Acute Radiation Syndrome — Disease Characteristics Research Report

Prepared: 2026-09-01 · Target: Acute Radiation Syndrome · MONDO:0033938 · Category: Complex (environmental / toxicologic, non-Mendelian)

⚠️ Read this before you paste anything into a snippet: field

Every quoted fragment below came out of PubMed's efetch records through an automated retrieval layer, which means a summarizing model touched them on the way here. They are almost certainly verbatim, but "almost certainly" is exactly the failure mode that eats a whole curation session three PRs later. So:

Run just fetch-reference PMID:<id> and re-verify every snippet against the cache before it goes into YAML. Treat this whole document the way the repo already tells you to treat a deep-research report — as leads, not as evidence.

Same deal with the ontology terms. I've split them into two buckets: - ✅ VERIFIED — I looked the CURIE up either in this repo's own cache/<prefix>/terms.csv (which is authority-backed and already passing term validation elsewhere in the KB) or live against EBI OLS4. The id↔label pair is real. - 🔎 LEAD — a plausible term I did not resolve. Do not bind it until just validate-terms says yes.

sup. Let's get into it.

1. Disease Information

What it is

Acute radiation syndrome is what happens when a body absorbs a big dose of penetrating ionizing radiation all at once. It isn't one disease so much as a scheduling problem: radiation preferentially kills cells that are dividing, and the body's fastest-dividing tissues are the ones constantly rebuilding themselves — bone marrow, gut lining, skin, gonads. The mature cells already made keep working until they hit their normal expiry date. Then nothing replaces them. So the illness arrives on a timetable set by the half-life of each cell type, which is why ARS has that eerie "feels fine for a week, then falls apart" shape. It's less like a poisoning and more like quietly firing the entire maintenance staff of a building and waiting.

Orphanet's definition (ORPHA:454831):

"Acute radiation syndrome is a rare radiation-induced disorder resulting from whole body exposure to large doses of penetrating radiation (>0.7 Gray) within a very short period of time (usually minutes) and characterized by bone marrow syndrome with pancytopenia, gastrointestinal syndrome, and cardiovascular/central nervous system syndrome."

Four conditions have to be met simultaneously for ARS (this is the CDC framing): the dose must be large, the radiation must be penetrating (gamma, X-ray, neutrons — an external alpha or low-energy beta source can't reach marrow), it must hit most or all of the body, and it must be delivered in a short time at high dose rate. Drop any one of those four and you get a different disease. Chronic low-dose exposure, internal contamination, and a localized radiotherapy field are all not ARS, which matters for scoping the KB entry.

Identifiers

MONDO synonym on record: "Acute radiation sickness". Other names in common clinical and lay use: radiation poisoning, radiation toxicity, radiation sickness, and in the older Soviet/Ukrainian literature acute radiation disease (ARD).

Where the knowledge comes from — and why this matters for evidence grading

This is the unusual bit and it should shape the whole entry. Human ARS knowledge is aggregated from a small number of catastrophes, not from cohorts or EHR. The total documented human experience is a few thousand people across ~80 years. Lazarus & Gale put it bluntly (PMID:40074513, Best Pract Res Clin Haematol 2024):

"Radiological accidents/incidents are common with nearly 400 reported since 1944 exposing about 3000 people to substantial doses of ionizing radiations with 127 deaths."

So the sources stack like this: - Accident registries and case series — REAC/TS Radiation Accident Registry (Oak Ridge), IAEA accident reports, the Ukrainian Chernobyl ARS survivor cohort. - Atomic bomb survivor data (RERF, Hiroshima/Nagasaki) — the largest human dose-response dataset, but confounded by blast, burns, and wartime lack of medical care. - Iatrogenic near-analogs — total body irradiation conditioning before transplant is essentially a controlled ARS, and is the reason we know anything at all about supportive care windows. - Animal-Rule preclinical studies — because you cannot ethically run an ARS efficacy trial, the FDA's Animal Rule means every approved drug for this disease was licensed on non-human primate and rodent data. Christy & Herzig say it plainly (PMID:39000080): "Since clinical trials for ARS cannot be ethically conducted, animal testing is extremely important."

Curation consequence: a large fraction of the mechanistic evidence for this entry will legitimately carry evidence_source: MODEL_ORGANISM, and that's not a defect — it's the epistemic structure of the field. But per the repo's own rule, model-organism evidence should not be the only support for a human phenotype, and here the human phenotypes genuinely are documented (Chernobyl, Goiânia, Tokai-mura). Keep the two layers distinct.

2. Etiology

Primary cause

One cause, no ambiguity: acute, high-dose-rate, whole-body or significant partial-body exposure to penetrating ionizing radiation.

Exposure term: ECTO:7000047 — "exposure to ionizing radiation" ✅ VERIFIED (local cache/ecto/terms.csv; already bound elsewhere in the KB, so it validates offline).

This is the entry's single initiating node, and it should be wired into the pathograph via environmental[].influences_mechanisms with environmental_effect: TRIGGERS, not left as a floating list item. There is no other etiology to model.

Exposure scenarios (the real-world routes)

Feldman & Kazzi (PMID:37280005) note the driver behind the recent surge in preparedness work:

"The conflict in Ukraine has raised the specter of radiological and nuclear incidents, including fighting at the Zaporizhzhia nuclear plant, the largest nuclear powerplant in Europe; concerns that a radiological dispersion device ('dirty bomb') may be used; and threats to deploy tactical nuclear weapons. Children are more susceptible than adults to immediate and delayed radiation health effects."

Dose–effect thresholds (the load-bearing numbers)

These are the numbers that structure the whole entry, so get them right and cite them individually rather than as a block.

A caution worth writing into notes:. The A-bomb LD50 is lower than the textbook "3.5–4.5 Gy" figure, and the difference isn't a contradiction — it's confounding. Hiroshima and Nagasaki survivors had concurrent burns, blast trauma, malnutrition, and essentially no medical infrastructure. The higher figure assumes an intact hospital. Any prevalence/prognosis record here should carry the care context in population or notes, or the two numbers will look like a data error to a later reader.

Modifying factors — physical

Not all Grays are equal, and the entry should say so:

• Dose rate. Same total dose delivered slowly is far less lethal; sublethal damage repair happens between hits.
• Dose uniformity / partial-body sparing. This is enormous. Shielding even a small marrow volume changes the outcome completely — the standard preclinical GI-ARS model deliberately spares 5% of marrow precisely so the animal survives long enough to have a GI syndrome (Mitra et al., PMID:39930324: "13 Gy partial body X-irradiation with 5% bone marrow shielding").
• Radiation quality / LET. Neutrons carry higher relative biological effectiveness than photons at equal absorbed dose. Royba et al. (PMID:35994701) found in the dicentric assay that "the frequency of dicentrics depends on the type of radiation" — which means dose estimation itself is quality-dependent.
• Geometry — anteroposterior vs. rotational exposure changes marrow dose for the same free-in-air dose.

Modifying factors — host

• Age. Children are more radiosensitive both acutely and for late effects (PMID:37280005). Elderly patients have less marrow reserve.
• Pregnancy. Fetal exposure is a separate injury class (teratogenesis, growth restriction) beyond this entry's scope but relevant to triage.
• Combined injury — the big one. Radiation plus burn or trauma is dramatically worse than either alone, and this is a real mechanistic interaction rather than additive risk. Kiang & Blakely (PMID:36947602) state: "At present, no FDA-approved drug to protect, mitigate, or treat CI is available." Glowacki et al. (PMID:34233299) confirm mouse models "show dose-dependent impairment of wound healing." This deserves its own pathophysiology node with EXACERBATES semantics, not a footnote.
• Nutritional status, baseline marrow reserve, comorbid immunosuppression.
• Microbiome composition — see §6; this turns out to be a genuine host modifier, not a curiosity.

Genetic risk factors

There is no causal gene. ARS is fully environmental. But germline DNA-repair defects are legitimate susceptibility modifiers, and the mechanism is clean: if you can't repair a double-strand break, the same physical dose becomes a larger biological dose.

The ATM heterozygote question is unsettled and should be curated as such. Carriers are roughly 1–2.5% of the population, and whether they have clinically meaningful intermediate radiosensitivity remains debated. Model this as a discussions entry with kind: KNOWLEDGE_GAP rather than asserting an effect.

Beyond single genes, radiogenomics — Story & Durante (PMID:30421807) define it as "the study of genomic changes that underlie the radioresponse of normal and tumor tissues" — is the field studying common-variant contributions to normal-tissue radiation toxicity. Almost all of it is radiotherapy-derived (fractionated, localized) and its transfer to single-fraction whole-body ARS is an inference, not a demonstration. Flag that explicitly if you use it.

Suggested inheritance modeling: no inheritance: block for the disease itself. Susceptibility genes go in genetic: with relationship_type: SUSCEPTIBILITY or MODIFIER. Do not use CAUSATIVE here — nothing in the genome causes ARS.

Protective factors

Physical, not biological, is where the real protection lives: time, distance, shielding; sheltering and evacuation; partial-body shielding of marrow.

Genetic protective factors: none established. 🔎 No published protective allele for ARS that I could find.

Pharmacological prophylaxis: none FDA-approved. Singh et al. (PMID:39160790) are explicit: "Currently, there is no radiation medical countermeasure approved by the U.S. FDA which can be used before radiation exposure to protect exposed individuals." Candidates in §12.

A distinction the entry must not blur: potassium iodide is not ARS prophylaxis. KI (CHEBI:8346 ✅) saturates the thyroid so inhaled/ingested radioiodine can't be taken up. It does nothing about external whole-body dose and nothing about marrow. Same for Prussian blue (cesium/thallium) and Ca-/Zn-DTPA (transuranics) — those are decorporation agents for internal contamination, a different disease process. Putting them in treatments: for ARS without a very clear scoping note is a modeling error I'd expect a reviewer to flag.

Gene–environment interaction

The core interaction is dose × DNA-repair capacity: identical physical dose produces a larger biological insult in a repair-deficient host, because unrejoined double-strand breaks are the substrate for cell death. That's the mechanistic through-line linking §2 and §6. Beyond the monogenic syndromes, GxE evidence for ARS specifically is thin — most of it is extrapolated from radiotherapy toxicity cohorts.

3. Phenotypes

The four-phase temporal architecture

Every ARS phenotype has to be anchored to a phase, or the entry will read as if these things happen at once. They don't. Arnautou & Garnier (PMID:39025400):

"This syndrome typically progresses through three stages with a prodromal phase, a latency phase and a critical phase. Each of them varies in intensity and duration depending on the absorbed dose of radiation."

Orphanet counts four: prodromal → latent → manifest illness → recovery or death. Use four; the recovery/death fork is a real branch. Higher dose compresses every phase — at very high dose the latent period disappears entirely, which is itself diagnostic.

Suggested progression: records (the phase field is the foreign-key target):

Prodromal phenotypes

Frequency: near-universal above ~2 Gy. Prodromal nausea/vomiting is the CDC's and REMM's frontline triage sign precisely because it shows up in essentially everyone with a meaningful dose. Use frequency bands cautiously — I did not find a clean cohort denominator to cite, and per the repo's frequency-band guidance, a definitional-sounding statement is not a frequency.

Hematopoietic subsyndrome (H-ARS) — ≥ ~2 Gy

The dominant subsyndrome and the one every approved drug targets. Yamaga et al. (PMID:38333215) on the kinetics:

"Depending on the dose of ionizing radiation, death of hematopoietic stem and progenitor cells and apoptosis of lymphocytes occur, and lymphopenia becomes apparent within 6-24 hours after radiation exposure."

Severity: dose-graded, from mild transient cytopenia to irreversible marrow failure. Progression: progressive to nadir, then either recovery or not — model as PROGRESSIVE through manifest illness. QoL: during aplasia, near-total functional dependence — protective isolation, transfusion dependence, high infection risk. Formal EQ-5D/SF-36 data for ARS specifically: 🔎 not found. Don't invent one.

Gastrointestinal subsyndrome (GI-ARS) — ≥ ~6 Gy

Winters & Marzella (PMID:38616048) on the consequence chain:

"Damage incurred in the latter can lead to nutrient malabsorption, dehydration, electrolyte imbalance, altered microbiome and metabolites, and impaired barrier function, which can lead to septicemia and death."

Timing: onset within days; death typically 5–10 days at ≥10 Gy. Prognosis: without marrow support, uniformly fatal above ~10 Gy — GI-ARS is a concurrent syndrome, not a replacement, so the patient has a wrecked gut and no white cells at the same time. That combination is why bacterial translocation across a denuded barrier is so reliably lethal.

Cutaneous radiation syndrome (CRS)

Iddins et al. (PMID:34488201) frame it: "Cutaneous radiation injuries (CRI) or local radiation injuries (LRI) may lead to cutaneous radiation syndrome."

Manifestations, per the search literature: "skin damages, erythema, altered sensation, itching, edema, blistering, desquamation, ulcer, necrosis, hair loss, and onycholysis."

⚠️ The dose thresholds in this table came from general secondary sources, not a single citable primary paper I verified. They're broadly consensus but you need a real citation per row before they go in as evidence-backed claims.

CRS is famously waves, not a single event — erythema recurs at intervals over weeks as successive cell populations fail, which is why it fools clinicians who expect a burn to behave like a burn. It's also the subsyndrome most likely to appear without whole-body ARS (industrial source handling), and the one most likely to demand surgery.

Neurovascular / cardiovascular / CNS subsyndrome — very high dose

"The neurovascular/cardiovascular/central nervous system syndrome occurs at doses >50 Gray and is characterized by watery diarrhea, convulsions, coma, and death within three days of exposure."

Prognosis: universally fatal. There is no treatment; management is palliative. Note the dose threshold varies across sources (>20 Gy in some, >50 Gy in others) — 🔎 pin this to a specific citation rather than averaging.

Delayed effects (DEARE) — survivors

Wu & Orschell's DEARE work (PMID:37014943) documented in mice, after a sub-threshold H-ARS dose: "residual bone marrow damage (RBMD) and progressive renal and cardiovascular DEARE", plus "physiological and neural function, progressive fur graying, ocular inflammation, and malignancy."

Important curation note: most of the specific DEARE phenotype list is rodent-derived. The Ukrainian ARS survivor cohort (Belyi et al., PMID:20445398) is the human counterpart and covers "over 20 years of health outcomes among confirmed ARS survivors." Cite the human study for human claims; keep the mouse detail tagged MODEL_ORGANISM.

4. Genetic / Molecular Information

Causal genes: none

State this explicitly in the entry rather than leaving genetic: empty and ambiguous. ARS has no causal gene, no pathogenic variant class, no inheritance pattern, no penetrance, no carrier frequency, no founder effect, and no consanguinity role. Every one of those template fields is not applicable, and saying so is more useful than silence.

Somatic / acquired genomic change — the actual "genetics" of ARS

What radiation does to the genome is the disease:

• DNA double-strand breaks — the lethal lesion. Huang & Zhou (PMID:32355263): "DNA double-strand breaks are the most lethal lesions induced by ionizing radiation."
• Clustered/complex lesions — multiple damages within one or two helical turns, much harder to repair than isolated breaks. This is why high-LET radiation is disproportionately lethal.
• Chromosomal aberrations — dicentrics, rings, translocations, micronuclei. These aren't just damage; they're the diagnostic assay (§10).

Genes as biodosimetry readout (not as cause)

This is the one place a gene list genuinely belongs in the entry — as biochemical: / BiomarkerReadout records, not as genetic: causal entries.

The canonical radiation-responsive transcript panel:

From the biodosimetry literature: "Specific radiation-sensitive genes such as FDXR, DDB2, WNT3, and POU2AF1 have become well established for biodosimetry purposes and acute radiation sickness prediction, with FDXR and DDB2 commonly used as biomarkers for retrospective dosimetry within 72 hours after radiation exposure."

At the protein level: "When combining FDXR, ACTN1, and DDB2 proteins to estimate radiation dose by linear regression, the combination showed the lowest mean absolute errors (≤0.13 Gy) and highest coefficients of determination (R² = 0.96)." ⚠️ Verify this against the primary paper before quoting — it came through a search-result layer, not efetch.

Mechanism-relevant genes (for pathophysiology node binding)

Epigenetics

Radiation-induced methylation and chromatin changes are reported in the multi-omics literature (Shakyawar et al., PMID:36368026, catalogs "radiation-induced changes across genomic, transcriptomic, proteomic, metabolomic, and microbiome profiles"), but I found no ARS-specific epigenetic mechanism with the evidentiary weight to be a pathophysiology node. 🔎 Treat as an open area, not a claim.

Chromosomal abnormalities

Acquired only: dicentric chromosomes (the biodosimetry gold standard), rings, reciprocal translocations (persistent — used for retrospective dosimetry years later), and micronuclei. None germline.

5. Environmental Information

This section is the etiology, so it carries most of the entry's causal weight. See §2 for the full scenario list.

Environmental factor to model:

environmental:
- name: Acute whole-body exposure to penetrating ionizing radiation
  exposure_term:
    preferred_term: acute whole-body exposure to penetrating ionizing radiation
    term:
      id: ECTO:7000047          # ✅ VERIFIED
      label: exposure to ionizing radiation
  influences_mechanisms:
  - target: <the DNA damage / energy deposition node>
    environmental_effect: TRIGGERS
    causal_link_type: DIRECT

Note the preferred_term is deliberately more specific than the ECTO label — that's exactly the pattern the repo's ontology contract endorses, since ECTO has no term for the acute high-dose-rate whole-body case.

Lifestyle factors: essentially none. Occupation is the real exposure axis (nuclear workers, industrial radiographers, radiotherapy staff, emergency responders, military). Smoking/diet/alcohol are not established modifiers of acute radiation lethality.

Infectious agents: not causal — but bacteria are absolutely part of the mechanism. The killing blow in both H-ARS and GI-ARS is usually endogenous bacterial translocation across a failed gut barrier into a host with no neutrophils. Yamaga et al. (PMID:38333215) name it: "lethal sepsis stands as a major contributor to the mortality in ARS." So the microbiome belongs in the pathograph, and the relevant taxa are commensals gone rogue rather than an outside pathogen.

Gut microbiome as a bidirectional modifier — genuinely one of the most interesting recent developments, covered in §6.

6. Mechanism / Pathophysiology

The causal chain, step by step

Here's the spine. Branches are marked. Where a step is inferred rather than demonstrated in humans, I say so.

Step 0 — Energy deposition. An ionizing photon or particle traverses tissue and deposits energy, both by directly ionizing macromolecules and — for about two-thirds of low-LET damage — by radiolysis of water, generating hydroxyl radicals, superoxide, and hydrogen peroxide. → leads to

Step 1 — Clustered DNA lesions, especially double-strand breaks. DSBs are the lethal lesion (PMID:32355263). Base damage and single-strand breaks are mostly repaired; complex clustered damage is not. → leads to

Step 2 — DNA damage response activation. MRN complex senses the break; ATM (and ATR, DNA-PKcs) phosphorylate H2AX and downstream substrates; CHK1/CHK2 relay; TP53 is stabilized. → branches into three fates: - 2a — Repair and survive. NHEJ or homologous recombination rejoins the break; cell cycle resumes. - 2b — Arrest / senescence. p53 → CDKN1A/p21 → durable arrest. GO:0090398 cellular senescence ✅ — observed as p21/waf1 upregulation in irradiated minipig ileum (PMID:38674120). - 2c — Death. p53 → BAX/PUMA → mitochondrial outer membrane permeabilization → intrinsic apoptosis (GO:0008630 ✅). Or, for cells that attempt to divide with unrepaired breaks, mitotic catastrophe — which is why proliferating compartments are hit hardest.

Step 3 (parallel arm, not downstream of Step 2) — the membrane/ceramide pathway. This one is important and easy to model wrong. Radiation triggers translocation of secretory acid sphingomyelinase (SMPD1) into cholesterol/glycosphingolipid rafts of the outer plasma membrane leaflet, where sphingomyelin is hydrolyzed to ceramide, which coordinates transmembrane apoptotic signaling. This happens within hours and does not require nuclear DNA damage signaling — it's a parallel initiating lesion, and if you draw it downstream of p53 you've inverted the biology.

From Paris et al. and the anti-ceramide follow-up work: "Extensive endothelial cell apoptosis occurred in the first 4 hours after administering radiation doses sufficient for inducing the GI syndrome (≥15 Gy), while epithelial cell apoptosis occurred several hours after endothelial cell death." And the causal test: "Genetic inactivation of ASMase in mice, or intravenous treatment with the endothelial cell survival factor bFGF prior to whole body irradiation, attenuated radiation-induced endothelial apoptosis of the intestinal microvascular system, preserved crypt stem cell clonogens, and protected mice against lethality from the GI syndrome."

⚠️ This is MODEL_ORGANISM evidence and the endothelial-first model has been contested in the field. Curate it as the leading mechanism with an explicit discussions entry noting the debate, not as settled fact.

Step 4 — Selective destruction of self-renewing compartments. The tissues that lose are the ones that must constantly rebuild: hematopoietic stem/progenitor cells, intestinal crypt stem cells, basal keratinocytes and follicular stem cells, spermatogonia. Plus lymphocytes, which are the exception that proves the rule — they die by interphase apoptosis without dividing, which is why lymphopenia is the earliest signal. → branches into the four subsyndromes

Branch A — Hematopoietic ARS (≥ ~2 Gy)

4A.1 HSPC apoptosis + mitotic death, plus damage to the bone marrow niche itself — endothelial and mesenchymal stromal cells, not just the stem cells. This is a relatively recent refinement and matters therapeutically. Vercellino et al. (PMID:38679747) established that a thrombopoietin mimetic works partly through the niche: "TPOm interacts with BM vascular and stromal niches to locally support hematopoietic reconstitution and systemically improve survival in mice after TBI." → leads to

4A.2 Loss of proliferative reserve. Nothing dramatic happens yet — this is the latent phase. → leads to

4A.3 Sequential cytopenias as each mature lineage runs out on its own clock: lymphocytes 6–24 h, neutrophils ~10–21 d, platelets ~2–4 wk, red cells latest. → leads to

4A.4 Two convergent failure modes — infection (no neutrophils + breached mucosal barriers) and hemorrhage (no platelets). → leads to

4A.5 Sepsis, multi-organ failure, death. Or, with support and cytokines, marrow reconstitution and recovery.

Branch B — Gastrointestinal ARS (≥ ~6 Gy)

4B.1 Endothelial apoptosis in the lamina propria microvasculature (Step 3) plus direct crypt stem cell clonogen death. → leads to

4B.2 Crypt depopulation. → leads to

4B.3 Villus denudation — enterocytes continue their normal migration up the villus and slough off, and nothing arrives to replace them. Takes ~3–5 days, which sets the syndrome's timing. → leads to

4B.4 Barrier failure + massive fluid/electrolyte loss + malabsorption. → leads to

4B.5 Dysbiosis and bacterial translocation, DAMP release, cytokine amplification. Freeman (PMID:40368913): the field's priorities are "Understanding intestinal stem cell contributions to recovery, vascular damage mechanisms, and crypt-villus regeneration following irradiation." → converges with 4A.4 →

4B.6 Septicemia and death, typically 5–10 days.

The regeneration counter-branch — genuinely exciting recent biology, and the reason GI-ARS may become treatable: - Revival stem cells (revSC) — Ayyaz et al. (PMID:31019301) identified "a distinct, damage-induced quiescent cell type that we term the revival stem cell (revSC)", marked by high clusterin, "extremely rare under homoeostatic conditions," which "reconstitute the LGR5+ CBC compartment and are required to regenerate a functional intestine" via YAP1. - Tuft cells as a human reserve pool — Huang et al. (PMID:39358509): "tuft cells survive irradiation damage and retain the ability to generate all other epithelial cell types," representing "a damage-induced reserve intestinal stem cell pool in humans." Human tissue evidence, which is rare here — weight it accordingly. - Isthmus progenitors — Malagola et al. (PMID:38848678) argue stemness "resides in the isthmus region" and that "neither de-differentiation nor reserve ISC are drivers of intestinal regeneration." ⚠️ This directly contradicts the revSC model. That's a real live controversy and belongs in mechanistic_hypotheses with competing hypothesis_group_ids, not resolved by fiat. - Niche signals — macrophages drive repair via NRG1 and osteopontin, and "macrophage ablation led to compromised regeneration" (PMID:40086603); a tissue-intrinsic IL-33/EGF circuit promotes epithelial regeneration (PMID:37669929).

Branch C — Cutaneous radiation syndrome

4C.1 Basal keratinocyte and hair-follicle stem cell death + dermal microvascular endothelial injury. → 4C.2 epilation, erythema in waves, dry then moist desquamation, blistering. → 4C.3 at higher dose, ulceration and necrosis, often requiring surgery. → 4C.4 late dermal fibrosis.

When CRS coexists with whole-body dose, you get radiation combined injury, where wound healing is itself radiation-impaired — a vicious circle with no approved therapy (PMID:36947602).

Branch D — Neurovascular / CNS syndrome (very high dose)

4D.1 Overwhelming endothelial death and vascular permeability + direct neuronal/glial damage. → 4D.2 cerebral edema, refractory hypotension. → 4D.3 prostration, ataxia, convulsions, coma. → 4D.4 death within ~3 days. Gorbunov & Kiang (PMID:33979447) note "cranial or total-body irradiation can cause a plethora of biochemical and cellular disorders in brain tissues."

Step 5 — Systemic amplification via DAMPs. Yamaga et al. (PMID:38333215) describe the loop: radiation causes cellular injury through DNA damage and oxidative stress, prompting DAMP release; these molecules then "interact with pattern recognition receptors, triggering inflammatory responses"; and "lethal sepsis stands as a major contributor to the mortality in ARS." This makes the syndromes non-independent — GI barrier failure feeds the systemic inflammation that worsens everything else. Radiation-induced multi-organ dysfunction syndrome is the terminal common path.

Step 6 — DEARE in survivors (months to years): residual bone marrow damage, progressive renal failure, pneumonitis→fibrosis, cardiovascular rarefaction and senescence, cataract, cognitive/behavioral change, secondary malignancy.

The microbiome layer — a real modifier, not decoration

This deserves its own treatment because the evidence is now strong enough to model.

Guo et al., Science 2020 (PMID:33122357) — the "elite survivor" study. Mice that survived high-dose radiation to live normal lifespans harbored "distinct gut microbiota that developed after radiation and protected against radiation-induced damage and death." Elevated Lachnospiraceae and Enterococcaceae correlated with restored hematopoiesis and GI repair, and — critically — "these bacteria were also found to be more abundant in leukemia patients undergoing radiotherapy, who also displayed milder gastrointestinal dysfunction", which is the human anchor. On the metabolite side: "Metabolomics revealed increased fecal concentrations of microbially derived propionate and tryptophan metabolites," and administering those metabolites "caused long-term radioprotection."

Supporting metabolite work: - Propionate (CHEBI:17272 ✅) and butyrate (CHEBI:17968 ✅) — short-chain fatty acids. - Valeric acid — Li et al. (PMID:31931652): "VA exerted the most significant radioprotection among the SCFAs." - Indole-3-carboxaldehyde (I3A) — Xie et al. (PMID:38706205): I3A "activated the AhR/IL-10/Wnt signaling pathway to promote intestinal epithelial proliferation." - Contrarily, Jiao et al. (PMID:40192235) found raffinose-metabolizing bacteria impair hematopoietic recovery via bile acid/FXR/NF-κB — so it cuts both ways. - And Cook et al. (PMID:36253079) found that antibiotic microbiome depletion did not protect against radiation carcinogenesis and actually "shortened the lifespan when Ab were administered before and after TBI" — a useful REFUTE-direction evidence item against naive "just sterilize the gut" reasoning.

Cell death modes beyond apoptosis

Ferroptosis is now implicated. Horseman et al. in the Göttingen minipig (PMID:38674120): "GPX-4 and SLC7A11 were downregulated post-irradiation, consistent with ferroptosis at 6 and 35 days post-irradiation in all groups." And the ATM–NCOA4–ferritinophagy link (PMID:36752571) connects the DDR kinase directly to iron-dependent death. Also relevant: necroptosis, autophagy, and pyroptosis via NLRP3 — the minipig study found radiation increased "IL1B, TNFA, CCL2, IL18, and CXCL8, and the inflammasome component NLRP3."

Suggested GO / CL bindings for pathophysiology nodes

Biological processes — all ✅ VERIFIED against cache/go/terms.csv:

On modifier choice: most of these are genuinely INCREASED/DECREASED (quantitative), and per the repo's guidance that's the default. Don't reach for LOSS_OF_FUNCTION — nothing here is escaping regulatory control; things are simply running above or below normal.

Cell types — all ✅ VERIFIED against cache/cl/terms.csv:

CL:0000037 hematopoietic stem cell · CL:0000049 common myeloid progenitor · CL:0000557 granulocyte monocyte progenitor cell · CL:0000556 megakaryocyte · CL:0000775 neutrophil · CL:0000542 lymphocyte · CL:0000084 T cell · CL:0000236 B cell · CL:0000115 endothelial cell · CL:0002139 endothelial cell of vascular tree · CL:0002250 intestinal crypt stem cell · CL:0002563 intestinal epithelial cell · CL:0000584 enterocyte · CL:0000510 paneth cell · CL:0002253 epithelial cell of large intestine · CL:0000312 keratinocyte · CL:0002559 hair follicle cell · CL:0000134 mesenchymal stem cell · CL:0000235 macrophage · CL:0000216 Sertoli cell · CL:0000023 oocyte

🔎 LEAD: tuft cell, revival stem cell (no CL term likely exists — use preferred_term and leave term: off rather than binding something wrong).

Chemical entities — ✅ VERIFIED: CHEBI:17761 ceramide · CHEBI:26523 reactive oxygen species · CHEBI:16240 hydrogen peroxide · CHEBI:15379 dioxygen · CHEBI:17272 propionate · CHEBI:17968 butyrate 🔎 LEAD: hydroxyl radical, superoxide, indole-3-carbaldehyde, valerate/pentanoate.

Molecular profiling

• Transcriptomics — the biodosimetry panels above; single-cell RNA-seq of the irradiated marrow niche (PMID:38679747) and regenerating intestine (PMID:31019301, PMID:38848678).
• Proteomics — FDXR/ACTN1/DDB2 plasma panel; a "plasma proteomic biodosimetry approach based on a panel of radiation-responsive biomarkers" for combined injury (PMID:36947602).
• Metabolomics — SCFAs, tryptophan/indole metabolites, bile acids.
• Microbiome — 16S and shotgun profiling across mouse, minipig, and human radiotherapy cohorts.
• Multi-omics integration — Shakyawar et al. (PMID:36368026): "multi-omic profiles obtained from high-resolution omics platforms offer a holistic approach" for identifying organ-specific damage biomarkers.
• Single-cell — see above. Spatial transcriptomics and CRISPR screens for ARS specifically: 🔎 not found in this sweep.

Datasets: I did not run just discover-datasets. Do that before adding any datasets: block, and remember the relevance-triage warning — searching "radiation" or a DDR gene will surface a mountain of radiotherapy and cancer datasets that resolve perfectly and have nothing to do with ARS. That's Named Entity Confusion reached through dataset search, and it's a live risk here more than for most diseases.

7. Anatomical Structures Affected

Organ level

Primary (in rough order of radiosensitivity):

Secondary / late (DEARE): lung UBERON:0002048 ✅ · kidney UBERON:0002113 ✅ · brain UBERON:0000955 ✅ · central nervous system UBERON:0001017 ✅ · lens of camera-type eye UBERON:0000965 ✅ · thyroid gland UBERON:0002046 ✅. 🔎 LEAD: heart, oral mucosa, salivary gland, vasculature/microcirculation.

Body systems: hematopoietic/immune, gastrointestinal, integumentary, reproductive, then cardiovascular, respiratory, renal, and nervous late.

Tissue and cell level

Epithelium (intestinal, epidermal, follicular), hematopoietic tissue, vascular endothelium, and lymphoid tissue. The cell list is in §6. The organizing principle worth stating in the entry's prose: radiosensitivity tracks proliferative rate (the old Bergonié–Tribondeau observation), with lymphocytes as the glaring exception — they're radiosensitive despite being quiescent, because they're primed for apoptosis.

Subcellular level

• Nucleus / chromosome — the DSB target. 🔎 LEAD: GO:0005634 nucleus, GO:0005694 chromosome (verify).
• Plasma membrane lipid rafts — the ASMase/ceramide platform, and the reason the membrane is a second independent target rather than an afterthought. 🔎 LEAD: GO:0045121 membrane raft (verify).
• Mitochondrion — MOMP, ROS amplification, GO:0051882 mitochondrial depolarization ✅.
• Lysosome — ferritinophagy/autophagy arm.

Localization / lateralization

Bilateral and systemic by definition for whole-body exposure. The clinically important exception is partial-body exposure, which is common in real accidents (a worker holding a source) and produces asymmetric, geometry-dependent injury: severe local cutaneous/deep-tissue damage over the exposed area with relative marrow sparing elsewhere. This isn't a footnote — partial-body geometry is the thing that determines whether biodosimetry estimates are even interpretable, and it's why the field built assays that can distinguish partial- from total-body exposure.

8. Temporal Development

Onset

Age: any. Not congenital, not age-dependent — it's whenever the exposure happens. Occupational cases skew adult; mass-casualty scenarios include all ages, with children at higher risk per unit dose.

Pattern: acute, sharply. temporality: ACUTE is the right qualifier throughout. Prodromal symptoms begin within minutes to 48 hours; the entire syndrome is defined by a single point-source exposure.

Progression and the phase structure

Covered in §3, but the two things worth restating for a progression: block:

1. Higher dose compresses everything. The latent period shortens and eventually disappears; time-to-emesis shortens; time-to-death shortens. The phase structure is dose-parameterized, not fixed.
2. The subsyndromes are concurrent, not sequential. At 8 Gy a patient has H-ARS and GI-ARS and CRS simultaneously. Modeling them as stages would be wrong. MacVittie & Farese's framing (PMID:32868706) — "concomitant multiple organ injury" — is the right one.

Duration: self-limited in one direction or the other within weeks, then either full recovery, recovery-with-DEARE (lifelong), or death.

Critical intervention windows — the actionable part

There's a real asymmetry here that the entry should capture: the drugs work if given early, and the diagnostics work if sampled early, but the patient feels fine during exactly that window. The latent phase is a trap, and it's the single most operationally important fact about this disease.

Remission

Recovery is treatment-assisted, not spontaneous above ~2 Gy — marrow reconstitution from surviving stem cells, accelerated by cytokines. Below ~2 Gy, spontaneous recovery is the norm. Above ~12 Gy, neither happens.

9. Inheritance and Population

Inheritance

Not applicable. No inheritance pattern, penetrance, expressivity, anticipation, mosaicism, founder effect, consanguinity role, or carrier frequency. Say so explicitly.

Epidemiology

Orphanet prevalence: 1–9 per 1,000,000. That maps to prevalence_class: BAND_1_9_PER_1000000 and a rate_per_100000 of roughly 0.1–0.9 — though honestly a "prevalence" for an accident-driven condition is a strange quantity, and the notes: field should say that plainly. It's a rare-disease registry artifact more than an epidemiological measurement.

Cumulative human experience (the more meaningful figure), from Lazarus & Gale (PMID:40074513):

"Radiological accidents/incidents are common with nearly 400 reported since 1944 exposing about 3000 people to substantial doses of ionizing radiations with 127 deaths."

Chernobyl — the reference case series. From Belyi et al. (PMID:20445398): "134 of those were confirmed, including 28 patients who died due to lethal total-body gamma-irradiation." The fuller picture from IAEA/WNA sources: 237 initially suspected, 134 confirmed by 1989, doses 0.8–16 Gy, 28 deaths within 11 to 96 days, and a further 19 deaths 1987–2004 from causes not necessarily radiation-attributable.

Goiânia, 1987 — a Cs-137 teletherapy source opened in a scrapyard: ~250 people contaminated, 4 deaths from ARS. The definitive account is the IAEA's The Radiological Accident in Goiânia.

Tokaimura, 1999 — criticality accident at a uranium conversion plant, 2 worker deaths, 436 people dose-assessed.

Hiroshima/Nagasaki, 1945 — the largest dataset, but combined injury throughout.

REAC/TS registry inclusion criteria (useful for understanding what "a case" means): ≥250 mSv to whole body/marrow/gonads, ≥6,000 mSv to skin of extremities, ≥750 mSv to eye or thyroid, or internal contamination above half the maximum permissible body burden.

Demographics

• Sex ratio: no biological sex difference in radiosensitivity is well established for ARS. Observed case series skew male because of occupational composition, not biology. ⚠️ Don't record a sex ratio as a disease property.
• Geographic distribution: wherever sources and reactors are. Historical clusters follow accidents, not populations.
• Age: children more sensitive; occupational cases adult.
• Ethnicity: no established difference.

10. Diagnostics

ARS has an unusual diagnostic problem: there is no confirmatory test that's fast, and the fast tests aren't confirmatory. So triage runs on clinical kinetics while the real dosimetry catches up.

Clinical dose assessment — the frontline

Time to emesis. From the REMM/AFRRI guidance: "For time to emesis less than 4 hours, the effective whole-body dose is likely to be at least 3.5 Gy" and "If time to emesis is less than 1 hour, the whole-body dose probably exceeds 6.5 Gy, and a very complicated and likely fatal medical course may be expected."

Lymphocyte depletion kinetics. Serial absolute lymphocyte counts, ideally q6h for 48 h. The rule of thumb: "if within the first two days of exposure, lymphocytes have decreased by 50% and are less than 1000 cells per μL, the patient has received at least a moderate dose." Two formal models exist — the Andrews Lymphocyte Nomogram and the Goans exponential-decline model (see PMID:17993851, "Estimating radiation dose from time to emesis and lymphocyte depletion").

METREPOL response categories — the structured severity system, and a natural fit for a dismech definitions: entry with definition_type: PHENOTYPE_ALGORITHM and derivation_basis: ESTABLISHED_CRITERIA. It grades four organ systems separately — H (hematologic), N (neurovascular), C (cutaneous), G (gastrointestinal) — each on a 1–4 scale, and "identifies the likelihood of 'irreversible' (H4), and 'reversible' (H3, H2, and H1) damage to the bone marrow", linking the four to predict multiorgan failure. Lebaron-Jacobs & Herrera-Reyes (PMID:34801995) revisited it 20 years on, asking "if proposed criteria are still relevant for the medical management of radiation-induced injuries" — so treat it as established-but-under-review, with validation_status.status: UNVALIDATED unless you find a validation study.

Biodosimetry — the confirmatory layer

Dicentric chromosome assay (DCA) — the gold standard. Jeong et al. (PMID:36543843): "The dicentric chromosome assay is the 'gold standard' in biodosimetry for estimating radiation exposure." Its weakness is throughput: it needs a 48-hour lymphocyte culture and manual scoring, which is fine for one patient and hopeless for ten thousand.

The whole recent field is about fixing that: - RENEB inter-laboratory comparison 2021 (Endesfelder et al., PMID:37018160): "In total 33 laboratories from 22 countries around the world participated", with all participants successfully classifying extreme doses into clinically relevant exposure groups. This is the network that would actually run a mass-casualty response. - RABiT-II high-throughput DCA (Royba et al., PMID:35994701) — automated, validated under complex exposures including mixed fields. - Deep-learning automated scoring (PMID:36543843, PMID:38687685) — "particularly advantageous in scenarios such as large-scale radiological incidents." - G0-PCC (premature chromosome condensation) multiwell assay (PMID:38854157): "Our method can address the need for a same-day cytogenetic biodosimetry test in radiation emergency situations." Same-day is the goal.

Other cytogenetics: cytokinesis-block micronucleus assay (faster, less specific); FISH translocation painting (for retrospective dose years later); ring chromosomes.

Transcriptional/protein biodosimetry: the FDXR/DDB2/CDKN1A/AEN qPCR panels (§4), valid within ~72 h. Machine-learning combinations — e.g. ACTN1 + DDB2 + FDXR plus B and T cell counts — can "quantify and distinguish between partial-body irradiation and total-body irradiation exposures", which no single marker does.

Physical dosimetry: personal dosimeters where worn; EPR spectroscopy of tooth enamel, fingernails, or toenails; OSL of personal electronics; activation analysis (Na-24 in blood) for neutron exposure.

The NIAID program overview is Satyamitra et al. (PMID:37742625), "The NIAID/RNCP Biodosimetry Program: An Overview."

Routine clinical workup

CBC with differential q6h × 48 h then daily; comprehensive metabolic panel; serum amylase (rises with salivary gland dose); C-reactive protein; blood/urine/stool cultures; HLA typing early (do it while there are still lymphocytes to type); type and screen. Imaging as clinically indicated. Skin photography with serial documentation and mapping for CRS. Biopsy is rarely needed for diagnosis but marrow biopsy documents hypocellularity.

LOINC-coded lab observations for a biochemical: block: absolute lymphocyte count, absolute neutrophil count, platelet count, hemoglobin. 🔎 I did not resolve specific LOINC codes — look them up rather than guessing, and note the repo's memory that LOINC does not carry reference ranges as a field.

Genetic testing

Not applicable for diagnosis. ARS is not a genetic disease. Genetic testing has one narrow role: if a patient shows radiation toxicity grossly disproportionate to estimated dose, consider a DNA-repair-deficiency syndrome (ATM, NBN, LIG4, etc.) — that's a rare-but-real scenario and worth a sentence, not a section. WGS/WES/panels/CMA/karyotype/FISH/mtDNA/repeat expansion: all not applicable. (Note the irony that karyotyping is central here — just as a dosimeter, not a genetic test.)

Differential diagnosis

This is where ARS actually gets missed, because a patient with nausea, vomiting, and falling counts and no exposure history looks like ten other things. Ebeling et al. built a whole simulation curriculum around exactly this (PMID:37538304): "ARS is a high-risk, low-frequency diagnosis that can be fatal and is difficult to diagnose without an obvious history of ionizing radiation exposure."

The tell that separates ARS from all of them is the combination of a compressed prodrome, a deceptively well latent period, dose-proportional lymphocyte decline, and dicentric chromosomes. Nothing else does all four.

Screening

No population screening — there's no pre-symptomatic state to screen for. "Screening" in the ARS context means mass-casualty triage biodosimetry: rapidly sorting thousands of worried-well from the genuinely exposed. That's the entire design driver behind RABiT-II, the G0-PCC assay, and the qPCR panels.

11. Outcome / Prognosis

Survival

Prognosis is a nearly pure function of dose, medical care, and combined injury.

Human LD50/60: ~2.7–3.1 Gy marrow dose without medical care (A-bomb data, DS02); commonly quoted as ~4.5 Gy with modern care, though as noted in §2 that higher figure is soft.

NHP LD50/60 with medical management: 7.52 Gy (Farese et al., PMID:22929469): "The study defined an LD30/60 of 7.06 Gy, LD50/60 of 7.52 Gy, and an LD70/60 of 7.99 Gy." ⚠️ Do not present this as a human number — the macaque with full supportive care is roughly twice as tolerant as an unsupported human, and conflating the two is an easy and serious error.

Cause of death: in H-ARS, infection and hemorrhage during aplasia. In GI-ARS, sepsis from bacterial translocation plus fluid/electrolyte collapse. In CNS syndrome, cerebral edema and cardiovascular collapse.

Morbidity and function

Survivors face DEARE — progressive renal impairment, pulmonary fibrosis, cardiovascular disease, cataract, endocrine dysfunction, infertility, cognitive/behavioral change, and elevated cancer risk. Gasperetti et al. (PMID:36688956) drive the point home: "These data show survivors of H-ARS are at risk for the development of delayed renal toxicity and emphasize the need for the development of medical countermeasures for delayed renal injury." The Ukrainian ARS survivor cohort (PMID:20445398) is the long human record.

There's a structural irony worth writing into the entry: as H-ARS treatment improves, DEARE becomes a bigger problem, because more people survive to develop it. Wu et al. (PMID:37014943) note that "while effective medical countermeasures (MCM) for the hematopoietic-acute radiation syndrome (H-ARS) have been identified and approved by the FDA, development of MCM for DEARE has not yet been successful." We got good at the first act and never wrote the second.

QoL instruments: 🔎 No ARS-specific validated instrument found. EQ-5D/SF-36/PROMIS data for this population: not located.

Prognostic factors

Absorbed dose and its uniformity; time to emesis; lymphocyte depletion slope; METREPOL response category (especially H4 = irreversible marrow damage); presence of combined injury (major adverse factor); age; comorbidity; internal contamination; time from exposure to cytokine administration.

Prognostic biomarkers: dicentric frequency; ALC nadir and slope; the FDXR/DDB2 transcript panel; 🔎 candidate protein panels not yet clinically qualified.

12. Treatment

The frame

There is no antidote. Treatment is: replace what the marrow can't make, push what's left to regenerate faster, keep infection out, and manage the gut and skin. Everything approved is in the second category.

FDA-approved medical countermeasures

All licensed under the Animal Rule — no human efficacy trials exist or can exist. Indication wording is consistently "acute exposure to myelosuppressive doses of radiation." From REMM's countermeasure page:

⚠️ Verify the biosimilar approval dates against FDA sources before committing them — they came from a single page fetch and dates are exactly the kind of thing that gets transcribed wrong.

Romiplostim is the mechanistically distinct one and worth its own treatment record. It targets the platelet arm rather than the neutrophil arm — Bussel et al. (PMID:34079225): "Romiplostim binds to and activates the TPO receptor on megakaryocyte precursors, thus promoting cell proliferation and viability, resulting in increased platelet production." The pivotal NHP data (Bunin et al., PMID:37224926) showed a "40% to 55% survival benefit compared with controls, less severe clinical signs, reduced incidence of thrombocytopenia and/or neutropenia." Single dose, which matters enormously for mass-casualty logistics.

Lazarus & Gale's clinical bottom line (PMID:40074513): "The favorable benefit-to-risk ratio of these drugs over hematopoietic cell transplants suggests giving them soon after exposure to acute high-dose and-dose-rate whole body ionizing radiations."

Suggested treatment YAML shape

treatments:
- name: Filgrastim
  therapeutic_modality: OTHER        # 🔎 no ideal value; G-CSF is a recombinant protein
  treatment_term:
    preferred_term: Pharmacotherapy
    term:
      id: NCIT:C15986                # ✅ VERIFIED
      label: Pharmacotherapy
    therapeutic_agent:
    - preferred_term: filgrastim
      term:
        id: NCIT:C1474               # ✅ VERIFIED in cache/ncit/terms.csv
        label: Filgrastim

⚠️ Two traps here. 1. There's a known pattern in this repo where NCIT drug terms fail therapeutic_agent dynamic-enum validation even when the CURIE and label are both correct — CHEBI is the safer default. But filgrastim/pegfilgrastim/sargramostim/romiplostim are recombinant proteins and peptibodies with no CHEBI terms, so NCIT is the only option. Run just validate-terms on this block specifically and be ready for it to fail. 2. therapeutic_modality has no value that cleanly fits a recombinant cytokine. PROTEIN_REPLACEMENT is wrong (nothing is being replaced), SMALL_MOLECULE is wrong. OTHER is honest. Don't stretch a value to look complete.

Verified NCIT terms for the rest:

Reminder from the repo's own guidance: NCIT:C15433 Nutritional Support should not be reflexively tagged BEHAVIORAL — here it usually means parenteral nutrition, which isn't a diet-pattern change.

Supportive care backbone

This is where survival actually comes from, and it deserves as much detail as the drugs: - Protective/reverse isolation during neutropenia - Broad-spectrum antimicrobial prophylaxis and treatment — antibacterial, antifungal, antiviral - Transfusion support — and here's the ARS-specific detail that matters: all cellular blood products must be irradiated to prevent transfusion-associated GVHD, because a profoundly lymphopenic irradiated patient cannot reject donor lymphocytes. Leukoreduced and CMV-safe as well. This is a genuine ARS-specific practice point, not generic transfusion medicine - Antiemetics — 5-HT3 antagonists; ondansetron CHEBI:7773 ✅ - Fluid and electrolyte replacement, especially in GI-ARS - Nutritional support, often parenteral when the gut is denuded - Analgesia and wound care for CRS; surgical debridement, grafting, sometimes amputation - Psychosocial support — the Chernobyl and Goiânia experiences both document severe psychological morbidity, including in the unexposed

Hematopoietic cell transplantation

Reserved and controversial. Arnautou & Garnier: HSCT "will be carefully considered on an individual basis, especially for patients who do not respond following 3 weeks of cytokine therapy." The Chernobyl transplant experience was poor — GVHD in the setting of combined injury, plus autologous recovery in patients who'd been transplanted anyway. The modern consensus leans strongly toward cytokines first.

Investigational — the pipeline

Hematopoietic: - TPOm (JNJ-26366821), a PEGylated thrombopoietin mimetic — works through the niche (PMID:38679747): "TPOm interacts with BM vascular and stromal niches to locally support hematopoietic reconstitution." - HemaMax (rHuIL-12) — Phase 1 completed, NCT01742221, "Hematopoietic syndrome due to acute radiation syndrome" - PLX-R18 placental cell therapy — Phase 1, NCT03797040, for post-exposure prophylaxis or treatment of HS-ARS - KMRC011 — Phase 1 completed, NCT03585803, listed condition "Acute radiation syndrome" - 16,16-dimethyl PGE2 — effective as radioprotector but notably not as a mitigator for DEARE (PMID:37014943): given after TBI it "enhances survival from H-ARS but has little impact on RBMD or other DEARE." A nice worked example of directness mattering - Eltrombopag post-TBI — NCT00903929, Phase 1, completed

Radioprotectors (pre-exposure): - BIO 300 (synthetic genistein nanosuspension) — NHP pilot (PMID:39160790): all four treated animals survived to day 60 vs 50% of vehicle controls, but notably "BIO 300 Injectable Suspension did not mediate an improvement in blood cell counts," and the authors conclude "protection against irradiation is attainable without much improvement in the complete blood count (CBC) profile." Oral suspension Phase 1 NCT06741345 completed Dec 2024. ⚠️ n=4 per arm — a pilot, not efficacy - 5-androstenediol / NEUMUNE (PMID:38097137) — IND status granted; "modulate[s] cell cycle progression, reduces radiation-induced apoptosis, and regulates DNA repair" - Gamma-tocotrienol, amifostine (CHEBI:2636 ✅, active metabolite WR-1065 CHEBI:72583 ✅ — but amifostine is approved only for radiotherapy-related xerostomia and cisplatin nephrotoxicity, not ARS)

GI-targeted: - MIIST305, a mucus-layer glycopolymer (PMID:39930324) — the standout result: "Approximately 85% of the animals survived the irradiation exposure... In contrast, no control, Vehicle-treated animals survived past day 10 at this radiation dose." - Anti-ceramide scFv (Nagesh et al., PMID:37815783) — mechanistically the most elegant thing in the pipeline, targeting the Step-3 ceramide arm directly. Mice at 15 Gy + BMT + scFv survived to day 90 with "normal appearance, behavior, and serum biochemistries, and surprisingly, at full autopsy, near-normal physiology in all 42 tissues examined" - Captopril (CHEBI:3380 ✅) — ACE inhibitor, mitigates H-ARS in Göttingen minipigs (PMID:34449797, PMID:38674120); "suppressed radiation-induced IL1B and TNFA" - Microbial metabolites — propionate, tryptophan metabolites, I3A, valeric acid (§6) - Liangxue-Guyuan-Yishen decoction — a TCM formulation with GI-ARS rat data via WNT and MEK/ERK (PMID:37697698)

Cell and EV therapies: - MSCs and MSC-derived extracellular vesicles (Miura & Fujii, PMID:39679884) — promising preclinically, but the authors are appropriately cautious: "the effectiveness of MSC transplantation in addressing acute radiation syndrome affecting organs in irradiated individuals is limited" and "further investigation is required" - Umbilical cord blood (Hurley et al., PMID:37967239) — improved survival, hematopoietic recovery, reduced GI damage, and mitigation of pneumonitis/fibrosis in preclinical work - Romiplostim-induced EVs (Yamaguchi et al., PMID:37238707) — a genuinely odd and interesting result: serum EVs from romiplostim-rescued mice, transferred to other irradiated mice, "drastically improved by 50-100%" their 30-day survival, with miR-144-5p found only in EVs from treated animals

Combined injury: nothing. Worth stating as an explicit gap (PMID:36947602).

Treatment algorithms

Follow the METREPOL response category. Broadly: decontaminate → assess dose (clinical + biodosimetry) → cytokines within 24 h if dose >2 Gy → supportive care scaled to RC → HSCT decision at ~3 weeks for non-responders → DEARE surveillance for survivors.

Pharmacogenomics: none established for ARS. NOT_APPLICABLE.

Personalized medicine: dose-guided rather than genotype-guided. The "biomarker" here is absorbed dose, and the whole biodosimetry enterprise exists to make treatment personalized in that sense.

13. Prevention

Primary prevention — where nearly all the real benefit lives

• Radiation protection fundamentals: time, distance, shielding. ALARA.
• Source security and regulatory control — the Goiânia accident happened because a teletherapy source was abandoned in an unsecured building. "A caesium-137 teletherapy unit was left unsecured after an institute moved to new premises, becoming totally insecure." Every orphan-source accident is a regulatory failure before it's a medical one.
• Reactor safety, criticality controls, transport regulation (IAEA framework)
• Emergency planning — sheltering, evacuation, exclusion zones
• Occupational dosimetry and dose limits
• Nuclear nonproliferation and counterterrorism — the actual primary prevention for the scenarios driving current funding

Pharmacological primary prevention: none approved. Restating Singh et al. (PMID:39160790): "Currently, there is no radiation medical countermeasure approved by the U.S. FDA which can be used before radiation exposure to protect exposed individuals."

The KI distinction, one more time

Potassium iodide (CHEBI:8346 ✅) is thyroid-blocking prophylaxis against radioiodine uptake. It is on every emergency-planning list and is not ARS prophylaxis. If the KB entry lists it under treatments: without a scoping note, a reader will reasonably conclude it protects against whole-body exposure. It does not. Same logic for Prussian blue (Cs/Tl decorporation) and Ca-DTPA/Zn-DTPA (transuranic decorporation) — these are for internal contamination, which is a different disease process with a different time course. Consider ECTO:9000084 "exposure to iodine" ✅ if you model the radioiodine arm at all, and consider whether internal contamination deserves a differentials: entry rather than a treatment entry.

Secondary prevention

Mass-casualty triage biodosimetry (§10) — rapid identification of who actually got a dose so countermeasures go to the right people and the worried-well don't consume the supply.

Tertiary prevention

Countermeasure administration within the therapeutic window; infection prophylaxis; irradiated blood products; DEARE surveillance — renal function, pulmonary function, ophthalmologic exam, cardiovascular risk management, cancer screening. The Ukrainian ARS survivor cohort is the model for this.

Immunization

Not applicable — no vaccine, no infectious etiology. (Standard vaccination status matters for a neutropenic patient's infection risk, but that's not ARS prevention.)

Genetic screening and counseling

Not applicable. No carrier screening, no PGD, no prenatal testing. NCIT:C15240 Genetic Counseling ✅ has a vanishingly narrow role — only if a DNA-repair syndrome is uncovered incidentally through disproportionate radiosensitivity.

Public health

Emergency preparedness training (the simulation curricula like PMID:37538304 exist because clinicians genuinely don't recognize this), stockpiling (Strategic National Stockpile holds the approved cytokines), risk communication, environmental monitoring and remediation, food/water controls after a release.

14. Other Species / Natural Disease

Taxonomy

Every mammal gets ARS, with the same subsyndrome structure and different dose thresholds. Verified taxon IDs from cache/ncbitaxon/terms.csv:

NCBITaxon:9606 Homo sapiens ✅ · NCBITaxon:10090 Mus musculus ✅ · NCBITaxon:10116 Rattus norvegicus ✅ · NCBITaxon:9544 Macaca mulatta ✅ · NCBITaxon:9615 Canis lupus familiaris ✅ · NCBITaxon:9823 Sus scrofa ✅ · NCBITaxon:9825 Sus scrofa domesticus ✅ · NCBITaxon:7955 Danio rerio ✅

🔎 LEAD: Macaca fascicularis (cynomolgus).

Comparative radiosensitivity

Species differ substantially in LD50 — dogs are notably more radiosensitive than mice, primates intermediate. ⚠️ I did not find a single citable modern comparative table, and the numbers floating around secondary sources vary with dose rate, strain, and supportive care. Don't tabulate species LD50s without a primary citation per row. One 2025 paper directly compares macaque species ("Comparison of sensitivity of rhesus and cynomolgus macaque for acute radiation effects", Sci Rep) — worth chasing for a proper citation.

Natural disease in other species

Yes, and it's documented. Cannon & Kiang (PMID:32663058) reviewed wildlife after Chernobyl and Fukushima: "Humans were evacuated from the immediate regions but the wildlife stayed and continued to be affected." Their review covers effects on "vegetation, insects, fish, birds and mammals," and notes that "adaptation to radiation is evident and the ecosystems have dynamically changed."

Domestic animals in the Chernobyl exclusion zone experienced ARS. There is no established naturally occurring animal ARS outside of accidental exposure — no OMIA entry, because it isn't a genetic condition.

VBO breed identifiers: not applicable — no breed predisposition.

Orthologous genes: the DDR is deeply conserved (ATM, TP53, the MRN complex, NHEJ machinery all have orthologs from yeast through mammals), which is exactly why model organisms work here. 🔎 Specific NCBI Gene IDs not retrieved.

Zoonosis / transmission

Not applicable. ARS is not transmissible. Worth stating explicitly, because there is a persistent public misconception that irradiated people are themselves radioactive. They are not — unless they carry internal or external contamination, which is a separate and real concern for responders. That distinction (irradiation ≠ contamination) is worth a notes: line; it's the single most consequential public-understanding error about this disease.

15. Model Organisms

Because of the Animal Rule, model organisms aren't a supporting line of evidence here — they are the regulatory basis for every approved therapy. That inverts the usual weighting and should be said in the entry.

The consortium context

MCART (Medical Countermeasures Against Radiological Threats), NIAID-sponsored, is the organizing body: "charged with developing medical countermeasures (MCM) to treat the key sequelae of acute radiation syndrome (ARS) and the delayed effects of acute radiation exposure (DEARE)." Its models are built "within the criteria of the FDA's 'animal rule.'" Key reference: MacVittie & Farese, PMID:32868706, "Defining the Concomitant Multiple Organ Injury within the ARS and DEARE in an Animal Model Research Platform."

Models

Mouse — NCBITaxon:10090 ✅ - C57BL/6J TBI — the workhorse H-ARS model, 30-day survival endpoint, LD50/30 typically ~7–8 Gy strain- and facility-dependent - Partial-body irradiation with 5% bone marrow shielding at 13 Gy — the standard GI-ARS model (PMID:39930324). The shielding is the whole trick: without it the animal dies of H-ARS before GI-ARS can be studied - 15 Gy WBI — the ~90% GI-ARS lethal dose for C57BL/6J (PMID:37815783) - ASMase-knockout (Smpd1−/−) — the genetic proof of the ceramide arm - DEARE longitudinal model — mice followed to 12+ months (PMID:37014943) - Recapitulation: excellent for H-ARS kinetics and mechanism. Limitations: small body mass changes dosimetry; different marrow reserve; supportive care not comparable to human ICU; 30-day endpoint misses DEARE by design

Rat — NCBITaxon:10116 ✅ - WAG/RijCmcr — Gasperetti et al. (PMID:36688956) established it "as an effective model for the evaluation of medical countermeasures (MCM) for acute hematologic radiation syndrome (H-ARS)." LD50/30 determined for adult and pediatric animals, both sexes. "87.5% and 100% of adult rats succumb to lethal hematopoietic acute radiation syndrome (H-ARS) at TBI doses of 8 and 8.5 Gy, respectively", and pegfilgrastim "improved 30 d survival from 12.5% to 83% at 8 Gy and from 0% to 63% at 8.5 Gy." - The best DEARE model — followed to 300 days with renal endpoints. Lung-DEARE at PBI >12 Gy (pneumonitis days 50–100); kidney-DEARE at >8 Gy (BUN >120 mg/dL)

Non-human primate — NCBITaxon:9544 ✅ (rhesus); cynomolgus 🔎 - The pivotal licensure model. Farese et al. (PMID:22929469): 48 rhesus macaques, blinded and randomized, LD30/60 7.06 Gy, LD50/60 7.52 Gy, LD70/60 7.99 Gy, with supportive care, "with a relatively steep slope of 1.13 probits per linear dose" - Different radiation sources give different curves — a LINAC 6MV photon study reported LD30/50/70 of 5.71 / 6.78 / 7.84 Gy. Source and dose rate are not interchangeable - Recapitulation: the closest to human physiology and supportive care. Limitations: cost, ethics, small n (the BIO 300 study had four animals per arm), and macaques with full care substantially out-tolerate unsupported humans

Minipig — NCBITaxon:9823 / NCBITaxon:9825 ✅ - Göttingen minipig H-ARS — captopril studies at 1.79–1.80 Gy Co-60 (PMID:34449797, PMID:38674120); skin physiology is the closest available to human, making it the preferred CRS model - Sinclair minipig GI-ARS — dose-finding with microbiome and inflammasome endpoints (PMID:39012765) - ⚠️ Provenance caution: the two Kenchegowda/Seed/Singh minipig methodology papers (PMID:34402700, PMID:32892657) came back with one flagged RETRACTED. Check retraction status before citing either

Canine — NCBITaxon:9615 ✅ — historical importance (much of the classical dose-response and marrow-transplant work), less used now.

Zebrafish — NCBITaxon:7955 ✅ — 🔎 not prominent in the ARS literature I sampled; more common for developmental radiobiology.

In vitro / NAM systems: - Human intestinal organoids — used to validate the macrophage/NRG1 regeneration finding (PMID:40086603) and the human tuft-cell reserve pool (PMID:39358509). These belong in experimental_models: (not animal_models:) - Human intestinal resection tissue — the strongest evidence tier available for GI mechanism - Hematopoietically humanized mice — used for candidate protein biodosimetry markers - Primary human lymphocyte culture — the substrate for the dicentric assay itself

Modeling these in dismech

Use ModelMechanismLink on animal_models: entries (this is what modeled_mechanisms is for — animal models never belong in experimental_models:). Sketch:

animal_models:
- name: Rhesus macaque total-body irradiation with medical management
  species: Rhesus macaque
  publication: PMID:22929469
  modeled_mechanisms:
  - target: <the H-ARS marrow failure node>
    relationship: RECAPITULATES
    fidelity: HIGH
    description: >-
      Dose-response for hematopoietic lethality with supportive care;
      the licensure model for every FDA-approved H-ARS countermeasure.
    limitations: >-
      LD50/60 of 7.52 Gy with full supportive care substantially exceeds the
      human LD50/60 of roughly 2.7-3.1 Gy estimated from atomic-bomb survivor
      marrow doses, so the dose axis does not transfer directly to humans.
    readouts:
    - name: 60-day survival
      target: <same node>
      direction: DECREASED

And there's at least one good FAILS_TO_RECAPITULATE / HUMAN_MODEL_MISMATCH candidate here: the behavioral and cognitive DEARE findings in mice (PMID:37014943 — anxiety changes in females, blunted shock flinch, altered exploratory behavior in males) have no established human counterpart. That's exactly the "evidence exists in a model, translational validity is the open question" shape the schema's HUMAN_MODEL_MISMATCH was built for.

Model resources

MGI, RGD, ZFIN, Alliance of Genome Resources for the standard organism databases; MCART for consortium models; NIAID/RNCP for the countermeasure program.

Gaps and honest non-findings

Things I looked for and did not find, so nobody re-runs the same search:

• A validated QoL instrument for ARS. None found. EQ-5D/SF-36/PROMIS data: absent.
• An ARS-specific epigenetic mechanism with enough weight to be a pathophysiology node.
• Spatial transcriptomics or CRISPR screens specific to ARS.
• A clean modern comparative species LD50 table with a citable primary source.
• A definitive ICD-11 code. One source said NF00, the template guessed NF06. Unresolved.
• A DOID identifier.
• Human efficacy data for any ARS drug. By construction — the Animal Rule exists because this data cannot exist.
• Recent ClinicalTrials.gov ARS-condition trials are sparse: NCT01742221 (HemaMax), NCT03585803 (KMRC011), NCT03797040 (PLX-R18), NCT00903929 (eltrombopag), NCT06741345 (BIO 300 oral, Phase 1). Most acute radiation syndrome query hits are TBI-conditioning transplant trials, which are a different thing wearing similar words — classic Named Entity Confusion, and worth a note in the entry so the next curator doesn't fall in.
• LOINC codes for the ARS lab panel — not resolved.

Suggested curation approach, briefly

Some structural calls that follow from the above:

1. entry_type: DISEASE. ARS is a real clinical entity with a defined mechanism, not a grouping — even though it bundles four subsyndromes.
2. Model the four subsyndromes as has_subtypes, not separate entries. They share one etiology, one dose axis, and one temporal architecture; they differ only in which tissue's threshold you crossed. Names should be short and slug-friendly: H-ARS, GI-ARS, CRS, CNS-ARS. Then use the subtype: foreign key on phenotypes and progression records.
3. The initiating node is the exposure, wired via environmental[].influences_mechanisms with TRIGGERS. If it isn't wired, the pathograph will have every mechanism node orphaned from its cause — one of the recurring reviewer findings in this repo.
4. Watch for module conformance opportunities. just list-modules before you build anything — cellular senescence, inflammaging, and any fibrosis or DAMP/inflammation module are plausible conformance targets, and there may be a toxicology-family module that already covers part of this chain under a name you wouldn't guess.
5. Give the "irradiation ≠ contamination" distinction and the "KI is not ARS prophylaxis" distinction each a notes: line. They're the two errors most likely to be introduced by a future well-meaning editor.
6. Expect heavy MODEL_ORGANISM grading and don't apologize for it — but keep the human accident evidence (Chernobyl, Goiânia, Tokai-mura, A-bomb) carrying the human phenotype claims.

Sources

Primary literature (PMIDs cited above): 40368913 · 39025400 · 38616048 · 39930324 · 39000080 · 40074513 · 37280005 · 38679747 · 39679884 · 37967239 · 39160790 · 37238707 · 36688956 · 38097137 · 37014943 · 38674120 · 37815783 · 37697698 · 37538304 · 37224926 · 20445398 · 38333215 · 36947602 · 34488201 · 34801995 · 34233299 · 33979447 · 32663058 · 37018160 · 36543843 · 38687685 · 35994701 · 38854157 · 38029728 · 37742625 · 33122357 · 38706205 · 31931652 · 40192235 · 36253079 · 36368026 · 34084131 · 31019301 · 39358509 · 38848678 · 40086603 · 37669929 · 22929469 · 32868706 · 34449797 · 39012765 · 32355263 · 32791865 · 36752571 · 30421807 · 34079225 · 1762100 · 2693695 · 17993851 · 18544701 · 39270512 · 37642199 · 39600027 · 37149389 · 38003561 · 34402700 · 39355046

Databases and guidance: - MONDO:0033938 via EBI OLS4 - Orphanet — Acute radiation syndrome (ORPHA:454831) - MeSH D054508 - ICD-10-CM T66 - REMM — Acute Radiation Syndrome · Myeloid Cytokines for H-ARS · Time Phases of ARS · Lymphocyte Depletion Kinetics · Time to Onset of Vomiting - CDC — Acute Radiation Syndrome: Information for Clinicians - AFRRI — Medical Management of Radiological Casualties, 4th ed. - IAEA — The Radiological Accident in Goiânia · IAEA — Tokaimura · UNSCEAR — Chernobyl - REAC/TS Radiation Accident Registry overview - NIAID — approval of romiplostim for acute radiation injury · FDA Nplate approval letter, 2021 - MCART Consortium Animal Models Series - ClinicalTrials.gov: NCT01742221 · NCT03585803 · NCT03797040 · NCT00903929 · NCT06741345

One last time, because it's the thing most likely to bite: every quote in here needs re-verification against just fetch-reference before it becomes a snippet:, and every 🔎 CURIE needs just validate-terms before it becomes a term:. The verified ✅ ones came from this repo's own caches or live OLS, so those should sail through — but the MONDO and ECTO terms in particular are worth a validation run in both the worktree and the primary checkout, since the write hook validates from the primary and that's exactly where a cache gap hides.

Reference Validation

Checked with linkml-reference-validator 0.2.1.

All extracted references resolved successfully.