• dose metrics (quantity and concentration of substance)

• degree of percutaneous absorption of substance

• degree of reactivity between substance and immune system.

Health status

Adverse skin reactions are determined by the health of the individual in a number of ways. In order to produce adverse effects, most irritants or allergens must cross the stratum corneum to reach the viable epidermis. If an essential oil is applied to damaged, diseased or inflamed skin the condition may be worsened, since larger amounts than normal may be absorbed and adverse reactions are more likely to occur.

Psychological stress can have an adverse effect on skin health. Several mechanisms have been proposed:

Interestingly, cutaneous allergic reactions were suppressed at both induction and elicitation phases in mice, by inhalation of the synthetic odorant ‘citralva’ (Hosoi & Tsuchiya 2000). It has been suggested that inhalation of anxiolytic odorants prevents the release of substance P and the consequent inflammatory cascade (Hosoi et al 2003). In mice, valerian oil inhalation reduced stress-induced plasma corticosterone levels (Hosoi et al 2001), presumably because it is anxiolytic.

Chronic illness and old age are often associated with reduced antioxidant status (see Ch. 12, p. 170) and this may increase skin reactivity. Endogenous antioxidants such as glutathione help prevent the formation of allergen-protein adducts, an important step in skin sensitization, and also help protect the skin against UVA-induced phototoxicity (Lutz et al 2001; Meewes et al 2001). Both scenarios involve reactions with free radicals, as does photocarcinogenesis. When oxidative stress overwhelms the skin’s antioxidant capacity, this leads to degenerative processes, which may include DNA damage (Briganti & Picardo 2003).

Age

In a survey of 57,779 adult dermatitis patients, the prevalence of fragrance allergy increased with advancing age, probably due to increasing cumulative exposure to fragrances (Uter et al 2001). In Europe, the frequency of fragrance allergy among dermatitis patients is low in the first two decades of life (2.5–3.4%). It gradually increases in females after the age of 20 to peak in the 60s at 14.4% of those tested, with a decline to 11.6% in the 80s. The prevalence in males rises more slowly and peaks at 13.7% in the 70s, declining to 10.8% in the 80s (Buckley et al 2003). The number of Langerhans cells is significantly reduced in aged skin, which may explain why it is less reactive to allergens (Kligman 1979; Fenske & Lober 1986).¹

Children up to three months old are theoretically at greater risk of skin sensitization due to the immaturity of their skin and its barrier function. However, contact allergies are in fact rare in young children, possibly due to a lack of exposure to allergens. Nickel and topical medications are among the most common causes of infant ACD (Carder 2005). In a survey of 304 unselected infants, no reproducible positive reactions to the FM were found at either 12 or 18 months (Jøhnke et al 2004). This does not mean that infant contact with fragrance materials or essential oils is devoid of risk; in fact intensive exposure may predispose a child to skin allergy problems later in life. In a Turkish survey of children with contact dermatitis, there was one reaction to the FM in 84 children aged 2–8 (1.2%) and four reactions in 89 children aged 9–16 (4.5%) (Onder & Adisen 2008).

Gender

According to Scheinman (1996) women are 3–4 times more likely to be affected by fragrance sensitivity than men. In one report the male/female distribution was 26.5/74.4% for the fragrance mix (ratio 2.8:1), and 17.6/82.4% for its constituents (ratio 4.7:1), in another the female/male ratio for the fragrance mix was 2.3:1 (Lunder & Kansky 2000; Temesvári et al 2002). A Swedish report of hand eczema found that ICD was much more common in women than men (Meding 1990).

The higher susceptibility of women may be due to a greater use of fragranced products, or because women are inherently more susceptible to skin allergies. Modjtahedi et al (2004) concluded that females are more at risk of ACD because of different exposure patterns, and not because of differences in intrinsic skin characteristics. However, women have more vigorous cellular immune reactions than men, and testosterone has been shown to suppress immune function (Darnall & Suarez 2009). Cytokine production decreased in men but increased in women, following a psychosocial stress test. Differences in response were also evident between women at different stages of the menstrual cycle (Kirschbaum et al 1999). These findings suggest that hormonal parameters do affect susceptibility, but because of differences in immune reactivity, not skin permeability.

Genetics and race

Individual susceptibility is in part dependent on genetic factors. For example, carriers of defective genes for certain isoenzymes of glutathione S-transferase or N-acetyltransferase are more susceptible to skin allergy (Lutz et al 2001; Nacak et al 2006). This may be because reduced antioxidant status leads to greater reactivity to allergens.

Atopic dermatitis is a highly heritable condition, with 81 reported genetic markers in people of Asian or Caucasian racial origin. Some of these genes are associated with skin barrier dysfunction, and more than half are associated with the immune response (Barnes KC 2010). There are conflicting data as to whether atopic dermatitis is a risk factor for ACD, though it is thought to be a risk specifically for type I allergic reactions. Those who consider it a risk factor include Larsen et al (1996b), Manzini et al (1998) and Mortz et al (2001). Those finding no association include Giordano-Labadie et al (1999), Mortz et al (2002) and Schafer T et al (2001). One study found atopic dermatitis to be a risk factor for hand dermatitis in massage therapists (Crawford et al 2004). The prevalence of atopic dermatitis varies considerably with age, race and location. In a pediatric practice in San Diego, California, incidence was 2% for Hispanics, 2.8% for non-Hispanic whites, 3.2% for people of mixed race, 3.7% for blacks, 5.6% for non-Filipino Asians and 8.5% for Filipinos (Baker RB 1999). In adults, prevalence varies from 0.2% in Scotland (> 40 years) to 3.1% in the Netherlands (Herd et al 1996; Verboom et al 2002).²

It is now recognized that a subset of the population is susceptible to multiple contact allergies, defined as positive patch tests to three or more allergens. It is estimated that 5% of contact dermatitis patients are ‘polysensitized’. These patients are more easily sensitized, 80% are women, and increased risk is associated with atopic dermatitis, leg ulcers and old age (Carlsen 2009). Polysensitization is thought to be due to genetic factors (Schnuch et al 2007b). The increased susceptibility to patch testing is evident at both induction and elicitation stages, and has been linked to genetic markers for TNF-α and IL-16 (Carlsen et al 2008; Schnuch et al 2008). Both of these cytokines are involved in the inflammatory response to an allergen, and corresponding genetic over-expression could therefore increase susceptibility to contact dermatitis (Reich et al 2003; Westphal et al 2003). It is not known whether polysensitization represents a distinct phenotype.

In a review encompassing 8,610 white and 1,014 black dermatitis patients, it was concluded that there were no differences in the overall response rate to a variety of allergens. Although some differences were seen in reactions to specific allergens, these are probably due to differences in exposure determined by ethnicity (Deleo et al 2002). There were no significant racial differences in adverse reactions to eugenol, cinnamaldehyde or cinnamic alcohol in 887 white and 114 black dermatology patients (Dickel et al 2001). Further to the particular sensitivities discussed in the section covering Pigmented contact dermatitis below, a report from Korea highlights a relatively high reactivity to sandalwood oil and cinnamic alcohol, both tested at 2%, in 422 dermatitis patients (80% female) with suspected fragrance allergy. A total of 35 fragrance chemicals and seven essential oils or absolutes were tested (An et al 2005).

• the substance applied (lipid and water solubility, molecular size, etc.)

• the circumstances of application (area of body, presence of occlusion, etc.)

• skin permeability (age, skin health, etc.).

Dose metrics

Skin reactivity to an allergen is highly concentration-dependent and this is clearly seen in Table 5.1. As the concentration of cinnamaldehyde is reduced, the number of people reacting decreases from 100% to zero. Concentration dependency is also seen in mixtures of fragrance materials, and in essential oils (Santucci et al 1987; Selvaag et al 1995; Frosch et al 2005b). Although there are inconsistencies, the data in Tables 13.1–13.4 and Tables 14.2–14.6 indicate that overall, the percentage of allergic reactions increases in relation to the concentration of the test substance used for test subjects with similar classifications of skin disease. (Note that concentrations used in patch testing are measured by weight, not by volume.)

However, the critical exposure determinant for evaluating skin sensitization risk is ‘dermal loading’, or dose per unit area of skin exposed, and not simply the total dose applied or the total area of application (Robinson et al 2000). Concentration and vehicle have been used in the theoretical risk assessment of cinnamaldehyde-containing products (Gerberick et al 2001a). Since all allergens show dose–response and threshold characteristics, it should be possible to use them safely, so long as they are incorporated at concentrations well tolerated by most individuals.

Frequency of exposure

Repeated exposure to the same substance over time is widely considered to increase the risk of adverse reaction. This probably explains why infants show no reactivity to the fragrance mix. However, frequent exposure to a low-risk fragrance allergen may not lead to a high incidence of ACD.

Fenn (1989) analyzed 400 fragranced products, and listed the most used fragrance constituents (bv volume %) and the percentage of products containing them in the USA and the Netherlands. These include, for example, linalool (90% USA, 91% Netherlands), 2-phenylethanol (82%, 79%), and linalyl acetate (78%, 67%). It is notable that these three constituents have very low reported incidences of sensitization. In a paper reporting on patch tests in eleven European dermatology clinics, with all 48 of the materials listed by Fenn (1989), only 10 tested positive and only one of these (citronellol) is in the top ten ranking for either country. Each material was tested on 100 or more dermatitis patients, and clinical relevance of the positive reactions was not established in a single case (Frosch et al 1995a). Therefore, frequency of exposure does not necessarily increase risk, even though it can be a factor in ACD.

The substance applied

Clearly, some essential oils are more likely to cause adverse reactions than others, and the presence and concentration of a relatively potent allergen is a major factor in ACD. The presence of adulterants or contaminants may also be a factor. The oxidation of essential oil constituents can increase risk (see Ch. 2, p. 10–11) because the oxides and peroxides formed are more reactive. This is seen with (+)-limonene, δ-3-carene and α-pinene and is due to the formation of oxidation products, some of which are more sensitizing than the parent compound (see Chapter 14, Constituent Profiles). However, the fact that an essential oil constituent is capable of being oxidized does not automatically mean that the use of an essential oil containing it presents a significant allergenic risk.

Once oxidation has begun it is difficult to halt, but the onset of oxidation can be prevented for a period of weeks, months or years by storage under nitrogen, by keeping a product cool, by screening it from UV rays, and by the addition of antioxidants such as (synthetic) butylated hydroxytoluene (BHT) or (natural) α-tocopherol. The efficacy of other natural antioxidants with regard to essential oil oxidation has not been much studied, but these include other tocopherols, rosmarinic acid, propyl gallate and ascorbyl palmitate. Combinations may be more effective than single chemicals. We recommend that effective antioxidant systems are used in preparations containing essential oils prone to oxidation.

In general, aged products are more likely to cause sensitization reactions. However, cinnamaldehyde may be markedly less allergenic in aged preparations, where it has oxidized to cinnamic acid, and in mixtures in which it can react with alcohols and/or amines to form non-allergenic compounds.

Interpreting patch test data

Flaws in research design are apparent in reports of fragrance allergy, among the most fundamental being a lack of clear and consistent criteria both for the diagnosis of ACD, and the absence of grading of the response to a test material. Many studies do not record reactions to the vehicle alone, as a control. Some studies have small numbers of participants and in many, unclear provenance and/or purity of the materials being tested is an issue (merely stating the name of the supplier does not clearly identify a substance. Single chemicals are available in various grades of purity, and essential oils can vary greatly with chemotype and origin). Almost none of the reports in which essential oils or absolutes are patch tested give a detailed compositional breakdown. Problems such as these considerably reduce the significance of the research, and yet such papers are widely cited as definitive, simply because they are published in peer-reviewed journals.

Most patch tests are conducted using petrolatum as a diluent, even though petrolatum is, albeit rarely, a clinically relevant cause of allergy (Kundu et al 2004; Rios Scherrer 2006; Tam & Elston 2006). In one study of skin reactions to fragrance materials, one person out of 167 (0.6%) had an allergic reaction to 100% petrolatum, and two (1.2%) had an irritant reaction to it (Larsen et al 1996b). In an analysis of data from 79,365 patients patch tested with petrolatum, 0.3% had unconfirmed positive reactions (Schnuch et al 2006). This suggests that reaction rates of 0.3% or less to fragrance materials have no significance. (The 0.3% figure does include doubtful reactions, but so do many of the reports of adverse reactions to fragrant substances.)

False-negative reactions may occur in maximation tests due to the small size of the test group (25 volunteers) and yet the rate of positive findings is sometimes so great as to suggest distortion with false positives (Marzulli & Maibach 1980). Albert Kligman, who designed the test, cautioned: ‘It must be thoroughly understood that the maximation procedure does not directly assess safety in use (except when negative). It does not predict the incidence of sensitization in a population of users’ (Kligman 1966). Approximately 50% of reactions to the fragrance mix are thought to be false positives (see Fragrance mix, below).

Patch testing is commonly carried out using a battery of potential allergens, not all of them fragrance materials. ‘Excited skin syndrome’ (ESS), or ‘angry back’ is characterized by multiple reactions to allergens (Bruynzeel & Maibach 1986). Patients with ESS may have a disposition to develop sensitivities to unrelated allergens, and the close proximity of the patches may be an exacerbating factor (Brasch et al 2006, Duarte et al 2002b). In one analysis, ESS developed in 39 of 630 dermatitis patients or 6.2% (Duarte et al 2002a). An ‘angry back’ reaction to patch testing may or may not suggest polysensitization (described above, under Genetics and race).

Another reason for false-positive reactions is the misidentification of irritants as allergens (Kligman 1998). A further problem is discordance between patch test systems, of which there are several. In using two different patch test systems, TRUE Test missed 50% of the clinically relevant FM reactions that were detected using Finn Chambers (Suneja & Belsito 2001). Although the Finn Chambers system appears to be more accurate in detecting clinically relevant fragrance allergy, TRUE Test is the only commercial system approved in the USA. When both TRUE Test and Trolab were used to test the FM simultaneously in 5,006 dermatitis patients, 218 (4.4%) reacted to TRUE Test, 464 (9.3%) reacted to Trolab, and 187 (3.7%) reacted to both (Mortz & Andersen 2010). There is clearly a significant problem when the results of patch testing are so dependent on which brand of patch is used.

‘Clinical relevance’ refers to whether a positive reaction to a patch test is relevant to the patient’s presenting condition. In one study of 702 patients, clinical relevance was not established in almost one third of cases with positive patch tests (Frosch et al 1995b). In a meta-analysis of ACD data that adjusted for clinical relevance, the prevalence of FM allergy in dermatitis patients was 3.4%, 3–4 times less than typical estimates (Krob et al 2004). Storrs (2007) found that, ‘many persons…have positive patch test reactions, but few of these individuals have clinical allergies to fragrances.’ Similarly, Hostýnek & Maibach (2003b, 2004b, 2004c) found that a clear cause–effect relationship has only infrequently or rarely been established for citronellol, geraniol and linalool, in spite of these being cited as frequent causes of ACD.

To summarize, a substantial percentage of patch tests using fragrant materials on dermatitis patients generate either false positives or are clinically irrelevant, and there are design flaws in much of the research. As regards the general population, because of threshold considerations, only a very small percentage of those who test positive to a fragrance material will develop related clinical manifestations (Kimber et al 1999). The patch testing data from the FM and its constituents do not readily extrapolate to actual use, because the severe conditions of patch testing are much more likely to result in allergic reactions. There are multiple reasons for this.

The incidence of patch test reactions to low molecular weight chemicals depends, not simply on the size of the patch used, nor on the total dose or concentration of substance applied, but on the dermal loading (the dose per area of exposed skin) expressed as μg/cm² (Friedmann 1990; Friedmann et al 1983a, 1983b; White et al 1986). Hostýnek & Maibach (2004a) suggest that 1% of an ingredient in a perfume spray results in a dermal loading of 26 μg/cm² of that ingredient (Gerberick et al 2001a). However, a patch test using 1% of the same ingredient has a dermal loading 11, 21, 38 or 68 times greater, depending on which of four patch test systems is being used: Finn Chamber 8 mm, Professional Products 1.9 cm × 1.9 cm, Webril 2 cm × 2 cm or Hill Top Chambers 19 mm, respectively (Robinson et al 2000). Therefore a patch test with 1% tea tree oil presents a higher risk than a product containing 1% tea tree oil. The concentrating effect of occlusion (see Ch. 4, p. 46), combined with the relatively long duration of exposure, adds considerably to the severity of patch testing.

The threshold concentration for eliciting a positive reaction varies inversely with the severity of the induction regime. So, the higher the percentage of substance used to initially induce the allergic response, the lower the percentage of the same substance that will elicit an allergic reaction. Therefore normal exposure to weak allergens, such as some fragrance materials, may induce ‘sub-clinical’ allergic states (a low-level induction state), in which no skin reaction occurs under normal conditions of use. However, the more severe conditions of patch testing may elicit a positive reaction. Consequently, dermatologists sometimes induce positive reactions that would not occur under everyday use conditions (Hostýnek & Maibach 2004a). For example, two dermatology patients became sensitized to sesquiterpene lactone mix when it was used on them in patch testing (Kanerva et al 2001b). In patients with multiple contact allergies, the risk of acquiring additional allergies increases with repeated patch testing (Carlsen 2009).

All of the above explains why patch testing with essential oils and fragrance materials suggests a much higher level of ACD incidence than is apparent from consumer complaints, or from the experience of most aromatherapists. It also means that realistic safe levels of fragrance materials are greater than a no effect concentration established in a series of patch tests would indicate. True prevalence or incidence rates from the use of naturally or synthetically fragranced products, whether in the general population or in dermatitis patients, cannot be determined from patch test data (Naldi 2002).

Patch testing is an indispensable tool for identifying the cause of a pre-existing skin allergy, and dermatologists consider the severe testing conditions appropriate for detecting susceptibilities that might otherwise have gone undetected. Unfortunately they may also be inducing or eliciting reactions that might not otherwise have occurred. Some consider that the risk of inducing sensitization is appropriate in diagnostic testing, but not when screening healthy volunteers (Basketter et al 1997; Uter et al 2004). There is debate about the circumstances in which patch testing should be used in clinical aromatherapy.

Induction

Elicitation

The priming of T lymphocytes in the lymph nodes is central to this process, and sufficient quantities of both hapten and Langerhans cells must be available for the allergenic response to be fully enacted (Kimber et al 2008). Finally, CD4 + T cells ‘call off’ the inflammatory reaction, and the skin lesions disappear, unless there is repeated exposure to the same haptens. The mechanistic basis of ACD is illustrated in Figure 5.2, and has been reviewed by several research groups (Basketter et al 1995, Belsito 1997, Fyhrquist-Vanni et al 2007). Fragrance allergens, such as isoeugenol and cinnamaldehyde, significantly alter the expression of Langerhans cell-surface proteins, while sodium lauryl sulfate, an irritant, does not. This shows a clear distinction between sensitization and irritation (Verrier et al 1999).

Essential oil constituents may be metabolized in the dermis to reactive metabolites, from pro-haptens to haptens. For example, isoeugenol is oxidized to a highly reactive paraquinone methide in mouse skin (Bertrand et al 1997). A proportion of cinnamyl alcohol (a pro-hapten) is converted by enzymes in human skin, to cinnamaldehyde (a hapten). Interestingly, cinnamaldehyde is in turn ‘detoxified’ in the skin to cinnamyl alcohol and cinnamic acid, and the extent of this detoxification is dependent on the amount of available enzymes (Smith et al 2000; Cheung et al 2003). This would explain why increasing concentrations of an allergen result in increasing incidence of sensitization.⁵

We know that thresholds exist for both induction and elicitation of allergic responses, although these are not absolute values (Kimber et al 1999; Andersen et al 2001). Variation between individuals at the induction stage may be due to differences in susceptibility to acquired sensitization (Kimber et al 2008). Thresholds in skin contact sensitization are both quantitative, for example, in the degree of protein haptenation, and qualitative, in terms of the type of immune response that is engaged. The existence of thresholds is evidenced by the observation that only a proportion of those exposed to a substance become sensitized and, of this group, only a proportion develop ACD (Kimber et al 1999).

In cross-sensitization, an individual reacts to a substance that is chemically and structurally similar to the one that caused the original sensitization (Basketter et al 1995).

Allergic contact dermatitis

Allergic contact dermatitis is the usual clinical consequence of delayed hypersensitivity. In addition to being elicited through direct skin contact, ACD may be elicited through airborne contact or ingestion. Airborne contact dermatitis has been reviewed by Dooms-Goossens & Deleu (1991), De Groot (1996) and Schaller & Korting (1995). It is more frequently caused by fine particles than by volatile molecules, but in any case is very rare: less than 0.2% of dermatitis patients (Komericki et al 2004). Although individuals who react to dermal testing do not always react when the same substance is given orally, foods such as citrus fruits and spices can play a role in ‘systemic’ contact dermatitis (Forsbeck & Skog 1977; Salam & Fowler 2001).

Skin reactions are generally limited to the area of skin to which the allergen is applied, but more widespread reactions occasionally occur (Bruynzeel et al 1984; Seidenari et al 1990a; Dharmagunawardena et al 2002). An individual with skin contact allergy to a fragrance material may have more frequent and more severe eye and airway symptoms to the same airborne substance (Elberling et al 2004).

Pigmented contact dermatitis

The term ‘PCD’ was coined by a Danish dermatologist, who described an epidemic of melanosis in Copenhagen (Osmundsen 1970). Although the cause was eventually found to be a whitener in a laundry detergent, PCD can also be precipitated by rubber products, azo dyes, cosmetics and fragrances. A distinguishing characteristic of PCD is a triggering of skin hyperpigmentation without the stimulus of UV light. Unlike photosensitivity, it occurs only in a very small percentage of individuals. Reactions are non-eczematous, are usually on the face, are more commonly seen in women than men, and are generally limited to darker-skinned individuals. It is thought that, in these cases, melanin passes into the upper dermis when the dermoepidermal junction is severely disturbed by inflammatory processes in the skin (Trattner et al 1999).

In a report from Spain, a 27-year-old Caucasian female developed dark brown hyperpigmentation on her face. Patch tests were positive for geraniol and lemon oil, and were not UV-dependent (Serrano et al 1989). In a review of 29 cases of PCD in Israel, four had positive and relevant reactions to the fragrance mix (Trattner et al 1999). In tests using guinea pigs with moderately colored skin, 100% jasmine ‘oil’ and 20% ylang-ylang oil caused hyperpigmentation that followed contact allergy, while 100% benzyl salicylate was a much less potent inducer of PCD. It was noted that it could take up to 30 days to reach a plateau of pigmentation, in comparison with about seven days for UVB irradiation. As part of the test procedure, the animals were injected with Freund’s complete adjuvant, an inflammatory substance. This test was said to resemble the hyperpigmentation often seen in Asian skin (Imokawa & Kawai 1987).

In Japan, in the 1960s and 1970s, there were reports of women developing areas of brown hyperpigmentation, invariably on the face. It was determined through systematic patch testing that the major causative agents were coal tar dyes and fragrances. Materials frequently implicated included jasmine absolute, the essential oils of ylang-ylang, cananga, geranium, patchouli and sandalwood, and the constituents benzyl alcohol, benzyl salicylate, geraniol and β-santalol. Major Japanese cosmetic companies stopped using various sensitizers in their products in 1977, and since 1978 the incidence of this condition is said to have decreased significantly.

The term ‘pigmented cosmetic dermatitis’ was coined by Nakayama et al (1984) to describe the cases seen in Japan. Biopsies suggested that the hyperpigmentation was due to melanin being released from cells in the basal layer of the epidermis when they were attacked by lymphocytes (Nakayama 1998). According to De Groot & Frosch (1998), the condition is virtually unknown in Western countries and is limited to central and eastern Asian races. However, pigmented cosmetic dermatitis is now seen either as a variant of PCD or as the same condition (Trattner et al 1999; Shenoi & Rao 2007).

Subsequent patch testing in Japan does not support the view that Japanese people are any more susceptible than Caucasians to ACD from the essential oils and constituents listed above, with the possible exception of sandalwood oil and benzyl salicylate (Itoh 1982; Sugai 1994; Sugiura et al 2000). However, there is an increased susceptibility to PCD, which is no doubt genetic. Hyperpigmentation is the most common cosmetic skin complaint in people of Asian ethnicity, who have a greater predisposition to congenital and acquired pigmentary skin disorders than other racial groups (Kurita et al 2009, Yu et al 2007).

The fragrance mix

A standard mixture of eight components in a base of petrolatum, known as the fragrance mix (FM), has been routinely used to screen for ACD in susceptible individuals since the late 1970s (Box 5.3). The components were originally used at 2% w/w each (totaling 16%) but, since this proved too irritating, the concentration was reduced to 1% each (Larsen 1977; Wilkinson JD et al 1989). Two FM ingredients, α-amyl cinnamic aldehyde (incorrectly reported as α-amyl cinnamic alcohol in some papers) and hydroxycitronellal are not found in essential oils. The emulsifier sorbitan sesquioleate is also present in the FM, at 5%. Since many fragrance-sensitive individuals react to Peru balsam (the raw material, not the essential oil), this is frequently tested in addition to the FM.

The average percentage of dermatitis patients reacting to the FM from ten sets of figures was 9.56%, with a low of 4.1% and a high of 15.0%.⁶ De Groot & Frosch (1998) estimated a range of 6–11%, and a multicenter study gave a figure of 10.5% for Central Europe (Schnuch et al 1997). White races are more susceptible than Asians to the FM, and to isoeugenol and oakmoss. Benzyl salicylate allergy is more common in Asia than in Europe or the USA (Larsen et al 1996a).

The percentage of the general population reacting to the FM is naturally lower than the 10% or so of dermatitis patients. FM allergy prevalence rates of 0.5%, 1.1%, 1.8%, 1.8% and 1.8% (mean 1.4%) have been found in five European studies that collectively patch tested 81,609 people (Seidenari et al 1990b; Nielsen & Menné 1992; Mortz et al 2002; Schnuch et al 2002a; Dotterud & Smith-Sivertsen 2007). This suggests that there is a sevenfold difference in the relative risk of fragrance allergy between dermatitis patients and the general population (10 ÷ 1.4 = 7.14). However, extrapolating FM data to the real-world risk presented by actual fragrances or essential oil use, is more complex than this, and can only ever be a rough approximation (see Interpreting patch test data above).

It is estimated that the FM identifies only 60–80% of people with fragrance allergy (Larsen et al 1998; Trattner & David 2003). An improvement has been reported by including additional substances in the mixture such as benzyl salicylate and benzyl alcohol (Larsen et al 1998). Citral has also been cited as one of the most common non-FM allergens (Frosch et al 2002a; Heydorn et al 2003b). Of the essential oils and absolutes tested, ylang-ylang oil has provided a relatively high percentage of responses, as has lemongrass. Table 5.6 compares reactions to the FM and to several essential oils. New fragrance materials not found in essential oils are also being identified as allergens. Hydroxyisohexyl 3-cyclohexene carboxaldehyde, a widely used aromachemical also known as Lyral, causes allergic reactions in 2–3% of dermatitis patients tested (Johansen et al 2003b), and it is now included in the patch test ‘standard series’ in Germany (Geier et al 2002). α-Hexylcinnamic aldehyde has also been cited as a potential allergen (Larsen et al 1998).

However, whether such testing truly identifies people with fragrance allergy is open to question. False-positive reactions to the FM are frequently cited, and the combination of its components may lead to a synergistic action (Santucci et al 1987). Johansen et al (1998a) found that a combination of two fragrance allergens in individuals allergic to both substances had a synergistic effect; the 1:1 mixtures elicited responses as if the doses were three to four times higher than those actually used. In four separate European studies, the probability that a patient with a positive reaction to the FM will have a reaction to any of its individual components was 50–56%, suggesting that one of every two FM reactions is a false positive (Frosch et al 1995b; Naldi 2002; Schnuch et al 2002b; Heydorn 2003b).

The emulsifier sorbitan sesquioleate (SSO), which is used at 5% in the FM, is itself an allergen (Tosti et al 1990) and it can increase reactions to the FM by 13.2%. In one small study, the addition of SSO increased allergic reactions to individual constituents by these amounts: eugenol 20%, isoeugenol 35%, oakmoss 39%, geraniol 66%, cinnamaldehyde 500% (Frosch et al 1995b). Orton & Shaw (2001) reported 12 clinically relevant cases of SSO allergy, which had caused false-positive FM results. Enders et al (1991) considered that SSO was not allergenic, but that it could cause false-positive results when mixed with the FM.

There have been reports of allergies to the FM increasing over the years 1979–1992 (Johansen & Menné 1995), 1992–1996 (Marks et al 1998) and 1985–1998 (Johansen et al 2000). These and other results suggest increasing reactions to fragrance materials (Becker et al 1994; Katsarou et al 1999). However, subsequent data point to a reversal of this trend. In a multicenter project (1996–2002), reactions to the FM increased from 10.2% in 1996 to 13.1% in 1999, and then showed a steady decline to 7.8% in 2002 (Schnuch et al 2004a). This may be due to a reduction of consumer exposure to FM ingredients in fragranced products in recent years.

Individual FM constituents have their own trends. Allergic reactions to cinnamaldehyde are decreasing over time, possibly because it is being used less in personal care products (Johansen & Menné 1995; Marks et al 1995, 1998; Buckley et al 2000; Nguyen et al 2008). Reactions to oakmoss absolute are high and have shown no signs of decreasing, either from 1979–1992 or from 1996–2002 (Johansen & Menné 1995; Schnuch et al 2004a). In a retrospective analysis of 3,636 patients, the incidence of reactions to patch testing with 1% isoeugenol showed a year-on-year increase in the UK. This may be due to an increased use of compounds chemically similar to isoeugenol (White et al 2007). There was no change in the frequency of reactions to geraniol in the UK from 1980 to 1996, in spite of a general upward trend for the period (Buckley et al 2000).

If we look at skin assays that address reactions to actual fragrances on the general population, the data show very much lower responses. From a total of 11,632 patch tests (on 10,400 subjects) with consumer products containing either eugenol as such, or clove leaf oil, there was only one instance of induced hypersensitivity at 0.05% eugenol, and one of pre-existing sensitization at 0.09% (Rothenstein et al 1983). Similarly, data from a total of 16,530 patch tests indicates that, as present in consumer products and fragrance blends, cinnamyl alcohol has no detectable potential to induce hypersensitivity (Steltenkamp et al 1980b), nor does citral, after 12,758 patch tests were conducted with products containing it (Steltenkamp et al 1980a). Cinnamaldehyde, when tested alone at concentrations up to 0.008%, or in consumer products at up to 0.6%, elicited no pre-existing hypersensitivity reactions in any of the 4,117 patch tests that constituted the survey (Danneman et al 1983).

In a Swedish study over 4.5 years (January 1994 through to June 1998) there was no significant difference in either frequency of complaints about products with and without fragrance, or in a paired comparison of 17 products marketed both with and without fragrance. The fragranced leave-on products contained up to 770 ppm, and the fragranced wash-off products up to 84 ppm of between 0 and 7 of the FM ingredients (Barany & Loden 2000). While this finding might not be very meaningful in many countries, in Sweden it is significant, as there is a simple and much-used system for such complaints. This suggests that a level of 0.07% would be safe for FM ingredients in leave-on products.

Enders et al (1989) found that only 69 of 142 (42.6%) patients who tested positive to the 8% FM in 1982/83, still tested positive to it in 1987. In 18 patients with previously positive patch tests, only 20 out of 26 tests (77%) were still positive on repeat testing (Soni & Sherertz 1997). This raises the possibility of eventual habituation to allergens in some people. It seems unlikely that the large difference seen by Enders et al could be totally explained by false positives in the earlier tests that did not show up on later testing.