The US EPA’s landmark draft TSCA risk evaluation of formaldehyde, published on 15 March, is based on science that has been the subject of heated debate for years, particularly within the context of the agency’s Integrated Risk Information System (IRIS) programme.
The draft evaluation is certain to pour fresh fuel on the fire, not least because now there is the prospect of direct regulatory impact. While influential, IRIS assessments are not directly relevant to any particular regulatory programme. TSCA evaluations, on the other hand, necessarily trigger risk management when the agency determines a substance, under its conditions of use (COU) breaches the "unreasonable risk" threshold.
The draft evaluation concluded that 58 out of 62 COUs contribute to formaldehyde’s overall unreasonable risk. If upheld in a final evaluation, the EPA will need to address each of those applications in a subsequent TSCA section 6(a) risk management rule.
Stakeholder input and engagement with the draft evaluation is expected to be considerable in the weeks to come, given that industry has said formaldehyde supports employment for nearly a million workers and generates over half a trillion dollars in sales in the US.
In the meantime, there are several elements of the draft evaluation likely to be focal points for debate.
Draft IRIS assessment provides key hazard values
As expected, the draft TSCA evaluation relies heavily on the 2022 draft IRIS assessment for key hazard values.
In particular, the IRIS assessment provides – with only minor adjustments – the TSCA values for chronic inhalation leading to adverse effects on human health. Those are:
- the cancer inhalation unit risk (IUR) of 1.1 x 10-5 per microgram-per-cubic-metre (μg/m3); and
- the non-cancer reference concentration (RfC) of 0.021mg/m3.
This is significant because the draft IRIS assessment has already been peer-reviewed by the National Academies of Sciences, Engineering, and Medicine (NASEM), which broadly supported the approach taken.
Given that work, the EPA has indicated that the forthcoming peer review of the draft TSCA evaluation could exclude consideration of aspects lifted directly from that assessment to avoid duplicating effort.
While such a decision would not alter the approach taken by stakeholders during the public consultation, the EPA is unlikely to make changes to the evaluation that have not been considered during peer review.
Fresh findings
Despite the reliance on certain IRIS values, the draft TSCA risk evaluation also includes a lot of original work.
The IRIS programme is exclusively focused on hazard, meaning the EPA had to look elsewhere for information on exposure.
Furthermore, the draft IRIS assessment focused on chronic inhalation exposure leading to human health effects. It did not consider any element of environmental risk, nor dermal or oral exposure over any time frame, and only considered acute inhalation qualitatively.
This means that a lot of what appears in the draft TSCA evaluation is not directly related to the treatment of formaldehyde under IRIS, which has been the source of so much controversy for so long.
Cancer mostly not significant for unreasonable risk
Historically, carcinogenicity – particularly relating to myeloid leukaemia – has formed a major part of the debate about formaldehyde.
For example, the 2022 draft IRIS assessment found that the available evidence "demonstrates" formaldehyde causes myeloid leukaemia, nasopharyngeal cancer (NPC) and sinonasal cancer (SNC).
In written comments on that draft, the American Chemistry Council (ACC) dedicated 23 pages, with 19 separate sub-headings, to criticism of the EPA’s handing of myeloid leukemia, plus a further five to NPC.
However, in the draft TSCA evaluation, cancer drives the unreasonable risk finding via just one COU – commercial use of automotive care products, lubricants, greases, fuels and related products – with nasopharyngeal cancer the relevant hazard endpoint.
Furthermore, for this COU, non-cancer effects also contribute to the unreasonable risk finding, and with more certainty (see below). That means the significance of cancer to the overall risk determination may be negligible.
Myeloid leukaemia effectively plays no part in the draft TSCA evaluation at all because – as in the draft IRIS assessment – the EPA concluded that it did not have sufficient confidence in the IUR owing to uncertainties in the dose-response data.
Instead, the overall cancer IUR is based solely on NPC. (The draft IRIS assessment does not include an IUR specifically for SNC.)
This approach was supported by the NASEM during peer review of the draft IRIS assessment.
The general public, however, may object given the EPA said "there is strong evidence that formaldehyde exposure causes myeloid leukemia in humans".
Non-cancer effects via inhalation are significant for almost all COUs
Beyond cancer, inhalation is an important exposure route driving the overall formaldehyde risk determination, with:
- 46 occupational COUs and seven consumer COUs flagged for sensory eye irritation associated with acute inhalation; and
- 45 occupational COUs and three consumer COUs flagged for a range of non-cancer effects associated with chronic inhalation.
The evaluation includes a no observed adverse effect concentration (NOAEC) of 0.5ppm for acute exposures.
The draft IRIS assessment, while focused on chronic inhalation, identified sensory irritation as an endpoint relevant for acute exposures because the effect occurs as an immediate response to an exposure. However, the programme did not calculate a risk value, meaning the NOAEC calculated for acute exposure originates in the draft TSCA evaluation.
The EPA said it based that NOAEC on four "high-quality" controlled exposure human studies that together represent a "robust" dataset.
By contrast, the agency lifted the non-cancer reference concentration (RfC) of 0.021mg/m3directly from the draft IRIS assessment. It derived this value via an unusual approach that warrants some consideration (see box).
Risk value derived from ‘suite of impacts’
Substances of concern are often associated with multiple effects for the same exposure scenario. Therefore, when conducting a risk assessment, for efficiency, it is normal to identify the ‘critical’ effect for a particular category of hazard and limit risk calculations to just that. The remaining effects are not considered further in terms of risk.
For example, if substance X causes effect Y at 5 milligrams per kilogram of bodyweight per day (mg/kg per day) and effect Z at 10mg/kg per day, it probably is not necessary to calculate the risk for both. Effect Y is always caused when effect Z is caused, and therefore it can be assumed that any risk level that is protective for Y will also be protective for Z.
According to the EPA, formaldehyde is associated with a range of chronic non-cancer effects via inhalation, including:
- reduced pulmonary function;
- increased asthma prevalence;
- reduced asthma control;
- allergy-related conditions;
- male and female reproductive toxicity; and
- developmental effects.
However, instead of selecting a critical effect from these, the EPA derived the risk value from "a suite of impacts to the respiratory system". The agency said the overall RfC was "chosen to reflect an estimate of continuous inhalation exposure to the human population (including sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime".
Dermal exposure is also significant
Many of the COUs identified for unreasonable risk on account of inhalation also contribute to formaldehyde’s risk with respect to acute dermal exposure.
In total, dermal exposure hit the unreasonable risk mark for 47 occupational COUs and seven consumer COUs based on skin sensitisation.
The corresponding risk levels were derived from an "extensive" dataset of human, animal, and in vitro studies, the EPA said.
The elicitation threshold of 10.5µg/cm2 was based on data from a 1997 study by Flyholm et al comprising patch testing of 20 formaldehyde-sensitive people.
The induction threshold of 100µg/cm2 was based on a 2003 study by Basketter et al comprising local lymph node assay (LLNA) testing in mice.
Wealth of monitoring data for inhalation exposure in the workplace
With respect to exposure, the EPA relied on a mix of monitoring and modeled data.
Monitoring data can be problematic for a range of reasons, for example older data may not reflect modern working conditions or job activities. Also, some sites and activities are more likely than others to be monitored, which may skew the data.
Nevertheless, monitoring data is almost always preferable to modeled data for exposure assessment because it is chemical specific and directly applicable to the exposure scenario.
The EPA used monitoring data for inhalation exposure assessment for all but three COUs. A key source of that monitoring data was samples collected by the US Occupational Safety and Health Administration (OSHA) during facility inspections.
Where inhalation monitoring data was not available, the agency used modeled data.
In contrast, there was no available monitoring data for dermal exposure in the workplace and the agency relied exclusively on modeled data.
Rulemaking uncertainty
Where the EPA determined that a COU contributes to formaldehyde’s overall unreasonable risk via a particular combination of exposure route, duration and adverse effect, the agency assigned one of two ratings. For each identified risk there is either:
- a high level of certainty; or
- less certainty.
The agency apparently arrived at these ratings by comparing a risk attributable to a TSCA-regulated exposure to the risk posed by background levels of formaldehyde in the air.
"EPA has high level of certainty of the contribution to the unreasonable risk of formaldehyde from a COU when the risk from such COU is much greater than the risk expected from the formaldehyde based on monitored concentrations in the indoor air," it said in the draft TSCA evaluation. "EPA is less certain of the contribution by the COU when the risk from the COU is within the expected risk based on monitored concentrations in the indoor air."
How the agency will account for these certainty ratings in any rulemaking is unclear.
The situation is made somewhat simpler by there being no COUs with only a "less certainty" finding. This likely renders moot the question of whether a COU must still be addressed by rulemaking if it contributes to the unreasonable risk but without high certainty.
For many COUs, however, there are both high certainty and less certainty contributions, raising questions about how the EPA will target any risk management measures it imposes. For example, domestic manufacturing has high certainty for acute inhalation exposure and less certainty for chronic inhalation exposure
Risk management might take a very different form if the EPA must address both of these contributions as opposed to just the acute exposure.
The road ahead for risk management is also muddied by the EPA’s derivation of an existing chemical exposure limit (ECEL) of 0.011ppm as an eight-hour time-weighted average (TWA), a level that may be below background levels in the environment. How that could translate into an occupational workplace safety threshold is another unanswered question.
The EPA will accept comments on the draft risk evaluation until 14 May, with plans to finalise it before the end of the year. It would then have two years to adopt a risk management rule.