16 — No data

A threshold for compounds no one has tested.

+ The questionWhat is done for a compound nobody has studied?

A compound without a study has no tolerable intake

A laboratory can measure how much of a compound left a part without knowing what the compound is. Chapter 13 ends with two such results from the extracts of the running device's silicone cushion: a peak in the isopropanol extract that matched no spectrum in the library, and a compound in the hexane extract with only a tentative name. A tolerable intake, as Chapter 14 builds it, starts from a study of the compound, and the unknown peak has no study to start from. The tentative name need not supply one either. In FDA's 2024 draft guidance on chemical analysis, a tentative name means only that "one or more candidate molecular structure can be assigned", and each candidate may be a compound that no one has ever dosed.

What the evaluation can do next depends on how much it knows of the structure. FDA's draft asks for one of the two highest identification levels of Chapter 13 before a toxicological risk assessment, and its flowchart sends an unknown peak back to the analysis, to revise the method or add another. Nelson Labs, a testing laboratory, set out its own practice in workshop slides in 2023: an unidentified compound is "considered carcinogenic/mutagenic" until a structure lets a toxicologist clear it. Until a study exists, the evaluation needs a number that holds for a compound no one has yet studied.

Safe doses of many chemicals set the threshold

The threshold of toxicological concern, or TTC, is that number: an intake low enough that harm is very unlikely even for a compound no one has tested. ISO/TS 21726, the 2019 technical specification that applies it to devices, defines it as a "level of exposure for constituents, below which there would be no appreciable risk to human health". The threshold is read from chemicals that have been tested. Their no-adverse-effect doses, the NOAELs of Chapter 14, are ranked from the most potent chemical to the least. The dose below which only one chemical in 20 falls is the fifth percentile, and that dose divided by 100 is the threshold. In a 2025 study of device constituents, Builee and colleagues read the 100 as two tenfold factors, one for animal to human and one for person to person.

On the same factor of 100, only a compound more potent than 19 tested chemicals in 20 would have a tolerable intake below the threshold. The statement holds only if the untested compound resembles the chemicals the percentile came from, and that condition is why far more potent groups are excluded by name (last section). The condition is this book's reading rather than a stated rule.

Munro and colleagues derived such thresholds in 1996 for three classes of chemical structure named after Cramer, as Builee and colleagues quote their work. A Cramer class ranks a structure by the concern it raises before any testing, and each class carries its threshold. Class III holds the structures whose tested members had the lowest safe doses. Divided by 100, the fifth percentile of each class gave 30 µg per kilogram of body weight per day for class I, 9 for class II and 1.5 for class III, each for exposure over a lifetime.

ISO/TS 21726 takes its non-cancer values from Munro, and they are lifetime values: its first edition lists none for shorter contact, Builee and colleagues note. They took 1,252 NOAELs for 790 device constituents from oral studies registered with the European Chemicals Agency, set genotoxic compounds aside, and grouped what remained by the length of the study. In 2025 they proposed 112 µg per kilogram per day for contact up to 30 days, 111 for 31 to 365 days and 41 beyond a year. Most of the authors worked for device makers and testing laboratories, and six companies in the field paid the publication fees. No standard has adopted the values.

39 — Many tested chemicals, ranked by their safe dose
0.001 0.01 0.1 1 10 100 1,000 10,000 0 5 % 50 % 100 % SHARE OF TESTED CHEMICALS WITH A LOWER SAFE DOSE TESTED CHEMICALS, EACH AT ITS LOWEST NO-ADVERSE-EFFECT DOSE fifth percentile: 4.1 mg/kg/d — 1 chemical in 20 lies to its left ÷ 100: 10 animal to human × 10 person to person threshold: 41 µg/kg/d (0.041 mg/kg/d) III 1.5 II 9 I 30 Cramer classes, lifetime, µg/kg/d (Munro, 1996) more potent NO-ADVERSE-EFFECT DOSE, NOAEL (mg PER kg OF BODY WEIGHT PER DAY) less potent only these, 1 in 20, would have a tolerable intake below the threshold Spread schematic; dot positions illustrative. Fifth percentile and threshold: the group of studies longer than a year, within 1,252 NOAELs for 790 device constituents in oral studies registered with ECHA (Builee and colleagues, 2025; a proposal, not adopted in a standard; industry authors). Cramer values: Munro 1996, as Builee 2025 quote them.
Released substanceLimit or thresholdFocal detail
Plate 39 — The threshold sits two tenfold steps below the safe dose that only one tested chemical in 20 undercuts. On a 2025 proposal built from studies longer than a year it is 41 µg per kilogram per day; the lifetime values Munro set for the three Cramer classes in 1996 sit lower on the same scale.

Damage to DNA sets the lowest value, and shorter contact a higher one

No dose of a compound that may damage DNA is assumed safe, for the reasons Chapter 11 gave, so its threshold is set at an accepted risk instead. ICH M7, the guideline on DNA-reactive impurities in medicines, sets 1.5 µg a day, an intake that corresponds in theory to an excess lifetime risk of cancer of 1 in 100,000. FDA issued the guideline's current revision as guidance in July 2023.

ICH M7 derives that value from the TD50, the dose that gives a tumor incidence of 50 % in the most sensitive species at the most sensitive site. From that point it draws a straight line to zero. A line from zero that passes through a risk of 1 in 2 at the TD50 reaches 1 in 100,000 at a dose 50,000 times smaller, so for one compound the method comes down to "simply dividing the TD50 by 50,000". The same line leaves no dose at which the risk disappears.

A shorter exposure may take a higher daily intake, because ICH M7 takes the risk of cancer to rise with the dose added up over the exposure, "uniformly distributed over the total number of exposure days". The guideline names the principle Haber's rule, by which concentration multiplied by time is a constant. ICH M7's table of acceptable intakes for one impurity allows 120 µg a day for less than a month, 20 for one to 12 months, 10 for one to ten years and 1.5 beyond ten years. A contact of under a month may therefore take 80 times the lifetime value each day. The steps are fixed values, each at or below what the rule's arithmetic alone would allow.

The device route takes over both sets of values. ISO/TS 21726 draws its cancer thresholds from ICH M7's principles and its non-cancer thresholds from Munro's classes, as Efor, a consultancy, set out in 2025. A device constituent takes the four steps above for its contact, each "of any one constituent" on FDA's town-hall slides of 7 October 2026. For a compound that does not cause cancer, the most protective class value is 90 µg a day, in ISO's summary of the specification as a standards seller reproduces it. FDA recognized the specification in full in July 2019.

In ISO's record, as the seller shows it, the specification's status has been "to be revised" since 10 October 2025, so as of November 2026 these are the 2019 edition's values. FDA staff at the town hall gave the specification's table per kilogram instead: "two micrograms per kilogram per day" for 30 days or less. For the 60 kg adult default of Chapter 1, 2 µg per kilogram gives the same 120 µg a day. For contact of 30 days or more, FDA's 2024 draft holds that a threshold based on 1.5 µg a day "would be conservatively protective". The draft calls this a dose-based threshold, and Chapter 17 turns it into a concentration.

40 — The daily threshold for a DNA-reactive compound, by length of contact
1 10 100 1,000 DAILY INTAKE (µg A DAY, OF ANY ONE CONSTITUENT) under 1 month 1 to 12 months 1 to 10 years over 10 years to lifetime limited (<24 h) prolonged (24 h to 30 d) long-term (>30 d) contact categories as FDA's slides set them 120 20 10 1.5 FDA's 2024 draft: 1.5 µg a day ‘conservatively protective’ for 30 days or more risk taken to rise with the dose added up over the exposure: Haber's rule under a month: 80 times the lifetime value each day Values: ICH M7(R2) Table 2, one impurity (FDA guidance, July 2023); the same four on FDA's town hall slides, 7 Oct 2026. ISO/TS 21726:2019 lists 120, 20, 10 and 1.5 µg a day, in ISO's summary as a standards seller reproduces it. Its status has been ‘to be revised’ since 10 Oct 2025. FDA staff stated the value for 30 days or less as 2 µg per kg per day.
Limit or thresholdFocal detail
Plate 40 — ICH M7 lets a DNA-reactive impurity take more each day when the contact is shorter, because it counts the dose added up over the exposure. FDA showed the same four steps beside the contact categories in October 2026; for 30 days or more, its 2024 draft takes the lifetime value.

A second screen works on the total amount rather than the daily rate. The toxicological screening limit of ISO 10993-17's 2023 edition is a "cumulative exposure dose" of an identified constituent "over a specified time period". The limit puts the book's invariant into one number: what the contact adds up to is what the body meets. In Efor's 2023 account, the limit is a daily threshold multiplied by a default number of days. For contact of 30 days or less it is 120 µg a day over one day, which gives 120 µg, and for longer contact it is 20 µg a day over 30 days, which gives 600 µg.

FDA staff gave the same two figures at the town hall, and they called a quantity "at or below the 120 microgram TSL value" low. A constituent whose dose stays below the limit needs no further systemic risk assessment, in the reading of Gruber of the Johner Institute, a consultancy. Efor's 2024 article on the limit excludes devices "intended for premature babies and infants under 6 months of age".

Structure decides which threshold applies

Which value a compound takes depends on its structure, so identification matters even where no study exists. A structural alert is a part of a molecule that marks a possible reaction with DNA. EFSA, the European Food Safety Authority, adopted its guidance on the threshold in food safety on 24 April 2019. Where the weight of evidence says a substance may be a DNA-reactive mutagen or carcinogen, the guidance puts it in its lowest class: 0.0025 µg per kilogram of body weight per day, or 0.15 µg a day for a person of 60 kg. Efor's 2025 article lists the same 0.15 µg a day for compounds with genotoxic alerts, after Kroes and colleagues in 2004.

The device route lists nothing that low. The lowest value in ISO's summary of ISO/TS 21726 is ICH M7's 1.5 µg a day, ten times the food-safety value for an alert. This book therefore reads a device compound with an alert as taking ICH M7's value for its contact, and one without an alert as taking the value of its Cramer class, unless it belongs to a group the threshold excludes. That rule joins the two routes the specification draws on.

Solved example: what identifying one compound from the cushion is worth
  • The contact. The cushion is in long-term contact after about a month of nightly use (Chapter 2), and for 30 days or more FDA's draft takes 1.5 µg a day as conservatively protective.
  • The unknown peak. With no structure, nothing rules out an alert. Nelson Labs' practice treats such a peak as carcinogenic or mutagenic, and for this contact that holds it to the 1.5 µg a day above. FDA's draft sends the same peak back to the analysis.
  • The tentative name. Suppose every candidate structure is free of alerts and of the excluded groups and falls in Cramer class III, whose lifetime value is 1.5 µg per kilogram per day. For the same 60 kg that EFSA's conversion uses, 1.5 times 60 is 90 µg a day, ISO/TS 21726's most protective non-cancer value.
  • The result. 90 divided by 1.5 is 60. If a confident or confirmed identity, as FDA's draft asks, shows a structure free of alerts, the intake the evaluation can accept rises sixtyfold, with no toxicity study of the compound.
  • Assumptions. The compound is illustrative. ICH M7's 1.5 µg a day is set per person and Cramer class III's value per kilogram, so this example scales both alike. For the lightest user, 30 kg (Chapter 1), class III gives 45 µg a day and the alert value 0.75, and the ratio stays sixtyfold.

Where the threshold stops

Some structures cause cancer at doses so low that the threshold excludes them by name. ICH M7 calls them the cohort of concern: "aflatoxin-like-, N-nitroso-, and alkyl-azoxy compounds". EFSA's 2019 guidance lists the same kinds of structure as high-potency carcinogens, with the benzidines. ISO/TS 21726 does not apply to the cohort. FDA's draft asks that a member known or suspected to be present "should be investigated", even below the reporting threshold that Chapter 17 derives from the TTC.

Lists drawn from food safety exclude whole groups of substances, and EFSA's 2019 guidance gives two grounds. Some substances are not represented in its database or lie outside its domain: inorganic substances, proteins, nanomaterials, radioactive substances, organosilicon substances, and metals in elemental, ionic or organic form, except organic salts whose counter ion is an essential metal such as sodium. Others have special properties: the high-potency carcinogens above, steroids, and substances that may bioaccumulate, the polyhalogenated dibenzodioxins, dibenzofurans and biphenyls among them. Builee and colleagues, citing the same guidance, also set aside polymers above 1,000 g/mol and substances with "known or suspected pharmacological activity".

EFSA excludes organosilicon substances because the toxicity database of Munro and colleagues, from which the class values come, does not represent them. On EFSA's terms, a short silicone chain from the cushion (Chapter 3) is an organosilicon substance. The threshold therefore does not cover it, and it needs a tolerable intake of its own (Chapter 14).

The threshold is an intake into the whole body, so it covers effects that follow absorption and leaves out the contact site. ISO/TS 21726 sets no threshold for harm to cells, irritation, sensitization or effects on blood. Sensitization acts through the immune memory of earlier contacts (Chapter 11), so an intake below a systemic threshold says nothing about the skin under the cushion. Particles fall outside the threshold for another reason: a particle's effect depends on its size and where it lodges (Chapter 9), not on an intake of dissolved molecules.

The air path falls outside the specification altogether, since its values "do not apply to potential exposure via gas pathways of medical devices". For the running device, that leaves the skin contact of the cushion and the headgear under ISO/TS 21726. The parts of the air path, from the intake to the mask, take the thresholds for what they release into the breathed gas from ISO 18562, the breathing-gas standard (Chapter 19).

In practice

For each compound in an extract report, record its identification level, any structural alert, its Cramer class, and whether it belongs to the cohort of concern or an excluded group. Take the threshold for the contact counted in Chapter 2, and cite ISO/TS 21726 by its 2019 edition with the date its status was checked. Send a compound with an adequate study to a tolerable intake, because the threshold "is not applicable to constituents with adequate toxicity data". Send gas-pathway parts to ISO 18562 (Chapter 19), and local effects to the tests of Chapter 18.

+ What this chapter established
  • A compound without a study has no tolerable intake, so it is held to a threshold of toxicological concern: the fifth percentile of many tested chemicals' safe doses, divided by 100.
  • A compound that may damage DNA takes the lowest threshold on the device route, 1.5 µg a day for lifetime contact, and ICH M7's steps allow 80 times that each day for contact of under a month.
  • Structure decides the value, so a confident or confirmed identity that rules out an alert can raise the accepted intake sixtyfold, to Cramer class III's 90 µg a day.
  • The threshold leaves out the cohort of concern, excluded groups, compounds with adequate data, local effects, particles and the gas pathway.
Created with Claude (Anthropic).
Educational reference material, not engineering, regulatory or clinical advice. Product and company names referenced are the property of their respective owners.