Exposomics

Established The exposome is the proposal that environmental exposure be measured across a lifetime with the seriousness the genome received, and after twenty years it is a concept with a measurement problem rather than a measurement programme with a concept problem. Christopher Wild named it in 2005 explicitly as a complement to the genome: every exposure from conception onward, chemical, physical, biological, occupational and social, together with the internal responses those exposures provoke. Established The definition has been stable ever since. What has not stabilised is any answer to the question a genome answers trivially — how much of it have you actually measured, and how would you know.

Established The structural asymmetry with genomics is a property of the objects, not of effort or money. A genome is finite, fixed, written in a four-letter alphabet and amplifiable — the polymerase chain reaction turns one molecule into a billion, which is why a cheek swab suffices. Established An exposome is unbounded, time-varying, measured in parts per trillion, and has no amplification step: the analyte in a blood sample is all there will ever be, and a missed exposure window is gone. Frontier Everything difficult here follows from those two sentences, including the claim this brief is most confident about — that environmental attribution lags genomic attribution by decades for reasons no funding increase alone repairs.

Established The field works where the exposure is single, cheaply measurable and legislable, and stalls where it is cumulative. Blood lead in the United States fell from a population geometric mean near 12.8 µg/dL in the late 1970s to under 1 µg/dL in the 2010s, tracked the whole way by one biomonitoring survey, and the exposure was removed by statute. Established No comparable chain runs from a mixture. Frontier Untargeted high-resolution mass spectrometry, the instrument the field’s coverage claims rest on, routinely reports tens of thousands of molecular features per sample and confidently identifies a low single-digit percentage of them, against an industrial chemical inventory of roughly 350,000 registered substances and mixtures worldwide.

Established The honest summary is that the exposome is a research agenda with excellent instruments, several national programmes, and very few replicated exposure–disease effects that classical epidemiology did not already have. Frontier That is roughly the position genomics occupied in 1998, before the significance threshold, the reference sequence and the consortium-replication habit arrived. Speculative Which of those three the exposome acquires first is the question worth watching.

Established A note on sourcing. This brief was commissioned in September 2026 from the Institute’s research base. Reading-list entries without links are cited from the bibliographic record rather than re-fetched, and claims are dated no later than early 2026 unless carried by a linked source.

Established Start with what the instrument can and cannot do, because every coverage claim in the field is downstream of it. A single untargeted liquid-chromatography high-resolution mass spectrometry run on human plasma typically yields between ten thousand and a hundred thousand molecular features, and a feature is a mass-to-charge ratio at a retention time, not a compound. Established Naming one requires an authentic reference standard or a library spectrum, graded on the five-level confidence scheme Schymanski and colleagues published in 2014, where level 1 means confirmed against a standard run in the same laboratory. Frontier Published annotation rates in human samples are commonly under ten per cent of features at any level and a low single-digit percentage at level 1. The bottleneck is identification, not detection.

Frontier The one formal test of how well laboratories identify unknowns was run by the US Environmental Protection Agency, and it produced the most useful discouraging number in the subject. Its non-targeted analysis collaborative trial circulated blinded synthetic mixtures built from a library of over a thousand substances to a few dozen laboratories, which correctly identified a minority of the spiked compounds — per-laboratory true positive rates ranging from a few per cent to roughly forty — while also reporting substances that were not there. Established This brief cites the trial by name from the bibliographic record and prints the shape of the result rather than one headline rate, because scoring depends heavily on mixture composition and on how a near-miss isomer is counted.

Established Against instrument coverage, put the size of the target. The CAS registry passed 200 million disclosed substances in the mid-2020s, which is the wrong denominator; the right one is the roughly 350,000 chemicals and mixtures registered for production and use across national inventories, a count that took a dedicated study to assemble because no single register exists. Established A substantial share of those entries are confidential or ambiguously identified, so the field cannot build the reference library it needs without a disclosure change — which is why chemical identity disclosure appears here as an institutional dependency rather than a scientific one.

Established Biomonitoring — the targeted, known-analyte half of the field — is where the measured record is genuinely strong. The US National Health and Nutrition Examination Survey has measured chemicals in a probability sample of the population in two-year cycles since 1999, and its national reports are why population exposure trends are known at all. Established Per- and polyfluoroalkyl substances are detected in the serum of about 97 to 98 per cent of sampled Americans, and serum perfluorooctane sulfonate fell roughly sevenfold between the 1999–2000 cycle and the late 2010s, from a geometric mean near 30 ng/mL to the low single digits, following the manufacturing phase-out rather than any clinical intervention. Frontier That is the field’s best causal demonstration and it is an ecological one: a production decision moved a population body burden, measured.

Frontier What that burden means clinically is a separate and much weaker claim, and the most careful authority says so in its own thresholds. The US National Academies’ 2022 guidance sets action tiers on the summed serum concentration of seven PFAS: below 2 ng/mL no adverse effects are expected, 2 to 20 ng/mL triggers limited screening, above 20 ng/mL a wider screen. Established Those are consensus screening triggers, not dose–response curves. Frontier The strongest human evidence behind them remains the C8 Science Panel, which studied roughly 69,000 people in a contaminated Ohio River valley water district and found a probable link between perfluorooctanoic acid and six conditions — exposures far above background, generalised downward to everyone else by assumption.

Frontier Microplastics are where body-burden numbers and their disputes are loudest, and where the measurement problem is most acute. The most cited clinical result is a 2024 New England Journal of Medicine cohort of 257 carotid endarterectomy patients, in which polyethylene was detected in about 58 per cent of excised plaques and patients with detectable particles carried a markedly elevated hazard of a composite cardiovascular endpoint over roughly thirty-four months. Frontier It is observational, single-region and non-randomised; the detection method infers polymer mass from thermal decomposition products that overlap with endogenous lipid fragments; and there are no certified reference materials for micro- and nanoplastics in tissue against which any laboratory can be calibrated. Speculative A 2025 report of very high polymer concentrations in decedent brain tissue, rendered in the press as a spoon’s worth of plastic per brain, drew immediate methodological objection on exactly those grounds, and this brief does not treat it as a measured body burden.

Established The upstream tonnage is not in dispute; the step from a tonnage to a dose is. The United Nations Environment Programme puts mechanical recycling at about nine per cent and projects plastic waste generation nearly tripling by 2060. Handwave Neither number tells anyone what is inside a person, and section 14 takes the specific body-burden claims apart.

Frontier Wearables and personal sensors are the field’s most oversold asset and its most underused one. Low-cost optical particle sensors now number in the hundreds of thousands worldwide and, once corrected against reference monitors, resolve neighbourhood-scale differences regulatory networks miss; uncorrected, they systematically over-read in smoke. Frontier Silicone wristband passive samplers accumulate hundreds of semi-volatile compounds over a week of ordinary wear and have been screened against libraries of over a thousand chemicals, and the most-cited personal-exposome demonstration followed a handful of participants with body-worn samplers and reported thousands of chemical and biological species around them. Established What none of these does is what consumer wearables are assumed to do: a smartwatch measures physiology and position, and infers exposure from the second through a model.

Frontier The exposure-wide association study is the field’s methodological signature, and its replication record is why this brief is cautious. The original 2010 environment-wide association study screened 266 environmental factors against type 2 diabetes in national survey data and surfaced a handful of hits, chiefly persistent organic pollutants. Frontier Later work established the two facts that govern the design: environmental variables are far more correlated with each other than genetic variants are, so the effective number of independent tests is unknown; and an exposure’s estimated effect varies enormously with the choice of covariates, a phenomenon named the vibration of effects. Established Genomics solved the analogous problem socially, by agreeing a genome-wide significance threshold near 5 × 10^-8 and refusing to publish without independent replication. Established There is no agreed exposome-wide threshold, and that absence obstructs more than any instrument limitation.

Established The programme state is real but small against the comparator. The US National Institutes of Health built a shared analytical service — a children’s exposure-analysis resource established in the mid-2010s and its successor covering all ages — that performs targeted and untargeted analysis for funded studies, which is an infrastructure model rather than a cohort. Frontier The environmental health institute has since made exposomics a strategic priority and moved towards a national coordinating network, and Europe funded a cluster of nine exposome projects under Horizon 2020 at a total in the region of €106 million, with a dedicated research infrastructure placed on the European roadmap. Established Set that against genomics: one reference genome cost on the order of $2.7 billion to 2003, and the successor population programme has assembled 245,388 clinical-grade genomes and returned more than 700,000 personalised results to over 277,000 participants. Frontier No exposomics programme anywhere is within an order of magnitude, and the gap is in cohort scale and specimen archives rather than in instruments.

Frontier Where exposure science does carry an attribution, it does so through modelled air pollution rather than through anything the exposome added. Global burden estimates attribute on the order of eight million deaths a year to air pollution, and the World Health Organization’s 2021 guideline put the annual fine-particulate target at 5 µg/m³, a level almost the entire world population exceeds. Frontier These are model outputs built on concentration–response functions from a few dozen cohorts, which is why they are flagged frontier here rather than because the direction is doubted. Frontier The most legible recent natural experiment is a congestion charge: New York’s transport authority reported fine particulates in the charged zone down 22 per cent in the first year alongside 27 million fewer vehicle entries — self-reported by the policy’s author and the recipient of its revenue, which is exactly why an independent exposure measurement is what would turn it into evidence.

Frontier What fraction of the blood exposome is identifiable even in principle? Estimates of the number of distinct small molecules circulating in human blood at detectable concentrations run from the low thousands to the tens of thousands, and the fraction of those that are exogenous rather than endogenous or microbial is not known to better than a factor of several. Frontier Until that denominator is agreed, coverage claims are unfalsifiable.

Frontier Does internal dose beat external proxy for predicting disease? The exposome’s founding argument is that measuring what is in the person beats modelling what was around them. Frontier For persistent compounds with long half-lives the case is strong; for short-lived compounds a single spot sample may carry less information than a residential postcode, because within-person variability swamps between-person variability. Speculative No study has yet run the head-to-head comparison across a panel of exposures in the same cohort with a common outcome.

Speculative Is the social exposome a dose or a metaphor? The definition includes psychosocial and socioeconomic exposure, and the biological embedding of chronic stress is a real literature. Handwave The step where a neighbourhood deprivation index becomes a dose with units, additive alongside a serum concentration, is where the framework works by assertion.

Established Annotation, not detection, is the binding technical constraint. Spectral libraries cover tens of thousands of compounds; the commercial chemical space runs to hundreds of thousands, with transformation products and metabolites multiplying that again. Established Every unmatched feature is a measurement the field paid for and cannot use.

Established There is no amplification step, and this is irreducible. A genomic assay can be repeated from vanishing material because the template is copied; an exposure assay consumes its analyte. Established That sets a hard floor on sample volume, pushes the cost per untargeted panel into the hundreds of dollars, and means a cohort that banked 200 µL aliquots in 1995 has already decided which questions can be asked of it.

Frontier The sample containers are made of the thing being measured. Blood collection tubes, pipette tips, storage vials and cap liners are polymers with plasticiser additives, which makes blank contamination the central methodological threat in both phthalate and microplastic work. Established Every credible microplastics laboratory now reports procedural blanks; not every published result did.

Established The field depends first on pre-diagnostic specimen archives it does not control. Any prospective exposure–disease claim requires a sample taken before the disease existed, which means the scientific value of an exposomics programme in 2035 was determined by freezer decisions taken in 1995. Established The large population biobanks are the asset, and their sampling protocols were designed for genetics, not for trace organics.

Frontier It depends on causal-inference machinery borrowed from genetic epidemiology. Mendelian randomisation, negative-control outcomes and sibling designs are what let an observational exposure association survive confounding, and each needs a genetic instrument or a family structure the exposure cohort may lack. Established The Precision Medicine brief records what within-family correction did to the polygenic literature: effects shrank by 47 per cent for education and 22 per cent for cognitive ability. Speculative The exposome has not yet had that audit.

Established The decisive test is a blinded, pre-registered, multi-cohort replication round for untargeted exposure signals in pre-diagnostic archived specimens, scored against a significance threshold agreed before anyone looks at the data. It is decisive because it tests the only claim that would make exposomics a distinct science rather than a collection of instruments: that untargeted measurement of stored specimens discovers exposure–disease relationships targeted epidemiology did not already know. Frontier The design is unexciting and available today — one discovery cohort, three independent replication cohorts, one blinded laboratory, a pre-specified multiplicity correction, and publication of the null. Established Nobody has funded it. The specimens exist, the instruments exist, and the missing item is an agreement between cohorts to be scored against each other.

Frontier Second: a blinded inter-laboratory trial on micro- and nanoplastics in spiked human tissue, with certified reference materials. Until laboratories can recover a known quantity of polymer from a known matrix within a stated uncertainty, the body-burden literature cannot distinguish a finding from a blank. Established The precedent exists: the non-targeted analysis collaborative trial did precisely this for chemical identification and produced a sobering, useful number.

Frontier Third: randomised exposure reduction with a health endpoint. Household air filtration trials are the model — randomise high-efficiency filtration into homes, measure both the achieved concentration reduction and a clinical outcome. Frontier The largest such trial in a heavily polluted city randomised filtration during pregnancy, missed its birth-weight endpoint and reported a modest difference in child cognitive scores at follow-up; both results are cited from the bibliographic record and both show the same thing, that achieved exposure contrast in these trials is small against ambient variation. Established A trial achieving a large, verified personal-exposure contrast would be worth more than ten observational cohorts.

Frontier Fourth: the natural experiments already running, measured properly. Manufacturing phase-outs of individual PFAS, national lead eliminations, and urban road-pricing schemes each move a population exposure on a known date. Frontier New York’s congestion charge is the cleanest current case and the least well measured — the 22 per cent fine-particulate reduction is reported by the scheme’s operator, with no independent personal-exposure panel running inside and outside the zone. Established Instrumenting a natural experiment costs a fraction of running a trial and the opportunity closes on a schedule.

Established The instruments are not the problem, which is unusual for a frontier subject. Orbitrap and time-of-flight analysers deliver the resolving power and mass accuracy needed to separate isobaric candidates, and ion-mobility separation adds an orthogonal dimension — collision cross-section — that narrows identification without a new standard. Frontier What is missing is throughput discipline: a cohort of 100,000 samples at twenty minutes of chromatography each is four years of continuous instrument time before any repeats.

Frontier Passive personal samplers are the most cost-effective unbuilt product in the field. A silicone band costs a few dollars, requires no power, integrates a week of semi-volatile exposure and ships in an envelope. Frontier Its weakness is quantification: uptake rates are compound-specific and temperature-dependent, so the band reports presence and relative abundance far better than it reports dose.

Speculative Automated annotation is the plausible force multiplier. Machine-learned spectral prediction from structure, and structure prediction from spectra, would let a library be generated computationally rather than synthesised compound by compound. Frontier Current systems propose candidates with useful recall and poor precision, which here means hypotheses a chemist must still confirm against an authentic standard.

Established This subject sits between two briefs that each own half of it and neither of which owns the join. Precision Medicine documents an internal-omics enterprise whose measured record splits cleanly: it predicts harm from a drug well and benefit poorly, ending at roughly 1.5 per cent end-to-end benefit in matched oncology. Frontier Its structural gap is that almost none of its predictors carry an environmental term, so a patient’s occupational history, water supply and air quality enter the model as noise. Speculative The exposome is the proposed repair; it has not yet demonstrated that adding an exposure term improves any clinical prediction.

Established At the other end, Planetary Stewardship owns the aggregate accounting and shows what happens to a quantified-limit framework whose control variables are not stable. Its novel-entities and chemical-pollution boundaries are the planetary-scale statement of the problem this brief meets in a blood sample. Established Its nitrogen case — a published boundary of 61 (35–84) Tg yr^-1, a factor of 2.4 end to end — is the clearest illustration of what a number inherits when it is downscaled, and a per-person chemical budget would inherit worse.

Frontier Analytical chemistry, environmental epidemiology and occupational hygiene are the parent disciplines, and the last of these is routinely forgotten. Industrial hygiene has a century of practice in personal dosimetry, with sampling trains, exposure limits and a legal duty attached, and it solved in one workplace the problem exposomics states in general. Established Occupational cohorts also carry the largest exposure contrasts available to human science, which is why most established chemical carcinogens were identified at work rather than in the general population.

Established The funding asymmetry is the fact to carry. Genomics received a flagship international project with a single deliverable, a completion date and roughly $2.7 billion; exposomics has received analytical service contracts, a European project cluster in the low hundreds of millions of euros, and strategic-priority language. Frontier That is not only a quantity difference: a project with a deliverable forces standardisation, and a service model does not.

Established The regulatory statutes were written substance by substance and cannot ingest a cumulative measurement. Chemical control in the United States and the European Union proceeds through individual substance dossiers, and although both regimes now contain mixture and grouping provisions, the operative unit of decision is still one substance. Frontier An exposome result — this population carries an elevated joint burden of forty compounds, none of which exceeds its own limit — has no receiving slot in either statute. Established Regrettable substitution follows directly, which is the documented history of both PFAS and phthalate regulation.

Frontier There is no institutional home for a reference exposome. Genomics has reference sequences maintained by named consortia; exposure science has no body chartered to maintain the equivalent — an open, versioned library of chemical identities with spectra, updated as the inventory changes. Speculative Whichever institution takes that on acquires the field’s central chokepoint, and none has volunteered.

Established The exposome makes inequality legible, which is its strongest ethical argument and its sharpest hazard. Exposure to air pollution, lead, noise, heat and occupational chemicals is patterned by income, race and housing, and measuring the gradient hands advocacy a number it did not have. Frontier The same number attached to an individual is a liability in insurance underwriting and employment screening, and no jurisdiction has an exposure-data analogue to genetic non-discrimination law.

Frontier Return of individual results is an unsolved problem with a live example. Participants with a measured serum PFAS concentration want to know what it means, and the National Academies’ guidance answers with screening tiers rather than with a risk estimate. Established Telling a person they carry 12 ng/mL of a compound with a multi-year half-life and no clearance therapy is an intervention with psychological effects and no clinical one.

Speculative If cumulative attribution worked, chemical regulation would change unit. Instead of asking whether a substance is safe, a regulator could ask whether a population’s total burden is within a budget, and allocate that budget across uses — which is the structure climate policy adopted when it moved from source permits to carbon budgets. Handwave Every element of that analogy is currently missing: there is no agreed total, no conversion between substances, and no allocation rule.

Frontier And there is a durable civilisational asset even if attribution fails: the archive. Banked specimens, measured and re-measurable, are a record of the chemical composition of human beings during the century in which synthetic chemistry saturated the environment. Speculative Its value to a future investigator exceeds its value to the cohort that collected it, which is exactly the kind of asset funding cycles handle worst.

These horizons track the measurement and attribution capability this brief assesses, not the popularity of the word.

• 10 yr: Frontier Open spectral libraries and computational annotation raise confident identification in human specimens from a low single-digit percentage of features into the low tens; certified reference materials for micro- and nanoplastics in tissue exist, and a share of the current body-burden literature does not survive them; one or two national exposomics networks reach cohort scales of a hundred thousand banked specimens with untargeted analysis.
• 25 yr: Speculative An exposome-wide significance threshold is agreed the way the genome-wide one was, and a small number of exposure–disease associations replicate across independent cohorts and enter clinical guidance; personal passive sampling is routine in cohort studies; cumulative-risk provisions become operative in at least one major chemical statute.
• 50 yr: Speculative Lifetime exposure histories, assembled from archived specimens plus continuous sensing, are a routine clinical input alongside family history, and prediction models carry an environmental term that measurably improves them.
• 100 / 250+ yr: Handwave A per-person or per-population chemical budget, allocated and enforced the way an emissions budget is, with substances traded against one another by a common risk unit. Every component of this — the total, the exchange rate, the enforcement body — is currently absent, and the framework is offered here as a destination rather than a forecast.

• Depends on Results neighbouring subjects produce for their own reasons. Precision Medicine supplies the linkage architecture, the companion-diagnostic regulatory form and the cautionary arithmetic — a funnel ending near 1.5 per cent end-to-end benefit — against which any exposure-stratified intervention will be scored, and its within-family shrinkage audit is the methodological debt this subject has not paid. Planetary Stewardship supplies the aggregate accounting frame and the worked demonstration of a quantified-limit framework carrying unstable control variables. What this subject is actually blocked on appears on no technology map: a library of chemical identities and spectra covering substances in commerce, which is a metrology and disclosure product rather than a scientific result.
• Requires (not on this map) Six things that are not experiments. An annotated spectral library covering substances actually in commerce, without which untargeted analysis keeps producing features nobody can name. An agreed exposome-wide significance threshold, the social object that made genome-wide association studies replicable and whose absence is why exposure-wide studies are not. Certified reference materials for micro- and nanoplastics in tissue, because a body-burden literature with no recovery standard cannot separate a finding from a blank. Mandatory disclosure of chemical identities and production volumes, since substances held as confidential business information cannot enter any library and are invisible to a survey claiming to be untargeted. Collection and storage consumables not themselves made of the analytes, which is a procurement rather than a discovery. And funding that keeps pre-diagnostic specimen archives frozen and catalogued across decades — the one input whose value was set a generation earlier.
• Enables Cumulative-risk regulation with a measured basis rather than a modelled one; environmental-justice claims with individual-level dosimetry behind them; exposure terms in clinical risk prediction; and retrospective reanalysis of any archived cohort each time the library improves, which is the only mechanism in this field that turns old money into new results. It also supplies a benchmark other fields can be held to: a proposal to explain disease by environment now has to state its identification confidence level, its multiplicity correction and its replication cohort.
• Adjacent Sits alongside Precision Medicine, which owns the internal-omics half and the reimbursement machinery, and Planetary Stewardship, which owns the planetary-boundary accounting into which any chemical-burden budget would have to fit. Occupational hygiene, which this map does not carry as a subject, is the closest working analogue at the scale of one workplace.

Handwave “The exposome has been mapped.” It has not, and the word mapped has no defined meaning here. Established A genome is complete when every base is called; an exposome has no denominator, so completeness is undefined, and the operative number — the share of detected features that can be named — remains in the single digits at the highest confidence level.

Handwave “You eat a credit card of plastic every week.” The figure of about 5 grams per week traces to a 2019 commissioned report and an underlying analysis whose own range spans roughly two orders of magnitude, aggregated from studies using incompatible size cut-offs and detection methods. Established It is a synthesis of ingestion estimates, not a measurement of anyone, and the authors of the critiques include researchers who publish microplastics findings themselves.

Speculative “There is a spoonful of plastic in your brain.” This derives from a 2025 report of high polymer concentrations in decedent brain tissue by thermal decomposition analysis. Frontier The objection raised immediately, and not yet resolved, is that the decomposition products used to quantify polyethylene overlap with those of endogenous lipids, which are abundant in brain. Established A blinded spiked-tissue trial would settle it; none has been published.

Handwave “Plastic pollution has pushed Earth past all nine planetary boundaries.” This circulates from a news item rather than a study, and it is included here so readers recognise it rather than treat it as evidence. Established The Planetary Stewardship brief shows why any boundary count needs its framework attached before it means anything: the same named boundary carries different control variables in different papers by overlapping author teams.

Frontier “Seventy to ninety per cent of disease risk is environmental.” The figure is the complement of twin-study heritability estimates, restated by the exposome literature’s founding papers, and it is not a measurement of environmental causes. Established It is a residual: everything not attributed to additive genetic variance, including measurement error, stochastic developmental noise and gene–environment interaction. Frontier Quoting it as though exposures had been measured to that share inverts what the number is.

Frontier “PFAS are in 98 per cent of people, so PFAS are harming 98 per cent of people.” Detection frequency is a statement about analytical sensitivity as much as about biology, and sensitivity has improved by orders of magnitude since the 1990s. Established The National Academies’ own tiering puts summed serum concentrations below 2 ng/mL in a no-expected-effects category, which is an explicit statement that detection is not harm.

Frontier “My wearable measures my exposome.” Consumer devices measure heart rate, movement, position and occasionally ambient temperature or ultraviolet index. Established Everything exposure-like they report is inferred by joining position to a modelled pollution surface, which is the external-proxy method the exposome concept was proposed to replace.

Frontier “Untargeted analysis finds any chemical present.” It finds ionisable compounds within the chosen chromatography and ionisation mode, in the concentration range the instrument reaches. Established Metals, most gases, many highly polar compounds and anything below the detection limit are simply absent from the result, and no single method covers the chemical space a claim of totality implies.