Australian therapeutic sunscreens are multi-filter mixtures: Single-ingredient margin of safety assessment does not address cumulative exposure.

Joseph Mizikovsky: Writing – review & editing, Validation. Abhinandan Chowdhury: Writing – original draft, Validation, Project administration, Methodology, Investigation, Data curation, Conceptualization.

The article processing charge for open-access publication was paid by the Australian Sunscreen Council , a registered charity focused on balanced UV exposure and Australia’s peak industry body for the sunscreen and SPF cosmetics sector. VeganicSKN Australia, the corresponding author’s employer, is a member of the Australian Sunscreen Council. Neither the Australian Sunscreen Council nor VeganicSKN Australia had any role in study design, ARTG data extraction, analysis, interpretation, manuscript preparation, or the decision to submit. This research did not receive any other funding from public, commercial, or not-for-profit sectors.

Of the 924 ARTG-listed therapeutic sunscreens, 187 (20.2%) were mineral-only formulations containing zinc oxide and/or titanium dioxide as the sole declared UV filters, with no organic UV filter present. The remaining 737 products (79.8%) contained at least one organic UV filter. Within this organic-containing subset, no product was formulated with a single organic filter; all used combinations of two or more ( Fig. 1 ). Four-organic-filter products accounted for 45.7% of the organic-containing subset (337 products), and 85.2% contained four or more organic UV filters. A further 39.5% contained five or more. Two-filter and three-filter formulations together made up 14.8% of the organic-containing subset. In other words, the typical Australian therapeutic sunscreen that contains any organic filter is not a single-filter product; it is a four-to-six-filter mixture. This distribution holds across general adult sunscreens and children’s sunscreens alike. Of the 48 children’s sunscreens identified in the dataset, five declare six organic UV filters in a single formulation. One such product (ARTG 430536; sponsor: Vitality Brands Worldwide Pty Ltd) declares homosalate, octocrylene, octyl salicylate, avobenzone, ethylhexyl triazone and bemotrizinol at a combined declared concentration of 24.8% w/w. Paediatric exposure scenarios are therefore not exceptions to the dominant multi-filter pattern. Comprehensive multi-pathway risk assessment has identified infants as the most vulnerable population for organic UV filter exposure [63] . Their estimated daily intake is approximately two to three times higher than that of adults. The descriptive analysis presented here reports population-level counts and prevalence from a complete enumeration of ARTG-listed therapeutic sunscreens, rather than estimates from a sample. Formal statistical uncertainty estimation is therefore not applicable; the figures reported are the actual population values for products listed on the ARTG at the date of extraction. The dataset is provided in full as Supplementary Table S1 to permit independent recalculation. Within the organic-containing subset, individual organic filter concentrations ranged from approximately 3–10% w/w (median values per filter type). Combined organic UV filter loads in four- to six-filter formulations commonly reached 15–25% w/w of the total formulation. The TGA Australian Sunscreen Exposure Model (ASEM) includes a default adult whole-body exposure scenario assuming application of 140 mL of sunscreen per day, at maximal-use conditions [61] . At an assumed density of approximately 1 g/mL, a 15–25% w/w combined organic UV filter load corresponds to 21–35 g of chemical UV filter applied to skin surface per day, distributed across four to six different molecules. This represents the applied mass under maximal-use conditions; the fraction systemically absorbed is filter-specific and formulation-dependent, and is not estimated here. Twenty distinct UV filter active ingredients were identified across the full dataset ( Fig. 2 ). The market was dominated by a number of recurring actives. Butyl methoxydibenzoylmethane (avobenzone) appeared in 641 products (69.4%), octocrylene in 600 (64.9%), octyl salicylate in 411 (44.5%) and homosalate in 410 (44.4%). Ethylhexyl triazone appeared in 290 products (31.4%) and bemotrizinol in 286 (31.0%). Zinc oxide, the dominant inorganic filter, appeared in 243 products (26.3%). The same handful of chemicals therefore recurs, in the same combinations, across hundreds of products. The implications for population-level exposure are that an Australian consumer switching between brands is, in the majority of cases, switching between formulations built from the same restricted set of organic filters. Fig. 2 Prevalence of UV filter active ingredients across 924 therapeutic sunscreens with active AUST L designation listed on the Australian Register of Therapeutic Goods (January 2026). Prevalence is calculated as the number of products in which the ingredient appears at any declared concentration, divided by 924. Absolute product counts (n) are shown at the end of each bar. Filter names follow common usage where available (avobenzone for butyl methoxydibenzoylmethane, oxybenzone for benzophenone−3). Abbreviations: BMDM, butyl methoxydibenzoylmethane / avobenzone; OCR, octocrylene; OCS, octyl salicylate; HMS, homosalate; EHT, ethylhexyl triazone; BEMT, bemotrizinol; DHHB, diethylamino hydroxybenzoyl hexyl benzoate; 4-MBC, 4-methylbenzylidene camphor; PBSA, phenylbenzimidazole sulfonic acid; MBBT, methylene bis-benzotriazolyl tetramethylbutylphenol; OMC, octyl methoxycinnamate; BP−3, oxybenzone; DTS, drometrizole trisiloxane; ECM, ecamsule; TBT, tris-biphenyl triazine; DPDT, disodium phenyl dibenzimidazole tetrasulfonate; IAMC, isoamyl methoxycinnamate; PS−15, polysilicone−15. Prevalence of UV filter active ingredients across 924 therapeutic sunscreens with active AUST L designation listed on the Australian Register of Therapeutic Goods (January 2026). Prevalence is calculated as the number of products in which the ingredient appears at any declared concentration, divided by 924. Absolute product counts (n) are shown at the end of each bar. Filter names follow common usage where available (avobenzone for butyl methoxydibenzoylmethane, oxybenzone for benzophenone−3). Abbreviations: BMDM, butyl methoxydibenzoylmethane / avobenzone; OCR, octocrylene; OCS, octyl salicylate; HMS, homosalate; EHT, ethylhexyl triazone; BEMT, bemotrizinol; DHHB, diethylamino hydroxybenzoyl hexyl benzoate; 4-MBC, 4-methylbenzylidene camphor; PBSA, phenylbenzimidazole sulfonic acid; MBBT, methylene bis-benzotriazolyl tetramethylbutylphenol; OMC, octyl methoxycinnamate; BP−3, oxybenzone; DTS, drometrizole trisiloxane; ECM, ecamsule; TBT, tris-biphenyl triazine; DPDT, disodium phenyl dibenzimidazole tetrasulfonate; IAMC, isoamyl methoxycinnamate; PS−15, polysilicone−15. Under the MoS framework reconstructed in §2.5, each UV filter is evaluated as if it were the only active ingredient on the skin. The framework provides a coherent characterisation of single-ingredient hazard at a defined SED. It does not, however, address the conditions of use established in §3.1–3.2. Specifically, no step in the current assessment (i) sums the contributions of co-formulated filters acting on the same biological pathway; (ii) accounts for transformation products generated on the skin under UV irradiation, or in chlorinated swimming-pool water; (iii) adjusts dermal absorption inputs for other ingredients in the same product that are known to drive chemicals deeper into skin; or (iv) integrates exposure from non-sunscreen sources such as indoor dust, human milk, or co-use of multiple UV-filter-containing products on the same day. A children’s sunscreen containing five organic UV filters at a combined load of 25% w/w therefore delivers a multi-chemical dermal dose for which no cumulative safety assessment exists in any jurisdiction. Each filter passes its individual MoS; the formulation as a whole has not been assessed. The principle that formulated products can differ toxicologically from isolated active ingredients is well established across multiple regulatory domains. For glyphosate-based herbicide formulations, surfactants and adjuvants both contribute additional toxicity and facilitate transport of the active across biological membranes [39] . Ingredient-by-ingredient assessment of the active alone does not capture the toxicity of the finished product. The same principle is recognised for transdermal pharmaceutical patches, where vehicle and penetration enhancers determine absorbed dose. These examples differ from sunscreens in exposure context, but the difference cuts in only one direction. Glyphosate exposure is largely occupational and selective. Transdermal patches are prescribed medical interventions delivered to defined patient populations. Australian therapeutic sunscreens are recommended for daily whole-body application across an entire population in the world’s highest-UV environment. They are the formulated-product class with the broadest sustained dermal exposure of any chemical category in routine consumer use. Whether ingredient-by-ingredient assessment captures the combined biological effect of these formulated products remains an open question. No reviewed regulatory assessment has demonstrated that the single-ingredient framework is adequate under the conditions of use documented here. Five of the nine filters reviewed in §2.4 show estrogenic activity in standard MCF−7 proliferation assays or in vivo estrogen-responsive endpoints (including uterotrophic and mammary-gland assays). These are oxybenzone, 4-methylbenzylidene camphor, octyl methoxycinnamate, homosalate and octocrylene [23] , [28] , [35] , [42] , [53] . Several of the same filters also engage the androgen receptor [33] , [56] or thyroid axis [26] , [51] . This is a pattern across the actives in routine use rather than an outlier property of one ingredient. Additive estrogenic effects have been demonstrated experimentally for mixtures of commonly used UV filters at concentrations relevant to actual sunscreen use [20] . The individual potencies of these filters differ, and rigorous cumulative assessment would apply potency weighting (for example, relative potency factors referenced to a standard estrogen). The current Margin of Safety framework, however, applies no mixture model at all. Where five filters act on the same hormone receptor in the same direction, current safety assessment treats this as five separate small effects, none of which alone crosses a regulatory threshold. The skin sees one combined dose. Several filters produce reactive chemicals under sunlight, through different mechanisms. Avobenzone generates reactive oxygen species under UV irradiation. PBSA produces oxidative DNA damage in human skin cells through photosensitisation [3] . Octocrylene degrades over time inside the bottle, generating benzophenone, itself the subject of regulatory concern [11] . Octyl methoxycinnamate degrades rapidly under UV exposure, forming cyclodimer and aldehyde photoproducts with cellular toxicity comparable to the parent compound [58] . These effects accumulate at the level of the skin but are not summed in any safety dossier, because each is attributed to a different parent ingredient. Octyl salicylate, present in 411 of the 924 Australian sunscreens analysed (44%), is a recognised dermal penetration enhancer in the pharmaceutical sciences literature [18] , [27] , [49] . It is not the only filter in the Australian dataset with this property. In an in vitro hairless mouse skin model, Pont et al. [47] tested seven UV filters individually at standard sunscreen concentrations. Four significantly increased the dermal penetration of a co-applied compound: octyl salicylate, octyl methoxycinnamate, oxybenzone and homosalate. Octocrylene was the only filter tested that did not enhance penetration. The penetration enhancement effect was confirmed in human skin for octyl salicylate. Octyl salicylate and octyl methoxycinnamate also physically damaged the stratum corneum barrier in the same study [47] . Pont et al. [48] further showed that UV filter combinations at standard sunscreen concentrations produced penetration enhancement greater than the average of individual filters acting alone [48] . When octyl salicylate, or any other documented penetration enhancer, is co-formulated with other systemically absorbed UV filters, it may increase their dermal uptake under conditions not reflected in the isolated-ingredient absorption studies that are used as inputs to the SED calculation. The single-ingredient framework also creates classification asymmetries within the same chemical family. Butyloctyl salicylate and octyl salicylate (ethylhexyl salicylate) are both esters of salicylic acid and share the salicylate chromophore that produces UVB absorption at approximately 300 nm [8] . In the Australian Permissible Ingredients Determination, octyl salicylate is regulated as a UV-filter active with a maximum-use limit of 5%, while butyloctyl salicylate is permitted at up to 7% as a non-active excipient. A product can therefore lawfully contain both ingredients, delivering substantially more salicylate chromophore to the user’s skin than the UV-filter maximum-use limit for octyl salicylate alone would suggest. The structural inadequacy of single-ingredient assessment for cumulative exposure thus extends to chemically related ingredients separated only by their listing classification. The Matta et al. [36] , [37] randomised clinical trials measured plasma concentrations of multiple UV filters after maximal-use application of commercial sunscreen formulations. The 2020 trial used commercial formulations containing all four penetration-enhancing filters identified above (octyl salicylate, octyl methoxycinnamate, oxybenzone and homosalate) [36] . Plasma concentrations of six filters exceeded the FDA 0.5 ng/mL toxicological-concern threshold on day 1 of typical use. This is consistent with, but does not prove, the inference that the isolated-ingredient absorption studies underlying current MoS calculations may underestimate real-world systemic exposure for filters routinely co-formulated with documented enhancers. The structural gaps above presume that modern human pharmacokinetic data exist for every filter under assessment. They do not. Maximal-use clinical pharmacokinetic studies under FDA protocol have been conducted for six of the nine filters reviewed in §2.4: avobenzone, oxybenzone, octocrylene, homosalate, octyl salicylate and octyl methoxycinnamate [36] , [37] . All six exceeded the FDA 0.5 ng/mL toxicological-concern threshold on day 1 of typical use. Three filters in current Australian therapeutic use have no comparable data: phenylbenzimidazole sulfonic acid, ethylhexyl triazone and 4-methylbenzylidene camphor. The dermal absorption inputs to their MoS calculations are derived from older study designs that do not capture the high-volume, whole-body, repeated-application exposure pattern recommended by Australian sunscreen guidelines. A detailed summary of the toxicological evidence base for these nine filters, including endpoints of concern, evidence type, key citations, and strength ratings, is provided in Table 1 . Table 1 Toxicological evidence summary for the nine most prevalent organic UV filters in the Australian therapeutic sunscreen market. Entries marked ‘data-gap exhibit’ indicate citations included to document regulatory or scientific data gaps rather than positive toxicological findings. Filter Endpoint Study type Strength Citation Avobenzone Photoinstability, radical/photoproduct formation, phototoxicity In vitro photochemistry; in vitro 3T3 (formulation) Moderate [10] , [40] Hepatic metabolism, systemic persistence (low extraction, high protein binding) In vitro HLM + IVIVE Low-Moderate [4] Membrane interaction (mixture-dependent) In vitro biophysical (monolayer) Low [44] OMC Endocrine: estrogenic (uterotrophic, ER proliferation) In vivo rodent + in vitro receptor/cell Moderate-High [53] Endocrine: thyroid (reduced T4, deiodinase; receptor-binding MoA) In vivo rodent + in silico docking Moderate [26] , [32] Photodegradation, photoproduct cytotoxicity In vitro photochemistry Moderate [58] Octyl salicylate Dermal penetration enhancement (co-applied compounds); confirmed human systemic absorption In vitro animal skin + human clinical PK + human biomonitoring Moderate-High [6] , [36] , [47] Oxidative stress, DNA damage, mitochondrial impairment, developmental toxicity In vitro (fibroblast) + in vivo zebrafish Moderate [19] Documented regulatory data gap: ECHA required reproductive/developmental toxicity (PNDT, EOGRTS) and endocrine-related fish sexual development testing for 2-ethylhexyl salicylate under REACH. The requirement was upheld by the ECHA Board of Appeal in 2020 and by the EU General Court in November 2023, with a compliance deadline of 25 February 2024 Regulatory determination (ECHA) data-gap exhibit [14] Octocrylene Benzophenone formation on aging (mutagen/carcinogen precursor); high dermal absorption of benzophenone In vitro accelerated-aging chemistry Moderate (significance disputed) [11] Confirmed human systemic absorption Human clinical PK High [36] Human metabolism, biomonitoring, rising population exposure Human oral kinetics + biomonitoring Moderate-High [5] , [6] Hepatic metabolism, longest half-life, bioaccumulation potential In vitro HLM + IVIVE Low-Moderate [4] Homosalate Estrogenic and anti-androgenic activity In vitro (MCF−7, MDA-kb2) + in vivo rodent Moderate [33] , [53] Thyroid disruption (raised Tg/TPO mRNA; altered T3/T4, TSH, GH) In vitro human + rat thyroid cells + in vivo zebrafish Moderate [7] , [25] Cytotoxicity, oxidative stress, DNA damage, developmental toxicity In vitro (fibroblast) + in vivo zebrafish Moderate [19] Confirmed human absorption + deep isomer-resolved biomonitoring; regulator-ordered repeat-dose/reproductive testing unfulfilled; safe limit cut 10–0.5% single-product, 7.34% face-only, combined-use MoS below 100 Human clinical PK + human metabolism + regulatory determination Moderate-High / data-gap ( [15] ; [36] ; SCCS (Homosalate), 2021) Oxybenzone Estrogenic activity at human-relevant concentrations; anti-androgen receptor antagonism In vitro (MCF−7 pS2, reporter gene) High [20] , [53] Endocrine bioactivity within 10-fold of human plasma (only filter of the nine) In vitro HTS (ToxCast/Tox21) vs human PK High [45] Confirmed human systemic absorption far exceeding FDA threshold; ubiquitous biomonitoring Human clinical PK + HBM meta-analysis High [36] , [37] , [43] Female reproductive effects (rodent) overlapping human internal exposure; human associations (menstrual, fibroids, endometriosis) In vivo rodent + human epidemiology (integrative review) Moderate-High [43] EHT 1997 SCCNFP basis predates modern OECD Regulatory determination data-gap exhibit [50] Confirmed human systemic presence (breast milk, every sample); indoor dust/air aggregate intake (toddlers up to 839 ng/kg/day) Human biomonitoring Moderate [12] , [59] Chlorination DBPs (at least 12), several more toxic to aquatic organisms than parent Environmental chemistry + ecotox Low-Moderate [29] PBSA Photosensitised ROS + oxidised guanine (8-oxodG) lesions under UV-A/UV-B In vitro/in cellulo human keratinocytes Moderate [3] Reproductive endocrine disruption + transgenerational F1 effects (reduced egg production, altered sex steroids, HPG genes) In vivo zebrafish life-cycle Moderate [60] Systemic uptake (plasma/liver/kidney), CYP1A induction, oxidative-stress biomarkers in fish In vivo fish/bivalve Low-Moderate [16] , [17] 4-MBC Eestrogenic activity (uterotrophic, ER-β preferential ligand, MCF−7 pS2) In vitro + in vivo rat uterotrophic High [42] , [53] Developmental/reproductive toxicity: delayed male puberty, altered reproductive organ weights, estrogen-target gene expression in prostate/uterus In vivo rat multigenerational (pre/postnatal) High [13] , [21] , [34] Thyroid axis interference In vivo rat Moderate [54] , [55] EU outright prohibition (Annex VI to Annex II banned list); Australia permits 4%, no current TGA safety assessment Regulatory determination data-gap / regulatory exhibit [51] Note: Strength reflects the quantity, consistency and human-relevance of available toxicological data for each endpoint, not the magnitude of risk. The scale characterises the state of the evidence base for each endpoint. It is not a quantitative weight-of-evidence determination under a formal framework. Ratings were assigned on a five-tier ordinal scale anchored on study type and replication. High denotes human confirmation, replicated in vivo findings across independent studies, or in vitro bioactivity benchmarked against measured human internal exposure. Moderate denotes in vitro findings supported by non-mammalian in vivo models, or a single mammalian in vivo study. Low denotes a single in vitro or biophysical system with no whole-organism confirmation. Hyphenated tiers denote intermediate positions. Entries drawn from regulatory determinations are labelled data-gap exhibits and are not rated, as they document absent or unfulfilled testing rather than primary findings. Where both apply, both labels are given. A parenthetical qualifier marks findings whose toxicological significance is contested in the literature. Low and Moderate ratings therefore identify endpoints characterised only by sparse or fragmentary data. They should be read as markers of incomplete toxicological characterisation, that is, regulatory data gaps, rather than as evidence of low hazard. Single-ingredient Margin of Safety assessment treats this incomplete base as sufficient, which is the limitation this review addresses. Toxicological evidence summary for the nine most prevalent organic UV filters in the Australian therapeutic sunscreen market. Entries marked ‘data-gap exhibit’ indicate citations included to document regulatory or scientific data gaps rather than positive toxicological findings. Note: Strength reflects the quantity, consistency and human-relevance of available toxicological data for each endpoint, not the magnitude of risk. The scale characterises the state of the evidence base for each endpoint. It is not a quantitative weight-of-evidence determination under a formal framework. Ratings were assigned on a five-tier ordinal scale anchored on study type and replication. High denotes human confirmation, replicated in vivo findings across independent studies, or in vitro bioactivity benchmarked against measured human internal exposure. Moderate denotes in vitro findings supported by non-mammalian in vivo models, or a single mammalian in vivo study. Low denotes a single in vitro or biophysical system with no whole-organism confirmation. Hyphenated tiers denote intermediate positions. Entries drawn from regulatory determinations are labelled data-gap exhibits and are not rated, as they document absent or unfulfilled testing rather than primary findings. Where both apply, both labels are given. A parenthetical qualifier marks findings whose toxicological significance is contested in the literature. Low and Moderate ratings therefore identify endpoints characterised only by sparse or fragmentary data. They should be read as markers of incomplete toxicological characterisation, that is, regulatory data gaps, rather than as evidence of low hazard. Single-ingredient Margin of Safety assessment treats this incomplete base as sufficient, which is the limitation this review addresses. Even where individual filter assessments exist, the framework treats the sunscreen tube as the only source of exposure. The real-world picture is wider. Triazine UV filters have been identified as an emerging class of abundant, ubiquitous pollutants in indoor dust and air from southern China [12] . Benzophenone−3, 4-methylbenzylidene camphor, homosalate and octocrylene have been measured at substantial concentrations in indoor dust, with indoor dust estimated to contribute approximately 8% of total organic UV filter exposure via combined ingestion and dermal absorption [2] . Indoor air has been identified as the dominant non-dermal exposure pathway for several filters in comprehensive multi-pathway risk assessments, with estimated daily intakes of EHMC and EHS via this route exceeding contributions from drinking water and indoor dust combined [63] . Several organic UV filters and their metabolites have additionally been detected in human milk, including benzophenones, their hydroxylated metabolites, and triazine UV filters [41] , [59] . Infant exposure, therefore, begins well before the first-time sunscreen is applied to the infant’s own skin. Organic UV filters undergo chemical transformation in chlorinated swimming-pool water, producing disinfection by-products whose toxicology is largely unstudied [22] , [30] , [38] . Finally, UV filters are present in many cosmetic product categories beyond sunscreens. A German market survey of 4447 cosmetic products found UV filters or zinc oxide in 22.5% of products examined [62] . The same survey reported a median of four chemical UV filters per sunscreen and three per perfume containing UV filters. A consumer using a daily SPF moisturiser, a separate body sunscreen, a tinted facial product and an SPF lip balm can therefore apply the same filter from four different products in a single morning. Each product’s compliance with concentration limits is assessed in isolation. None of these exposures are subtracted from, or added to, the dermal budget used in any current MoS calculation. They are not edge cases; they are the conditions under which Australian sunscreens are actually used. Single-ingredient Margin of Safety assessment is not arbitrary. The framework applies a default uncertainty factor of 100. This is derived from a tenfold factor for interspecies extrapolation and a further tenfold factor for interindividual variability [52] . Each filter currently in routine Australian use has passed this threshold under an individual exposure scenario. The argument advanced here is not that the current assessment has produced demonstrated population harm. It is that the framework’s assumptions are inconsistent with the conditions of use established in §3.1-§3.2. The analysis that follows assumes the continued necessity of effective broad-spectrum photoprotection in the Australian context. The question is not whether sunscreens should be used, but whether the regulatory framework adequately characterises the chemicals they contain under conditions of recommended use. The [61] Safety Review concluded that five of seven reviewed sunscreen actives were low risk at current permitted concentrations, with restricted use recommended for homosalate and oxybenzone [61] . The present analysis does not contest those findings within their own assumptions. The TGA Safety Review itself acknowledged several limitations of its own scope. Aggregate exposure from non-sunscreen consumer products containing the same active ingredients was not considered. Metabolite and impurity toxicology were not considered. Available data on endocrine-modifying potential were noted as inadequate to draw causal conclusions in humans. The present analysis addresses precisely those acknowledged gaps. The mixture-toxicology and cumulative-risk methods required to address the gaps identified in §3.3–3.5 are not hypothetical. They are methodologically established in adjacent regulatory domains. The European Food Safety Authority routinely conducts cumulative risk characterisations for groups of substances sharing a common adverse outcome on the thyroid, the liver and other targets, using common assessment groups, relative potency factors, and hazard indices summed across substances sharing a mode of action [9] . The same principle is already operationalised within the cosmetic safety framework itself. The SCCS Notes of Guidance require aggregate exposure assessment for substances classified as carcinogenic, mutagenic or reproductively toxic (CMR) at category 1A and 1B [52] . For substances suspected of endocrine activity, the SCCS provides age-specific aggregate cosmetic exposure methodology. The methodological precedent for cumulative exposure assessment of cosmetic ingredients, therefore, exists within the same regulatory framework that currently approves UV filters individually. Translating this machinery to organic UV filters in therapeutic sunscreens requires five elements. • First, pathway grouping: filters acting on the same biological target (estrogen receptor, androgen receptor, thyroid axis, oxidative stress) should be grouped into common assessment groups, with hazard indices summed across co-formulated members of each group. • Second, human pharmacokinetic data for all permitted filters: as documented in §3.3.4, modern maximal-use clinical pharmacokinetic designs have been applied to six of the nine filters reviewed and not to PBSA, EHT or 4-MBC. Extending the design to the remaining permitted filters is a matter of investment, not methodology. • Third, transformation-product toxicology: chemicals formed when UV filters break down on the skin under sunlight or in chlorinated water should be characterised and included in the SED for the parent filter. • Fourth, formulation-aware dermal absorption: where a permitted filter is also a penetration enhancer, or where commercial formulations routinely contain such excipients, dermal absorption inputs should reflect the actual formulation rather than the isolated ingredient. • Fifth, a formal re-evaluation cycle: several filters in current use were approved based on regulatory opinions issued in the 1990s or early 2000s. A defined re-evaluation cycle (for example, every ten years) tied to the Permissible Ingredients Determination would ensure that listing decisions reflect current science rather than legacy approvals. First, pathway grouping: filters acting on the same biological target (estrogen receptor, androgen receptor, thyroid axis, oxidative stress) should be grouped into common assessment groups, with hazard indices summed across co-formulated members of each group. Second, human pharmacokinetic data for all permitted filters: as documented in §3.3.4, modern maximal-use clinical pharmacokinetic designs have been applied to six of the nine filters reviewed and not to PBSA, EHT or 4-MBC. Extending the design to the remaining permitted filters is a matter of investment, not methodology. Third, transformation-product toxicology: chemicals formed when UV filters break down on the skin under sunlight or in chlorinated water should be characterised and included in the SED for the parent filter. Fourth, formulation-aware dermal absorption: where a permitted filter is also a penetration enhancer, or where commercial formulations routinely contain such excipients, dermal absorption inputs should reflect the actual formulation rather than the isolated ingredient. Fifth, a formal re-evaluation cycle: several filters in current use were approved based on regulatory opinions issued in the 1990s or early 2000s. A defined re-evaluation cycle (for example, every ten years) tied to the Permissible Ingredients Determination would ensure that listing decisions reflect current science rather than legacy approvals.

The authors are employed by VeganicSKN Australia, a manufacturer of mineral (zinc oxide) sunscreen products. JM is the Chief Executive Officer of VeganicSKN Australia and holds equity in the company. AC also holds equity in VeganicSKN Australia. The authors’ employer manufactures products in the mineral class and does not manufacture organic UV filters. Neither author holds equity in any other sunscreen-related entity. Neither author holds patents related to sunscreen formulations or UV filters. JM is a Director of the Australian Sunscreen Council, and AC is an expert member of the same body (unpaid, with no financial benefit). The Australian Sunscreen Council is a registered charity focused on balanced UV exposure and Australia’s peak industry body for sunscreen, whose published goal is to promote safe UV exposure and to ensure that the benefits of sunscreen ingredients outweigh their risks. The authors have no financial relationship with any organic UV-filter manufacturer, distributor or industry association. The complete ARTG dataset analysed in this study is provided in Supplementary Table S1 to permit independent verification of all numerical claims. These interests are disclosed in full; readers should weigh the arguments on the cited primary evidence.

A complete extraction of therapeutic sunscreens with an active AUST L designation was performed from the publicly accessible Australian Register of Therapeutic Goods (ARTG) database in January 2026. The extraction captured each listing’s ARTG identifier, sponsor, product name, ingredient list with declared concentrations (where available), claimed sun-protection factor (SPF) and intended-use statement. Inclusion was restricted to active AUST L therapeutic sunscreen listings classified by the TGA as primary sunscreens (i.e., the principal purpose of the product is photoprotection). Secondary sunscreens (cosmetic products carrying an incidental SPF claim) and ARTG entries that had been cancelled or suspended at the time of extraction were excluded. Duplicate listings, those representing the same finished formulation under multiple ARTG identifiers (private-label) re-listings, pack-size variants, and sponsor transfers, were collapsed to a single record of an identical declared active ingredient profile and concentration. Where the ARTG record listed a single ingredient more than once for a given product, the ingredient was counted as present once per product. After application of these criteria, 924 unique therapeutic sunscreen products remained. For each retained product, the declared UV filter actives were tabulated and classified as organic UV filters or inorganic (mineral) UV filters. Zinc oxide and titanium dioxide were classified as inorganic. The organic-filter count per product was used as the primary descriptor of formulation complexity. Of the 924 products, 187 (20.2%) contained only inorganic UV filters, with no organic UV filter declared. The remaining 737 products (79.8%) contained at least one organic UV filter. Subsequent multi-filter analysis was performed on the organic-containing subset (n = 737), with the mineral-only subset (n = 187) reported separately. Combined organic UV filter load was calculated by summing the declared concentrations (% w/w) of all organic UV filters within each formulation. All 924 products in the final dataset declared specific numerical concentrations for each active UV filter ingredient; no concentration ranges or undeclared concentrations were encountered, and no midpoint substitution was required. Per-filter prevalence was calculated as the number of products in which each active ingredient appeared at any declared concentration, divided by the total of 924. Per-filter prevalence ( Fig. 2 ) is calculated against the full 924-product population. Fig. 1 plots the organic-filter count distribution against the full 924-product population, including the mineral-only subset shown as the leftmost bar. Multi-filter claims in the text (e.g., 85.2% of organic-containing products contain four or more organic filters) are scoped to the organic-containing subset (n = 737) where indicated, and the appropriate denominator is named at each occurrence. Fig. 1 Distribution of organic UV filter count per product across 924 therapeutic sunscreens with active AUST L designation listed on the Australian Register of Therapeutic Goods (January 2026). The leftmost bar (grey) represents 187 mineral-only sunscreens containing zinc oxide and/or titanium dioxide as the sole declared UV filters, with no organic UV filter present. The remaining bars (brown) represent the 737 products containing one or more organic UV filters; no product is formulated with a single organic filter, and 628 (85.2%) contain four or more. Zinc oxide and titanium dioxide are recorded separately and not included in the organic-filter count. Distribution of organic UV filter count per product across 924 therapeutic sunscreens with active AUST L designation listed on the Australian Register of Therapeutic Goods (January 2026). The leftmost bar (grey) represents 187 mineral-only sunscreens containing zinc oxide and/or titanium dioxide as the sole declared UV filters, with no organic UV filter present. The remaining bars (brown) represent the 737 products containing one or more organic UV filters; no product is formulated with a single organic filter, and 628 (85.2%) contain four or more. Zinc oxide and titanium dioxide are recorded separately and not included in the organic-filter count. The toxicological synthesis presented here is a narrative regulatory-perspective review, not a systematic review. Filters were selected based on formulation prevalence in the Australian Register of Therapeutic Goods, identifying the nine organic UV filters that recurred in the dominant multi-filter combinations from §3.1–3.2 and that are subject to recent or ongoing re-evaluation by the TGA, the SCCS, the UK Scientific Advisory Group on Chemical Safety (SAG-CS) or the US Food and Drug Administration (FDA): octocrylene, homosalate, octyl salicylate, octyl methoxycinnamate (OMC), avobenzone, phenylbenzimidazole sulfonic acid (PBSA), ethylhexyl triazone (EHT), 4-methylbenzylidene camphor (4-MBC) and oxybenzone. Searches were run in PubMed, Scopus and Google Scholar using each filter’s INN/INCI name and CAS number combined with the terms toxicology, endocrine, photodegradation, dermal absorption and mixture, covering the period 2000–2026. Regulatory opinions (TGA, SCCS, FDA, AICIS) and published cumulative risk frameworks for pesticide residues (EFSA) were retrieved directly from the issuing bodies. Inclusion was prioritised for peer-reviewed primary studies on endocrine activity, photochemical stability, dermal penetration, and pharmacokinetics. Mechanistic findings of relevance to mixture behaviour, including shared molecular targets, photochemical convergence, and formulation-dependent absorption, were extracted in preference to single-ingredient hazard descriptors already reflected in current MoS dossiers. The review does not follow PRISMA guidelines and was not designed to be exhaustive. Its purpose is to characterise documented mechanistic concerns for the recurring filters in the Australian market, not to deliver a quantitative weight-of-evidence assessment. The standard MoS calculation as applied by the TGA and SCCS was reconstructed from the SCCS Notes of Guidance for the Testing of Cosmetic Ingredients and their Safety Evaluation [SCCS/1647/22] [52] : identification of a no-observed-adverse-effect level (NOAEL) from animal studies; estimation of a systemic exposure dose (SED) from dermal absorption data and standard product-use assumptions; and calculation of MoS = NOAEL / SED, with an acceptance threshold of MoS ≥ 100.

A systematic extraction of 924 therapeutic sunscreen products listed on the Australian Register of Therapeutic Goods shows that organic-filter sunscreens are predominantly four-to-six-filter mixtures. Combined organic UV filter loads commonly reach 15–25% w/w of the finished product. A small group of recurring actives dominates the market. The filters that recur most frequently converge on shared biological targets. These include estrogen, androgen and thyroid signalling, and oxidative stress. One of the most common, octyl salicylate, is a known dermal penetration enhancer co-formulated with other systemically absorbed filters. The current single-ingredient Margin of Safety assessment does not sum co-formulated contributions on shared pathways. It does not account for transformation products formed on the skin. It does not adjust for penetration-enhancing excipients in the same product. It does not integrate exposure from indoor air, indoor dust, human milk, chlorinated water or concurrent multi-product co-use. These structural gaps mean that single-ingredient MoS assessment does not capture cumulative exposure in the multi-filter products that dominate the Australian therapeutic sunscreen market. A pathway- and mixture-based evaluation framework is needed. Such a framework is already methodologically established in adjacent regulatory domains and remains feasible with existing methods. The continued reliance on a single-ingredient framework, in the face of documented co-formulation patterns and shared mechanistic targets, is itself a finding that warrants regulatory attention.

During the preparation of this work the authors used Claude (Anthropic) in order to assist with literature search and primary-source verification, manuscript structuring, and language refinement. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

This study analyses ARTG-declared formulation data and published toxicological evidence. The toxicological literature review was targeted at filters identified by their prevalence in §3.1 and §3.2 and does not follow a systematic review protocol such as PRISMA. It does not include new in vivo or in vitro experimental work, and it does not provide a quantitative cumulative risk estimate. Filter concentrations were taken from declared values on the ARTG; product-by-product analytical verification was outside scope. Inorganic filters (zinc oxide, titanium dioxide) are characterised only with respect to their market prevalence; their separate safety profile is not the subject of this paper. The analysis is not an argument against sunscreen use: ultraviolet radiation is a proven carcinogen, effective photoprotection saves lives, and Australians should continue to use sunscreen. The argument is structural. The sunscreens Australians use can, and should, be assessed for safety as the multi-ingredient mixtures they actually are.

Australia regulates sunscreens as therapeutic goods rather than cosmetics, applying safety, quality, and efficacy requirements prior to market entry [61] . National guidelines recommend applying sunscreen to all exposed skin whenever the UV index reaches 3 or above. For much of the Australian population, this translates into daily, whole-body, repeated application for several months of the year. Under the Therapeutic Goods Administration’s own Australian Sunscreen Exposure Model, an adult following these guidelines may apply approximately 140 mL of sunscreen per day to around 75% of body surface area [61] . No other consumer product delivers this volume of synthetic chemicals to this much skin surface this frequently. Achieving high-SPF, broad-spectrum protection with acceptable cosmetic properties is difficult to obtain with a single UV filter. Modern formulations therefore combine multiple organic filters, each selected for its absorption spectrum, photostability and solubility. A UVB absorber such as octocrylene or homosalate is typically paired with a UVA absorber such as avobenzone, stabilised by a third filter acting as a triplet quencher, and boosted by a fourth or fifth filter to reach the SPF target [57] . Each filter is approved individually under the Margin of Safety (MoS) framework used by the Therapeutic Goods Administration (TGA), the European Commission’s Scientific Committee on Consumer Safety (SCCS) and comparable bodies. The combined mixture that the consumer actually applies is not assessed as a mixture by any regulator. A growing body of work has identified concerns relating to individual organic UV filters. Several show estrogenic, androgenic and thyroid-axis activities in vitro [20] . Some cause oxidative DNA damage through photosensitisation [3] . Degradation products accumulate in commercial formulations over time [11] . Systemic absorption exceeds the FDA’s toxicological-concern threshold under maximal-use conditions [36] , [37] . Substantial non-dermal exposure via indoor dust and air has been documented [1] , [12] , [63] . Mixture-toxicology methods adequate to address the additive contributions of co-formulated filters have been developed and routinely applied in adjacent regulatory domains, most notably pesticide cumulative risk assessment [9] . Application of these methods to sunscreens requires a prior question: what filter combinations are actually on the market, and at what combined doses? Recent international work has begun answering that question at the population level. Jo et al. [24] analysed 2183 Korean sunscreen products and applied association rule mining to identify multi-filter co-occurrence patterns as a regulatory priority for hazard and risk assessment [24] . Pniewska and Kalinowska-Lis [46] characterised 200 sunscreens in the Polish cosmetic market and reported that most products contained four or five UV filters [46] . Levanaityte et al. [31] analysed 467 sunscreens in Lithuanian community pharmacies and found a mean of 4.29 filters per formulation [31] . These studies converge on a consistent finding: most marketed sunscreens are multi-filter products, and current single-ingredient safety assessment does not reflect this composition. No comparable analysis has been conducted for Australian therapeutic sunscreens. The Australian market differs from European and Asian cosmetic-route products in regulatory framework, product population, and application guidance. The authors do acknowledge that ultraviolet radiation is a Group 1 carcinogen and sunscreen is a proven public health intervention in Australia. The present analysis is not an argument against sunscreen use; it concerns the regulatory assessment framework applied to the actives those products contain. The objectives of this study were: (i) to systematically extract and characterise the full population of therapeutic sunscreens listed on the ARTG; (ii) to quantify the distribution of organic UV filter count per product and the prevalence of individual filters across the market; (iii) to review the published toxicological literature on the filters that recur most frequently in the resulting multi-filter combinations; and (iv) to evaluate whether the single-ingredient MoS framework currently applied by the TGA and comparable regulators is fit for purpose for products of this composition. To our knowledge, this is the first comprehensive characterisation of UV filters in Australian therapeutic sunscreens at the formulation level. The analysis extends recent international market analyses [24] , [31] , [46] . It applies a comparable approach to Australia’s therapeutic sunscreen regulatory framework. It links the formulation findings to an explicit evaluation of the current single-ingredient Margin of Safety assessment.

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Dr. Abhinandan Chowdhury reports financial support was provided by Australian Sunscreen Council. Abhinandan Chowdhury reports a relationship with VeganicSKN Australia that includes: employment and equity or stocks. Joseph Mizikovsky reports a relationship with VeganicSKN Australia that includes: board membership, employment, and equity or stocks. Joseph Mizikovsky reports a relationship with Australian Sunscreen Council that includes: board membership. The authors are employed by VeganicSKN Australia, a manufacturer of mineral (zinc oxide) sunscreen products. JM is the Chief Executive Officer of VeganicSKN Australia and holds equity in the company. AC also holds equity in VeganicSKN Australia. The authors’ employer manufactures products in the mineral class and does not manufacture organic UV filters. Neither author holds equity in any other sunscreen-related entity. Neither author holds patents related to sunscreen formulations or UV filters. JM is a Director of the Australian Sunscreen Council, and AC is an expert member of the same body (unpaid, with no financial benefit). The Australian Sunscreen Council is a registered charity focused on balanced UV exposure and Australia’s peak industry body for sunscreen, whose published goal is to promote safe UV exposure and to ensure that the benefits of sunscreen ingredients outweigh their risks. The authors have no financial relationship with any organic UV-filter manufacturer, distributor or industry association. The complete ARTG dataset analysed in this study is provided in Supplementary Table S1 to permit independent verification of all numerical claims. These interests are disclosed in full; readers should weigh the arguments on the cited primary evidence. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper