EPA/600/A-97/042
Proceedings of healthy buildings '95
Mixtures of Volatile Organic Compounds: Detection of Odor, Nasal
Pungency, and Eye Irritation
J, Enrique Cometto-Muniz and William S. Cain
Chemosensory Perception Laboratory, Department of Surgery (Otolaryngology),
University of California, San Diego, La Jolla, CA 92093-0957, USA.
INTRODUCTION
Human beings can typically sense, in sufficient concentration, almost any organic
vapor via smell, though some substances have lower thresholds than others by as many as
nine orders of magnitude (see Cain, 1988; Devos, Patte, et al., 1990). The reasons for this
spread in potency remain surprisingly obscure. In part, it seems due to physicochemica!
properties, such as solubility. In aliphatic series, for example, more lipid- soluble members
generally have greater potency (Cometto-Muniz and Cain, 1994). This holds only up to a
certain point in the series, where potency shows either no further increase or an actual
decrease. Such an outcome suggests that, in addition to solubility or some correlated
property, molecular size may also determine odor potency. That is, even with sufficient
vapor pressure to become airborne in quantity, molecules can Be too big to stimulate
effectively.
At some concentration, typically above the odor threshold, organic vapors may also
trigger chemesthesis, i.e., may be felt rather than smelled. The chemesthetic modality or
common chemical sense (CCS) is mediated in the mucosac of the face by the trigeminal
nerve. The difference in concentration between when a substance can be smelled and when
that same substance can be felt depends in part upon its reactivity Towards mucosal tissue.
Aggressive vapors, such as formaldehyde, acrolein, ozone, and others, may damage tissue
and thereby be felt. In these cases, the vapors often have chemesthetic thresholds in the
rough vicinity of their odor thresholds. For formaldehyde, for example, nasal irritation
occurs at a concentration only a few times higher than the odor threshold (Berglund and
Shams Esfandabad, 1992).
Most organic vapors in the environment do not react readily with mucosal tissue,
yet these too may have some ability to stimulate the chemesthetic sense. In such cases, the
concentration required for detection via feel lies at least an order of magnitude, and in some
instances many orders of magnitude, above that required for detection via odor. In recent
investigations, we have charted the difference between olfactory and chemesthetic
thresholds for various aliphatic series of chemicals (Cometto-Muniz and Cain, 1990;
Cometto-Muniz and Cain, 1991; Cometto-Muniz and Cain, 1993; Cometto-Muniz and
Cain, 1994). In general, the chemesthetic outcome of interest was nasal pungency, though
171
-------
in some cases it was ocular pungency (eye irritation) (Comctto-Mun^ and Cain, 1995). As
with odor potency, chcmesthetic potency increased with chain length, though not in exactly
(he same way. Chemesthetic potency increased somewhat more predictably. For example, it
increased with chain length in much the same way from one series to another. In this
behavior, chemesthesis mimicked phenomena such as narcosis and anesthesia that these
same organic vapors can induce under appropriate conditions.
Our results have also implied that, to a first approximation: a) chemesthctic
detection via the eye occurs at about the same concentration as chemcsthetic detection via
the nose, and b) the threshold of detection via feel occurs at the same level of saturated
vapor concentration irrespective of the substance or the series it comes from. Decades ago,
Ferguson (Ferguson, 1939; Ferguson, 1951) articulated a principle that various phenomena,
e.g., anesthesia, occurred at fixed percentages of saturated vapor, or, as he expressed it, at
criterion levels of thermodynamic activity. This in turn suggested an equilibrium between
concentration in the vapor phase and concentration in the biophasc where biological
activation occurred. Insofar as a biological effect takes place at a criterion percentage of
saturated vapor only for members within a particular scries, it is not sufficient to imply the
veracity of the Ferguson principle. Insofar as the effect takes place at a criterion level
across all series, it does. For chemesthesis, it seems to hold roughly (Abraham, Andonian-
Haftvan, Comctto-Muniz, and Cain, 1995).
Nielsen and associates (Nielsen and Alarie, 1982; Nielsen, 1991) argued for a
protein receptor as the site of transduction for chcmesthetic responses to nonrcactive
irritants, though they admitted that such compounds might exert their action through
nonspecific physical adsorption. It is tempting to speculate that the rules of chemesthesis
indicated in our various studies imply stimulation without need to invoke a receptor.
Certainly, high quantities of organic chemicals can have direct effects on membranes. For
example, high quantities can change fluidity and can allow ion flux across a membrane.
Such phenomena can generally account for narcosis and anesthesia, two of the very
biological properties seen as analogous to sensory irritation. Regarding olfaction, recent
studies point towards the existence of many protein receptors (Buck and Axel, 1991).
Olfaction distinguishes itself from chemesthesis in remarkable variation in sensory quality
such as skunky, fishy, floral, minty, ethereal, and so on. The many receptors of olfaction
presumably permit the discriminations upon which such qualitative variation is based.
Chemesthesis exhibits only trivial qualitative variation in comparison.
A quantitative structure-activity relationship (QSAR) based upon a linear solvation
energy relationship (LSER) has described the relative potency of organic vapors for a
number of biological effects (e.g., anesthesia, respiratory depression) in animal models
(Abraham, 1993). Abraham and colleagues have demonstrated that a five-parameter
equation that describes the biophasc as a solvent for impinging vapors can account for 97%
or more of variance in potency (Abraham, Whiting, ct a!.. 1990; Abraham, 1993; Abraham.
Nielsen, ct al., 1994). A roughly analogous four-parameter solution of an LSER for nasal
irritation in humans has accounted for 96% of the variance (Abraham, Andonian-Haftvan.
etal., 1995).
Our cxpcrimcnial strategy has entailed testing subjects diagnosed (Cain, I9S9) as
anosmic (i.e., lacking olfaction) to chart thresholds for nasal pungency independently from
smell. We also test normosmic subjects (i e., persons with normal olfaction) matched' to the
anosmics in age, gender, and smoking-status to map corresponding odor thresholds. In the
present extension of our series of studies, we explored thresholds for odor, nasal pungency,
and eye irritation for mixtures of the relatively nonrcactive compounds studied previously
individually.
Most studies of mixtures of odoriferous vapors have examined suprathrcshold
perception via ratings of intensity (e.g., Olsson, 1994). Such studies have found that the
perceived intensity of mixtures falls short of the sum of the intensities of their (unmixed)
components (e.g., Bcrglund and Olsson, 1993). Perceived intensity of suprathreshold
mixtures also falls somewhat below predictions made from "addition" of concentration but
172
-------
much closer to it than addition of sensory magnitude (Cain, Schict, et al. 1995).
Suprathreshold experiments with pungent odorants at concentrations that clearly appealed
also to the nasal CCS yielded a concentration-dependent degree of addition of perceived
intensity (Comctto-Muniz, Garcfa-Medina, ct al., 1989). Whereas suprathreshold odor in
such mixtures displayed hypoadditivity of sensation, nasal pungency displayed additivity or
even hyperadditivity (Cometto-Mufiiz and Hemdndez, 1990).
Exploration of differences in "additivity" between odor and pungency at
suprathreshold levels has had no counterpart at the threshold level. Few investigations
have explored potency of odor in mixtures, and none has apparently explored potency of
nasal pungency in mixtures. The outcomes for odor have uncovered simple stimulus
agonism (Guadagni, Buttery, et al., 1963; Patterson, Stevens, et al., 1993) and some
evidence of synergistic stimulus agonism, particularly for mixtures of three or more
components (Rosen, Peter, et al., 1962; Baker, 1963; Laska and Hudson, 1991). Simple
agonism in, for example, a balanced three-component mixture implies that when the
substances are presented mixed, each needs to be at only 1/3 its individual threshold
concentration for the mixture to be perceived. Synergistic agonism implies — following
the same example of a three-component mixture — that individual substances need to be at
a concentration even lower than 1/3 of their respective thresholds for the mixture to be
perceived; whereas partial agonism implies that they need to be at a concentration higher
than 1/3 of their respective thresholds but less than the threshold value itself. Independence
implies that at least one of the components in the mixture needs to be at its individual
threshold for the mixture to be perceived. Finally, antagonism implies that the componenls
need to be at concentrations higher than their individual thresholds for the mixture to be
perceived.
In view of earlier research in olfaction and in view of the success of an LSER to
predict nasal pungency, we expected both olfaction and chemesthesis to show simple
agonism with some possibility of an increase in agonism toward synergy as the mixtures
increased in number of components.
METHODS
Stimuli
The stimuli included members of homologous series of alcohols (1-propanol, 1-
butanol, 1-hexanol), esters (ethyl acetate, hexyl acetate, heptyl acetate), ketones (2-
pentanone, 2-heptanone), and alkylbenzenes (toluene, ethyl benzene, and propyl benzene).
All were analytical-grade reagents. Single chemicals tested for the three relevant sensory
responses of odor, nasal pungency, and eye irritation were: 1-propanol, 1-hexanol, ethyl
acetate, heptyl acetate, 2-pcntanone, 2-heptanone, toluene, ethyl benzene, and propyl
benzene. Mineral oil served as solvent to prepare three-fold dilution steps of the pure (100
%v/v) substance (i.e., 33, 11, 3.7, 1.1, etc., %v/v). Due to the limited solubility of 1-
propanol in mineral oil, the first two members of its series, 33 and 11 %v/v, were prepared
in deionizcd water.
Five mixtures were prepared using mineral oil as solvent: two 3-component
mixtures (labeled A and B, below), two 6-component mixtures (labeled C and D), and one
9-component mixture (labeled E). Mixture A comprised: l-propanol, ethyl acetate, and 2-
pcntanone. Mixture B comprised: 1-hexanol. heptyl acetate, and 2-heptanone. Mixture C
comprised: 1-propanol, 1-butanol, ethyl acetate, 2-pentanone, toluene, and ethyl benzene.
Mixture D comprised- 1-hexanol, I-heptanol, hexyl acetate, heptyl acetate, 2-heptanone,
and propyl benzene. Mixture E comprised: 1-propanol, 1-hexanol, ethyl acetate, heptyl
acetate, 2-pentanone, 2-heptanone, toluene, ethyl benzene, and propyl benzene. These
particular mixtures were chosen to give one 3-component (i.e., A) and one 6-component
('•e., C) mixture of relatively low molecular weight, high vapor pressure chemicals; one 3-
173
-------
component (i.e., B) and one 6-component (i.e., D) mixture of higher molecular weight,
lower vapor pressure chemicals; and one 9-component mixture (i.e., E) having both kinds
of chemicals.
Substances in each liquid mixture were present in proportions thai reflected their
odor thresholds measured in our previous studies (Cometto-Muniz and Cain, 1990;
Cometto-Muniz and Cain, 1991; Cometto-Muniz and Cain, 1993; Cometto-Muniz and
Cain, 1994). This was done by using a "reference" that contained, for each liquid mixture,
components at their individual odor thresholds as found previously. For example, the odor
thresholds for 1-propanol, ethyl acetate, and 2-pentanone equaled 0.0051, 0.0017, and
0.0017%v/v, respectively, and the reference for this three-component mixture therefore had
these concentrations. Based on such a reference, threefold steps in concentration were
prepared for each mixture both above and below the reference. Accordingly, the first step
for mixture A above the reference was: 0.015 (0.0051 X 3), 0.0051 (0.0017 X 3), and
0.0051 (0.0017 X 3)%v/v, respectively, for 1-propanol, ethyl acetate, and 2-pentanone. The
first step below the reference was: 0.0017 (0.0051 -K 3), 0.00056 (0.0017 H- 3), 0.00056
(0.0017 •*• 3)%v/v, respectively, for the same chemicals in the same order. Steps above the
reference continued until they reached the maximum value that could be presented in a
mixture of the specified proportions (e.g., for mixture A: 33, II, and 11 %v/v cf 1 -propanol,
ethyl acetate, and 2-pcntanone, respectively). Steps below the threshold continued until
they fell definitively below the odor threshold for even the most sensitive subject.
We prepared duplicate series for each of the nine single chemicals and each of the
five mixtures. Stimuli were delivered from cylindrical, squeezable polyethylene bottles
(270 ml capacity) (Cain, 1989) containing 30 ml of solution. For measurements of odor and
nasal pungency, the bottle closure had a pop-up spout that fit into the nostril being tested.
Each nostril was tested separately. For measurements of eye irritation, the bottle caps held a
tube that led to a 25-ml measuring chamber (of the type used in variable volume
dispensers), the rim of which was placed around the eye. Each eye was tested separately.
The tube that fed the chamber was connected to the headspace of the bottle. A squeeze of
the bottle delivered a puff of vapor into the measuring chamber where the eye was exposed.
A polyethylene dust cover closed the open end of the measuring chamber when the bottle
was not in use.
The concentration of each compound in the headspace of every bottle was measured
by a gas chromatograph (photoionization detector) equipped with a gas sampling valve,
allowing direct sampling of the headspace. For every single or mixed stimulus, repeated
chromatographic readings were taken at each dilution step. Readings were also taken from
bottles containing pure chemicals (100 %v/v). The headspace of a bottle with a pure
chemical contains vapor saturated with the chemical at room temperature (23 °C). The
concentration of the saturated vapor from each substance was retrieved from the literature.
Knowledge of saturated vapor concentration (at 23°C) and its associated average
chromatographic reading allowed conversion of the readings from the other bottles into
concentration units (ppm by volume), and derivation of a calibration curve.
Subjects
Eight subjects, four males and four females, provided values for odor and eve
irritation thresholds. The subjects covered a wide range of age (21 to 60 years) in order to
match the group of anosmics available for testing nasal pungency thresholds. Participants
included a male and a female in each of the following categories: early twenties, early
thirties, early forties, and late fifties/sixty. A 21 year-old male became unavailable after
being tested for odor thresholds. In order to complete the group, he was replaced by a 22
year-old male tested only for eye irritation thresholds.
Four anosmic subjects, two males and two females, provided values for nasal
pungency thresholds. One participant was a head-trauma anosmic (male, 66 years old), the
174
-------
other three were congenital anosmics (a male, 43 years old, and two females, 20 and 62
years old, respectively),
Procedure
On odor and nasal pungency trials, participants delivered the stimulus by inserting
the pop-out probe inside the specified nostril and squeezing the bottle as they sniffed. On
eye irritation trials, they squeezed the bottle while the eye was exposed in the measuring
chamber. Subjects quickly learned to squeeze and sniff with equal vigor on every trial.
We used a forced-choice ascending method of limits to measure threshold. The
subject had to choose the stronger smelling or stronger feeling of two stimuli. One was a
blank of mineral oil and, at the start, the other was a high dilution-step, low concentration
of the stimulus, either single chemical or mixture. If the choice was correct, testing
continued with the same step from the duplicate set, also paired with a blank. If the choice
was incorrect, testing continued with the next step — a liquid-phase concentration three
times higher — paired with a blank. In this way, correct choices entailed the presentation of
the same concentration, whereas errors triggered step-wise increments in concentration.
The procedure continued until five correct choices were made in a row, in which case that
step was taken as threshold. Once the threshold was measured for one nostril or eye, the
other nostril or eye was tested. After this, testing began again with another stimulus. Both
(he ascending-concentration approach to the threshold and the separate testing of each
nostril helped to minimize effects of adaptation, frequently encountered in olfactory
investigations (see Cometto-Muniz and Cain, 1995).
Sessions typically lasted between one and two hours and were repeated until eight
thresholds per subject, four for each nostril or eye, were obtained for each single compound
and each mixture. This equaled a total of 64 odor or eye irritation thresholds per stimulus
and 32 nasal pungency thresholds per stimulus.
Data analysis
Geometric means served to summarize threshold concentrations within and among
subjects. The geometric mean acknowledges that threshold data for chemosensory stimuli
conform to log normal distributions (Brown, Mac Lean, et al., 1968; Amoore, 1986; Gain
and Gent, 1991).
For analysis of agonism in mixtures, consider the case of a three-component
mixture with a formulation that reflects perfectly a subject's relative sensitivity to its three
components unmixed. Such a mixture would be perfectly balanced. If the components
were simply to add their individual effects, detection would occur when each component
fell at one-third the concentration of its unmixed threshold. Hence, if component A had a
threshold of 3 ppm, component B of 9 ppm, and component C of 27 ppm, and if the
mixture were made up in the proportions 1:3:9, respectively, for balance, then we would
expect detection for the mixture to occur with component A at 1 ppm, component B at 3
ppm, and component C at 9 ppm.
The relative contribution of components in a just-detectable, three-component
mixture can be reflected in the formula: WS - a(RA) + b(Ru) + c(Rc), where WS stands
for weighted sum; RA = concentration of A in mixture •*• threshold A; RB = concentration
of B in mixture -s- threshold B; RC — concentration of C in mixture * threshold C; and with
a, b, and c as weighting coefficients to reflect the degree of balance in the mixture. These
weighting coefficients in our three-component mixture are defined as follows: a = [RA/(RA
+ RB + RC)] * 3; b = [RB/(RA + RB + Rc)l * 3; c = [RC/(RA + RB + RC)] * 3. The
second factor (3 in our case) of each coefficient represents the number of components in the
mixture.
175
-------
In our example, WS = 1(1/3) + 1(1/3) +1(1/3) = 1, which implies complete
agonism of stimulating power. In an unbalanced mixture, which occurs more commonly
than not because no single subject has exactly the same sensitivity as the reference group
used to formulate the mixture, a, b, and c will take on different values, but will always add
up to 3 for a three-component mixture, to 6 for a six-component mixture, and to 9 for a
nine-component mixture. Irrespective of the relative weights of the components, whenever
WS lies at 1.0 simple agonism will have held. Whenever WS lies significantly above or
below 1.0, then departure from simple agonism will have held, as follows: 1) WS above 1.0
implies that the components do not add their sensory potency completely when mixed
(partial agonism or even antagonism) and 2) WS beiow 1.0 implies that the components
have gained sensory potency when mixed (synergistic agonism).
to"
10*
D • 'A
D
D
Q .
*
A
i.
•
*
A •
. A
a
a .
' 0
• u
• A
a
a
D D
&
. 1
A
a a • •
. ' A A
" D
a .
•
" A
• *
D D
" ' O
d
M
A A
' •
A*
0 D
O O
A
3 D
a
I
• •
A
0 D
a
8
* •
. &
o •
* ° a
e
a.
c
t-
10' i
10
10"
« «*-•«. «>*. t. c -* Ml TO
£ X ? >• 1 a •- > >
«• , x: « ® e £a
-"•Sg-^-i iSg
X « « a
Figure 1. Thresholds (ppm ± SD) for nasal pungency (filled squares), eye irritation (triangles), and odor
(empty squares). The SD is indicated by the dots. The nine sections of the graph correspond lo nine
substances. Each section lists, first, the threshold for the substance by itself (e.g., 1-propanol), then,
consecutively, the level at which that substance was present when the threshold was achieved for mixtures
of increasing complexity, e.g., l-propanol in mixture & (3 components) when A achieved threshold, I-
propanol in C (6 components) when C achieved threshold, 1-propane! in E (9 components) when E
achieved threshold.
RESULTS
Figure 1 presents thresholds (ppm by volume) for all three sensory responses
plotted for each of the nine substances. The graph is divided in nine sections one per
substance. Within a section, the first value corresponds to the threshold (odor, eye irritation,
and nasal pungency) of the substance when presented alone. Subsequent thresholds
correspond to the concentration at which the substance was present when that particular
176
-------
mixture reached threshold (e.g., l-propanol in A corresponds to the level of 1-propanol
found in mixture A when the mixture achieved threshold). A common trend in the three
sensory responses and all chemicals is for thresholds to decline with increasing number of
components in the mixture. This indicates that some degree of stimulus agonism is taking
place in the mixtures to help precipitate odor, eye irritation, and nasal pungency when each
component is below its individual sensory threshold. It should be pointed out that pungency
thresholds for hcptyl acetate and propyl benzene presented alone could only be measured in
two to four of eight repetitions per subject (depending on the anosmic). Probably as a result
of this, the pungency threshold for mixture D could only be measured in two to five of the
eight repetitions. Hence, values given for the pungency of these two chemicals and of
mixture D do not represent the average of all anosmics on all repetitions, but do represent
the average of those cases where a threshold was obtained.
Table I.
Weighted sum (WS) of ratios of mixed thresholds/unmixed thresholds.
Odor
Subject 1
2
3
4
5
6
7
8
Geo. Mean
Eye Irritation
Subject I
2
3
4
5
6
7
8
Gco, Mean
Nasal Pungency
Subject 1
2
3
4
Geo. Mean
1.0
4.9
6.4
3.7
2.6
5.9
5.0
2.8
3.6
2.2
1.6
1.9
1.9
1.0
2.1
3.3
2.0
1.9
2.1
2.2
1.3
2.4
1.9
Three Components
Mixture A Mixture B
6.6
2.8
1.9
3.0
6.3
1.4
0.3
4.7
2.5
6.0
0.9
1.4
1.6
0.6
1.1
2.2
4.6
1.7
0.7
2.3
2.0
0.9
1.3
Six Components Nine Components
Mixture C Mixture D Mixture E
0.9
6.3
3.7
2.3
6.5
2.0
7.2
4.2
3.4
5.8
7.0
6.6
6.7
6.S
1.8
1.6
0.4
3.2
1.4
0.6
2.3
1.1
1,3
3.1
2.0
3.0
2.6
2.9
1.4
2.6
i.l
2.2
0.6
0.4
0.2
0.6
0.4
0.03
0.2
0.1
0.2
0.8
0.5
1.0
1.1
0.8
IA
1.3
1.2
2.0
1.8
1.0
2.0
3.0
1.6
O.S
0.5
0.6
0.6
0.7
0.3
0.9
0.3
0.6
1.7
1.5
1.5
1.8
1.6
Table 1 shows, subject-by-subject, values of WS from the formula given in the
section on data analysis. For averages across subjects, 12 out of 15 cases yielded WS's
above 1.0 which implies a general tendency for components to act, at most, as partial
177
-------
agonists in mixtures. This varied, however, both with sense modality and with complexity
of a mixture.
To compare odor with eye irritation, an analysis of variance (ANOVA) was
performed on the logarithm of the values in Table I with the two variables modality (two
levels: odor vs eye irritation) and mixtures (five levels) in a repeated-measures design
(recall that seven subjects were common to all measures, with the eighth subject in each
group matched to a counterpart in the other). The results revealed a significant difference
between modalities (F[l,7]=25.86, p=0.001), with eye irritation showing higher stimulus
agonism; a significant difference among mixtures (F[4,28]=14.01, p<0.00005), with the
more complex mixtures tending to show greater agonism; and a significant interaction of
modality by mixture (F[4,28]=4.37, p=0.007) which reflected the tendency for complexity
to have more leverage for eye irritation.
ANOVA on the same values, but excluding mixture E, which contained substances
of both low and high lipophilicity gave a view of the variables modality, complexity of
mixture, and lipophilicity. The results again revealed significant differences for modality
(F[l,7]=18.98, p=0.003), as well as for three vs six components (F[l,7]=9.38, p=0.02), with
six showing greater agonism, and for lipophilicity (F[l,7)=l 1.18, p=0.01), with the
mixtures (B and D) of the more lipophilic substances snowing greater agonism. The two-
way interaction of number of components by lipophilicity and the three-way interaction
were also significant (F[I,7]=23.97, p=0.002 and F[ 1,7]=8.71, p=0.02, respectively).
Table 2. Results of 95% confidence intervals on log WS. Expected value for agonism = 0, i.e., log of 1.0.
Modality Mixture n Mean S.D Result
Odor
Eye
Irritation
Nasal
Pungency
A
B
C
D
E
A
B
C
D
E
A
B
C
D
E
8
8
8
8
8
8
8
8
8
8
4
4
4
.4
4
0.551
0.390
0.532
0.115
0.207
0.281
0.238
0.344
-0.645
-0.261
0.285
0.113
0.813
-0.093
0.203
0.263
0.436
0.317
0.296
• 0.151
0.145
0.343
0.167
0.444
0.180
0.124
0.246
0.035
0.157
0.046
Partial agonism
Partial agonism
Partial agonism
Agonism
Partial aponism
Partial agonism
Agonism
Partial agonism
Syncrgistic agonism
Svncrgistic agonism
Partial agonism
Agonism
Partial agonism
Agonism
Partial agonism
Two ANOVAs were analogously performed on log WS for nasal pungency. The
first was a one-way ANOVA that showed significant differences among mixtures
(F[4,12]=18.40, p<0.00005). The second was a two-way ANOVA that again excluded
mixture E and showed significant differences for lipophilicity (F[l,3]=50.26, p=0.006),
with the more lipophilic substances showing greater agonism, and for the interaction
178
-------
number-of-components by lipophilicity (F[1,3]=I2.10, p=0.04), but not for number of
components.
Ninety-five percent confidence intervals around the log WSs allowed us to
determine if the values for each mixture and modality differed significantly from 0 (simple
agonism), i.e., if WS differed significantly from 1.0. The results, shown in Table 2,
reinforce the picture that mixtures increase their degree of stimulus agonism with
increasing number of components and increasing lipophilicity of components. Eye irritation
even reached syncrgistic agonism for the most lipophilic (D) and the most complex (E)
mixtures.
DISCUSSION
Studies of odor mixtures at threshold levels have been very few and have generally
entailed no vapor-phase calibration of the stimuli, typically liquid solutions of odorant
presented in open containers. Without such calibration, the data could in any instance
reflect departures from Raoult's law (proportionality between the vapor-pressure of a solute
and its mole fraction in solution) as readily as they do the biological rules of how humans
detect mixtures.
A study of odor thresholds for three chemicals (1-butanol, p-cresol, and pyridine)
relevant to off-flavors in drinking water yielded simple agonism most commonly in binary
mixtures and synergistic agonism in the ternary mixture (Rosen, Peter, et al., 1962). A
study of food-related chemicals by Guadagni et al. (Guadagni, Buttery, et al., 1963) also
yielded simple agonism among components in all but one of 20 mixtures composed of two
to 10 components. The components included various saturated and unsaturated aliphatic
aldehydes, an alcohol, a carboxylic acid, a sulfide, and an aminc. Another study motivated
by the flavor of drinking water yielded synergistic agonism for many binary mixtures of
eight starting materials (m-cresol, pyridine, 1-butanol, acrylonitrile, n-amyl acetate, 2,4-
dichlorophenol, acetophenone, n-butyl mcrcaptan) and for the eight-component mixture
(Baker, 1963). The investigation yielded antagonism for the pair m-cresol—acetophenone.
More recently, Laska and Hudson (Laska and Hudson, 1991) measured odor
thresholds for various single chemicals (isoamyl acetate, alpha-pinene, cyclohexanone,
cineole, linalool, (-)-carvone, t-butylcyclohexyl acetate, mcthylpropylkctone, and dccyl
acetate) and for three-component mixtures. The general outcome suggested simple-to-
synergistic agonism.
The trend of the data therefore lay towards simple-to-synergistic agonism, with
greater synergism for more complex mixtures. When, in a recent investigation, we studied
a psychophysically balanced, rigorously-calibrated ternary mixture of 1-butanol, 2-
pentanone, and n-butyl acetate, we also obtained simple agonism (Patterson, Stevens, Cain,
and Cometto-Muniz, 1993). Hence, when the mixture achieved odor threshold, the
headspacc concentration of each component was approximately one third that of the
individual odor threshold of the component. This outcome lent credence to earlier results
with non-calibrated stimuli insofar as it indicated that the olfactory system could display
simple agonism.
The present data argue for partial as opposed to simple agonism, though degree of
agonism tended toward simpte as the number of components increased. Differences among
stimuli and among people (see Table 1) both need further attention in the quest for
generalities. For at least the present stimuli, it seems reasonably certain that irritation
exhibits more agonism than odor and that eye irritation exhibits the most. Lipophilicity
seems important in mixtures for both modalities.
We have yet to develop any theory to explain why agonism would increase with
complexity. Theory might need to await definitive dismissal of the possibility that the
tendency for agonism to increase with complexity results from a complexity-dependent
vulnerability towards imbalance in mixtures, whether for odor or irritation. For simple
mixtures (e.g., two or three components), small errors of measurement could mimic just
179
-------
about any outcome, from synergistic agonism to antagonism. Scrutiny of our data revealed
no significant correlation between degree of imbalance, measured as root-mean-squared
deviation from perfect balance, and degree of agonism. Nevertheless, the matter merits
continued attention in the choice of conditions in future investigations. Despite the penalty
of additional work, complex mixtures will reveal more about agonism than will simple
mixtures.
As if to add insult to injury, the proper study of agonism should also include
variation in the proportions of components in mixtures. Measurements should therefore
include psychometric (stimulus-response) functions for each component unmixed and
against the many backgrounds of various proportions of the components. A ten-component
mixture, studied in a few subjects, could consequently entail millions of sensory and
analytical measurements. Would it be worth Ihe effort of a few person-years of work to
make such measurements in a strategically-chosen mixture. In our view, it would, both for
theoretical and practical understanding. We see our present work and our work on single
chemicals chosen -from homologous chemical series as relevant to the choice of. that
strategic mixture.
What are the practical implications of what we know now for human detection of
VOCs in indoor air, where scores may be present simultaneously? In any such mixture,
the more lipophilic substances will have much more weight as sensory stimuli. The
LSER developed by Abraham, Andonian-Haftvan, Comctto-Muniz, and Cain (1995)
gives quantitative predictions of exactly how much. In mixtures, lipophilicity takes on
even more importance, for it may increase the "gain" in detection, but agonism would in
any case reduce the amount by which any single substance needed to be present to
contribute to detection of irritation or odor. Conservatively predicted, the "gain" from
complexity would be just proportional to the level of complexity. In that case, a 100-
component mixture, balanced for the effect of its components, would require 1/100th of
the threshold of each component to cause a sensory effect. The possibility thai
complexity-dependent agonism may actually lead to synergistic agonism, no matter what
the sensory channel, would make the prediction of a 100-fold gain conservative possibly
by orders of magnitude. In reality, though, mixtures may be dominated by just a few
components and until we know just how agonism operates in such unbalanced cases via
the study of varying proportions we will be left with an uncomfortable degree of
uncertainty.
Summary
Thresholds for detection of odor, nasal pungency (irritation), and eye irritation were
measured for single volatile organic compounds (VOCs) (I-propanol, 1-hexanol, ethyl
acetate, heplyl acetate, 2-pentanone, 2-heptanone, toluene, ethyl benzene, and propyl
benzene) and certain mixtures of them (two three-component mixtures, two six-component
mixtures and one nine-component mixture). Nasal pungency was measured in persons
lacking a functional sense of smell to avoid interference from olfaction. The results showed
the existence of various degrees of stimulus agonism (additive effects) in the three sensory
channels. Such agonism increased with the complexity of the mixtures and with the
lipophilicity of their components. Eye irritation even showed synergistic stimulus agonism
for the most lipophilic (one of the six-component) and the most complex (the nine-
component) mixtures. The results indicate that complex chemical environments may
enable chemosensory, and particularly irritative, detection when single VOCs lie far below
their individual thresholds. Even the present rather rudimentary state of knowledge of how
lipophiticity influences detection of single VOCs and how it and complexity (number of
components) influence detection of mixtures would presumably allow refinement of a
measure such as total mass (or concentration) of volatile organic compounds (TVOC) into a
much more meaningful index that we might call the perceptually weighted level of VOCs,
180
-------
or PWVOC. Such an index, like the index dB(A) for sound, could make an otherwise
strictly physical measurement at least somewhat predictive of human responses.
Acknowledgments
The research described in this article was supported by NIH Grant DC00284 and
by the United States Environmental Protection Agency through Cooperative Agreement N°
CR-816362-02 awarded to the John B. Pierce Laboratory. It has been reviewed by the
Agency and cleared for publication, but does not necessarily reflect the views of the
Agency and no official endorsement should be inferred. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use. Thanks
are due to Mrs. Robin.Babbitt and Mr. Todd O'Hearn for their excellent technical
assistance. Research performed while J.E.C.-M. was a member of the Carrera de!
Investigador Cientifico, Consejo Nacional de Investigaciones CientiTicas y Tecnicas
(CONICET), Republica Argentina, -
References
Abraham, M. H. (1993). "Application of solvation equations to chemical and biochemical
processes." Pure & Appl. Chem. 65( 12): 2503-2512.
Abraham, M. H. (1993). "Scales of solute hydrogen-bonding: Their construction and
application to physicochemicai and biochemical processes." Chem. Soc. Rev. 22:
73-83.
Abraham, M. H., J. Andonian-Haftvan, J. E. Cometto-Muniz, and W. S. Cain (1995,
submitted). "An analysis of nasal irritation thresholds using a new solvation
equation." Fundam. Appl. Toxicol.
Abraham, M. H., G. D. Nielsen, et al. (1994). "The Ferguson principle and an analysis of
biological activity of gases and vapors." J. Pharm. Sci. 83: 680-688.
Abraham, M. H., G. S. Whiting, et al. (1990). "Hydrogen bonding 12. A new QSAR for
upper respiratory tract irritation by airborne chemicals in mice." Quant. Struct.-
Act. Relat. 9: 6-iO.
Amoore, J. E. (1986). Effects of chemical exposure on olfaction in humans. Toxicology
of the Nasal Passages. C. S. Barrow. Washington, Hemisphere Publishing
Corporation: 155-190.
Baker, R. A. (1963). "Odor effects of aqueous mixtures of organic chemicals." J. Water
Pollution Control Federation 35: 728-741.
Berglund, B. and M. J. Olsson (1993). "Odor-intensity interaction in binary and ternary
mixtures." Percept. Psychophys. 53:475-482.
Berglund, B. and H. Shams Esfandabad (1992). Humans as discerners of odor and
irritation. Fechner Day '92. Proceedings of the Eighth Annual Meeting of the
International Society for Psychophysics, Stockholm, Sweden, International
Society for Psyehophysics-
Brown, K. S., C. M. Mac Lean, et al. (1968). "The distribution of the sensitivity to
chemical odors in man." Hum. Biol. 40: 456-472.
Buck, L. and R. Axel (1991). "A novel multigene family may encode odorant receptors:
A molecular basis for odor recognition." Cell 65: 175-187.
Cain, W. S. (1988). Olfaction. Stevens' Handbook of Experimental Psychology. Jj_
Perception and Motivation: R. C. Atkinson, R. J. Herrnstein, G. Lindzey and R.
D. Luce. New York, John Wiley & Sons. 409-459.
Cain, W. S. (1989). "Testing olfaction in a clinical setting." ENT...J. 68: 316-328.
181
-------
Cain, W. S. and J. F. Gent (1991). "Olfactory sensitivity: Reliability, generality, and
association with aging." J. Exp. Psycho!.: Hum. Pcrcep. & Perform. 17: 382-391.
Cain, W. S., F. T. Schict, ct al. (1995, in press). "Comparison of models of odor
interaction." Chem. Senses.
Comctto-Muniz, J. E. and W. S. Cain (1990). "Thresholds for odor and nasal pungency."
Phvsiol. Bchav. 48: 719-725.
Cometto-Muniz, J. E. and W. S. Cain (1991). "Nasal pungency, odor, and eye irritation
thresholds for homologous acetates." Pharmacol. Biochem. Behav. 39: 983-989.
Cometto-Muniz, J. E. and W. S. Cain (1993). "Efficacy of volatile organic compounds in
evoking nasal pungency and odor." Arch. Environ. Health 48: 309-314.
Cometto-Muniz, J. E. and W. S. Cain (1994). "Perception of odor and nasal pungency
from homologous series of volatile organic compounds." Indoor Air 4: 140-145.
Cometto-Muniz, J. E. and W. S. Cain (1994). "Sensory reactions of nasaJ pungency and
odor to volatile organic compounds: The alkylbenzenes." Am. Ind. Hyg. Assoc. J.
55:811-817.
Cometto-Muniz, J. E. and W. S. Cain (1995). Olfactory adaptation. Handbook of
Olfaction and Gustation. R. L. Doty. New York, Marcel Dekker, Inc.: 257-281.
Cometto-Muniz, J. E. and W. S. Cain (1995). "Relative sensitivity of the ocular
trigeminal, nasal Irigeminal, and olfactory systems to airborne chemicals." Chem.
Senses 20: 191-198.
Comctto-Muniz, J. E., M. R. Garcia-Medina, et al. (1989). "Perception of pungent
odorants alone and in binary mixtures." Chem. Senses 14: 163-173.
Cometto-Muniz, J. E. and S. M. Hernandez (1990). "Odorous and pungent attributes of
mixed and unmixed odorants." Percept. Psychophys. 47: 391 -399.
Devos, M., F. Patte, et al., Eds. (1990). Standardized Human Olfactory Thresholds.
Oxford, IRL Press at Oxford University Press.
Ferguson, J. (1939). "The use of chemical potentials as indices of toxicity." Proc R Soc
(Lond) Ser B 127: 387-404.
Ferguson, J. (1951). Relations between thermodynamic indices of narcotic potency and
the molecular structure of narcotics. Proc Symp sur le Mechanisme de la Narcose,
Paris, CNRS.
Guadagni, D. G., R. G. Buttery, et al. (1963). "Additive effect of sub-threshold
concentrations of some organic compounds associated with food aromas." Nature
200: 1288-1289.
Laska, M. and R. Hudson (1991). "A comparison of the detection thresholds of odour
mixtures and their components." Chem. Senses 16: 651-662.
Nielsen, G. D. (1991). "Mechanisms of activation of the sensory irritant receptor by
airborne chemicals." CRC Grit. Rev. Toxicol. 21: 183-208.
Nielsen, G. D. and Y. Alarie (1982). "Sensory irritation, pulmonary irritation, and
respiratory stimulation by airborne benzene and alkylbenzenes: Prediction of safe
industrial exposure levels and correlation with their thermodynamic properties."
Toxicol Appl Pharmacol 65: 459-477.
Olsson, M. J. (1994). "An interaction model for odor quality and intensity." Percept. &
Psychophys. 55: 363-372.
Patterson, M. Q., J. C. Stevens, W. S. Cain, and J. E. Comctto-Muniz (1993). "Detection
thresholds for an olfactory mixture and its three constituent compounds." Chem.
Senses 18: 723-734.
Rosen, A. A., J. B. Peter, et al. (1962). "Odor thresholds of mixed organic chemicals." L_
Water Pollut. Control Fed. 35: 7-14.
182
-------
NRMRL-RTP-P-202
TECHNICAL RiPORT DATA
(Please read Instructions on she reverse before comple\
1. REPORT NO.
:PA/600/A-97/042
2.
4. TITLE AND SUBTITLE
5. REPORT DATE
Mixtures of Volatile Organic Compounds: Detection of
Odor, Nasal Pungency, and Eye Irritation
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
J. Enrique Cometto-Muniz and William S. Cain
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Chemosensory Perception Laboratory
Department of Surgery (Otolarygology)
University of California, San Diego
La Jolla, California 92093-0957
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA CR 816362-02
NIH Grant DC00284
12, SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 9/95
14. SPONSORING AGENCY CODE
EPA/600/13
15. SUPPLEMENTARY NOTES A PPCD project officer is Mark A. Mason, Mail Drop 54, 919/541-
4835. Presented at Healthy Buildings '95, Milan, Italy, 9/10-14/95.
16. ABSTRACT
The paper gives results of the measurement of threshold responses of
odor, nasal pungency (irritation), and eye irritation for single chemicals (l-propa-
nol, 1-hexanol, ethyl acetate, heptyl acetate, 2-pentanone, toluene, ethyl benzene,
and propyl benzene), and their mixtures (two three-component mixtures, two six-
component mixtures, and one nine-component mixture). Nasal pungency was mea-
sured in subjects lacking a functional sense of smell (i. e., anosmics) to avoid inter-
ference from olfaction. Various degrees of stimulus agonism (additive effects) were
observed for each of the three sensory channels when testing mixtures. As the num-
ber of components and the lipophilicity of such components in the mixtures increa-
sed, so did the degree of agonism. Synergistic stimulus agonism characterized the
eye irritation response for the most complex (the nine-component) and the most lipo-
(one of the six-component) mixtures. Physicochemical properties play a large
role in the determination of sensitivity to airborne chemicals, particularly to their
ability to evoke irritation. Whereas this has revealed itself previously with respect
to single chemicals, it seems to have relevance to mixtures as well.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Pollution
Measurement
Organic Compounds
Volatility
Smell
Odors
Olfactory Nerve
Senses
Pollution Control
Stationary Sources
Nasal Pungency
Nasal Irritation
Eye Irritation
Trigeniinal Nerve
13 B
14G
07C
20 M
360,05J
36P
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73J
------- |