiring in pharmacological/genetic experiments, with associated histone changes.

**Does not show:** Most tests target induction or broad metabolic enzymes, not delayed locus-specific histone maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2.

### [6] Glutaminolysis and Fumarate Accumulation Integrate Immunometabolic and Epigenetic Programs in Trained Immunity.

**Shows:** Glutaminolysis/fumarate and mevalonate-related metabolism contribute to β-glucan induction; fumarate inhibits KDM5 and induces similar marks.

**Does not show:** Metabolite sufficiency and broad inhibitors do not demonstrate exclusive mark-based inheritance or autonomous maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2.

### [7] Metabolic Induction of Trained Immunity through the Mevalonate Pathway.

**Shows:** Mevalonate/IGF1R/mTOR perturbations regulate training; mevalonate-kinase-deficient patients have constitutive trained phenotypes.

**Does not show:** Constitutive metabolite excess is ongoing upstream stimulation, not antigen-independent memory; pleiotropy limits a selective epigenetic instrument.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

### [8] Modulation of Myelopoiesis Progenitors Is an Integral Component of Trained Immunity.

**Shows:** Sorted LT-HSCs taken 28 days after β-glucan transmit myeloid bias after 12 weeks in recipients with matched chimerism; heterologous LPS and chemotherapy responses improve.

**Does not show:** Transfer lineage bias is not direct mark copying; IL-1/glycolysis/statin/GM-CSF interventions occur on days 0–1 and test induction, not delayed maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph24, 1:Results:paragraph28, 1:Results:paragraph38, 1:Results:paragraph103, 1:Results:paragraph108, 1:Results:paragraph109, 1:Results:paragraph113. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [9] BCG Educates Hematopoietic Stem Cells to Generate Protective Innate Immunity against Tuberculosis.

**Shows:** BCG-educated HSCs generate protective macrophages; chimeric/parabiotic and transfer designs support sustainable lineage training.

**Does not show:** BCG bone-marrow access and potential persistence constrain full ligand independence; exact original transfer/clearance details inaccessible here; Mtb challenge is mycobacterial, not a distant heterologous pathogen.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.

### [10] C/EBPβ-Dependent Epigenetic Memory Induces Trained Immunity in Hematopoietic Stem Cells.

**Shows:** Acute LPS produces persistent HSC enhancer accessibility and improved Pseudomonas defense; C/EBPβ deletion erases long-term marks/expression.

**Does not show:** Exact deletion timing and viability/transfer controls are not examined; deletion of a transcription factor is not selective proof of mark copying; two corrections linked.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID32386557; ErratumIn: PMID36608675. Subclaims: SC2, SC3, SC4.

### [11] Maladaptive innate immune training of myelopoiesis links inflammatory comorbidities.

**Shows:** After 21-day periodontitis and 14-day resolution, purified LT-HSCs transmit myeloid bias and increased arthritis/LPS responses after 12–16 weeks; both CD41 strata transfer effects.

**Does not show:** IL-1R deletion precedes training; no delayed maintenance-stage IL-1 subtraction, ligand quantification or exclusive epigenetic mechanism; irradiated niche and finite duration limit scope.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph4, 1:RESULTS:paragraph13, 1:RESULTS:paragraph14, 1:RESULTS:paragraph17, 1:RESULTS:paragraph18, 1:RESULTS:paragraph21, 3:LIMITATIONS OF THE STUDY:paragraph0. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.

### [12] Western Diet Triggers NLRP3-Dependent Innate Immune Reprogramming.

**Shows:** Western-diet withdrawal normalizes serum inflammation but retains myeloid responses; NLRP3 loss prevents diet-induced reprogramming.

**Does not show:** Ldlr deficiency and remaining lipid/niche signals may persist; constitutive knockout tests induction and associated inflammation rather than late maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [13] Hyperglycemia Induces Trained Immunity in Macrophages and Their Precursors and Promotes Atherosclerosis.

**Shows:** Diabetic donor marrow produces macrophage changes in physiological glucose and increases atherosclerosis in normoglycemic Ldlr-null recipients.

**Does not show:** Whole marrow and recipient Western diet allow selection/niche feedback; RUNX1 inhibition is broad; human signatures are associative.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [14] BCG Vaccination in Humans Elicits Trained Immunity via the Hematopoietic Progenitor Compartment.

**Shows:** 15 BCG recipients versus five placebo controls show day-90 HSPC/monocyte remodeling and increased heterologous IL1β; marrow M. bovis testing is reported negative.

**Does not show:** Assay data/sensitivity not shown, a skin lesion remains, and no human lineage tracing or causal HNF1 deletion; no significant adult cell-frequency shift despite transcriptional bias.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph0, 1:Results:paragraph10, 1:Results:paragraph16, 1:Results:paragraph40, 1:Results:paragraph43. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [15] BCG vaccination alters the epigenetic landscape of progenitor cells in human bone marrow to influence innate immune responses.

**Shows:** Single-cell human marrow before/90 days after BCG/placebo shows HSPC expression/accessibility changes correlated with paired PBMC IL1β.

**Does not show:** Correlation across compartments does not demonstrate literal mark inheritance or no persistent ligand; indexed preprint is the same study, not replication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. UpdateOf: PMID38077046. Subclaims: SC2, SC3, SC4.

### [16] BCG Vaccination Induces Long-Term Functional Reprogramming of Human Neutrophils.

**Shows:** BCG alters neutrophil antimicrobial function and H3K4me3 profiles for at least three months.

**Does not show:** Repeated circulating cells cannot prove persistence in the same mature cell or autonomous progenitor mechanism; no clinical incidence endpoint.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [17] Induction of Autonomous Memory Alveolar Macrophages Requires T Cell Help and Is Critical to Trained Immunity.

**Shows:** Respiratory-virus-experienced alveolar macrophages retain antibacterial memory; T-cell help is reported required for priming rather than maintenance.

**Does not show:** Original detailed depletion/antigen controls unavailable; innate maintenance after adaptive induction differs from completely innate induction; normal niche requirements remain.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [18] Trained immunity of alveolar macrophages requires metabolic rewiring and type 1 interferon signaling.

**Shows:** LPS/IFN-β train cultured alveolar macrophages after rest; macrophage IFNAR loss blocks induction, while transferred trained AMs worsen pneumococcal infection despite whole-animal LPS protection.

**Does not show:** Methylation/acetylation inhibitors during induction do not alter IL6 training here; six-day context and transfer niche differ from clinical vaccination; pooling technical replicates limits sample interpretation.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:Results:paragraph5, 1:Results:paragraph7, 1:Results:paragraph8, 1:Results:paragraph13, 1:Results:paragraph14, 1:Results:paragraph21, 1:Results:paragraph22. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC5.

### [19] M. tuberculosis Reprograms Hematopoietic Stem Cells to Limit Myelopoiesis and Impair Trained Immunity.

**Shows:** Mtb IFN-I/iron signaling impairs myelopoiesis and trained defense, contrasting BCG/β-glucan IFN-II/IL1 programs.

**Does not show:** Chronic pathogenic infection and progenitor death are distinct from autonomous beneficial memory; ongoing infection is a confounder for maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [20] β-Glucan Reverses the Epigenetic State of LPS-Induced Immunological Tolerance.

**Shows:** β-glucan partly reverses LPS tolerance in vitro and restores cytokines in monocytes from experimental endotoxemia.

**Does not show:** Chromatin recovery accompanies reactivation; it does not prove the marks exclusively encode memory or clinical sepsis rescue.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [21] The Itaconate Pathway Is a Central Regulatory Node Linking Innate Immune Tolerance and Trained Immunity.

**Shows:** IRG1/itaconate/SDH processes distinguish tolerance from β-glucan training, including human genetic variants.

**Does not show:** Induction-stage metabolic control does not establish autonomous maintenance or a universal glycolytic phenotype.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [22] Transmission of stimulus-induced epigenetic changes through cell division is coupled to continuous transcription factor activity.

**Shows:** Immortalized macrophages retain changing molecular/function states after washout and 14 divisions; late Salmonella control parallels newly induced interferon programs.

**Does not show:** TF necessity is inferred from expression/motif coupling, not directly tested by delayed TF loss; initial responder sorting and immortalization limit generalization; mark persistence does not prove mark copying.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 2:Results:paragraph2, 2:Results:paragraph4, 2:Results:paragraph12, 2:Results:paragraph16, 2:Results:paragraph19, 2:Results:paragraph20, 2:Results:paragraph24. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [23] Long-term histone lactylation connects metabolic and epigenetic rewiring in innate immune memory.

**Shows:** H3K18 lactylation persists after washout and at 90 days after BCG; LDH/p300 inhibition attenuates training.

**Does not show:** Exact inhibitor timing and viability controls unexamined; metabolic and p300 effects are broad, and correlation with H3K18la does not isolate that mark as the memory carrier.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

### [24] Human macrophages encode stimulus-specific information of prior exposures through trained immunity.

**Shows:** Primary human macrophages retain stimulus-specific responses at six days, with stronger recall to similar challenges; some memories are lost by day 11.

**Does not show:** Finite and stimulus-dependent persistence contradicts a uniform durable state; TF licensing/chromatin are associated rather than direct maintenance subtraction; same authors/design as 2025 preprint counted once.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

### [25] Single-cell RNA sequencing reveals induction of distinct trained-immunity programs in human monocytes.

**Shows:** Four inducers yield distinct monocyte programs/subpopulations, replicated with in vivo signatures.

**Does not show:** Single-cell heterogeneity and patient signatures do not isolate self-maintenance or clinical protection.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [26] Establishment, maintenance, and recall of inflammatory memory.

**Shows:** In epidermal stem cells, delayed STAT3 loss after resolution preserves recall; post-induction AFOS leaves memory domains accessible with homeostatic TF support.

**Does not show:** Nonimmune tissue is a mechanistic comparator, not proof in myeloid cells; homeostatic maintenance factors are incompletely dissected.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph19, 1:RESULTS:paragraph22, 1:RESULTS:paragraph26, 1:RESULTS:paragraph28. No complete-paper or supplement audit claimed. none_found. Subclaims: SC2, SC3.

### [27] Innate immune memory in the brain shapes neurological disease hallmarks.

**Shows:** Peripheral inflammatory schedules induce six-month microglial training or tolerance with opposite AD/stroke pathology effects.

**Does not show:** Schedules and tissue context determine response; disease-transgene and persistent niche changes constrain intrinsic antigen-free maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

### [28] Macrophages acquire a TNF-dependent inflammatory memory in allergic asthma.

**Shows:** HDM inflammation produces TNF/2-HG/PGE2-related macrophage memory; TNF blockade or myeloid Tnf loss prevents imprinting.

**Does not show:** Allergen and disease milieu can continue; prevention does not establish delayed signal-independent maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

### [29] A distinct metabolic and epigenetic state drives trained immunity in HSC-derived macrophages from autoimmune mice.

**Shows:** Stringently purified autoimmune-exposed LT-HSCs yield trained macrophages 18 weeks after transfer without spontaneous recipient autoimmune activity or lineage skewing.

**Does not show:** Transferred progeny have reduced metabolism/accessibility despite increased killing/cytokines; irradiation and selection remain; not direct mark-copying or universal metabolic necessity.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 1:RESULTS:paragraph9, 1:RESULTS:paragraph11, 1:RESULTS:paragraph12, 1:RESULTS:paragraph13, 1:RESULTS:paragraph14, 1:RESULTS:paragraph17. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [30] Epigenetic memory of coronavirus infection in innate immune cells and their progenitors.

**Shows:** Severe-COVID HSPC/myeloid epigenetic changes persist months to one year; early IL6 contributes in a mouse model.

**Does not show:** Human infection/severity and residual inflammation are not randomized; cross-compartment signatures are not clonal inheritance or complete ligand exclusion.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3, SC4.

### [31] Heart failure promotes multimorbidity through innate immune memory.

**Shows:** Heart-failure donor marrow transmits cardiac dysfunction and altered susceptibility, with HSC TGFβ changes.

**Does not show:** Whole marrow/niche and disease carryover can coexist; TGFβ perturbation is not locus-specific epigenetic inheritance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.

### [32] Innate immune memory after brain injury drives inflammatory cardiac dysfunction.

**Shows:** Post-stroke myeloid changes transmit cardiac dysfunction; IL1β neutralization/CCR2/5 blockade prevent remote cardiac injury.

**Does not show:** Intervention timing and whole-cell milieu do not isolate late autonomous memory; patient concordance is observational.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.

### [33] Trained immunity causes myeloid cell hypercoagulability.

**Shows:** β-glucan/heme training increases myeloid procoagulant responses and four-week splenic activity via sphingomyelinase/tissue-factor decryption.

**Does not show:** Enhanced responsiveness can be harmful; systemic thrombotic clinical outcomes and autonomous maintenance are not demonstrated.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [34] Interleukin-1β induces trained innate immunity in human hematopoietic progenitor cells in vitro.

**Shows:** Four-hour IL1β exposure of human marrow progenitors produces monocyte offspring with enhanced heterologous cytokines and metabolism.

**Does not show:** Culture expansion/differentiation and normal growth signaling remain; no in vivo transplant or delayed maintenance blockade.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.

### [35] Trained immunity is regulated by T cell-induced CD40-TRAF6 signaling.

**Shows:** T-cell CD40-TRAF6 regulates myeloid functional/metabolic/chromatin training; combined blockade promotes graft acceptance.

**Does not show:** Adaptive crosstalk and combined CTLA4-Ig effects prevent attributing outcome solely to cell-autonomous innate memory.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

### [36] Lymphocyte crosstalk is required for monocyte-intrinsic trained immunity to Plasmodium falciparum.

**Shows:** Malaria-trained purified monocytes fail without lymphocytes; transwell CD4/IFNγ restores heterologous cytokine responses.

**Does not show:** This tests induction crosstalk, not proof that every established memory requires ongoing lymphocyte signals.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [37] Parenteral BCG vaccine induces lung-resident memory macrophages and trained immunity via the gut-lung axis.

**Shows:** Subcutaneous BCG induces AM memory via gut microbiome/metabolite changes and mycobacterial dissemination, independently of circulating monocytes.

**Does not show:** Distal persistent upstream changes provide an alternative to direct cell-intrinsic induction; linked correction unexamined.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. ErratumIn: PMID37161093. Subclaims: SC1, SC3.

### [38] BCG vaccination provides protection against IAV but not SARS-CoV-2.

**Shows:** BCG protects against influenza but not SARS-CoV-2, with different tissue tropism and cytokine responses.

**Does not show:** Pathogen-specific animal benefit cannot establish universal human heterologous protection or purely innate mediation.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [39] LPS-induced lung tissue-resident trained innate immunity provides differential protection against pneumococci and SARS-CoV-2.

**Shows:** Acute LPS generates persistent resident-AM phenotypes and pneumococcal protection with minimal SARS-CoV-2 protection.

**Does not show:** Whole-lung response and stable population size do not isolate same-cell autonomous memory or universal protection.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC5.

### [40] Influenza-trained mucosal-resident alveolar macrophages confer long-term antitumor immunity in the lungs.

**Shows:** Influenza-experienced AMs exert long-lasting tissue-specific antitumor effects with IFNγ/NK induction dependence.

**Does not show:** Human tumor association is observational; induction dependence is not continued maintenance dependence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [41] Trained immunity of alveolar macrophages enhances injury resolution via KLF4-MERTK-mediated efferocytosis.

**Shows:** Lineage tracing and AM transfer identify a KLF4/MERTK efferocytic trained subset that limits lung injury.

**Does not show:** Subset expansion/differentiation and tissue niche contribute; anti-inflammatory resolution differs from simply higher cytokine secretion.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3.

### [42] β-Glucan Induces Protective Trained Immunity against Mycobacterium tuberculosis Infection: A Key Role for IL-1.

**Shows:** β-glucan protects against Mtb and expands myeloid progenitors; IL1R deficiency abolishes protection.

**Does not show:** Constitutive receptor deficiency and early stimulus do not distinguish induction/maintenance; human ex vivo response is not clinical efficacy.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [43] Defective trained immunity in patients with STAT-1-dependent chronic mucocutaneaous candidiasis.

**Shows:** STAT1 CMC patient cells have Candida-specific training defects while STAT3 HIES cells train normally.

**Does not show:** Genetic pleiotropy and pathogen specificity prevent a universal innate-memory instrument or proof of antigen-free maintenance.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [44] β-Glucan-induced reprogramming of human macrophages inhibits NLRP3 inflammasome activation in cryopyrinopathies.

**Shows:** β-glucan-experienced human macrophages reduce NLRP3/caspase1/IL1β responses, including CAPS patient cells.

**Does not show:** Cytokine attenuation can coexist with memory; ex vivo anti-inflammatory effects do not establish clinical disease trajectory.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [45] The anti-inflammatory cytokine interleukin-37 is an inhibitor of trained immunity.

**Shows:** IL37 reverses training-associated metabolic/mark changes and reduces Candida protection in models.

**Does not show:** Broad anti-inflammatory effects and treatment timing do not isolate a specific self-maintaining histone carrier.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [46] β-Glucan Induces Distinct and Protective Innate Immune Memory in Differentiated Macrophages.

**Shows:** β-glucan-trained differentiated macrophage transfer protects Pseudomonas; sustained metabolic/chemokine state is independent of Dectin1/TLR2 here.

**Does not show:** Constitutive chemokine secretion differs from a fully resting state; ligand clearance and feedback remain unresolved; receptor dependence differs by model.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC3.

### [47] BCG-Derived Outer Membrane Vesicles Induce TLR2-Dependent Trained Immunity to Protect Against Polymicrobial Sepsis.

**Shows:** BCG outer-membrane vesicles induce TLR2-dependent progenitor expansion/metabolism and sepsis protection.

**Does not show:** Vesicle cargo/duration and experimental sepsis do not establish long-term autonomous maintenance or human safety/benefit.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC4.

### [48] Trained Immunity Confers Prolonged Protection From Listeriosis.

**Shows:** β-glucan nine weeks before Listeria increases antimicrobial myeloid responses and survival.

**Does not show:** Elapsed time alone does not exclude retained particulate ligand or ongoing niche stimulation; multiple population/metabolic changes coexist.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC3, SC4.

### [49] Control of nutrient uptake by IRF4 orchestrates innate immune memory.

**Shows:** IRF4 controls metabolic uptake and memory expansion of virus-specific NK cells.

**Does not show:** Antigen-specific NK memory is a comparator, not demonstration of nonspecific trained myeloid memory; generic survival effects constrain mechanism.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2.

### [50] Outcomes of controlled human malaria infection after BCG vaccination.

**Shows:** Randomized human BCG/malaria challenge alters activation and yields earlier/more severe symptoms; parasitemia inversely correlates with activation.

**Does not show:** 20 volunteers and correlational mediation do not establish clinical protection; enhanced activation can accompany adverse effects.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [51] BCG Vaccination Protects against Experimental Viral Infection in Humans through the Induction of Cytokines Associated with Trained Immunity.

**Shows:** Randomized BCG attenuated-yellow-fever challenge reduces viremia with IL1β-associated monocyte changes.

**Does not show:** Vaccine-strain viremia is not ordinary clinical disease; correlated cytokines do not prove innate mediation or autonomous persistence.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC2, SC5.

### [52] Activate: Randomized Clinical Trial of BCG Vaccination against Infection in the Elderly.

**Shows:** ACTIVATE interim report of 198 elderly patients finds longer time to first infection and fewer infections with BCG.

**Does not show:** Small interim analysis/probable viral diagnoses and clinical heterogeneity limit generalization; no selective trained-memory mediation test.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [53] Effect of BCG Danish and oral polio vaccine on neonatal mortality in newborn babies weighing less than 2000 g in India: multicentre open label randomised controlled trial (BLOW2).

**Shows:** BLOW2 randomizes 5420 NICU newborns under 2000 g to early BCG plus OPV and reduces neonatal all-cause mortality.

**Does not show:** Combined vaccines cannot isolate BCG or trained innate mechanism; infection-death secondary outcome added during trial; setting/timing-specific clinical benefit.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [54] Heterologous immunological effects of early BCG vaccination in low-birth-weight infants in Guinea-Bissau: a randomized-controlled trial.

**Shows:** Randomized low-birthweight infant BCG substudy shows increased heterologous cytokines four weeks later.

**Does not show:** Same Guinea-Bissau trial platform and immune substudy do not establish cytokine mediation of mortality or independent clinical replication.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC1, SC5.

### [55] Nonspecific effect of BCG vaccination at birth on early childhood infections: a randomized, clinical multicenter trial.

**Shows:** Danish neonatal randomized trial finds no overall benefit in parent-reported infections through 13 months.

**Does not show:** Maternal-BCG early subgroup is not the overall endpoint; low-mortality context may differ from NICU/LBW settings.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [56] Using BCG Vaccine to Enhance Nonspecific Protection of Health Care Workers During the COVID-19 Pandemic: A Randomized Controlled Trial.

**Shows:** 1221 Danish healthcare workers show no overall BCG benefit for absenteeism, COVID or hospitalization.

**Does not show:** Secondary revaccination analyses diverge and have wide intervals; clinical null does not disprove all cellular memory.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [57] Randomized Trial of BCG Vaccine to Protect against Covid-19 in Health Care Workers.

**Shows:** BRACE stage-2 randomized 3988 healthcare workers shows no lower six-month symptomatic/severe COVID risk.

**Does not show:** Early recruitment stop and vaccine-related censoring limit precision; severe chiefly means incapacitated three days, not hospital disease; no verified selective innate engagement in this report.

**Access/status:** Complete primary abstract and linked-status metadata examined. Selected body sections: 0:Methods, 1:Results. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [58] BCG vaccination of healthcare workers for protection against COVID-19: 12-month outcomes from an international randomised controlled trial.

**Shows:** BRACE 12-month follow-up shows increased symptomatic COVID and no severe/breakthrough benefit.

**Does not show:** Same trial as six-month report, not independent replication; severity definition and evolving vaccination/exposure limit universal conclusions.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [59] BCG Vaccination of Health Care Workers Does Not Reduce SARS-CoV-2 Infections nor Infection Severity or Duration: a Randomized Placebo-Controlled Trial.

**Shows:** Dutch healthcare-worker randomization with serologic detection shows no infection/duration/severity benefit despite early antibody effects.

**Does not show:** Small hospitalized counts and clinical context constrain severe-disease estimates; antibody enhancement is not clinical or innate-memory mediation.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [60] Bacillus Calmette-Guérin vaccine for prevention of COVID-19 and other respiratory tract infections in older adults with comorbidities: a randomized controlled trial.

**Shows:** BCG-PRIME randomizes 6112 older comorbid adults and finds no COVID or clinically relevant respiratory benefit.

**Does not show:** Six-month regimen-specific null does not eliminate every clinical context or demonstrate selective mechanistic engagement.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [61] Bacille Calmette-Guérin vaccination to prevent febrile and respiratory illness in adults (BRACE): secondary outcomes of a randomised controlled phase 3 trial.

**Shows:** BRACE combined-stage secondary analysis reports modestly more febrile/respiratory illness without severe-disease benefit.

**Does not show:** 6828 combined stages differs from 3988 stage-2 primary denominator; same trial, not independent cohort.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [62] Early BCG-Denmark and Neonatal Mortality Among Infants Weighing <2500 g: A Randomized Controlled Trial.

**Shows:** Third Guinea-Bissau early-BCG trial has lower infection-related neonatal mortality and favorable pooled neonatal estimate.

**Does not show:** Its standalone primary neonatal all-cause contrast is nonsignificant; polio-campaign censoring and pooled trials must be separated; mechanism unisolated.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.

### [63] Randomized trial of BCG vaccination at birth to low-birth-weight children: beneficial nonspecific effects in the neonatal period?

**Shows:** Early BCG in 2320 low-birthweight infants has a favorable four-week secondary mortality contrast.

**Does not show:** Primary infant mortality is nonsignificant; neonatal secondary effect and subgroup weights do not turn the primary endpoint positive or establish autonomous training.

**Access/status:** Complete primary abstract and linked-status metadata examined. No complete-paper or supplement audit claimed. none_found. Subclaims: SC5.