Inappropriate Use of Risk Assessment in Addressing Health Hazards Posed by Civil Aircraft Cabin Air

Toxicological reviews of data on Aerotoxic Syndrome that have been widely referenced by the airline industry tend to use a toxicological endpoint, Organo-Phosphate Induced Neuropathy (OPIDN), that is acknowledged to be the result of a very high dose of organophosphate exposure. Additionally, the reviews tend to only address one chemical, Tri-Ortho Cresyl Phosphate (TOCP), ignoring the presence of other toxic compounds in a complex mixture. In using this to justify the safety of the continued use of unfiltered engine bleed air to ventilate civil aircraft cabins, this represents a misuse of toxicological risk assessment. The approach totally ignores the scientific literature on repeated low-dose exposure to OPs over extended periods, the constant presence of a complex mixture of OPs in engine bleed air and their overall toxicity and the variable susceptibility of individuals to toxicological damage. This paper lists the above-mentioned studies and reviews a sub-set. We present the scientific literature that should be considered to make a realistic risk assessment of the hazards of aircraft engine bleed air.

further complicated by the production of pyrolysis breakdown molecules as the oil ages during use [2,4,6,7]. Oil utilised in turbine engines is consumed in normal engine operation and a small proportion of this used oil migrates over oil seals used within the engine [8][9][10]. All current pressurised commercial aircraft, with the exception of the Boing 787, use engine bleed air to ventilate and pressurise the cockpit and the cabin. Because the engine bleed air take-off port is aft of the front engine bearings, fugitive emissions from the engine bearing chamber escapes into the cabin breathing air via the oil seals [11,12]. Lower levels of oil can migrate past the seals during transient engine power changes and with changes in air supply configurations as part of normal engine operation. Far less frequently, events such as oil overfill or oil seal failure conditions may occur, leading to higher dose exposure to engine oil fumes. Both scenarios are referred to as 'fume events', with the latter scenarios possibly leading to a visible haze in the cabin. However, there is evidence that there is chronic continual low-dose exposure of aircrew and passengers to fugitive engine emissions during normal flying conditions [10,13]. The toxicity of tricresyl phosphate first came to light in the 'Ginger Jake' scandal in the 1920s. Jamaican Ginger was a medicine (also known as Jake). Because of a price increase in castor oil, one of the constituents, the manufacturer substituted it with Tri-Ortho Cresyl Phosphate (TOCP). Consumers of Ginger Jake received very high doses of TOCP. This proved to be neuro-toxic and caused paralysis, part of a syndrome that has subsequently become known as Organophosphate Induced Neuropathy (OPIDN). Subsequent considerations of the toxicity of cresyl phosphates in aircraft cabin air have tended to concentrate solely on the TOCP as the toxic agent and singly on the toxicological endpoint of OPIDN. This approach ignores a large body of science which considers the toxicity of other cresyl phosphate congeners [4,[14][15][16][17][18][19][20] and toxicological endpoints far more sensitive than OPIDN [21][22][23][24][25]. A critique of this approach in the consideration of aircraft cabin air safety is the motive for writing this paper.

Comparative toxicology of cresyl phosphates
If we solely consider the Ortho-Cresyl Phosphate (OCP) isomers, there are Mono-(MOCP), Di-(DOCP) and Tri-(TOCP) ortho-isomers present in jet engine oils. TOCP is said to be at very low levels, however the mono ortho & di-ortho isomers of cresyl phosphate have been generally ignored in toxicological assessments and are present at higher concentrations.
MOCP is 10 x more toxic than TOCP [3,17,26,27] DOCP is 5 x more toxic than TOCP [3,17,26,27] The technical mixture of the commercial formulation of TCP used in the most widely used engine oil, Mobil Jet Oil II contains [5]: MOCP 3070ppm in the TCP DOCP 6ppm in the TCP TOCP 5ppb (0.005ppm) in the TCP

Using Equivalent toxicity
Ortho TCP content in the oils with TCP at 3% in oils. The following was published in a 2001conference proceedings in Australia by Winder & Balouet [28]. Therefore, by solely concentrating on TOCP in deliberating the toxicity of oil in fugitive jet engine fumes, the toxicity of the ortho-isomers in cresyl phosphate is being underestimated by about 6 millionfold, as independently determined by Winder and Balouet. To this the toxicity of the meta and para isomers of cresyl phosphate would need to be added alongside the other (generally unspecified) impurities present in the technical mixture and, additionally, the pyrolysis products which appear as the oil ages in use [29][30][31][32].

Risk assessment
A risk assessment goes through several steps: Hazard identification: Which requires insight into the system or process under scrutiny.
Hazard assessment: Which can only be performed on hazards that have been considered, requires the application of scientific experimentation (e.g. toxicology) and is therefore costly and time consuming Exposure assessment: Is required when human exposure to toxic substances is being considered and requires further scientific investigation (e.g. measuring contamination levels). It can be expensive to perform adequately. The route of entry of the toxicant(s) is critical to the exposure assessment.
Risk assessment: Finally, the information collected in steps (1-3) can be combined into a risk assessment. The validity of this final step is totally reliant on the rigour and completeness of the earlier steps. Incorrect conclusions can be caused by not choosing the most sensitive toxicological endpoint, not considering the appropriate route of entry of the toxicant into the body, failure to address the problem of mixtures, etc. There is a pattern in following examples of toxicological assessment of the health risks (see (Table 3)) from inhalation of aircraft cabin air. The majority only consider TOCP as the sole toxic substance. They almost all adopt OPIDN as the toxicological endpoint, which is known to require a very high dosage. The research literature concerning OPIDN is all based on ingestion studies and not on inhalational exposure. The publications all consider high dose acute effects only, there is no mention of chronic repeated low dose exposure to cabin fumes. There is no mention of the extensive scientific literature on the mechanisms of low-dose OP exposure toxicity.

Three examples of unsustainable statements resulting from inappropriate risk assessment in the studies
In Table 3. These are representative of the majority of the studies in Table 3. (Example 1) The CAA Report (2004) [34][35]. Exposure to air crew is by inhalation, not ingestion, and therefore this represents a failure of hazard characterisation. All TCPs are assumed to be of the same toxicity as TOCP. This ignores the MOCP and DOCP isomers, which are more toxic and more abundant in jet engine oil, a failure of exposure assessment. The toxicological endpoint assessed is OPIDN, representing a failure of hazard identification. Therefore, this approach fails on all three of the preliminary steps of Risk Assessment.
A further point of interest is that in this CAA Report (2004) is information about the amount of TOCP found on the sooty deposits recovered from ventilation ducts on civil aircraft. In Table 1 of Appendix A of the report it is shown that, on a weight basis, the TOCP content of the deposits is hundreds of times higher than is found in the oil. The majority of particles passing through such ventilation ducts, during the lifetime of the aircraft, will not adhere to the duct walls but pass through into the breathing air in the cabin. The role of OPs adhering to respirable particles in the aetiology of Aerotoxic Syndrome has been discussed by Howard et al., [10]. The main point is that such adsorbed OPs will not be detected if the concentration of OPs in gaseous air is the only thing measured. The CAA report makes no mention or provision in their risk assessment of this further cause of exposure underestimation, despite their inclusion of the evidence of Table 1 in Appendix A.

Open Access Journal of Toxicology
Example 2) Dr Pleus [36] presents a risk assessment based on the oral dosage hen studies of Freudenthal [20] in which they administered oils 0.5%, 1%, or 3% TCP (with 0.4% orthoisomer) by gavage 5 days per week for 10 weeks [20]. Using the toxicological endpoint of OPIDN, he directly extrapolates from an LOAEL in hens of 20 mg/kg TCP. It should be noted that in the hens in this dosage group there was an over 60% inhibition of Neuropathy Target Esterase (NTE), indicating very high dosing. This oral dosage LOAEL finding is then subjected to a direct linear scaling to an equivalent human inhalational dose, taking into account body weight, breathing rate etc. Dr Pleus applies none of the usual Uncertainty Factors (UFs) commonly applied in regulatory toxicology (typically 10 for species variability, 10 for human intra-species variability). This simplistic linear transform leads Dr Pleus to calculate and opine that there would need to be 7,000mg of oil per cubic metre of cabin air for there to be any risk of OPIDN in air crew. A grahic is then presented of the Seattle tower with an LOAEL of oil of 7,000mg/m 3 equating to a column of oil reaching the top of the tower, while the oil odour threshold is <0.45mg/m3 and a visible oil haze is assumed to occur at 7mg/ m 3 [35,37], leading to much smaller columns of oil in the graphic. This is all a total irrelevance, as OPIDN simply doesn't occur. It is widely taught in undergraduate toxicology courses that ingestion of a toxic substance is almost invariably less harmful than if exposure is by inhalation. There are several reasons for this. The uptake from the alimentary tract is slower than by inhalation. Any toxin that is assimilated across the intestinal wall is taken to the liver by the portal vein. This keeps the initial bolus of toxin away the systemic circulation. The liver is the principal detoxification organ in the body where both Phase 1 and Phase 2 (conjugation) bio-transformations take place [38][39][40][41][42][43]. This makes the toxic substance more soluble and therefore more likely to be excreted by the kidney. This all leads to a lower sustained concentration of toxin. Not all orally administered toxin is usually assimilated, and some will pass out with the faeces. The following US EPA document summarises this very well [44].

"The only reliable way to characterize inhalation toxicity and to quantify inhalation risk is through the use of inhalation toxicity studies. Chemicals tend to be more toxic by the inhalation route than by the oral route due to rapid absorption and distribution, bypassing of the liver's metabolic protection (portal circulation), and potentially serious portal-of-entry effects, such as irritation, edema, cellular transformation, degeneration, and necrosis. An inhalation risk assessment that is based on oral data generally underestimates the inhalation risk because it cannot account for these factors."
Inhalational toxicology usually makes the, conservative, assumption that 100% of the inhaled dose is assimilated. Once across the blood/air barrier the toxic substance is in the systemic circulation with direct access to the vital organs without the chance of being detoxified in the liver. This is called the 'First Pass' effect. Therefore, Dr Pleus's approach represents a clear failure of exposure assessment. As with most of the above studies, the assumption of the toxicological endpoint of OPIDN is a failure of hazard characterisation. None of the published evidence on Aerotoxic Syndrome [10,13,45] describes anything like OPIDN and indeed it is safe to say that OPIDN is not seen in air crew. There is a different set of signs and symptoms that are happening at repeated low dose exposures. While OPIDN is associated with changes in cholinesterase levels, the chronic low dose effects in Aerotoxic Syndrome are not [21,25].
Example 3) De Ree [32] take a similar approach. Their 2 nd paragraph of Section 4 contains the core of the risk assessment: "Based on the detection limit of ToCP (0.5ng/m3 ) maximum uptake via inhalation with a 100% bioavailability would amount up to 0.02ng/kg body weight per day for a crew member of 70kg Step 1, Figure 2. This level of exposure was compared with the available lowest No-Observed Adverse Effect Level (NOAEL) of ToCP that was established in chickens and amounts to 1.25mg/ kg/ d after a repeated daily oral dose for 90 days Craig & Barth [27]. With respect to toxicological effects of ToCP it should be noted that NOAELs and Lowest Observed Adverse Effect Levels (LOAELs) were determined for two animal species; chicken and cat (reviewed in Craig & Barth [27]; Johannsen, 1977; Ehrich & Jortner. For the model, 1 mg/kg body weight per day was used as a NOAEL. It is recognized that the NOAELs and LOAELs have been obtained from toxicity studies done several decades ago that focused exclusively on major clinical symptoms, such as neuropathology, and not more recently developed neurobehavioral tests. More subtle neurobehavioral changes are usually seen at lower dose levels than those associated with neuropathology. Therefore, an Uncertainty Factor (UF) of 5 was applied to the selected NOAEL. The combined toxicity studies with two non-rodent animal species-chicken and cat-indicate a rather close similarity in NOAELs and LOAELs for ToCP. From a neurotoxicity point of view these two species are also considered to represent the human sensitivity rather well. It is therefore not necessary to add an additional uncertainty factor for comparison with the human situation (UF = 1). This approach is shown in step 2 in Figure 2." Thus, the regulatory limit they are using to establish safety is based on hen oral exposure toxicology. For 'dose' they assume that TOCP (and nothing else) is present at the LOD (so they maintain that is very conservative because no measurements they made reached LOD) and they assumed 100% assimilation by inhalation, which is standard toxicology. While the paper does acknowledge the possibility of lower dose, more subtle neurological effects it only assigns an uncertainty factor of 5 to address this. In this paper we see the same adoption of an unrealistic toxicological endpoint, OPIDN, based on ingestion rather than inhalation. The Open Access Journal of Toxicology paper ignores the complex mixtures problem. The paper avoids citing of the latest science on low dose repeat exposure to OPs. Thus, the use of OPIDN as the toxicological endpoint represents, at best, a failure of hazard assessment and at worst a complete misunderstanding of the clinical picture presenting in aircrew. The failures of risk assessment illustrated in these three examples are repeated in the majority of the other studies listed in Table 3.

Conclusions
By failing to take into account the increased levels and increased toxicity of the cresyl phosphate ortho isomers other than TOCP, the toxicity of the ortho isomers in TCP are underestimated by a factor of around 6 million. Additionally, toxicity of the non ortho isomers has been ignored, with a sole focus on OPIDN. To this should be added the exposure to OPs that are adherent to the particles in the aerosols in cabin air, which are generally not measured. This latter could be of considerable importance, as previously discussed by Howard [10]. There are widespread reports of illness among aircrew, which have been categorised by Michaelis [13]. The consequences of repeated low dose exposures to OPs have been reviewed by Terry [21]. In that review he states "….. there is now substantial evidence that this canonical (cholinesterase-based) mechanism cannot alone account for the wide variety of adverse consequences of OP exposure that have been described, especially those associated with repeated exposures to levels that produce no overt signs of acute toxicity. These include covalent binding of OPs to tyrosine and lysine residues, which suggests that numerous proteins can be modified by OPs. In addition, the mechanisms of oxidative stress and neuroinflammation and the known OP targets of motor proteins, neuronal cytoskeleton, axonal transport, neurotrophins and mitochondria. This type of exposure has been associated with prolonged impairments in attention, memory, and other domains of cognition, as well as chronic illnesses where these symptoms are manifested (e.g., Gulf War Illness, Alzheimer's disease)." This is precisely the spectrum of symptoms reported for air crew by Michaelis, Burdon & Howard [13], who often achieve cumulative career-long flying hours exceeding 20,000 hours. The significance of this pattern of exposure in neurotoxicology has been highlighted by Harris & Blain [46].
It is widely acknowledged in toxicological science that exposure via inhalation is more toxic, dose for dose, than by ingestion. Inhaled toxicants are assumed to be 100% assimilated. They pass directly into the systemic circulation, thus avoiding the liver-the major detoxification organ in the body, and having direct access to the brain and heart, possibly assisted by the presence of ultrafine particles [10]. The USEPA advice on this is clear and the failure of the studies in Table 3 to address this represents a major weakness in their use of toxicology in risk assessment. The toxicology of mixtures is more or less completely ignored in the above studies. OPs have been demonstrated to be capable of acting synergistically [47]. In addition chronic pre-exposure to repeated low dose OPs may pre-dispose increased vulnerability to subsequent higher dose OP exposure [48]. This acute-onchronic effect is repeatedly seen in the differential rate of referrals to hospital following fume events on civil aircraft. While a high proportion of aircrew frequently attend hospital after a fume event, passengers do not appear to require this. This is consistent with the published science on the topic. The use of OELs is inappropriate, they are not meant for regulating exposure of the public to toxic substances and they are not appropriate to apply to situations at altitude or for complex mixtures [10,49,50]. It is clear that the studies listed in Table 3 are of no relevance in the determination of potential harm to aircrew.