Brought to you by:
Topical Review

A mechanistic study and review of volatile products from peroxidation of unsaturated fatty acids: an aid to understanding the origins of volatile organic compounds from the human body

, , , , and

Published 13 May 2020 • © 2020 IOP Publishing Ltd
, , Citation Norman Ratcliffe et al 2020 J. Breath Res. 14 034001 DOI 10.1088/1752-7163/ab7f9d

Download Article PDF

Article metrics

1579 Total downloads

Share this article

Article and author information

Author e-mails

Norman.Ratcliffe@uwe.ac.uk

Author affiliations

1 Institute of Biosensor Technology, University of the West of England, Coldharbour Lane, Frenchay, Bristol BS16 1QY, United Kingdom

2 Institute of Animal Reproduction and Food Research of Polish Academy of Sciences, Tuwima 10 Str., 10-748, Olsztyn, Poland

ORCID iDs

Dates

  1. Received 5 November 2019
  2. Accepted 12 March 2020
  3. Published

Check for updates using Crossmark

Peer review information

Method: Single-anonymous
Revisions: 1
Screened for originality? Yes

Buy this article in print

Journal RSS
Sign up for new issue notifications
1752-7163/14/3/034001

Abstract

The assessment of volatile compounds (VOCs) for disease diagnosis is a growing area of research. There is a need to provide hard evidence i.e. biochemical routes, to justify putative VOC biomarkers, as in many cases this remains uncertain, which weakens their authenticity. Recently reports of volatile hydrocarbons and or aldehydes in bodily fluids and breath have been attributed to oxidative stress, although as discussed here, fewer compounds have been reported than expected from a mechanistic examination. Oxidative stress can result from many disease states which produce inflammation, and a better understanding of the interconnection between oxidative stress and the release of VOCs from target diseased and healthy organs could greatly help diagnoses. It is generally considered that oxidation of unsaturated fatty acids are a major source of these VOCs. An investigation listing the many possible volatile oxidation products has not been undertaken. This is described here using a mechanistic analysis (based on the literature) of the compounds derived from molecular cleavage and the results compared with a recent review of all the VOCs emanating from the human body, which satisfactorily explains the presence of at least 100 VOCs. Six important unsaturated fatty acids, oleic, palmitoleic, linoleic, linolenic, arachidonic, and cervonic acids have been shown to be capable of producing up to 18 n+6 unique breakdown products (where n = the number of alkene double bonds in the fatty acid hydrocarbon chain), in total 299 compounds. In many cases these have not been reported. We suggest several reasons for this: these VOCs have not been expected, so researchers are not looking for them and importantly some are not present in the mass spectral libraries, or they are too low a concentration to have been detected, or are not present. Furthermore a theoretical explanation for the origins of branched aldehydes and other compounds arising from bacterial oxidative metabolism of unsaturated fatty acids are described.

Export citation and abstract BibTeX RIS

Previous article in issue
Next article in issue
Supplementary data

1. Introduction

The human body emits a very wide range of volatile organic compounds (VOCs) that can be categorised into many chemical groupings such as alcohols, aldehydes, amines, aromatics, halides, ketones, saturated and unsaturated hydrocarbons, short chain fatty acids, sulphides and combinations of these groups [1]. Some of these compounds are putative biomarkers of disease and it is becoming apparent that it is important to establish sound biochemical routes to possible volatile biomarkers [2].

Aerobic cells naturally produce reactive oxygen species (ROS) such as hydrogen peroxide (H2O2), hydroxyl radicals (OH•), and organic peroxides [3]. Some VOCs linked to disease are considered to be end products of ROS interacting with cellular components particularly resulting in lipid peroxidation. A significant part of our bodies is made of unsaturated lipids, and prone to attack by ROS. The oxidative products can build up extensively as they are a result of self-propagating peroxidation chain reactions (see figure 1).

Figure 1.

Figure 1. Description of radical oxidation of an unsaturated fatty acid, showing propagation which explains the build-up of oxidation products.

Download figure:

Standard image High-resolution image

The elevation of ROS can be considered a generic response to disease processes. The hypothesis of ROS and the host response to disease is supported by the therapeutic benefits of exposure to focused doses of ionizing radiation [4]. These interventions generate reactive species that have the desired therapeutic effect by augmenting the host response. Since aerobic organisms are able to control the endogenous levels of ROS by antioxidant species, it has been hypothesized that aerobic organisms rely on these short-lived reactive oxygen species for molecular signaling [5]. A classic example of this is the ROS signaling for recruitment, differentiation and activation of monocytes in response to the presence of foreign organisms [6]. Furthermore, activated monocytes rely on the NADPH-dependent oxidase system to produce reactive oxygen species to kill foreign organisms [7]. There have been studies [8] that show that there are temporal relationships between exposure to dose dependent ionizing radiation or UV radiation and lipid peroxidation but no comparable relationships between the onset of disease and increases in lipid peroxidation.

There are studies [9] that have shown that reperfusion injury—one of the determinants of primary non-functioning of transplanted tissue—can be measured by lipid peroxidation. Cigarette smokers have been shown [10] to have elevated evidence of lipid peroxidation produced by the smoke and the presence of activated macrophages but cigarette smoking does not cause disease immediately. Smoking does produce heart disease COPD, emphysema, chronic bronchitis, after a lifetime of chronic exposure (pack years) but these diseases are caused by other combustion products from cigarette smoke [11].

Oxidative stress (OS) represents an imbalance between the synthesis of ROS and a biological system's ability to detoxify the reactive intermediates (produced by ROS reactions). The resulting damage due to excess ROS has been stated to play an important role in the development of many conditions [12, 13] e.g. chronic pulmonary disease, tumours, acute myocardial infarction, neurodegenerative disorders and some skin diseases [14]. OS has also been considered as a potential causative agent in tumour genesis, with enhanced oxidative activity in tumour tissues, and elevated aldehyde concentrations in the exhaled breath of patients with cancer [15, 16] probably generated through OS resulting from lipid peroxidation of cell membranes particularly the peroxidation of polyunsaturated fatty acids (PUFAs), such as arachidonic, linoleic acid and linolenic acid (which are cell membrane components) and to a lesser extent monounsaturated fatty acids (MUFAs) e.g. oleic. Under sustained OS, ROS are produced over a long time, with potential significant damage to cell structure and functions with proteins and lipids particularly significant targets for oxidative attack.

PUFAs are very susceptible to oxidation and can form hydroperoxides upon contact with oxygen in air, even without the need for enzymatic mediation [17]. Lipid radicals (LR) are initially generated by a hydrogen atom being initially removed from a fatty acid molecule, by a radical. Typical radicals that can be involved in the extraction of hydrogen atoms from lipids include the hydroxyl radical (HO•), the hydroperoxyl radical (HOO•), the lipid peroxyl radical (LOO•), and the alkoxyl radical (LO•) [18]. Lipid radicals thus generated are very reactive and can react with other radicals and particularly with oxygen, resulting in the formation of LOO• radicals, which can attack another lipid with a removal of a hydrogen atom, resulting in the formation of a lipid hydroperoxide (LOOH) and another LR. This new LR also reacts with oxygen and forms LOO•, which attacks another lipid to generate a lipid hydroperoxide, and so on, so lipid hydroperoxide accumulates as the chain reaction proceeds. It is well known that the lipid hydroperoxides can then breakdown to produce hydrocarbons, aldehydes etc, and a range of these classes of compounds is described [13, 19]. In the VOC field, it was Riely et al in 1974, who first reported hydrocarbons gases (particularly ethane and pentane) in mice breath after inducing oxidative stress by treatment with carbon tetrachloride [20]. Since then pentane in breath has been proposed as a marker of lipid peroxidation based on the assumption that it is produced but not metabolized [21]. However it is now known that the hydrocarbon can be further metabolized to 2-pentanol in the liver, and factors affecting the liver could alter pentane concentrations. Apart from pentane and ethane, many markers of oxidative stress have been proposed, e.g. lipid hydroperoxides, 4-hydroxynonenal (4-HNE), 4-hydroxyhexenal, malondialdehyde (MDA) and isoprostane [22]. The biomarkers that can be used to assess oxidative stress in vivo have been attracting interest because the accurate measurement of such stress is important for investigation of its role in diseases as well as to evaluate the efficacy of treatments.

There are many VOCs reported in the literature, such as in breath, where the origins are obscure, and their origins have not been speculated upon. This paper promotes an understanding of these origins by tabling a range of VOCs that theoretically might be expected to be produced from oxidation of the most common unsaturated fatty acids, particularly PUFAs. Extensive tabling of these VOCs has not been undertaken. The list of compounds is larger than might be expected, due to multiple oxidation sites combined with double bond migration. Previous VOC work has centred on reporting the presence of low molecular weight hydrocarbons and aldehydes, e.g. in breath, stool [12] and in cancer cells for instance [23]. This work compiles a wider range of other compounds expected to be produced e.g. very many unsaturated aldehydes, unsaturated short chain fatty acids, saturated fatty acids, aldehyde carboxylic acids, hydroxy alkene carboxylic acids, a range of alkenes and also branched chain derivatives from bacteria in particular. This will help other researchers to look for and detect these compounds to help explain the origins of VOCs in their studies and perhaps to strengthen the determination of oxidative stress, from oxygenation of lipids in organisms. It is also suggested that oxidative stress induced VOCs could be used to help determine disease and which organ is diseased, in selected cases, based on the premise that different organs tend to have different PUFA compositions and different concentrations of lipoxygenases (enzymes that facilitate hydroperoxide formation), particularly in the diseased state. The authors have proposed novel products of lipid peroxidation that may or may not be found in bodily fluids, including exhaled gaseous breath and exhaled breath condensate (EBC). In future work we intend to look for evidence that these novel products are to be found in bodily fluids, urine, blood and breath etc and thus show whether they could cross the alveolar junction into exhaled breath. Oxygenated compounds/oxidative stress compounds are expected from analyses of bodily fluids, yet relatively few have been reported. Reasons for this are discussed.

2. Method

To assess VOC peroxidation products of unsaturated acids, important unsaturated fatty acids in human were selected and examples of mono, di-, tri- tetra and hexa unsaturated acids were chosen: oleic and palmitoleic (one double bond), linoleic (two double bonds), linolenic (three double bonds), arachidonic (four double bonds) and cervonic acid (six double bonds), see figure 2 for structures.

Figure 2.

Figure 2. Structures of important unsaturated acids used in the theoretical study of products of oxidative stress from unsaturated fatty acids.

Download figure:

Standard image High-resolution image

To assess the oxidative stress breakdown products from bacteria, a literature search of branched, unsaturated fatty acids was undertaken, a significant source of which appears to be from bacterial origins. 14-Methylhexa-10-decenoic acid and 14-methylpentadec-10-enoic acid were chosen as typical examples for investigation.

Peroxidation was considered to be the main route of oxidation of unsaturated fatty acids.

The stepwise approach below was used to determine oxidation products:

  • abstraction of a H radical from an allylic position (from the alkyl side), e.g. Figure 3, giving a radical at the allylic carbon
  • reaction of the allylic carbon radical with oxygen and then hydrogen abstraction giving a radical, giving a hydroperoxide
  • cleavage of the hydroperoxide to give a hydroxyl radical and a carbon-oxygen radical
  • splitting of the carbon-carbon bond from the right hand side (RHS) of the carbon-oxygen radical bond, e.g. see figure 3, to give an aldehyde group and alkene carbon radical
  • reaction of alkene carbon radical with a hydrogen to give an alkene, OR with a hydroxyl group to give the enol form of an aldehyde which then rearranges to an aldehyde
  • then (from step 3) splitting the carbon-carbon bond from the left hand side (LHS) of the carbon-oxygen radical bond, to give an aldehyde and a carbon radical
  • reaction of the carbon radical (formed in step 6) with a hydrogen to give an alkane, OR with a hydroxyl group to give an alcohol
  • repeat (as appropriate) with allylic positions on other side of the double bond(s), see figure 4 for an example
  • after radical formation at step 1, 'double bond migration' is assessed with subsequently, a new carbon radical position produced two carbons away and available for hydroperoxide reaction, with subsequent molecular cleavage as in steps 3–7 above. Bond migrations will be undertaken multiple times for PUFAs.
  • the products obtained from steps 1–9 were then compared with data from the literature
  • check molecules in NIST library for double oxidative radical reactions (applicable for PUFAs, not MUFAs):
  • H abstraction of allylic hydrogen, to give an allylic radical (abstraction of an allylic H adjacent to 2 double bonds)
  • migration of double bond(s)
  • peroxidation of allylic carbon radical to give a hydroperoxide, by reaction with oxygen, or reaction with a hydroxyl radical group to give an alcohol
  • reduction of hydroperoxide, formed in step 14, to give an alcohol
  • then steps 1 to 11 above
  • depending on conditions, partial formation of dienes, by dehydration of hydroxyl alkenes
  • dienes (created in step 17) cyclising to furans

Figure 3.

Figure 3. Mechanism of oxidative stress products from a MUFA when the initial radical is on the alkane side, showing the many different classes of compounds, all of which are volatile or semi-volatile, LHS refers to left hand side and RHS, right hand side of the molecule as drawn.

Download figure:

Standard image High-resolution image
Figure 4.

Figure 4. Mechanism of oxidative stress products from a MUFA when the initial radical is on the carboxylic acid side, showing the many different classes of compounds, all of which are volatile or semi-volatile, LHS refers to left hand side and RHS, right hand side of the molecule as drawn.

Download figure:

Standard image High-resolution image

An example of steps 1–9 is given (figures 3 and 4), and this would be typical of MUFAs as described in figure 2. To emphasise, the radical A and B is capable of double bond migration, to give 2 other radicals which are then capable of steps 2–8 to produce even more compounds (this is well known in the literature). Step 9 shows double bond migration, as also shown in figures 3 and 4. 22 compounds can be made from just 1 double bond in a mono unsaturated fatty acid. It can be appreciated that these chemical processes will happen multiple times for PUFAs because of the extra double bonds and the numbers of hypothesised VOCs will therefore increase considerably with more double bonds in the fatty acid.

Following radical based oxidation/bond migration, a mixture of cis and trans compounds would be expected, with the latter predominating. As a mixture is expected, it is implicit in supplementary table 1 (stacks.iop.org/JBR/14/034001/mmedia) that both geometric isomers are expected and for brevity all the possibilities are not described.

Double oxidative radical reactions (steps 12–17 in the method above) can also occur (and is known in the literature [15]), see figures 5 and 6 giving a description of the classes of compounds from a PUFA with a general formula, for instance, n = 3 and m = 7 for linoleic acid, and n = 3 and m = 11 for docosadienoic acid. The literature describes double oxidative radical reactions to explain the origins of cytotoxic products, particularly 4-hydroxynonenal (4-HNE), which has seen much attention in the scientific literature [24], as arising from omega-6 fatty acids, see figure 6, for an example where n = 4. The nomenclature used was taking the CO2 H numbered as 1. The steps to create 4-HNE, are most likely: H abstraction to give a radical at carbon 11, then double bond migration (nominally from position 12 to 11), to give a radical at carbon 13, then reaction with a hydroxyl radical to give a 13-alcohol, (or peroxidation to give a 13-OOH, then reductive conversion to 13-OH), then H abstraction to give a carbon radical at the 8 position, then radical isomerisation to give the 10-radical, then peroxidation to give a 10-OOH and then cleavage to give a hydroxyl alkenal, amongst other chemical classes of compounds. The same type of methodology can be used to describe oxidation products from other PUFAs.

Figure 5.

Figure 5. Example of the classes of double oxidation products, using a general formula dienoic acid with initial oxidation at the carboxylic acid side, right hand side (RHS) of the PUFA.

Download figure:

Standard image High-resolution image
Figure 6.

Figure 6. Example of the classes of double oxidation products, using a general formula dienoic acid with initial oxidation at the alkyl, LHS side of the PUFA.

Download figure:

Standard image High-resolution image

PUFAs are considered much more susceptible to radical oxidation than monounsaturated acids, therefore this double oxidation was only considered for PUFAs. This process results in a cascade of extra hydroxy compounds. These hydroxyl products, particularly allylic hydroxy compounds, could also dehydrate to some extent in vivo, and also in the analysis stage especially if heating is involved.

Figure 7 shows how some hydroxy aldehydes are formed which are capable of cyclising to furans. The literature also describes the oxidation of unsaturated fatty acids to make a small number of other products, which are not considered volatile, therefore are not included here, as this paper focuses on volatile and semi-volatile compounds. Also the oxidation of fatty acids can make epoxides, as these compounds have figured sparsely in the VOC literature from humans, they have not been considered here.

Figure 7.

Figure 7. Selected furan products from double oxidation followed by cyclisation.

Download figure:

Standard image High-resolution image

Two bacterial acids: 14-methylpentadec-10-enoic acid and 14-methylhexadec-10-enoic acid were selected for comparison of their oxidation products with the human volatilome, particularly as there are considered to be more bacterial cells than human cells in our bodies).

The end product of this study is a table (supplementary table 1) listing extensive hypothetical end-products of oxidation using selected MUFAs and PUFAs, with products classified as saturated alcohols, saturated aldehydes, aldehyde alkene carboxylic acids, aldehyde alkene hydroxyl carboxylic acids, aldehyde hydroxyl alkenes, alkanes, alkenes, dialdehydes, alkene aldehydes, hydroxyl carboxylic acids, aldehyde carboxylic acids, hydroxyalkene aldehydes and furans. The compounds are separately described for a range of unsaturated fatty acids and include chemical formulae, chemical name and m. wts. to 4 decimal places. The latter is to aid identification by comparison with high resolution mass spectrometry (HRMS). For those not familiar with HRMS, it aids differentiation between molecular masses, which would appear similar to low resolution mass spectrometers [25]. The table also includes all compound CAS numbers when there is one ascribed to that compound, and also states whether it is in the 2014 MS NIST database. Supplementary table 1 also describes whether the hypothetical oxidation compounds of lipids have been previously discovered, by referring to the review of all known VOCs separated into VOCs from a range of sources breath, saliva, blood, milk, skin, urine and faeces for easy comparison [1].

3. Results and discussion

Supplementary table 1 shows the compounds, (a total of 299 VOCs/semi-VOCs ) listed together, hypothesized from the radical mediated oxidation processes (as described in the algorithm in the method section) of ROS .acting on selected MUFAs and PUFAs. Tables 1 and 2 give a sample of how the numerous compounds are generated.

Table 1. Hypothetical oleic acid (18:1 n-9) peroxidation breakdown products (22 unique compounds).

•H radical abstraction site Double bond migration Site of •OOH bonding Saturated and unsaturated oxocarboxylic acids Alkenes/Alkanes Saturated and unsaturated carboxylic acids Saturated and unsaturated aldehydes Alcohols and hydroxy-carboxylic acids
8 None 8 CHO(CH2)6CO2H CH2:CH(CH2)7CH3 CH3(CH2)5CO2H CHOCH:CH(CH2)7CH3, CHOCH:CH(CH2)8CH3a CH2OH(CH2)5CO2Ha
11 9 to 10 9 CHO(CH2)7CO2H CH2:CH(CH2)6CH3 CH3(CH2)6CO2H CHOCH:CH(CH2)6CH3, CHO(CH2)7CH3a CH2OH(CH2)6CO2Ha
8 9 to 8 10 CHOCH:CH(CH2)6CO2H, CHO(CH2)7CO2Ha CH3(CH2)6CH3 CH2:CH(CH2)6CO2H CHO(CH2)7CH3 CH3(CH2)6CH2OHa
11 None 11 CHOCH:CH(CH2)7CO2H, CHO(CH2)8CO2Ha CH3(CH2)5CH3 CH2:CH(CH2)7CO2H CHO(CH2)6 CH3 CH3(CH2)5CH2OHa

aCompounds from attack by OH radical.

Table 2. Hypothetical linoleic acid (18:2n-6) peroxidation breakdown products (38 unique compounds).

•H radical abstraction site Double bond migration Site of •OOH bonding Saturated and unsaturated oxocarboxylic acids Alkenes/alkanes Saturated and unsaturated carboxylic acids Saturated and unsaturated aldehydes Alcohols and hydroxy-carboxylic acids
8 None 8 CHO(CH2)6CO2H CH2:CHCH2CH:CH(CH2)4CH3 CH3(CH2)5CO2H CHOCH:CHCH2CH:CH(CH2)4CH3, CHO(CH:CH)2CH:CH(CH2)4CH3a CH2OH(CH2)5CO2Ha
11 9 to 10 9 CHO(CH2)7CO2H CH2:CHCH:CH(CH2)4CH3 CH3(CH2)6CO2H CHOCH:CHCH:CH(CH2)4CH3, CHOCH2CH:CH(CH2)4CH3a CH2OH(CH2)6CO2Ha
8 9 to 8 10 CHOCH:CH(CH2)6CO2H, CHO(CH2)2CO2Ha CH3CH:CH(CH2)4CH3 CH2:CH(CH2)6CO2H CHOCH2CH:CH(CH2)4CH3 CH2OHCH:CH(CH2)4CH3a
11 None 11 CHOCH:CH(CH2)7CO2H, CHO(CH2)8CO2Ha CH2:CH(CH2)4CH3 CH2:CH(CH2)7CO2H CHOCH:CH(CH2)4CH3, CHO(CH2)5CH3a None
14 12 to 13 12 CHOCH2CH:CH(CH2)7CO2H CH2:CH(CH2)3CH3 CH3CH:CH(CH2)7CO2H CHOCH:CH(CH2)3CH3, CHO(CH2)4CH3a CH2OHCH:CH(CH2)7CO2Ha
  11 to 10 13 CHOCH:CHCH:CH(CH2)7CO2H, CHOCH2CH:CH(CH2)7CO2Ha CH3(CH2)3CH3 CH2:CHCH:CH(CH2)7CO2H CHO(CH2)4CH3 CH2OH(CH2)3CH3a
14 None 14 CHOCH:CHCH2CH:CH(CH2)7CO2H, CHO(CH2)2CH:CH(CH2)7CO2Ha CH3(CH2)2CH3 CH2:CHCH2CH:CH(CH2)7CO2H CHO(CH2)3CH3 CH2OH(CH2)2CH3a

aCompounds attack by OH radical.

By studying these tables, it should be noted that for a single oxidation process i.e. steps 1–9 of the algorithm in the method section: the number of possible oxidation products (x), can be found from x = 18n+6, where n is the number of double bonds, see table 3. On some occasions there is more than 1 route to the same compound, leading to several duplicates in the tables of oxidation products. For example, in the case of oleic acid (table 1) there are 2 routes to 9-oxo-nonanoic acid (CHO(CH2)7CO2 H) and 1-nonanal (CHO(CH2)7CH3), therefore 22 different compounds are hypothesised.

Table 3. Number of theoretical derived VOCs/semi-VOCs from oxidation of MUFAs and PUFAs.

Acid No. double bonds No. possible oxidation products
Oleic 1 24
Palmitoleic 1 24
Linoleic 2 42
Linolenic 3 60
Arichodonic 4 78
Cervonic (DHA) 6 114

Double oxidation of PUFAs can lead to further new classes of compounds compared to a single oxidation, particularly hydroxy alkenes, hydroxyl aldehydes hydroxyl alkenals and furans, a further 50 such compounds are described in supplementary table 1, this is not exhaustive.

There are numerous alkene products derived from oxidation of the PUFAs, sometimes with the product resulting in a double bond in a different location. It is appreciated that although bond 'migration' appears to happen, it is not strictly true in molecular orbital terms, although the end product is certainly a double bond in a different location. Also, the PUFAs themselves are mainly all cis configuration. however the oxidised products would be expected to permit a partial change in configuration to the trans isomer so a mixture is expected, for the double bond where a radical is conjugated with the double bond, (trans-isomers are more stable thermodynamically than cis-isomers, because of reduced steric crowding) [26, 27]. Therefore where supplementary table 1 describes non-terminal alkenes, it will be appreciated that both geometric isomers will be expected to be present, although speculation on the precise ratio is beyond the remit of this work.

3.1. Comparison of supplementary table 1, with a review of known VOCs

A previous review [1], listing all the then identified VOCs from the apparently healthy human body, described 1840 VOCs assigned to bodily fluids and breath. A comparison with the study (supplementary table 1) here shows that peroxidation of unsaturated fatty acids can result in the formation of at least 100 of these compounds. More than one biochemical route to these compounds is of course possible. A large number of acids are listed in supplementary table 1, with hydroxy, aldehyde and alkene groups, hydrocarbons, alcohols and a range of aldehydes etc Many are not in the NIST library and also not found in the previous review [1]. There are occasions of the same compound being made by different oxidation routes, e.g. in the case of oleic and palmitoleic acids it can be seen that there are two routes to CHO.(CH2)7.CO2H.

Supplementary table 1 shows homologous series of simple VOCs are present, with occasional breaks in the sequences. These 'gaps' are a consequence of the double bond location, and the length of the unsaturated fatty acid. It may be that other unsaturated fatty acids in the human body will produce more compounds that will 'fill' these gaps. A description of supplementary table 1 is given, in terms of chemical classes:

3.2. Alkanes

Alkanes, from methane to octane were all present in supplementary table 1, CnH2n+2, n = 1–8 in the homologous series, and these also have all been reported in breath and stool, in the review of VOCs from healthy volunteers [1]. There are reports of increased hydrocarbons in several medical conditions e.g. pentane showed some correlation in the breath samples of IBD patients compared to those of the healthy controls [28]. It is interesting that many researchers consider that the source of methane in breath is from the gut as ca. 1 in 3 subjects possess gut methanogens [29]. However lipid oxidation is clearly another potential source. The authors are unaware of any studies undertaken to assess methane lipid origins, in breath, although methane, ethane, propane, butane and pentane have been well described as autoxidation products e.g. from linoleic acid [30]. Straight chain aliphatic hydrocarbons have been considered as non-invasive markers of free-radical induced lipid peroxidation in liver damage, especially breath ethane and pentane, which appear to be better correlated with alcohol induced hepatic injury than to other aetiologies [16]. Several other studies support the involvement of oxidative stress in liver disease such as alcoholic and non-alcoholic hepatotoxicity, infections etc

3.3. Alkenes (one double bond)

Alkenes, from ethene, and propene to decene were all present in supplementary table 1, CH3(CH2)n.CH:CH2, a homologous series from n = 0 to 7. All these compounds were found in the review of volatiles from the human body [1]. The 2-isomers, 2-pentene, 2-hexene, and 2-octene were present in supplementary table 1 and these also have all been reported in breath in the same review. Interestingly a study of IBD patients (in the alveolar air of children) found that the values of three specific VOCs (1-octene, 1-decene and E-2-nonene) could discriminate between IBD and controls [31, 32]. As examples in the literature, ethene has been shown in the volatilome of humans and can be formulated from oxidation of omega-3 acids e.g. linolenic acid, by disproportion of ethyl radicals [33]and 1-pentene has been reported to be generated by decomposition of ω-6 unsaturated fatty acid hydroperoxides e.g. from linoleic and arachidonic acid [34].

3.4. Alkenes (multiple double bonds)

1,3-Butadiene, 1,3-hexadiene and 2,5-undecadiene were found in our study here (supplementary table 1) and in the review [1] the first 2 were found in breath and the latter in skin. In our study here, many alkenes were found with multiple double bonds, e.g. 1,3-nonadiene, 1,3,6-nonatriene, 1,4-heptadiene, 1,4-decadiene, including tetra-enes, penta-enes and hexa-enes. None of these have been reported in the review [1] and in some cases have not been reported with CAS numbers or as being in the NIST library.

3.5. Straight chain alcohols

Alcohols, from methanol to octanol were all present in supplementary table 1, CH3(CH2)nOH from n = 0–7 except for propanol (n = 2) omitted in the homologous series. These were all reported in the review [1], breath has a homologous series of alcohols from methanol to pentanol with more extensive alcohols in faeces.

3.6. Straight chain aldehydes

Straight chain aldehydes from ethanal to decanal, CH3(CH2)nCHO from n = 0–8, were all present in supplementary table 1. These were all reported in the review in breath and faeces [1].

Aldehydes are capable of oxidation to acids, by oxygen, even without the mediation of a catalyst and these aldehydes could contribute to an increase of concentration of carboxylic acids, and a concomitant decrease in aldehyde concentration.

3.7. Dialdehydes

MDA has been reported in supplementary table 1 and is considered an end-product generated by decomposition of arachidonic acid and larger PUFAs, through enzymatic or non-enzymatic processes [35]. It has been extensively reported in association with oxidative stress and has recently been reported in the human volatilome in an article providing irrefutable information on the link between lipid peroxidation and post-operative complications [36].

3.8. Mono-unsaturated aldehydes

16 Mono-unsaturated aldehydes were described in supplementary table 1. Ten were described in the review of volatiles from the healthy human body[1]. Of the other six compounds, three had CAS and NIST numbers, two had CAS numbers with no NIST number and one compound, 6-methyl-2-heptenal, had no NIST or CAS numbers. More recently, 2-decenal and 2-dodecenal (and others) have been reported in breath condensate [37].

3.9. Polyunsaturated hydroxyl aldehydes

27 polyunsaturated hydroxyl aldehyde compounds were described in supplementary table 1, 13 had no CAS and NIST reference numbers and are not reported in the review [1], while 4-hydroxy-2,6,9-dodecatrienal, 4-hydroxy-2,6-dodecadienal and 4-hydroxy-2,6-nonadienal had CAS numbers but no NIST reference or report in the review [1]. Since then 4-hydroxy-2,6-nonadienal and 4-hydroxy-2,6-dodecadienal have been reported in breath condensate [37].

3.10. Monounsaturated hydroxyl aldehydes

Eleven monounsaturated hydroxyl aldehydes were described in supplementary table 1, only seven had CAS numbers, and only two had CAS numbers and NIST numbers: 4-hydroxy-2-hexenal and 4-hydroxy-2-nonenal (4-HNE). None were reported in the review [1], 4-HNE in particular has been extensively reported in association with OS and lipid oxidative breakdown, especially from n-6 PUFAs, mainly arachidonic and linoleic acids [38]. Since then 4-HNE has been reported in breath condensate, including 4-hydroxy-2-octenal, 4-hydroxy-2-decenal and 4-hydroxy-2-undecenal [37]. To further add to the series, 4-hydroxy-2-pentenal has been found in smokers breath using SESI-MS [39].

3.11. Straight chain carboxylic acids

Carboxylic acids, from ethanoic acid to octanoic acid were all present in supplementary table 1, CH3(CH2)n CO2H, n = 0–6 except for pentanoic (n = 3) and hexanoic acid (n = 4) in the homologous series. These were all reported in stool, in the review (1) with just ethanoic and butanoic acid in urine [1]. Inspection of the oxidation of the minor PUFAs, 7,10,13,16,19-docosapentaenoic acid and 8,11,14,17-eicosatetraenoic acid (marine derived lipids), show pentanoic acid and hexanoic acid would also be products, and could become part of the human volatilome by way of the food chain.

Acids can also be biosynthesised in the human body from aldehydes. Aldehyde oxidase (AO) is very concentrated in the liver, where it oxidizes multiple aldehydes [40]. Some AO activity has been located in other parts of the body including the lungs (epithelial cells and alveolar cells), the kidneys, and the gastrointestinal tract (small and large intestines). It should be pointed out that catalysts are not essential, air oxidation can oxidise aliphatic aldehydes into carboxylic acids [41]. A recent report, 2017, showed nonanoic acid presence with statistically significant differences (almost 9-fold) between a lung cancer group and control group [16], nonanoic acid is listed in supplementary table 1, its origins may be due to oxidative stress.

3.12. Multi-functional acids (3 groups)

3.12.1. Alkene acids

Eleven mono alkene acids are listed in supplementary table 1, with a range of carbon lengths and alkene positions, plus a range of multi alkene acids. 9-decenoic and 10-undecenoic acids have been previously reported from skin, and all the mono alkenes have been reported in the NIST library except for 8-nonenoic acid. In contrast, no NIST reports, or presence in the human volatilome have been reported for dienoic, trienoic and tetraenoic acids.

3.12.2. Aldehyde acids(oxo-acids)

They are 116 aldehyde acids are listed in supplementary table 1. This large number is interesting, as is the lack of reported compounds in the literature. Only one compound was in the NIST library, 9-oxonanoic acid, and only ten compounds had CAS numbers. Oxo-heptanoic, oxo-octanoic, oxo-nonanoic and oxo-decanoic acids have been reported in breath, in a paper studying smokers [39].

3.12.3. Alcohol acids

Thirty-seven hydroxy acids are listed in supplementary table 1, of which 32 have no CAS and NIST number and 5 compounds have both CAS and NIST numbers, 2 of which were found in reference 1, 2-hydroxyethanoic acid and 8-hydroxyoctanoic acid. There are a range of quite complicated substituted acids in this section such as 20 aldehyde-hydroxy-alkene acids which have no NIST, or CAS number, and which are not in the review [1].

3.13. Branched chain compounds

Six compounds are listed in supplementary table 1, as regards branched chain saturated aldehydes, the following compounds were described: 3-methylbutanal, 3-methylpentanal, 4-methylhexanal, 4-methylpentanal, 5-methylheptanal and 5-methylhexanal, none were reported in the review [1] and 5-methylheptanal is not in the NIST library. However many branched compounds have been reported in the human volatilome. To rationalize their potential origins, the oxidation of unsaturated branched-chain fatty acids as found in marine animals and microbial lipids, such as, B. cereus were considered, and two branched MUFAs selected as representative examples: 14-methylhexadec-10-enoic acid, 15-methyl hexadec-10-enoic acid and 14-methylpentadec-10-enoic acid [42].

Allylic oxidation of 14-methylpentadecenoic acid results in formation of 3-methylbutanal, which has been previously observed in faeces, urine breath and saliva, which is consistent with microbial origins [43]. Allylic oxidation produces decanoic acid, which has been observed (and can be synthesized by other mechanisms) and 6-methylhex-2-enal. Although the latter has not been reported, the isomer 2-methylhexenal has been observed before in milk [1], with methyl substitution in position 2. Furthermore 4-methylpentanal has been predicted, and although not reported in the review [1], the 2 and 3 substituted isomers have been observed.

3.14. Esters

Although the hydroperoxide oxidation process would not produce esters directly, a very large number of esters could be produced from the precursors produced i.e. from the reaction between alcohols and acids, even 35 esters could be made from the simple straight chain acids and alcohols described above, while many more acids and alcohols are shown in supplementary table 1. A very large number of esters were found in the human body, 213 were reported in the review of volatiles from the healthy body [1], so more discussion of the routes to their origins would be interesting. Work has been reported on the ease in which faeces can be used to synthesise esters from carboxylic acids [44]. Therefore acids and alcohols from oxidation of unsaturated fatty acids and in particular under OS conditions could react with themselves or with other gut acids/alcohols to synthesise a wide range of esters.

3.15. Epoxides and furans

Some epoxide/furans have been reported in the volatilome literature [45, 46], and figures 3 and 4 gives a mechanism to furan syntheses. Mono and di-substituted furans have been reported in the review [1]: 2-acetylfuran and 2-acetyl-5-methylfuran and recently significant differences between prostate cancer and benign prostatic hyperplasia patients were found in urinary levels of certain VOCs, furan (and xylene) both before and after prostate massage, supporting the proposal that VOCs may serve as prostate cancer-specific biomarkers [45].

Monosubstituted alkyl furans, specifically 2-pentylfuran has been suggested [46] to arise from linolenic acid and these can be considered to come from hydroxyl alkene aldehydes, the syntheses of these can be ascribed to double oxidation process of PUFAs.

It is known unsaturated fatty acids can give rise to epoxides, by oxidation with oxygen in the air. This subject hasn't been given much attention as here as epoxides have been rarely found (as yet) in the human volatilome [1].

3.16. MUFAs and PUFAs in the healthy and diseased human body

Products from oxidation of unsaturated fatty acids and in particular OS molecules were theoretically determined from oleic, palmitoleic, linoleic, linolenic, arachidonic and cervonic acids. These acids were chosen because they occur in the human body and, generally, have relative differences in concentrations in the disease state of different organs (albeit that there are gaps in the published literature on this aspect) and also a review of many papers shows that there are differences in lipids (fatty acids) in cancers from different organs.

Bacteria typically contain unsaturated fatty acids such as palmitoleic and linoleic acid and other PUFAs and in situations of plentiful oxygen are capable of similar products of oxidation as mammalian cells, particularly in oxidative stress situations [47]. Compared to mammals, bacteria can contain more unusual branched chain unsaturated fatty acids which will lead to different VOCs and semi-VOCs e.g. branched alcohols, aldehydes and alkanes in contrast to straight chain unsaturated fatty acids, as described for 14-methylpentadec-10-enoic acid and 14-methylhexadec-10-enoic acid in supplementary table 1. It can be considered that some of the branched compounds that have been found as VOCs in the human body (see reference 1), arise due to bacteria in the human body [42–44, 47].

Palmitoleic acid (1 double bond), oleic (single double bond) and linoleic acids (2 double bonds) were chosen for OS production studies as they are the commonest acids in adipose tissue, and in blood plasma [48]. Linolenic acid (3 double bonds) was included, although it occurs at a low concentration, to assess oxidation products from a 3 double bonded PUFA and also because it is one of two essential fatty acids (the other being linoleic acid), so called because they are necessary for health and cannot be produced within the human body. Arachidonic acid (4 double bonds) was the most common in blood plasma, and then the next most common, after this, at a much lower concentration was docosahexaenoic acid (cervonic acid), (6 double bonds).

Even in healthy humans not undergoing oxidative stress due to disease, oxidation occurs due to radical oxygen species occurring, in all cells. Although aerobic mitochondrial oxidative phosphorylation involves the concerted 4e reduction of molecular oxygen to water, it is considered that 1%–3% of molecular oxygen (from the air we breathe), undergoes a stepwise reduction to reactive oxygen species, (superoxide, O2- production is ca. 2 kg yr−1) and patients with infections, may make much more [49]. The products of ROS attacking MUFAs and PUFAs results in many products, as discussed here, and although some molecules may stay contained in the cell, some will diffuse out into the blood stream and hence to the many organs, and then excreted by breath, urine, stool, and sweat etc (along with their further metabolized daughter bi-products. Given that molecular oxygen is required to sustain aerobic metabolism, aerobic organisms have developed sophisticated protective mechanisms to defend against 1-e intermediates, however oxidised products will still be formed. VOC analyses could have a place for subtle studies to locate the origins of OS within the body and the OS phenomenon could well be exploited for disease diagnoses e.g. cancer by measuring/identifying volatile OS products as they are amplified numerous times due to the chain reaction/propagation processes.

As stated the chain reaction nature of OS means there is a magnifying of changes, so an initial localised event can trigger VOCs from many areas, thus complicating the assessment of the initial origins of OS; nevertheless it will flag that the body is in a diseased state. For instance, gastric mucosal injury is known to occur in patients with extensive burns [50] i.e. local injury to cell membranes and lipid peroxide production can affect even remote organs.

Numerous aldehydes can be observed to be synthesised in supplementary table 1, some previously reported in the volatile compound literature, some not. Compared with free radicals, aldehydes are relatively stable and can diffuse within or even escape from the cell and attack targets far from the site of the original event. They therefore are not only end products and remnants of lipid peroxidation processes but also may act as 'second cytotoxic messengers' for the primary reactions [51].

Some of these aldehydes have been shown to exhibit facile reactivity with various biomolecules, including proteins and DNA generating stable products at the end of a series of reactions that are thought to contribute to the pathogenesis of many diseases. Concentrations of 22 known aldehydes (by-products of lipid peroxidation), before the initiation of chemotherapy, have been reported to be significantly elevated in cancer patients compared to controls [52]. Aldehydes such as hexanal, heptanal, and MDA were strikingly higher in samples from cancer patients. In addition, in each form of cancer the pattern of aldehydes appeared to be unique when compared to controls, or to others forms of cancer.

During heavy oxidative stress, e.g. in patients with severe rheumatologic diseases such as rheumatoid arthritis, systemic sclerosis, lupus erythematosus, chronic lymphedema, or chronic renal failure, serum 4-HNE is increased to concentrations up to 3- to 10-fold higher than physiologic concentrations. 4-HNE (and MDA) have been nominated as major final products of lipid peroxidation [53].

Phillips et al evaluated exhaled breath samples from patients who received a heart transplant [54]. The study suggested that several hydrocarbons related to lipid peroxidation found in breath, could be used as biomarkers of heart transplant rejection supported by a more recent study by Pabst et al [55] with respect to pentane. The breath of angina patients were also found to be associated with 4-methyloctane, 4-methyldecane, hexane, 5-methylpentadecane, 7-methylhexadecane, 2-methylpropane, pentane, 2-methybutane [56]. There are a very large number of hydrocarbons in the environment and there is the possibility of erroneous correlations. This paper describes the range of other products expected to be synthesised apart from hydrocarbons and would strengthen the hypothesis of oxidative stress and VOCs if they could be identified.

In terms of gastro-intestinal disease, Crohn's disease or ulcerative colitis is characterized by acute intestinal inflammation of the mucosa and the intestinal lumen. Overproduction of ROS with subsequent lipid peroxidation has been proposed as one possible mechanism for these diseases. Pelli et al [57] showed that exhaled breath from patients with Crohn's disease and ulcerative colitis contains elevated levels of ethane, propane and pentane when compared to controls. At the same time, isoprene and butane had the same concentrations as in healthy volunteers. The results were only partially confirmed by Sedghi et al [58] who showed that ethane, but not pentane levels, were significantly increased in patients with ulcerative colitis and correlate with disease activity. Lipid peroxidation has been proposed repeatedly in the pathophysiology of IBD, and breath alkanes have been studied as a measure of lipid peroxidation and have been correlated with disease activity [59, 60]. In an initial animal study, Ondrula et al [61] showed increased exhalation of pentane by rats after induction of colonic inflammation. Pentane levels in exhaled air rapidly normalized with the resolution of inflammation. This was later supported by a human study undertaken by Kokoszka et al [60] who demonstrated a good correlation of breath alkanes with IBD activity. In general, it has not been established that increases in the levels of reactive oxygen species are the cause of any particular disease. It is possible that increased concentrations of reactive species could be the host response to the onset of disease. Therefore the elevation of reactive oxygen species would be a generic response to any disease process, the VOCs from humans may give more specificity.

This paper focuses on VOCs and oxidation of PUFAs, which are of interest to the scientific community working on volatiles and disease diagnoses. There will undoubtedly be some compounds not included, and compounds which arguably should not be included, depending on the definition of what is a VOC, or semi-VOC; there are more papers than can be read in a lifetime in the general area of oxidative stress and diseases.

4. Conclusion

This study has resulted in a large table of compounds of volatile/semi-volatile compounds generated by consideration of oxidation of a range of unsaturated fatty acids (299 in total). These compounds have been correlated with compounds in the previously published review summarizing the volatiles from the human body [1], satisfactorily explaining the origins of many VOCs from humans which hitherto have not been recognised. It would be comforting if many of the papers on correlation of VOC biomarkers with disease had firm biochemical origins. From the use of tables herein it can be determined that many VOCs have a plausible unsaturated fatty acid origin.

While many of the OS molecules in this paper have been reported, many have not. Are they in the volatilome and simply have not been detected yet? There could be several reasons for this, they may not appear in the NIST library, which is certainly the case for many compounds described herein, or they simply are not volatile enough (and/or below limit of detection). There is even the possibility that they are not present, although this would not seem likely.

There can be differences in the unsaturated fatty acids in different organs and peroxidation of these may be able, in selected cases to give a diagnostic organ specific VOC fingerprint. Furthermore, there is evidence of up and down regulation of lipoxygenases in cancers, monitoring of these is difficult as an invasive biopsy would be needed, however the action of these up and down regulated enzymes would be expected to produce characteristic VOCs, potentially monitorable in breath and bodily fluids. This aspect requires more scrutiny, as there are many sources of VOCs in the human body. Future work will be targeted at the detection of predicted metabolites in patients with conditions likely to have changed levels of VOCs/ semi-VOCs due to oxidative stress.

Acknowledgment

The University of the West of England financially supported this work.

10.1088/1752-7163/ab7f9d