Multiple Chemical Sensitivities: Addendum to Biologic Markers in Immunotoxicology (1992)

Iris R. Bell

The purpose of this paper is to review the clinical and research psychiatric and psychophysiologic literature on multiple chemical sensitivity patients (MCS) and to develop an integrative neuropsychiatric and biopsychosocial systems model of possible mechanisms. Much of the controversy over MCS has focused on hypotheses that the clinical syndromes must derive exclusively either from physiological or from psychogenic sources. However, such a dichotomous view of MCS is overly simplistic. A large body of data from both human and animal studies supports a more complex biopsychosocial approach involving an interplay of multiple influences in the expression of all human illnesses (Engel, 1977; Schwartz, 1982), including the clinical phenomenology of chemical sensitivities (Bell, 1987). This model means that MCS would involve a continuum and interaction of mechanisms in terms of the relative contributions of biological, psychological, and social factors in a given patient. Genetically-based neurochemical and/or receptor vulnerabilities in the central and autonomic nervous systems would make certain subsets of the population more likely to experience adverse effects of low dose chemical exposures.

Furthermore, even psychological and social "stress" can express health effects via biological mechanisms (Shavit et al., 1984; Sapolsky et al., 1990). Most of the major psychiatric disorders have genetic and neurochemical components (e.g. monozygotic twins have a 65-75% concordance vs dizygotic twins with a 14-19% concordance for major depression-Talbott et al., 1988). Thus, evidence for one type of factor (e.g. psychological) cannot rule out the contribution of any other type of factor (e.g. biological) to a clinical syndrome. Appropriate research in MCS must test plausible hypotheses by which a particular factor could causally contribute to or ultimately produce the clinical picture. Phenomenological labels alone fail to meet the latter requirement.

A logical site to examine the potential convergence of biological, psychological, and social factors in MCS is the central nervous system (CNS), especially the interconnected olfactory and limbic systems (Bell, 1982; Ashford and Miller, 1991). The CNS receives input from and sends output to other subsystems of the body, including behavioral, endocrine, immune, and autonomic functions, each of which may be disturbed in certain MCS patients.

The CNS is an integrative center that transduces biological and psychosocial experiences into changes in neural activity; in turn, these send biological, psychological, and social behaviors back out into the environment. Many pesticides and solvents, which are putative triggers for MCS, have effects on central nervous system function (Isaacson et al., 1990; Llorens et al, 1990; Tham et al., 1990) that could interact with individual vulnerabilities.

The olfactory system is the usual pathway by which airborne chemicals from the external environment interact with the brain. The limbic system is the phylogenetically older part of the CNS that receives extensive input from the olfactory system and regulates numerous functions including mood and social behaviors, cognitive function, and eating/drinking/ reproductive behaviors (Cain, 1974). Dysfunctions of the limbic system may play a key role in unipolar and bipolar depression, panic and other anxiety disorders, and schizophrenia. Damage to the hippocampus ill the limbic system is an important feature of dementia, especially of the Alzheimer type (Sapolsky, 1990).

At the outset, a model for MCS must be able to account for the clinical phenomenology. Patients with MCS report multiple symptoms in multiple systems. Clinical observations also include initial sensitization with acute high dose exposure, chronic low-dose sensitivity, spreading of number of sensitivities from one to many chemicals, adaptation to chronic exposures, and concomitant multiple food sensitivities (MFS). Little rigorous data are available to characterize the MCS population and their responses to chemical challenges (Tables 1a and 1b).

Despite extensive debate, only one controlled study has examined the effects of inhaled chemicals on individuals who report the MCS syndrome. Dory et al. (1988) compared the olfactory thresholds and selected autonomic responses to inhalation of phenyl ethyl alcohol and methyl ethyl ketone of 18 MCS patients with a matched control group of 18 normals. They found that the MCS group did not differ from normals on olfactory thresholds or cardiovascular responses to the two test chemicals, but that the MCS patients showed increased respiration rate and decreased nasal airway patency during chemical challenges. This study presented the chemicals in an environmentally controlled test chamber, but did not mask the identity of the chemical exposures or manipulate possible adaptation to the test items. Thus, these data confirm the presence of autonomic differences in MCS patients in terms of respiration and nasal airway function to specific chemical challenges, but do not address the larger clinical syndrome under consideration.

The remainder of the controlled data describes nonspecific phenomenology in subsets of

MCS patients. Most of these studies did not document tests of whether or not chemicals trigger illness in MCS patients. For example, Simon et al (1990) compared cohorts of plastics workers who did and did not develop chronic disabling MCS from low dose exposures to phenol, formaldehyde, and methyl ethyl ketone. They found that 54% of the 13 workers who developed MCS had a prior history of anxiety or depressive disorders in comparison with 4% of control workers. Staudenmayer and Selner (1990) observed higher levels of scalp electromyographic activity in MCS patients and similar patterns of electroencephalographic beta activity in MCS and mixed psychological outpatients in comparison with normal controls; this group reported performing negative controlled chemical challenges on the MCS subjects, but did not present their specific dependent measures or objective findings on this key point. Black's group (Black et al., 1990) reported that 65% of 23 MCS patients had histories of current or past mood, anxiety, or somatoform disorder in comparison with 28% of matched community controls. Although Black suggested that the psychiatric diagnoses could "explain some or all of their symptoms," they did not challenge these patients with chemicals to rule out chemically-induced illness. By consensual agreement within the American Psychiatric Association (Diagnostic and Statistical Manual III-revised, 1987), psychiatric diagnoses are descriptive labels only for phenomenology, not etiological or mechanistic explanations for syndromes. Thus, a psychiatric diagnosis labels a pattern of signs and symptoms, but offers no hypothesis concerning the mechanism(s) of the clinical phenomena (Davidoff et al., 1991). Finally, studies that have found psychiatric diagnoses in MCS patients have used non-random samples of the potentially chemically-sensitive population. Subject selection of MCS patients for research (Simon et al., 1990) and clinical series (Terr, 1986) has often derived from disabled subsamples with worker's compensation and other litigation claims, those self-identified with the field of clinical ecology (Doty et al., 1988; Black et al., 1990), or even those seen in psychiatric clinics (Black et al., 1990). Even in more recognized clinical syndromes involving chronic pain or traumatic injury, those individuals with associated legal and monetary cases represent a unique subset of any given chronically ill population. The discovery of psychiatric problems in a sample drawn in part from those seen in a psychiatric clinic (Black et al., 1990) is clearly a skewed representation of the true prevalence of psychiatric problems in the overall M CS population. Proper epidemiological research on characteristics of MCS will require case identification with more representative sampling from the general, chemically-exposed population.

Moreover, none of the studies that found increased depression, anxiety, and/or somatoform disorder histories compared MCS patients with chronically-ill control groups as opposed to healthy normals. Since chronic illnesses of all types, including cancer, heart disease, and autoimmune disorders, have a rent or past prevalence of depression and anxiety disorders ranging from 5% to as high as 45% (general average, 20%) (Katon and Sullivan, 1990), the finding of depression and anxiety would be expected in chronically ill M CS patients and would not establish a specific causal link between the affective disturbance and the chronic illness. In addition, somatic concerns, which constitute clinical criteria for depression and somatoform disorders, are often excluded in research studies of medically ill populations in order to avoid false positive diagnoses of depression on largely somatic symptomatology. In contrast with other investigations on geriatric and medically-ill populations-which use modified interview and/or self-rating scales free of somatic bias or adjust statistically for medically-related somatic complaints, no study of the psychiatric syndromes associated with subsets of MCS patients has ever made any such methodological adjustments.

Ironically, the finding of psychiatric dysfunctions may nonetheless offer clues to the biology of the disorder in a subset of MCS cases. That is, the mechanisms of depression and anxiety, as well as other psychiatric symptoms, involve putative dysregulation of brain chemistry and neurotransmitter receptors in specific neural pathways (Reiman et al., 1986; Talbott et al., 1988). Thus, one might expect individuals vulnerable to major psychiatric disorders be among the most susceptible to low doses of those environmental chemicals that could worsen their inherent dysfunctions in brain chemistry, either by direct action or by activation of endogenous mediators (Bell, 1982).

For example, animal models of depression include drug-reduced elevation of brain acetylcholine, drug-reduced depletion of brain serotonin and/or norepinephrine, and destruction of the olfactory bulbs (Jesberger and Richardson, 1985, 1988; van Riezen and Leonard, 1990). Psychiatric investigators have already raised the possibility that the presence of major depression in certain patients indicates an incased neurochemical vulnerability to certain environmental chemicals such as organophosphate pesticides (Rosenthal and Cameron, 1991). The latter agents raise brain acetylcholine levels, an effect which can induce depression in susceptible persons (Dilsaver, 1986).

The overlap between the biology of depression and certain MCS and MFS syndromes is further supported by converging evidence that many of the same disorders that some clinicians have claimed to be triggered by adverse food and chemical reactions also improve during chronic treatment with antidepressant medications. The disorders within medicine that share responsivity to these drugs include major depression, panic disorder, bulimia, obsessive-compulsive disorder, attention-deficit disorder with hyperreactivity, migraine headache, irritable bowel syndrome, and cataplexy (Hudson and Pope, 1990). Notably, 87% of patients with the somatization disorder known as Briquet's syndrome (hysteria) also experience major depression during the course of their illness; and 76% of such patients report multiple food intolerances as a criterion for diagnosis (Purtell et al., 1951), another overlap with MCS/MFS phenomenology. In addition, Bell et al. (1990) have recorded tinnitus in 48% Of one sample of community-recruited MFS/MCS patients, a condition also known to be responsive to antidepressant medications (Sullivan, et al., 1989). In terms of environmental reactivity, it is striking that a number of studies have found persons with a history of major depression to have an inordinately high prevalence of atopic allergy histories in comparison with both normals and other types of psychiatric patients (Nasr et al., 1981; Sugerman et al., 1982; Bartko and Kasper, 1989; Marshall, 1989; Bell et al., 1991). Such date contrast with the current lack of evidence for atopic mechanisms in MCS or MFS, despite the depression histories. More thorough epidemiological investigation of this point is needed to clarify the prevalence of atopy in depressed versus nondepressed MCS/MFS patients. Interestingly, however, in addition to other mechanisms, several tricyclic antidepressants exert also significant antihistamine effects at both H1 and H2 receptors (Schatzberg and Cole, 1991).

From the above list, controlled studies have indicated that adverse food and ingested chemical reactions may worsen certain cases of ADD with hyperreactivity (Egger et al., 1985; Kaplan et al., 1990; Swanson and Kinsbourne, 1980), migraine headache (Egger et al., 1983; Egger et al, 1989), and irritable bowel (Jones, et al., 1982), though some researchers have had negative findings in the latter disorder (Bentley et al., 1983). Thus, it is possible that some

brain neurochemical dysfunction susceptible to worsening by adverse food and chemical reactions and to improving with antidepressant drugs is common across a range of disorders, not all purely "psychogenic" in nature. This hypothesis raises two possibilities in identification of possible MCS cases: (a) choosing subjects with persona and family histories loaded for more than one of the above disorders may optimize chances of finding true cases; for example, Egger's group observed that the MFS dietary responders were those with histories of both migraine and seizures, but not with seizures alone (Egger, et al., 1989); (b) screening subjects for those who improve from 4-6 week trials with antidepressant medications, regardless of clinical diagnosis, should assist selection of biologically more homogeneous patients.

It is also striking that one of the best animal models of depression currently used by drug companies to identify new antidepressant medications is that of olfactory bulbectomy (Jesberger and Richardson, 1988; van Riezen and Leonard, 1990). As in human depression, animals with olfactory bulb damage demonstrate neuroanatomic dysregulation of the limbic-hypothalamic axis, neurochemical imbalances of serotonin, norepinephrine, gamma-amino-butyric acid, and acetylcholine, and improvement from chronic but not acute treatment with antidepressant drugs (Jesberger and Richardson, 1988). Furthermore, olfactory bulbectomy increases the vulnerability of another limbic structure, the amygdala, to kindling (see below). The amygdala is especially important in regulation of affect (e.g. fear-avoidance; rage), drive, and related autonomic/endocrine/immune functions (Mesulam, 1985). Given the role of the olfactory bulb in transmitting sensory and nonsensory information concerning airborne chemicals to the rest of the brain (Cain, 1974), it is reasonable to hypothesize that individuals with dysfunctional olfactory bulb pathways secondary to inherent neurochemical and/or receptor alterations may be the population most sensitive to developing MCS.

Persons prone to anxiety may also have inherent neurochemical defects that make them particularly susceptible to certain environmental chemicals. For example, drugs and certain pesticides such as lindane that may antagonize receptors for the inhibitory neurotransmitter gamma-aminobutyric acid (GABA) have apparent anxiogenic properties in animals and humans (Llorens et al., 1990), opposite to those of antianxiety drugs such as the benzodiazepines. Thus, persons with benzodiazepine-responsive anxiety disorders that place them closer to threshold for anxiety due to relatively impaired GABAergic function at baseline might be expected to experience increased anxiety on a biological basis from lindane or related exposure. Another affective spectrum anxiety disorder that a small subset of MCS patients may experience is panic disorder (Dager et al., 1987). Panic disorder is a chronic disorder associated with a broad range of symptoms, especially recurrent acute attacks of difficulty breathing and hyperventilation, palpitations, dizziness, fatigue, and an impending sense of doom. Chemical agents such as low concentrations of carbon dioxide (e.g. 5.5%) in room air and as excess lactate trigger panic attacks in panic patients much more often than in normals under open and blind conditions (Woods, et al., 1986; Roy-Bryne, et al., 1988). In view of the olfactory-limbic model for MCS, it is notable that panic disorder patients have hemispheric asymmetry of blood flow in parahippocampal brain regions on positron emission tomography

scans (Reiman et al, 1986). The parahippocampus is a limbic site that receives input from the main olfactory bulb, a major way-station for input from the nose and olfactory nerves. Furthermore, Dager et al. (1987) have hypothesized that occupational solvent exposures could induce panic disorder, possibly by activating a kindling process in the limbic system. Thus, a plausible hypothesis for some eases of environmentally-activated panic disorder is a biologically-based dysfunction of limbic pathways in response to olfactory system input, triggered by certain environmental chemicals.

Kindling occurs when repeated subthreshold stimuli summate to trigger seizure activity in brain cells with previously normal activity. The limbic system is especially susceptible to kindling, which may also play a role in the developing chronicity and treatment resistance of certain bipolar affective disorders (Post et al., 1984). In view of the hypothesized responsiveness of affective spectrum disorders to antidepressants, it is notable that these drugs reportedly raise the threshold for or even prevent kindling (cf. Jesberger and Richardson, 1985). Therefore, another, more mechanism-oriented label for the spectrum of disorders that is common to both affective spectrum and MFS/MCS might be kindling-related syndromes potentially responsive to antikindling medications.

Another property of limbic brain cells perhaps related to kindling is that of long-term potentiation (LTP). LTP involves persisting enhancement of synaptic response initiated by brief high-frequency stimulation of excitatory pathways at subictal levels (Racine et al., 1983; Stripling et al., 1988; lung et al., 1990; Kanter and Haberly, 1990). LTP is hypothesized to subserve memory formation and retrieval in the hippocampal region of the limbic system, but also occurs in olfactory cortex following high-frequency stimulation of the olfactory bulb (Stripling et al., 1988).

Notably, the nature of kindling-and perhaps long-term potentiation-suggests that these biological mechanisms might participate in development and/or expression of MCS-that is, by either initial acute high doses (cf. high doses=high frequency stimuli) or chronic summated low doses (cf. low doses=low frequency stimuli) of environmental chemicals stimulating the firing, long-term potentiation, and/or kindling of olfactory pathways-and from there, many other regions of the limbic system in genetically-predisposed individuals. Indeed, Bokina et al. (1976) reported a rabbit model of low dose environmental pollutant intolerance that matches the latter hypothesis. Their group noted that low concentrations of pollutants combined with a functional load (rhythmic flashing light) triggered abnormal paroxysmal activity in the olfactory bulb and corticomedical amygdala after an initial exposure to high concentrations of the same pollutants. Furthermore, Bokina reported alterations in visual evoked potentials during chronic exposure to low concentrations of formaldehyde and ozone.

In summary, olfactory pathway kindling and LTP might participate in registering information concerning past high dose and/or cumulative low dose chemical exposures that would increase the likelihood of limbic responsivity to subsequent low dose exposures. Resultant dysfunctions in numerous behavioral, autonomic, and endocrine subsystems under limbic regulation might then lead to multiple symptoms through a range of end-organ mechanisms. Finally, it is also possible that chemical exposures which kindle certain limbic pathways could damage capacity for long-term potentiation in the hippocampus (Racine et al., 1983; Armstrong et al., 1990). Such events might in turn disrupt capacity for normal memory

formation. These hypotheses offer the possibility of developing animal models for MCS at the level of the central nervous system that would permit greater experimental control and eliminate concerns regarding subject expectation and bias.

Although we have no further EEG data on MCS patients or chemically-exposed animals, Lorig and colleagues (Lorig and Schwartz, 1988; Lorig et al., 1988; Lorig, 1989; Lorig et al., 1990a; Lorig et al., 1990b) have performed a series of studies on the effects of low concentration odors on electroencephalographic responses, mood, and cognitive activity in normal human subjects (Lorig et al., 1990a; 1990b). Their findings suggest that chemical odors below the olfactory threshold (i.e. that are consciously undetectable) nonetheless produce distinct EEG responses, worsened mood, and poorer performance on a visual search task (Lorig et al., 1990b). They also examined evoked potentials during a common test of auditory attention and observed increased midline P200 wave amplitude even at an undetected low concentration of a common perfume constituent (galaxolide)(Lorig et al., 1990b). The P200 findings indicated that even undetected airborne chemicals may have a disruptive influence on attention processes. Moreover, in another investigation, they reported that EEG alpha frequency activity in the left hemisphere was lower and EEG beta activity showed greater spatial diversity between posterior and anterior regions during nose breathing than during mouth breathing of unfiltered indoor room air (Lorig et al., 1988). They concluded that undetected odors in indoor air inhaled via the nose exert a distinct effect on brain function outside conscious awareness or changes due to sensory perception (Lorig et al., 1988). Therefore, despite the subcortical location of the olfactory system, its broad connections into the forebrain provide the neuroanatomical and neurophysiological substrate for broad effects of low levels of environmental chemicals on EEG and behaviors regulated by the limbic system.

If such effects are measurable in normals, the olfactory-limbic system model for MCS would predict that the EEG, evoked potential, and cognitive findings would be even more pronounced in MCS patients. Preliminary evidence does support the possibility of abnormalities of attention and concentration in MCS. For example, Ben et al. (1990; submitted for publication, 1991) noted significantly slower performance on a timed mental arithmetic task in outpatients with multiple food and chemical sensitivities than in normal controls. Over 70% of this same sample of patients reported the symptom of difficulty concentrating in comparison with 13% of the controls (p<0.008). In addition, Feidler and Kippen (this volume) have reported evidence of poorer overall performance by MCS patients on the California Verbal Learning Test, a measure of immediate, delayed, and recognition memory also sensitive to attention processes.

In brief, the limbic-olfactory model suggests that even unperceived low concentrations of airborne chemicals can induce changes in brain electrical activity and resultant behaviors and physiology in both animals and human subjects. In certain individuals with particular neurophysiological and/or neurochemical predisposition to limbic system dysfunction (e.g. depression or anxiety disorders), such chemical exposures would activate clinical syndromes described as MCS. The environmental chemicals enter the brain via the olfactory system (Cain, 1974; Shipley, 1985) and would mobilize the abnormal brain processes in the limbic brain with or without sensory awareness via kindling and/or long-term potentiation with initial high dose exposures or with repeated low concentration exposures. One important marker of exogenous chemical olfactory-limbic effects may be abnormalities of attention processes, which

could be quantified with neuropsychological and EEG evoked potential tests during undetected chemical as well as sham exposures. Animal models may permit depth electrode recordings with behavioral learning tests, while MCS patients may provide surface computerized EEG and evoked potential data with attention and memory tests to evaluate the above hypotheses (cf. James et al., 1987). Time factors in terms of sequencing high to low dose exposures and of giving repeated low dose exposures may be necessary to elicit the proposed phenomena.

Within its descriptive labels, modem psychiatry has embraced a biopsychosocial perspective for diagnosis. The DSM-IIIR (American Psychiatric Association, 1987), for example, employs a multiaxial system to maximize appreciation of the multidimensionality of psychiatric disorders. Thus, axis I covers clinical syndromes, axis II personality disorders, axis HI medical disorders, axis IV severity of psychosocial stressors, and axis V global assessment of functioning. Obviously, patients with psychotic delusions of chemical sensitivity, those with only personality disorders, or those with completely incorrect, though nonpsychotic, misattribution of symptoms to chemical reactions may exist within the continuum of patients claiming to have MCS. At least in the case of psychosis, however, none of the studies that assessed psychiatric status reported finding any psychotic type disorders in certain MCS samples (Doty et al., 1988; Black et al., 1990; Simon et al., 1990; Staudenmayer and Selner, 1990). More refined and objective measures for identification of true cases will assist in screening out such individuals as well as any cases of nondelusional misattribution from MCS study populations.

Nevertheless, illnesses generally have psychosocial modulators, which interact with underlying biological vulnerabilities. These modulators include: premorbid personality and coping style (McCrae and Costa, 1986; Jamner et al., 1988); perceived social support (Cohen and Syme, 1985); perceived familiarity of the surrounding environment (Siegel et al., 1982); sense of control over the illness (Rodin, 1986; Sanderson et al., 1989); and classical conditioning of responses to both triggering and treating agents (Eikelboom and Stewart, 1982; Bolla-Wilson, et al., 1988; MacQueen et al., 1989). As emphasize above, identification of these factors in patients with MCS would be expected, For example, MCS patients have organized local and national support groups, which may have both positive benefits in terms of providing buffering effects from social support networks and negative effects in terms of reinforcing illness behaviors (Bell, 1987). They merit specific study to optimize clinical care and to understand the M CS syndromes, but demonstration of these psychosocial phenomena would not by itself ''disprove'' a role for biological causes of the syndromes.

At the same time, coincidental or deliberate manipulation of these factors would be likely to exert a marked influence on course and outcome in various individuals. Even in the case of term cancer, for example, a defiant and demanding versus submissive personality style (Levy, 1985) and participation in a cancer patient support group therapy versus no group therapy (Spiegel et al., 1989) are associated with longer survival times. Providing nursing home residents with a sense of control over their immediate physical surroundings and daily activities substantially reduces subsequent mortality rates over those of residents without such sense of control (Rodin, 1986). In the case of affective spectrum panic disorder, providing patients with an illusion of control over exposure to carbon dioxide will attenuate the severity of the carbon dioxide-induced anxiety (Sanderson et al., 1989), even though in other blinded studies the CO2 clearly can trigger panic over and above that set off by sham inhalation

(Woods et al., 1986). At the same time, perceived stress leads to immune dysfunctions in human (Kiecolt-Glaser and Glaser, 1987) and animal (Shavit et al., 1984) subjects. The administration of opiate drugs to opiate-tolerant animals in a strange cage versus a familiar cage doubles the mortality from the same high dose (64% versus 32%; whereas 96% of drug-naive animals die) (Siegel et al., 1982). The mediating mechanisms of these types of effects may involve hormonal, central nervous, and autonomic nervous system elements. Without accounting for the entire clinical situation in MCS, many similar factors might still contribute to the day-to-day course of the syndrome.

In modern health psychology and psychosomatic medicine, it is becoming apparent that the mind/brain and body have a two-way communication system that utilizes biological messengers to implement the effects of "stress" on the body and to change mood and behavior. These include common hormones and other mediators such as cortisol (Sapolsky et al., 1990), interleukins (Sapolsky et al., 1987; Spadaro and Dunn, 1990), vasopressin (Weingartner et al., 1981), substance P, vasoactive intestinal peptide, prostaglandins, kinins, histamine, and opiate-like peptides (enkephalins, endorphins) (Sloviter and Nilaver, 1987; Wiedermann, 1987). For instance, prostaglandins are elevated in major depression (Calabrese et al., 1986; Ohishi et al., 1988), interleukin 1 in allergy and infection (Walter et al., 1989), and opiate-like peptides in sleep apnea (Gislason et al., 1989) and depressive disorders (Agren et al., 1982). PGE2 induces wakefulness (Matsumura et al., 1989); IL-1 produces somnolence and reduces exploratory behavior (Walter et al., 1989; Spadaro and Dunn, 1990); opiate-like peptides facilitate antigen-induced histamine release and lower natural killer cell activity (Shavit et al., 1984; Mediratta et al., 1988).

Many of these same agents participate in more purely physiological events (e.g. inflammation, infection, atopy). However, multiple pathways that originate with biological as well as psychosocial triggers can set off release of these mediators. Once they act at their end-organs, the clinical effects appear similar, regardless of the nature of the original trigger. It is reasonable to hypothesize that many of these mediators contribute to symptom production in MCS from multiple initiating pathways. Clinical symptom pictures, in MCS patients (e.g. MFS/MCS daytime somnolence-see Bell et al., 1990 versus somnolence-inducing properties of prostaglandin D2 and interleukin 1) and data on mediators in syndromes overlapping MCS can guide the selection of possible endogenous agents for study.

The hypothesis of classical conditioning of symptom flares to olfactory stimuli represents a potential intersection between psychological and biological mechanisms. For example, antigen exposure in a sensitized animal will mobilize histamine release on a physiological basis. If this event is paired in time with a previously neutral event (e.g. a physiologically-inactive odor), subsequent release of histamine in response to the classically conditioned odor without antigen will occur (Russell et al., 1984; MacQueen et al., 1989). Numerous drug responses including those to opiates and insulin can also be classically conditioned, although the direction of the conditioned response is often opposite from that to the direct biological stimulus (Eikelboom and Stewart, 1982).

In all classical conditioning, it is important to note that the phenomenon requires an initially biologically active stimulus to pair with the previously inactive stimulus. As stated earlier, the existence of classically conditioned responses does not rule out the co-existence of more directly biologically-initiated responses in the same individual. Thus, while it is important to explore classical conditioning of chemical response in MCS patients and normals (cf. Kirk-Smith et al., 1983; Lorig and Roberts, 1990), the first priority of MCS research must be the biological effects of chemicals. Once the parameters of the biological effects are clarified, then possible psychosocial modifiers and their associated biological mediating mechanisms should be examined experimentally.

Methodological issues in the study of MCS are critically important. While the scope of this paper is limited to multiple chemical sensitivities, the body of controlled literature on multiple food sensitivities offers valuable clues to the proper design of studies on chemicals and chemically-sensitive patients (Pearson et al., 1983; Egger et al., 1985; Kahn et al., 1988, 1989; Kaplan et al., 1989). Table 2 summarizes the concerns regarding design, subject prescreening, sample size, placebo controls, blinding, duration and timing of avoidance and/or challenges, number of independent measures, and sensitivity of dependent measures. Importantly, the intense debate on the testing and treatment technique known as provocation-neutralization (King, 1988) should be held separately from the more fundamental questions concerning the nature and mechanisms of adverse chemical reactions. For the present, research on MCS should emphasize delivery of real-life-like concentrations via customary mutes of exposure (i.e. nose inhalation, dietary ingestion) rather than shifting focus to any particular testing technique or unusual route of administration.

From the MFS literature, we have learned that it is necessary to prescreen subjects for current-not merely historical-reactivity to the proposed test agents. Under experimental conditions, researchers have generally not detected effects when studying whole populations-but have found effects when studying subsets of prescreened populations. The latter point

raises statistical power considerations in the experimental design (King, 1985). If the effects are limited to a subset of a population, the less certainty we have in identifying these cases on entry into the study necessitates the use of larger sample sizes to avoid Type H error (missing an effect that is actually present by failing to choose enough responders as subjects).

It is also clear that blinding and selection of placebo conditions are difficult issues in a highly sensitive sample of subjects. It would be unfortunately easy to generate a placebo that is physiologically active, though otherwise indistinguishable from the intended active agent. Lorig's research on normal human subjects suggests a possible strategy, which involves a two-step procedure: (i) determination of olfactory thresholds for each test substance in the study subjects, (ii) experiments using sub-threshold concentrations of the test agent (Lorig et al., 1990a, b). This approach avoids sensory awareness and expectations, as well as the complication of possible biological reactivity to added masking odors intended to be inactive while hiding the actual active test agent.

With the background noise of chronic symptomatology and chronic everyday chemical exposures, the experimental approach in MFS also suggests ways to maximize emergence of test reactions under research conditions. That is, investigators need (a) to use repeated, multiple challenges over extended periods of time rather than a single challenge on one day (Kahn et al., 1988, 1989; Kaplan et al., 1989); and (b) to employ multiple rather than single chemical agents for tests (Kaplan et al., 1989). Patients who are in fact chemically sensitive to various substances may nonetheless hold incorrect beliefs about their reactivity to any specific chemical on a single day for a wide range of reasons, including the reality that no chemical exposure in real life is experienced separate from thousands of others in ambient air. Thus, Molhave's approach of starting investigation of tight building subjects and normals with mixtures of volatile organic compounds similar to those found in indoor air may be the most appropriate first step in MCS research as well (Molhave et al., 1986; Molhave 1990; Otto et al., 1990). MCS studies performed in environmentally controlled units with measurably low levels of ambient chemical noise may encounter fewer of these issues of background exposures in their design, especially with subjects resident for weeks at a time.

Finally, it is extremely important to select appropriate and sensitive dependent measures. It is wasteful to perform a controlled study, only to ask subjects whether or not they believe they are having adverse reactions (Pearson et al., 1983). Such dichotomous measures are notoriously insensitive, lower statistical power, and are likely to create Type II error of missing an actual effect of the chemical challenges. In addition to objective laboratory tests such as EEG, evoked potentials (Lorig, 1989), polysomnographic sleep studies (Kahn et al., 1988), blood and cerebrospinal fluid changes in putative mediators and their receptors, and neuropsychological cognition tests (Swanson and Kinsbourne, 1980; Molhave et al., 1986; Lorig et al., 1990b), researchers have a host of standardized self-report and observer-rated scales from medicine and psychology to grade changes in physical and emotional symptoms, behaviors, and mood along a continuum. All such measures offer the possibility of more sensitive continuous rather than dichotomous dependent variables for study.

In view of the obvious role of the olfactory system in communicating information about the chemical environment to the brain and thereby to the rest of the body, the olfactory-limbic model outlined above is a plausible alternative to the previous emphasis on putative immunological mechanisms for MCS. As indicated, neurons of the limbic brain have

excitability properties that could provide the basis for amplification and spreading of adverse reactions to low dose chemical exposures. Furthermore, numerous psychosocial modifiers could participate in the overall clinical course of the illness via their transduced effects on the brain and the bodily functions it regulates. Overall, as is common in the clinical practice of biopsychosocial patient care, the first priority should be to study the direct biological effects of low dose chemical exposures on health in the subset of individuals vulnerable to multiple chemical sensitivity. The next priority is to examine interactions between the biological and psychosocial dimensions of MCS. Eventually, well-done investigations will permit the objective evaluation of the relative proportions of the variance in MCS symptomatology to which biological, psychological, and social factors contribute.

Despite the controversies surrounding multiple chemical sensitivities, it is possible to develop experimentally testable hypotheses concerning the nature and mechanisms of MCS. This work requires synthesis of approaches and information from multiple fields, including occupational medicine and toxicology, internal medicine, health psychology, basic and applied neurosciences, and pharmacology. In this area, it is ultimately insufficient to frame questions in terms of exclusively biological versus psychosocial causes or to show a fit or lack of fit between known clinical entities and MCS. The problems raised by chemical sensitivities demand scientific thought beyond the boundaries of existing fields. The solutions are likely to expand our knowledge base in novel ways toward understanding interactions between human beings and their environment.

Agren, H., L. Terenius, A. Wahlstrom 1982. Depressive phenomenology and levels of cerebrospinal fluid endorphins. Annals N.Y. Acad. Sci. :388-398.

American Psychiatric Association 1987. Diagnostic and Statistical Manual of Mental Disorders. Third Edition, revised. Washington, D.C.: American Psychiatric Association.

Armstrong, D.L., J.L. Polan-Curtain, T. Osaka, M.R. Murphy, S.Z. Kerenyi, S.A. Miller, S.L. Hartgraves 1990. Central nervous system protection against effects of long-term low-dose nerve agents. Report for USAF School of Aerospace Medicine, Human Systems Division (AFSC), Brooks Air Force Base, Texas (USAFSAM-TR-90-31), U.S. Naval Medical Research and Development Command .

Ashford, N.A., C.S. Miller 1991. Chemical Exposures. Low Levels and High Stakes. N.Y.: Van Nostrand Reinhold.

Bartko, J.J. and S. Kasper 1989. Seasonal changes in mood and behavior. A cluster analytic approach. Psychiatry Res. 28:227-239.

Bell, I.R. 1982. Clinical Ecology: A New Medical Approach to Environmental Illness. Bolinas, CA: Common Knowledge Press.

Bell, I.R. 1987 Environmental illness and health: the controversy and challenge of clinical ecology for mind-body health. Advances 4(3):45-55.

Bell, I.R., Markley, E. J., B. Zumalt, D.S. King 1990. Psychological and somatic symptom and illness patterns in individuals with self-reported adverse food reactions. Biol. Psychiatry 27:168A.

Bell, I.R., M.L. Jasnoski, J. Kagan, D.S. King 1991. Depression and allergies: survey of a nonclinical population. Psychother. Psychosom. 51, in press.

Bentley, S.J., D.J. Pearson, K.J.B. Rix 1983. Food h sensitivity in irritable bowel syndrome. Lancet 2:295-297.

Black, D.W., A. Rathe, and R.B. Goldstein 1990. Environmental illness. A controlled study of 26 subjects with '20th century disease'. J. Amer. Med. Assoc. 264:3166-3170.

Bokina, A.I., N.D. Eksler, A.D. Semenenko, and R.V. Merkur'yeva 1976. Investigation of the mechanism of action of atmospheric pollutants on the central nervous system an comparative evaluation of methods of study. Environ. Health Perspectives 13:37-42.

Bolla-Wilson, K., R.J. Wilson, M.L. Bleecher 1988. Conditioning of physical symptoms after neurotoxic exposure. J. Occupational Med. 30:684-686.

Cain, D.P. 1974. The role of the olfactory bulb in limbic mechanisms. Psychological Bull.

81:654-671.

Calabrese, J.R., R.G. Skwerer, B. Barna, A.D. Gulledge, R. Valenzuela, A. Butkus, S.Subichin, N.E. Krupp 1986. Depression, immunocompetence, and prostaglandins of the E series. Psychiatry Research 17:41-47.

Cohen, S. and S.L. Syme, eds. 1985. Social Support and Health. New York: Academic Press, Inc.

Dager, S.R., J.P. Holland, D.S. Cowley, D.L. Dunner 1987. Panic disorder precipitated by exposure to organic solvents in the work place. Am. J. Psychiatry 144:1055-1058.

Davidoff L.L., T.J. Callender, L.A. Morrow, G. Ziem 1991. Letter to the Editor. J. Psychosom. Res. 35(4):1-3.

Dilsaver, S.C. 1985. Cholinergic theory of depression. Brain Res. Rev. 11:285-316.

Doty, R.L., D.A. Deems, R.E. Frye, R. Pelberg, A. Shapiro 1988. Olfactory sensitivity, nasal resistance, and autonomic function in patients with multiple chemical sensitivities. Arch. Otolaryngol-Head & Neck Surgery 114:1422-1427.

Egger, J. C.M. Carter, J. Wilson, M.W. Turner, J.F. Soothill 1983. Is migraine food allergy? A double-blind controlled trial of oligoantigenic diet treatment. Lancet 2:865-869.

Egger, J., C.M. Cater, J.F. Soothill, J. Wilson 1989. Oligoantigenic diet treatment of children with epilepsy and migraine. J. Pediatr. 114:51-58.

Eikelboom, R. and I. Stewart 1982. Conditioning of drag-induced physiological responses. Psychological Rev. 89:507-528.

Engel, G.L. 1977. The need for a new medical model: a challenge for biomedicine. Science 196:129-136.

Gislason, T., M. Almquist, G. Boman, C. Lindholm, L. Terenius 1989. Increased CSF opioid activity in sleep apnea syndrome. Regression after successful treatment. Chest 96:250-254.

Hudson, J.I. and H.G. Pope 1990. Affective spectrum disorder: does antidepressant response identify a family of disorders with a common pathophysiology? Am. J. Psychiatry 147:552-564.

Isaacson, L.G., S.A. Spohler, D.H. Taylor 1990. Trichloroethylene affects learning and decreases myelin in the rat hippocampus. Neurotoxicology Teratology 12:375-381.

James, L., A. Singer, Y. Zurynski, E. Gordon, C. Kraiuhin, A. Harris, A. Howson, R. Meares 1987. Evoked response potentials and regional cerebral blood flow in somatization disorder. Psychother. Psychosom. 47:190-196.

Jamner, L.D., G.E. Schwartz, H. Leigh 1988. The relationship between repressive and defensive coping styles and monocyte, eosinophil, and serum glucose levels: support for the opioid peptide hypothesis of repression. Psychosom. Med. 50:567-575.

Jesberger, J.A. and J.S. Richardson 1985. Animal models of depression: parallels and correlates to severe depression in humans. Biol. Psychiatry 20:764-784.

Jesberger, J.A. and J.S. Richardson 1988. Brain output dysregulation induced by olfactory bulbectomy: an approximation in the rat of major depressive disorder in humans. Intern. J. Neuroscience 38:241-265.

Jones, V.A., P. McLaughlan, M. Shorthouse, E. Workman, J.O. Hunter 1985. Food intolerance: a major factor in the pathogenesis of irritable bowel syndrome. Lancet 2:1115-1117.

Jung, M.W., J. Larson, G. Lynch 1990. Long-term potentiation of monosynaptic EPSPs in rat piriform cortex in vitro. Synapse 6:279-283.

Kahn, A., G. Francois, M. Sottiaux, E. Rebuffat, M. Nduwinana, M.J. Mozin, J. Levitt 1988. Sleep characteristics in milk-intolerant infants. Sleep 11:291-297.

Kahn, A., M.J. Mozin, E. Rebuffat, M. Sottiaux, M.F. Muller 1989. Milk intolerance in children with persistent sleeplessness: a prospective double-blind crossover evaluation. Pediatrics 84:595-403.

Kanter, E.D., L.B. Haberly 1990. NMDA-dependent induction of long-term potentiation in afferent and association fiber systems of piriform cortex in vitro. Brain Research 525:175-179.

Kaplan, B.J., J. McNicol, R.A. Conte, H.K. Moghadam 1989. Dietary replacement in preschool-aged hyperactive boys. Pediatrics 83:7-17.

Katon, W., M.D. Sullivan 1990. Depression and chronic medical illness. J. Clin. Psychiatry 51[6, suppl]:3-11.

Kiecolt-Glaser, J.K. and R. Glaser 1987. Psychosocial moderators of immune function. Annals Behavioral Medicine 9:16-20.

King D.S. 1985. Statistical power of the controlled research on wheat gluten and schizophrenia. Biol. Psychiatry 20:785-787.

King D.S. 1987. The reliability and validity of provocative food testing: a critical review. Meal. Hypotheses 25:7-16.

Kirk-Smith, M., C. Van Toiler, G. Dodd 1983. Unconscious odor conditioning in human subjects. Biological Psychology 17:221-231.

Levy, S.M. 1985. Behavior and Cancer. San Francisco: Jossey-Bass.

Liskow, B., E. Othmer, E.C. Penick, C. DeSouza, W. Gabrielli 1986. Is Briquet's syndrome a heterogeneous disorder? Am. J. Psychiatry 143:626-629.

Llorens, J., J.M. Tusell, C. Sunol, E. Rodriguez-Farre 1990. On the effects of lindane on the plus-maze model of anxiety. Neurobiology Teratology 12:643-647.

Lorig, T.S. 1989. Human EEG and odor response. Progress in Neurobiology 33:387-398.

Lorig, T.S. and G.E. Schwartz 1988. Brain and odor: I. Alteration of human EEG by odor administration. Psychobiology 16:281-284.

Lorig, T.S., G.E. Schwartz, K.B. Herman 1988. Brain and odor: II. EEG activity during nose and mouth breathing. Psychobiology 16:285-287.

Lorig, T.S., K.B. Herman, G.E. Schwartz 1990a. EEG activity during administration of low-concentration odors . Bulletin Psychonomic Soc. 28:405-408.

Lorig, T.S., E. Huffman, A. DeMartino, J. DeMarco 1990b. The effects of low concentration odors on EEG activity and behavior. J. Psychophysiology 5, in press.

Lorig, T.S., M. Roberts 1990. Odor and cognitive alteration of the contingent negative variation. Chemical Senses 15:537-545.

MacQueen, G., J. Marshall, M. Perdue, S. Siegel, J. Bienenstock 1989. Pavlovian conditioning of rat mucosal mast cells to secrete rat mast cell protease II. Science 243:83-85.

Marshall, P. 1989. Attention deficit disorder and allergy: a neurochemical model of the relation between the illnesses. Psychol. Bulletin 106:434-446.

Matsumura, H., K. Honda, W.S. Choi, et al. 1989. Evidence that brain prostaglandin E2 is involved in physiological sleep-wake regulation in rats. Proc. Natl. Acad. Sci. 86:5666-5669.

McCrae, R.R. and P.T. Costa 1986. Personality, coping, and coping effectiveness in an adult sample. J. Personality 54:385-405.

Mediratta, P.K., N. Das, V.S. Gupta, P. Sea 1988: Modulation of humoral immune responses by endogenous opioids. J. Allergy Clin. Immunol. 81:27-32.

Mesulam, M.M. 1985. Principles of Behavioral Neurology. Philadelphia: F.A. Davis Co.

Molhave, L. 1990. Volatile organic compounds. Indoor air quality and health. Indoor Air 5:15-33.

Molhave, L., B. Bach, and O.F. Pedersen 1986. Human reactions to low concentrations of volatile organic compounds. Environment International 12:167-175.

Nasr, S., E.G. Altman, H.Y. Meltzer 1981. Concordance of atopic and affective disorders. J. Affective Disord. 3:291-296.

Ohishi, K., R. Ueno, S. Nishino, T. Sakai, O. Hayaishi 1988. Increased level of salivary prostaglandins in patients with major depression. Biol. Psychiatry 23:325-334.

Otto, D., L. Molhave, G. Rose, H.K. Hudnell, D. House 1990. Neurobehavioral and sensory irritant effects of controlled exposure to a complex mixture of volatile organic compounds. Neurotoxicology Teratology 12:649-652.

Pearson, D.J., K.J.B. Rix, S.J. Bentley 1983. Food allergy: how much in the mind? A clinical and psychiatric study of suspected food hypersensitivity. Lancet 1:1259-1261.

Post, R.M., D.R. Rubinow, J.C. Ballenger 1984. Conditioning, sensitization, and kindling: implications for the course of affective illness. Pp. 432-456 in Neurobiology of Mood Disorders, R.M. Post and J.C. Ballenger, eds. Baltimore: Williams and Wilkins.

Purtell, J.J., E. Robins, M.E. Cohen 1951. Observations on clinical aspects of hysteria. A quantitative study of 50 hysteria patients and 156 control subjects. J. Am. Med. Assoc. 146:902-909.

Racine, R. J. N.W. Milgram, S. Hafner 1983. Long-term potentiation in the rat limbic forebrain. Brain Research 260:217-234.

Reiman, E.M., M.E. Raichle, E. Robins, F.K. Butler, P. Herscovitch, P. Fox, J. Perlmutter 1986. The application of positron emission tomography to the study of panic disorder. Am. J. Psychiatry 143:469-477.

Rodin, J. 1986. Aging and health: effects of the sense of control. Science 233:1271-1275.

Rosenthal, N.E., Cameron, C.L. 1991. Exaggerated sensitivity to an organophosphate pesticide. Am. J. Psychiatry 148:270.

Roy-Bryne, P.P. and T.W. Uhde 1988. Exogenous factors in panic disorder: clinical and research implications. J. Clin. Psychiatry 49:56-61.

Russell, M., K.A. Dark, R.W. Cummins, et al. 1984. Learned histamine release. Science 216:436-437.

Sanderson, W.C., R.M. Rapee, D.H. Barlow 1989. The influence of an illusion of control on panic attacks induced via inhalation of 5.5% carbon dioxide-enriched air. Arch. Gen. Psychiatry 46:157-162.

Sapolsky, R., C. Rivier, G. Yamamoto, P. Plotsky, W. Vale 1987. Interleukin-1 stimulates the secretion of hypothalamic corticotropin-relasing factor. Science 238:522-524.

Sapolsky, R.M., H. Uno, C.S. Rebert, C.E. Finch 1990. Hippocampal damage associated with prolonged glucocorticoid exposure in primates. J. Neuroscience 10:2897-2902.

Schatzberg, A.F., J.O. Cole 1991. Manual of Clinical Psychopharmacology, 2nd ed. Washington, D.C.: American Psychiatric Press, Inc.

Schwartz, G.E. 1982. Testing the biopsychosocial model: the ultimate challenge facing behavioral medicine? J. Consult. Clin. Psychol. 50:1040-1053.

Shavit Y., J.W. Lewis, G.W. Terman, R.P. Gale, J.C. Liebeskind 1984. Opioid peptides mediate the suppressive effects of stress on natural killer cell cytotoxicity. Science 223:188-190.

Shipley, M.T. 1985. Transport of molecules form nose to brain: transneuronal anterograde and retrograde labeling in the rat olfactory system by wheat germ agglutinnin-horseradish peroxidase applied to the nasal epithelium. Brain Res. Bull. 15:129-142.

Siegel, S., R.E. Hinson, M.D. Krank, J. McCully 1982. Heroin ''overdose'' death: contribution of drug-associated environmental cues. Science 225:733-734.

Simon, G.E., W.J. Katon, and P.J. Sparks 1990. Allergic to life: psychological factors in environmental illness. Am. J. Psychiatry 147:901-906.

Sloviter, R.S., G. Nilaver 1987. Immunocytochemical localization of GABA-, cholecystokinin-vasoactive intestinal peptide-, and somatostatin-like immunoreactivity in the area dentata and hippocampus of the rat. J. Compar. Neurol. 256:42-60.

Spiegel, D., J.R. Bloom, H.C. Kraemer, E. Gottheil 1989. Effect of psychosocial treatment on survival of patients with metastatic bleat cancer. Lancet 2(8673):888-91.

Spadaro, F. A. J. Dunn (1990). Intracerebroventricular administration of interleukin-1 to mice alter investigation of stimuli in a novel environment. Brain, Behavior, and Immunity 4:308-322.

Staudenmayer, H. and J.C. Selner 1990. Neuropsychophysiology during relaxation in generalized, universal 'allergic' reactivity to the environment. A comparison study. J. Psychosomatic Res. 34:259-270.

Stripling, J.S., D.K. Patneau, C.A. Gramlich 1988. Selective long-term potentiation in the pyriform cortex. Brain Research 441:281-291.

Sugerman, A.A., D.L. Southern, J.F. Curran 1982. A study of antibody levels in alcoholic, depressive, and schizophrenic patients. Annals Allergy 48:166-171.

Sullivan, M.D., R.A. Dobie, C.S. Sakai, W.J. Katon 1989. Treatment of depressed tinnitus patients with nortriptyline. Ann. Otol. Rhinol. Laryngol. 98:867-872.

Swanson, J.M., M. Kinsbourne 1980. Food dyes impair performance of hyperactive children on a laboratory learning test. Science 207:1485-1487.

Talbott, J.A., R.E. Hales, S.C. Yudofsky, eds. 1988. The American Psychiatric Press Textbook of Psychiatry . Washington, D.C.: American Psychiatric Press, Inc.

Terr, A.I. 1986. Environmental illness: a clinical review of 50 case. Arch. Intern. Med.

146:145-149.

Tham, R., B. Larsby, B. Eriksson, M. Niklasson 1990. The effect of toluene on the vestibulo-and opto-oculomotor system in rats, pretreated with GABAergic drugs. Neurotoxicology Teratology 12:307-311.

van Riezen H. and B.E. Leonard 1990. Effects of psychotropic drugs on the behavior and neurochemistry of olfactory bulbectomized rats. Pharmac. Ther. 47:21-34.

Walter, J.S., P. Meyers, J.M. Krueger 1989. Microinjection of interleukin-1 into brain: separation of sleep and fever responses. Physiol. & Behavior 45:169-176.

Weingartner, H., P. Gold, J.C. Ballenger, S.A. Smallberg, R. Summers, D.R. Rubinow, R.M. Post, F.K. Goodwin 1981. Effects of vasopressin on human memory functions. Science 211:601-603.

Weidermann, C.J. 1987. Shared recognition molecules in the brain and lymphoid tissues: the polypeptide mediator network of psychoneuroimmunology. Immunol. Lett. 16(3-4):371-378.

Woods, S.W., D.S. Charney, J. Loke, W.K. Goodman, D.E. Redmond, G.R. Heninger 1986. Carbon dioxide sensitivity in panic anxiety. Arch. Gen. Psychiatry 443:900-909.