10.3 Non-Cancer Systemic Toxicity Endpoints & Target Organ Critical Effects

Why non-cancer endpoints still need a route and a critical effect

Handbook III.A.3–4 sits in hazard identification: what adverse effects can this agent produce, in which tissues, by which routes, and which finding will later drive a reference dose (RfD) or reference concentration (RfC)? Independent OpenExamPrep teaching here is about interpreting surrogates and selecting the critical effect, not repeating chapter 1’s study-design recipe for how to pick a route when you still have a blank protocol. Chapter 12 will attach NOAEL/BMD math; chapter 13 will attach uncertainty factors. This section decides which endpoint is even a candidate. It is not an ABT or EPA product and does not claim official approval, review, or partnership with ABT or EPA.

Non-cancer hazards include target-organ toxicity (liver, kidney, lung, nervous system, reproduction), irritation and corrosion, sensitization, phototoxicity, immunotoxicity, endocrine disruption short of a tumor, and ecological apical effects. Cancer classification was chapter 9; mutagenic versus non-mutagenic cancer MOAs were section 10.1. Many of the same organs appear in both ledgers. The skill is to keep non-cancer critical effects from being steamrolled by a species-specific tumor you already judged not human-relevant.

Route surrogates: oral

Gavage is a precise mg/kg bolus. Hazard identification must ask whether the high Cmax, esophageal/forestomach contact, and possible reflux/aspiration are properties of the human ingested pattern or artifacts of the surrogate. A gavage-only gastric erosion at a volume that humans never see is a local finding that may not represent sipping a water contaminant. Conversely, gavage can reveal a Cmax-driven arrhythmia or seizure that a slow dietary intake would miss. When the human takes a discrete pharmaceutical dose, gavage (or capsule) is often the right surrogate, not a second-best.

Diet mixes closer to food-use chemicals and feed-through pesticides. Achieved dose equals concentration × food intake / body weight. Palatability-driven starvation produces body-weight loss, delayed ossification, and “liver enzyme” stories that are secondary to wasting (chapter 8’s secondary-effect language). For hazard ID, a dietary NOAEL that is really a food-refusal NOAEL is not a clean systemic critical effect. Homogeneity failures create over- and under-dosed animals in the same group—noise that can hide or fake a target organ.

Drinking water is the surrogate for many environmental contaminants. Taste aversion, photolysis, hydrolysis, and bottle leakage change achieved dose. Insoluble or reactive chemicals do not become water-study candidates because gavage staffing is scarce. A water study that holds concentration constant while rats grow will drop mg/kg/day over time; report both concentration and achieved dose before you pick a critical effect.

Gavage, diet, and water can all be “oral” and still disagree on the same milligram-per-kilogram nominal dose because kinetics and local GI contact differ. Route-to-route extrapolation after the fact is a risk-assessment patch (chapters 11–13), not proof that the three oral surrogates identified the same hazard.

Route surrogates: respiratory and mucosal

Vapor (molecular gas phase) deposits by diffusion and solubility. Water-soluble vapors are scrubbed in the nose of obligate nasal-breathing rodents; humans who mouth-breathe at work may deliver more of that vapor to the larynx and bronchi. A rodent olfactory-only lesion from a soluble irritant vapor can overstate human nasal cancer or understate lower-airway irritation, depending on the stem. Do not call the olfactory finding “systemic neurotoxicity” without absorption evidence.

Aerosol (particles or droplets) deposits by mass median aerodynamic diameter (MMAD) and hygroscopic growth. A 4 µm MMAD workplace dust is not identified by a vapor-only chamber. Poorly soluble particles raise overload and local inflammation questions (chapter 9 nanomaterials). Hazard ID must say region: extra-thoracic, tracheobronchial, pulmonary.

Intratracheal instillation is a liquid or slurry bolus into the airway. It can place a known lung load for mechanism studies. It is not a 6-hour/day, 5-day/week inhalation exposure: distribution is patchy, local concentration spikes, and upper-airway scrubbing is skipped. Using an instillation granuloma as the RfC critical effect for a workplace vapor is a surrogate error. Chapter 1 forbade selling instillation as OECD 413; here you forbid selling it as the human inhalation hazard without a serious dosimetry bridge (chapter 11’s HEC/RGDR preview).

Sublingual (and buccal) absorption feeds the systemic circulation while largely bypassing first-pass hepatic metabolism that oral gavage encounters via the portal vein. A parent-driven CNS or cardiovascular effect can look stronger sublingually (nitroglycerin and nicotine are the teaching pair). A hepatotoxic reactive metabolite generated on first pass can look worse after gavage than after sublingual dosing of the same milligram. If the clinical or abuse route is sublingual and the only toxicology is oral gavage, you may have identified the wrong first-pass hazard. The reverse error is using sublingual animal data to represent swallowed environmental residues.

Anatomical species differences as interpretation, not trivia

Rodent forestomach (squamous, keratinized, no human equivalent) sees prolonged gavage irritants. Forestomach-only hyperplasia or tumors are local rodent-compartment findings; they do not automatically become human gastric cancer. Obligate nasal breathers (rat, mouse) versus oronasal humans change vapor maps, as above. Rats lack a gallbladder; cholestatic patterns and bile-volume physiology differ from dogs and humans—use it when the critical effect is biliary, not as a slogan on every liver file. Mouse club (Clara) cell CYP2F2 bioactivates some lung toxicants (naphthalene, styrene in teaching discussions) more aggressively than human lung; a mouse-only bronchiolar necrosis can be a metabolic species difference rather than a default human RfC driver. None of these differences lets you ignore a systemic finding that does not depend on that anatomy.

Local versus systemic

Local effects occur at the portal of entry or application site: skin corrosion, respiratory irritation, gavage-site erosion, injection-site myositis, eye damage. They can still be the critical effect for an RfC (a sensory irritant that limits concentration before any liver enzyme moves). Systemic effects occur after absorption: hepatocellular necrosis after dermal uptake, methemoglobinemia after ingested aniline, delayed neuropathy after an absorbed OP. A local lesion can limit dose so that a systemic hazard is never expressed in that study—absence of systemic findings is then censoring, not proof of systemic safety. Conversely, a high-dose gavage ulcer is not a systemic liver hazard.

Anticipated on-target versus unanticipated off-target—organism to ecosystem

Handbook language in this neighborhood tracks II.5 (on- versus off-target) up to populations and ecosystems.

Anticipated on-target at the organism. The effect follows the intended interaction: kinase-inhibitor myelosuppression; AChE inhibition by an OP insecticide in a pest insect; β-blockade bradycardia. It is still a hazard. On-target does not mean “acceptable.”

Unanticipated off-target at the organism. A different receptor, enzyme, or organ than the design intended: hERG block from a non-cardiac drug; PPARα activation from a plasticizer; forestomach irritation from a chemical meant to be a systemic enzyme inhibitor.

Population. Reduced fecundity, altered sex ratio, or increased mortality that changes population trajectory—wildlife, a fishery, or a human birth cohort. Individual key events still sit underneath (chapter 8); hazard ID at this scale is the apical population endpoint.

Ecosystem. Community structure or ecosystem services: an insecticide’s on-target AChE story in insects becomes an unanticipated aquatic-invertebrate kill that lets algae bloom. A rodent oral study will not identify that hazard; III.A.2’s receptor-species logic from chapter 9 still applies. On-target in the pest is not a free pass at the pond.

Critical-effect selection for a later RfD

The critical effect is the first adverse, human-relevant effect that appears as dose (or concentration) increases, or a known precursor that is itself adverse in the decision context. It is the endpoint that will usually supply the point of departure for an RfD/RfC (chapter 12) after uncertainty factors (chapter 13).

Selection rules that cost items when ignored:

• Adversity. Adaptive centrilobular hypertrophy with CYP induction, quiet leakage enzymes, and reversal (chapter 9) is a weak critical-effect candidate. Necrosis, fibrosis, lasting functional loss, or a true precursor (sustained TSH with follicular hyperplasia if you have already judged that loop human-relevant) is stronger.
• Human relevance. Male-rat α2u nephropathy, rodent forestomach-only irritation, and PPARα-only rodent liver tumors (section 10.1) are poor RfD drivers when those MOAs are the whole story. The next relevant effect—often a different organ or a different species—becomes critical.
• Sensitivity. Among remaining relevant effects, the one with the lowest POD after a fair comparison (same study duration logic, not a 14-day finding beating a 2-year finding without thought) typically wins. Do not pick the flashiest lesion at a high dose if a quieter relevant effect occurs lower.
• Route. A gavage GI ulcer is a poor critical effect for an inhaled dust RfC. Match the POD’s route to the guidance value you intend to derive, or plan a dosimetry conversion explicitly.

Example: a 90-day rat study shows male-only hyaline-droplet nephropathy at all doses, female-rat hepatocellular necrosis at 50 mg/kg/day, and adaptive thyroid hypertrophy without TSH change at 150 mg/kg/day. The α2u kidney finding is a weak human RfD driver. The female liver necrosis is the better critical-effect candidate. The thyroid change needs the section 10.1 TSH-loop test before it competes.

Sensitization, phototoxicity, and immunotoxicity as non-cancer hazards

These endpoints were built as study packages in chapter 6 (LLNA/OECD 497, OECD 432/ICH S10, ICH S8 TDAR). In III.A they reappear as hazards that can outrank a quiet 90-day liver.

Skin (and respiratory) sensitization is a non-cancer hazard: once primed, low re-exposure can elicit dermatitis or asthma. There may be no useful “NOAEL” in the RfD sense for an already-sensitized person; classification and exposure control often matter more than a milligram-per-kilogram chronic value. Still, a sensitizer impurity can be the decision-driving hazard in a dossier whose systemic organs look clean.

Phototoxicity is a light-activated local (sometimes systemic) hazard at skin and eyes. A colorless, non-absorbing molecule is not a phototoxin. A 340 nm absorber destined for sun-exposed skin can have phototoxicity as the critical clinical hazard even if oral rat liver is unremarkable.

Immunotoxicity (unintended suppression or stimulation) can be the critical effect when TDAR or host resistance fails at doses below those that move ALT. Stress involution of the thymus is not automatically that hazard (chapter 5–6 interpretation). Intended immunomodulators are on-target immune hazards; they still need to be named in hazard identification.

Do not re-run LLNA stimulation-index arithmetic here. Do name these as non-cancer apical concerns when a stem tries to make “no liver necrosis” equal “no hazard.”

Scenario

Workers inhale a poorly soluble irritant aerosol. The sponsor submits a gavage 90-day with forestomach hyperplasia as the only finding and proposes that hyperplasia as the RfD critical effect for workplace air. Two failures: the surrogate route missed portal-of-entry lung, and forestomach anatomy is a species-local compartment. The critical effect for an RfC, once inhalation data exist, is more likely sensory irritation or pulmonary inflammation.

A sublingual cardiovascular drug produces parent-driven tachycardia in a buccal dog study at low milligrams; oral gavage rats show only first-pass-driven hepatocellular enzyme induction at much higher milligrams. Hazard ID for the clinical route should not hide behind the rat liver story. The anticipated on-target CV effect is still a hazard.

An OP insecticide’s AChE inhibition is anticipated on-target in insects. A mesocosm shows cladoceran collapse and an algal bloom. That is an ecosystem off-target (relative to the pest) hazard that mammalian oral cholinesterase inhibition does not replace.

Traps

• Treating gavage, diet, and drinking water as identical oral hazards at the same nominal mg/kg/day.
• Calling instillation “inhalation” for RfC selection.
• Using forestomach or α2u kidney as the default human critical effect.
• Equating “no systemic liver finding” with absence of local, sensitization, photo, or immune hazard.
• Treating on-target pharmacology as non-adverse, or treating an ecosystem kill as solved by a rodent NOAEL.