Epidemiological studies of taste function: Discussion and perspectives

Abstract

Efforts to quantify the public health impact of chemosensation present significant challenges, including a strong need for testing methods suitable for field assessment. This discussion highlights several promising approaches to the population-based study of taste function; it also identifies key principles that should be considered when adapting laboratory-based taste tests for field use.

Several decades of clinical and laboratory research have sought to explain the relationship between gustatory sensation, intake behavior, and long-term health outcomes. This story is extremely complex and far from complete, but one important idea is quite clear: The capacity to perceive taste sensations significantly influences food choice, which contributes in a cumulative manner to health status. This idea implies that individual differences in taste sensation contribute to population differences in dietary health risk; it also implies that changes in taste sensation over time modify such risk. As a result, measures of taste function (and dysfunction) may play an important role in predicting dietary health susceptibility.

Generally, population data on chemosensory function are scant, hampering efforts to determine its role in public health. For taste function in particular, large-scale epidemiological analysis is sorely needed. Clearly, if taste function plays a role in long-term health, estimates of its variability – as well as the consequences of such variability – are desirable. Moreover, if the impact of taste dysfunction is ever to be identified, a clear definition of “normal function” must be established. To accomplish these goals, suitable methods for population-based taste assessment must 1) allow for valid comparisons of oral sensation across individuals, 2) distinguish sensory impairment vs. normal function that varies substantially across the population, 3) incorporate dietary measures associated with long-term health outcomes, and 4) be rapid, reliable, convenient, cost-effective, and user-friendly.

This symposium presents several approaches to the population-based study of taste function and its subsequent impact on health. Cruickshanks et al. describe their efforts to translate laboratory-based methods to field study, highlighting both the promise and the pitfalls of such modifications. Golding et al. take a known phenomenon that has been characterized in detail – individual differences in the bitter taste of 6-n-propylthiouracil (PROP) – and verify its genetic basis in a large population; they also show preliminary data linking it to dietary intake. Duffy et al. explore the utility of surveys of food hedonics (i.e., “liking” and “disliking”) in both laboratory and community settings, demonstrating that dietary preferences mediate the relationship between taste sensation and health-related measures such as adiposity and blood pressure.

As Hoffman and Davis observe, the innovations presented here represent ongoing efforts to integrate taste testing into the cluster of sensory examination tools used in clinical trials and epidemiological studies (i.e., the NIH Toolbox). This process has largely involved the translation of laboratory-based methods for field use, which has required substantial protocol modification. As these methods are further refined, we present three key topics that warrant continued consideration and discussion.

What attributes of taste function should be measured to assess the health impact of chemosensory impairment?

In our view, the identification of health-related aspects of chemosensory experience appears to be easier for olfaction than for taste. To begin, validated tests of olfactory function exist (e.g., UPSIT, San Diego Odor Identification Test),¹ but validated tests of gustatory function do not. This is largely because the tasks most typically used in diagnostic olfactory testing – odor recognition and odor thresholds – are effective measures of real-world olfactory experience. In contrast, taste thresholds and taste quality recognition are not effective assessments of real-world taste experience;² taste thresholds and suprathreshold taste intensities dissociate, often quite substantially. For example, radiation therapy typically impairs suprathreshold taste sensation, but taste thresholds often remain stable.³ On the other hand, taste thresholds are elevated with age, but suprathreshold taste remains essentially normal.⁴ So, in crafting a diagnostic tool to measure taste function, the challenge at hand is to assess real-world taste experience – which requires suprathreshold testing.

How can suprathreshold taste function be assessed in a manner that permits valid comparisons between normal individuals and those with impaired taste?

Because taste function varies so dramatically across individuals, another requirement of an effective taste test is that it employ psychophysical methods that permit valid comparisons across groups of interest – in this case, between normal and taste-impaired individuals. Category and visual analogue scales are commonly used to assess suprathreshold perceived intensities, but our research suggests that these methods often fail to provide valid comparisons.^5–7

Invalid comparisons distort results, but valid comparisons are possible

To illustrate this problem, consider an example involving pain sensation: Two people may report that they are experiencing the most intense pain they have ever felt, but this hardly implies that they are experiencing the same pain intensity. Pain intensity varies with the source of the pain and individuals experience pain from different sources. Accordingly, the label “most intense pain ever felt” cannot denote the same pain intensity to everyone.^{5, 7}

Can this problem be solved? In a recent study, we revealed differences in the most intense pain sensations experienced by two groups of individuals: men vs. women.⁸ We asked these individuals to rate a variety of everyday sensations from different modalities, including “brightest light ever seen” (typically the sun) and “most intense pain ever experienced.” We also asked subjects to note the source of the pain. For women who selected childbirth as their most intense pain, that pain was 23% more intense than the brightest light they had seen. For men, the most intense pain was about equal to the brightest light they had seen. Assuming that the perception of brightness does not vary between women and men, we concluded that childbirth-related pain for these women was 23% more intense than the worst pain the men had ever experienced. (For the small number of men whose most intense pain came from especially painful sources – such as kidney stones – their ratings of the most intense pain ever experienced were also higher than ratings for the brightest light they had ever seen.)

To the best of our knowledge, this technique was first used in the chemical senses to study differences in taste sensation between nontasters and tasters of PROP and/or its chemical relative, phenylthiocarbamide (PTC);^9–10 the technique was later formalized as “magnitude matching” by Marks and Stevens.^11–12 In essence, valid differences are obtained between two groups by comparing the sensation of interest (e.g., PROP bitterness) to an unrelated sensation (e.g., brightness), which acts as a standard. We assume that the perceived intensity of the standard across the two groups is, on average, the same. Expressing the perceived intensity of the sensations of interest relative to the standard allows us to see absolute differences in the sensation of interest.

Conventional category and visual analogue scales are traditionally labeled in terms of sensations of interest; a category scale to measure salty taste might be labeled “not at all salty” at one end and “extremely salty” at the other end.¹³ Scales like these carry two major flaws. First, they do not permit cross-modality matching; that is, we cannot compare a sweet substance to a salty substance since the scale only permits ratings of salty. However, we can create scales that permit cross-modality matching by making the top label refer to all sensory experience; such a scale would extend from “no sensation” to the “strongest sensation of any kind ever experienced.” Second, the labels on category scales are not spaced to give the scale ratio properties. For example, on the scale of salty taste just described, a rating of 6 is not twice as salty as a rating of 3. This problem can be resolved by re-spacing (and labeling) the categories, a task performed by several investigators.^14–17 Combining these “fixes” of conventional scales resulted in the general Labeled Magnitude Scale (gLMS).¹⁸ Further testing with variants of the gLMS shows that the key feature allowing valid comparisons across individuals and groups is the label at the top of the scale.⁷ (All of the taste assessments described in this symposium used scaling methods similar to the gLMS.)

Invalid comparisons arising from faulty scaling are not simply statistical artifacts; they may carry real-world implications. For example, medical professionals typically evaluate patient pain using a category scale ranging from 0 to 10, in which 10 denotes “the most intense pain ever experienced”.¹⁹ Suppose that analgesics are administered when patients report an intensity rating of 4 or above. Our data on sex differences in pain ratings (described above) suggest that a rating of 4 does not indicate the same degree of pain to everyone; for those women who selected childbirth as their most painful experience, it indicates a more intense pain sensation. Consequently, these women would endure more intense pain before receiving medication. Obviously, this is not a wise practice.

In concert with measures of oral anatomy, valid scaling methods have revealed the true breadth of taste function. Individuals who experience the most bitterness from PROP experience more intense sensations from virtually all taste substances,²⁰ in part because they tend to express the most taste buds.²¹ Taste buds are surrounded by nerve fibers mediating pain sensation,^22–24 while fungiform papillae (i.e., the structures housing taste buds on the anterior tongue) are more diffusely innervated by nerve fibers mediating tactile sensation.^25–27 Accordingly, those who experience the most bitterness from PROP also experience the most intense irritation on the tongue (e.g., the burn of chili peppers)²⁸ and the most intense oral viscosity (e.g., the creaminess of fats).^29–30 Studies using sensory scales derived from magnitude matching reveal these associations, but those using older methodologies often miss them.

The logic of sensory scaling shows similar promise for hedonic measurement

As with taste sensation, the experience of liking or disliking foods has intensity, and comparisons of the intensity of liking or disliking are subject to the same problems noted for sensory intensity. Just as sensory scales are usually labeled in terms of sensations of interest, conventional scales for food hedonics are usually labeled in terms of the specific area of interest. For example, the Natick 9-point preference scale, one of the most widely used tools to assess food hedonics, is a category scale where 1 = “extremely dislike”, 5 = “neutral” and 9 = “extremely like”.³¹ This scale cannot provide valid comparisons across groups unless we can assert that the labels denote the same intensities of liking and disliking to all. To illustrate, suppose that food liking rises with body mass index (BMI). This would mean that “extremely like” would denote a hedonic intensity that rises with BMI. As a result, the association between liking and BMI would be obscured by the Natick scale because it treats “extremely like” as if it denoted the same intensity of liking to everyone.

We do not have a solution to the hedonic problem parallel to that suggested for sensation because we have not yet identified a standard hedonic experience – that is, one that is unrelated to food liking. However, using the insights gained from sensory scaling, we have designed a hedonic scale which permits measurement of pleasure of all kinds (i.e., not just the pleasure evoked by foods). In brief, this scale consists of a line running from −100 (“the most intense disliking of any kind you have experienced”) through neutral to +100 (“the most intense liking of any kind you have experienced”). Using this scale (known as the “hedonic gLMS”), we have shown that, in fact, the intensity of food liking rises with BMI.³² In addition, Duffy et al.³³ have used the hedonic gLMS in studies described in this symposium; hedonic ratings for fat and sweet foods produced better associations with body measures of adiposity than did food intake measures. These reports suggest that valid comparisons of food liking/disliking can make substantial contributions to epidemiological studies in the chemical senses.

What are the advantages of spatial taste testing in assessments of taste function and impairment?

An examination of the cranial nerves mediating olfaction and taste reveals the need for spatial tests of taste function. Olfactory sensation is mediated by a single cranial nerve (CN I), while taste sensation is mediated by branches of three cranial nerves (CNs VII, IX, and X). As such, olfaction has no back-up mechanism if CN I is damaged, particularly if such damage is bilateral. However, taste perception is much more stable because of a phenomenon known as “taste constancy”, in which central interactions among cranial nerve inputs ensure that whole-mouth taste sensation remains robust, even in the face of localized taste damage. Some of these central interactions involve inhibitory feedback, suggesting that taste constancy occurs via spatial disinhibition: Damage to one taste nerve releases inhibition on others, resulting in intensification of sensations from other regions of the mouth. This intensification compensates for the loss due to localized damage; it is further facilitated by the fact that taste cues are referred perceptually to sites in the mouth that are touched. As a result, whole-mouth taste sensation remains largely unaffected.²

Knowledge of taste constancy goes back at least two centuries. Brillat-Savarin³⁴ reported on an unfortunate man who had his tongue cut out as punishment for a plot to escape from prison: “He replied … that he still possessed the ability to taste fairly well; that he could tell, with other more normal men, what was pleasant or unappetizing; but that very sour or bitter things caused him unbearable pain.” This report illustrates both taste constancy and the underlying disinhibition induced by localized taste damage; subsequent reports from patient cohorts indicate that chorda tympani damage results in increased glossopharyngeal taste intensity, particularly for bitter stimuli.³⁵ Further support for oral disinhibition comes from studies in which the chorda tympani was anesthetized to simulate localized damage, resulting in intensified taste sensations elsewhere in the mouth.^36–38 In other words, taste loss on the anterior tongue (i.e., chorda tympani) leads to enhanced taste on the posterior tongue (i.e., glossopharyngeal nerve), thus supporting taste constancy. Oral sensations beyond taste may be affected as well: Chorda tympani anesthesia also leads to more intense oral pain in some individuals.³⁹ In fact, the “unbearable pain” experienced by Brillat-Savarin’s subject may have resulted from intensification of the stinging sensations evoked by acid.

Brillat-Savarin’s description also demonstrates the utility of spatial testing to evaluate localized losses in epidemiological studies. One might argue that localized losses need not be assessed because of taste constancy, but these losses also intensify the non-taste sensations of oral pain and oral touch. For example, the chorda tympani passes through the middle ear on its way to the brainstem, rendering it vulnerable to damage via ear infection. Our recent data show that ear infections are associated with BMI gain.^40–41 We believe that, as with oral pain, chorda tympani damage intensifies the tactile sensations evoked by fats. This intensification appears to render high-fat foods more palatable,⁴² which would promote increased caloric intake and BMI gain.

Importantly, taste damage not only intensifies some oral sensations, but also produces oral phantoms.^38–39 The term “dysgeusia” has had a variety of definitions – including, in its broadest form, any pathology involving taste sensation – but we restrict its use to the experience of taste sensation in the absence of any obvious cause. Some cases of dysgeusia are genuine taste sensations in which the stimulus is actually present, but its source is not recognized by the patient; for example, medications can enter saliva and produce taste sensations. However, in our experience, most patients reporting chronic dysgeusia show localized taste losses upon spatial testing,⁴³ suggesting that these taste sensations are phantoms resulting from release of inhibition. In addition, oral burn phantoms, which are often equally distressing to patients, may arise from localized taste damage: Burning mouth syndrome appears to be an oral pain phantom induced by oral disinhibition following taste loss.⁴⁴

A brief evaluation of clinical disorders that compromise taste function supports the concept of oral disinhibition. These disorders often mention both taste and oral burn sensations that occur in the absence of stimulation (i.e., phantoms). To wit, in a classic paper on chorda tympani damage induced by stapedectomy (i.e., middle ear surgery), Bull⁴⁵ reported that “[t]he most constant symptom referable to taste was that of a metallic sensation on the tongue”, but “salty”, “bitter”, and “sore” sensations were also described.

To date, spatial taste testing has not been adapted for use in epidemiological studies because it is time-consuming and labor-intensive. However, recent work suggests that the identification of pathology in individual taste nerves may be accomplished with simpler spatial taste tests – tests that may be more suitable for epidemiological testing. Based on the logic of oral disinhibition, damage that is restricted to either the chorda tympani or the glossopharyngeal nerve will produce localized taste loss, but it should not substantially reduce whole-mouth taste. Consequently, ratios of taste intensity on the tongue tip vs. whole-mouth taste intensity are useful for evaluating chorda tympani damage. In a similar manner, ratios of taste intensity on the circumvallate papillae vs. whole-mouth taste intensity are useful for evaluating glossopharyngeal nerve damage. More widespread types of damage, such as combined chorda tympani + glossopharyngeal nerve loss, should reduce whole-mouth taste sensation; as normative data emerge, such damage should be apparent with whole-mouth testing (i.e., sip-and-spit).