Sensitization to 26 fragrances to be labelled according to current European regulation

Results of the IVDK and review of the literature


Professor Dr med. Axel Schnuch
Zentrale des IVDK
Von Siebold Sr. 3
Tel: +49 551 39 64 39
Fax: +49 551 39 60 95


To study the frequency of sensitization to 26 fragrances to be labelled according to current European regulation. During 4 periods of 6 months, from 1 January 2003 to 31 December 2004, 26 fragrances were patch tested additionally to the standard series in a total of 21 325 patients; the number of patients tested with each of the fragrances ranged from 1658 to 4238. Hydroxymethylpentylcyclohexene carboxaldehyde (HMPCC) was tested throughout all periods. The following frequencies of sensitization (rates in %, standardized for sex and age) were observed: tree moss (2.4%), HMPCC (2.3), oak moss (2.0), hydroxycitronellal (1.3), isoeugenol (1.1), cinnamic aldehyde (1.0), farnesol (0.9), cinnamic alcohol (0.6), citral (0.6), citronellol (0.5), geraniol (0.4), eugenol (0.4), coumarin (0.4), lilial (0.3), amyl-cinnamic alcohol (0.3), benzyl cinnamate (0.3), benzyl alcohol (0.3), linalool (0.2), methylheptin carbonate (0.2), amyl-cinnamic aldehyde (0.1), hexyl-cinnamic aldehyde (0.1), limonene (0.1), benzyl salicylate (0.1), γ-methylionon (0.1), benzyl benzoate (0.0), anisyl alcohol (0.0). 1) Substances with higher sensitization frequencies were characterized by a considerable number of ‘++/+++’ reactions. 2) Substances with low sensitization frequencies were characterized by a high number of doubtful/irritant and a low number of stronger (++/+++) reactions. 3) There are obviously fragrances among the 26 which are, with regard to contact allergy, of great, others of minor, and some of no importance at all.

Contact allergy (CA) to fragrance compounds is a well-recognized problem. The fragrance mix I (FM I), containing 8 different compounds, ranks second in the statistics of CA for many years (1) and during the 1990s an increase in sensitization frequency caused considerable concern (2). It has been estimated that 2–4% of the general population suffers from CA to fragrances contained in the FM I (3). However, fragrance CA is only partially diagnosed by patch testing with the FM in patients with a history of adverse reactions to fragrances, because further fragrance compounds are capable of causing CA to fragrance products (4–6). At least 2 of them have already been proven to be frequent sensitizers (7–9), namely, farnesol, and hydroxymethylpentylcyclohexene carboxaldehyde (HMPCC, Lyral®). This situation prompted the EU to address the issue of prevention. It was decided that, if consumer products contain one or several fragrances out of a list of 26 fragrances considered as contact allergens and compiled by the Scientific Committee on Cosmetics and Non-Food Products (SCCNFP) ( en.pdf), these fragrances should be labelled (10), while the remaining fragrance compounds are still globally labelled as ‘perfume’ as before.

We were interested in the actual frequencies of CA to these 26 fragrance compounds (see Table 2), which were therefore patch tested in consecutive, unselected patients by the IVDK network during a 2-year period.

Table 2.  Results of patch testing 26 fragrances
SubstanceFrequencies of sensitizationReaction pattern
n testn pos% Pos% Pos std.95% CIirr./?f++++++RIPR
  1. HMPCC, hydroxymethylpentylcyclohexene carboxaldehyde.Number of patients tested (n); number and proportion of patients reacting allergic (n pos/% pos), frequency of allergic reactions, standardized for age and sex (% Pos. std; column 5), together with the 95% confidence interval (95% CI); column 6) and reaction pattern of the patch test preparations (irr: irritant; f: follicular; ?: doubtful; RI: reaction index; PR: positivity ratio).

Group IUpper CI > 1.0
 Tree moss abs 1%1658452.72.41.7–3.2294281610.262.2
 HMPCC 5%213255022.42.32.1–2.614919361116250.571.9
 Oak moss abs. 1%2063462.22.01.4–2.6183291430.463.0
 Hydroxycitronellal 1%2063271.31.30.7–1.89020610.574.1
 Isoeugenol 1%2063261.31.10.7–1.617219700.273.1
 Cinnamic aldehyde 1%2063211.01.00.5–1.515119110.190.5
 Farnesol 5%4238380.90.90.6–1.23863071−0.178.9
Group IIUpper CI: 1.0 to >0.5
 Cinnamic alcohol 1%2063130.60.60.2–1.014012100.092.3
 Citral 2%2021130.60.60.3–1.01129310.069.2
 Citronellol 1%200390.40.50.1–0.8618100.188.9
 Geraniol 1%2063100.50.40.1–0.7181820−0.380.0
 Eugenol 1%2065110.50.40.2–0.7112920−0.181.8
 Coumarin 5%202080.40.40.1–0.8607100.187.5
 Lilial 10%200480.40.30.1–0.6316110.375.0
 Amyl-cinnamic alcohol 1%197770.40.30.1–0.671610−0.185.7
 Benzyl cinnamate 5%204260.30.30.1–0.670411−0.166.7
Group IIIUpper CI: ≤0.5
 Benzyl alcohol 1%216670.30.30.1–0.5121331−0.342.9
 Linalool (stab.) 10%240170.30.20.0–0.4121610−0.385.7
 Methylheptin carbonate 1%240160.20.20.0–0.4120501−0.383.3
 α-Amyl-cinnamic aldehyde 1%206240.20.10.0–0.3131310−0.675.0
 α-Hexyl-cinnamic aldehyde 10%201930.10.10.0–0.3101300−0.6100.0
 Limonene 2%239630.10.10.0–0.361210−0.466.7
 Benzyl salicylate 1%204120.10.10.0–0.271200−0.6100.0
 γ-Methylionon 1%200410.00.10.0–0.2114100−0.9100.0
 Benzyl benzoate 1%200310.00.00.0–0.180100−0.8100.0
 Anisyl alcohol 1%200410.00.00.0–0.151100−0.7100.0


The multicentre project IVDK (Information Network of Departments of Dermatology) is an instrument of epidemiological surveillance of CA and has been described in detail elsewhere (1, 11). Patch tests are performed in accordance with the recommendations of the International Contact Dermatitis Research Group (12) and the German Contact Dermatitis Research Group (DKG) (13). Patch test material is obtained from Hermal/Trolab, Reinbek, Germany. Patch test preparations are applied for 24 or 48 hr. Readings are done until at least 72 hr using the following grading based on international standards (14), further refined by the German Contact Dermatitis Group (13): neg, ?, +, ++, +++, irritant, follicular. The patch test results of every reading, a standardized history (including age, sex, atopic diseases, current and former occupation(s), presumptive causal exposures), along with final diagnoses and site(s) of dermatitis are assessed and documented. All data are transferred to the data centre in Göttingen in an anonymized format every 6 months.

During 4 periods of 6 months each, from 1 January 2003 to 31 December 2004, 25 fragrances (Table 1) were successively patch tested additionally to the standard series, i.e. in unselected patients, by departments of the IVDK. In the first period 8, in the second 6, in the third 3, and in the last period 8 compounds were added to the standard series, the number of patients tested with each preparation ranging from 1658 (tree moss) to 4238 (farnesol; tested during 2 periods). HMPCC was tested in the standard series in 21 325 patients throughout the study period.

Table 1.  Demographic description of the test populations of the different test periods using the items of the MOAHLFA index, and substances tested (for number of patients tested see Table 2)
  1. HMPCC, hydroxymethylpentylcyclohexene carboxaldehyde.

 Tested2349 2170 2599 2268 
OOccupational dermatitis34914.930914.236414.034615.3
AAtopic dermatitis43218.436817.045717.639117.2
A≥40 years150564.1146467.5174567.1160570.8
Substances tested during periodsBenzyl alcohol, coumarin; citronellol, benzyl salicylate, citral, benzyl cinnamate, α-hexyl-cinnamic aldehydeTree moss, lilial, γ-methylionon, amyl-cinnamic alcohol, anisyl alcohol, benzyl benzoate,Linalool, limonene, methylheptin carbonateOak moss abs, isoeugenol, hydroxycitronellal, cinnamic aldehyde, cinnamic alcohol, eugenol, geraniol, α-amyl-cinnamic aldehyde
HMPCC (Lyral®)

For the description of the demographic characteristics of patients tested the MOAHLFA index is used. MOAHLFA is the acronym for male, occupational dermatitis, atopic dermatitis, hand dermatitis, leg dermatitis, face dermatitis, and age >40 (1).

Frequencies of sensitization (as % of patients tested) were calculated both as crude proportions and proportions standardized for sex and age (15). Subgroups of patients defined by sensitization to an index allergen were analysed for concomitant reactions (crude proportions). The reaction index (RI) (16), relating the number of allergic reactions to the number of doubtful or irritant reactions, ranging from RI = −1 (all reactions nonallergic) to RI = +1 (all reactions being allergic), and the positivity ratio (PR), as the proportion (%) of + reactions out of the total number of allergic reactions (17), were calculated as parameters to assess the patch test preparation. A low RI (e.g. −0.8) together with a high PR (e.g. 100%) is indicative of a ‘problematic’(17) patch test preparation (see Table 2, group III), where a number of the ‘+’ reactions may be suspected to be falsely positive. For data management and analysis, the statistical software package SAS (version 9.1, SAS Institute, Cary, NC, USA) was used.



The population patch tested is described by the relative proportion of the following characteristics (Table 1), which differed slightly between the 4 periods: male (M) 37–39%, occupational dermatitis (O) 14–15%, atopic dermatitis (A) 17–18%, hand dermatitis (H) 26–29%, leg dermatitis (L) 11–15%, face dermatitis (F) 12–16%. The greatest variation during the study period was observed for ‘age 40 and above’ (A), namely between 64% and 71%, underscoring the need for standardizing the patch test results to enable comparisons. In contrast, regarding other potentially confounding factors, the patient population can be regarded as sufficiently stable (data not shown).

The MOAHLFA index of groups of patients reacting positive to certain compounds shows that patients reacting to fragrances may differ with regard to some demographic and clinical data: most of the patients were older (higher % of age >40), between 70% and 90%. Only patients reacting to coumarin (38% with age >40), citral (62%) and farnesol (66%) were considerably younger. In citral and coumarin positive patients men were somewhat overrepresented (77% and 63%) and the hands were by far the leading localization of eczema (54% and 75%). This corresponds to a higher percentage of suspected occupational dermatosis (31% and 25%), however, without any hints on specific related occupations or exposures (data not shown). Sensitization to benzyl alcohol, eugenol, geraniol, and cinnamal was (strongly) associated with leg dermatitis (29%, 46%, 40%, and 33%), indicating an important role of this risk factor, in contrast to low proportions in the case of patients reacting positively to HMPCC (7%), citral (8%), or farnesol (5%). Face dermatitis, which is often caused by cosmetics, was generally increased (>20%) in patients reacting to this group of fragrances, in particular to geraniol (50%), cinnamic alcohol (39%), and isoeugenol (23%), however, it was not increased in patients with positive reactions to citral, coumarin, eugenol, and benzyl alcohol (15%, 0%, 9%, and 14%, respectively). Finally, the very low number of men reacting to cinnamic alcohol (8%) is surprising.

Frequencies of sensitization

The frequencies of sensitization as expressed by the + to +++ reactions to each of the 26 fragrance compounds are presented in Table 2. The follicular reactions were considered to be nonallergic when calculating the percentages of positive reactions. Leading allergens with the upper confidence interval (CI) > 1.0% are oak moss, tree moss, HMPCC, hydroxycitronellal, isoeugenol, cinnamic aldehyde, and farnesol.

A second group of compounds with an upper CI between 1.0% and >0.5% comprises cinnamic alcohol, citral, citronellol, geraniol, eugenol, coumarin, lilial, amyl-cinnamic alcohol, and benzyl cinnamate.

The third group with upper CI of less than 0.5% assembles 10 compounds: benzyl alcohol, linalool, methylheptin carbonate, α-amyl-cinnamic aldehyde, α-hexyl-cinnamic aldehyde (AHCA), limonene, benzyl salicylate, γ-methylionone, benzyl benzoate (BB), and anisyl alcohol.

Sensitization to allergens of the first group is significantly more frequent than sensitization to allergens of the third group (c.f. CIs not overlapping).

Sex differences The following compounds caused sensitization in women more often than in men: HMPCC (2.7 versus 1.6) and cinnamic alcohol (0.9 versus 0.3), and (nonsignificantly) cinnamic aldehyde (1.2 versus 0.7), and eugenol (0.6 versus 0.2), whereas citral and coumarin appeared to be more frequent in men (1.2 versus 0.3 and 0.7 versus 0.3), respectively.

Reaction pattern Allergens of the first group exhibit a ‘favourable’ reaction pattern with a positive RI (except farnesol) and a low PR (except cinnamic aldehyde), i.e. with a considerable number of stronger (++/+++) reactions. Allergens of the third group have a negative RI throughout, with more irritant/doubtful than allergic reactions indicating the possibility of false positive reactions. However, stronger allergic reactions (lower PR of benzyl alcohol, limonene, α-amyl-cinnamic aldehyde) did occur even in this group, indicative of these substances to be – albeit rare – sensitizers.

Concomitant reactions Frequent concomitant reactions (crude rates) in subgroups of patients defined by a sensitization to an index allergen are presented in Table 3. For comparison, the reactions to the FM are disclosed also. With decreasing frequencies of sensitization the number of concomitant reactions increases (Table 3), oak moss and α-amyl-cinnamic aldehyde being the extremes. Concomitant reactions unrelated to fragrances emerged in a higher frequency also, namely nickel (40% and 33.3%) in farnesol and cinnamic alcohol positives and Lanolin (43%) in benzyl alcohol positives.

Table 3.  Frequent (>30%) concomitant reactions in patients with positive patch test reactions to fragrance compounds
Allergenn posFMFrequent concomitant reactions [% of those reacting positively to the index allergen (left column)]
  1. HMPCC, hydroxymethylpentylcyclohexene carboxaldehyde; FM, fragrance mix.

  2. aSeparate analysis of IVDK-data (unpublished) on 65 tree moss positives.

Oak moss4680.0None 
Tree moss4557.1Colophony34.9
Oak moss53.8*
HMPCC (Lyral®)50250.7Hydroxycitronellal30.2
Isoeugenol2683.3Balsam of Peru62.5
Oak moss30.8
Hydroxycitronellal2766.7Oak moss44.4
HMPCC (Lyral®)29.6
Cinnamic aldehyde2170.0Cinnamic alcohol42.9
Cinnamic alcohol1391.7Cinnamic aldehyde69.2
Balsam of Peru41.7
Balsam of Peru30.8
Eugenol1180.0Balsam of Peru70.0
Oak moss36.4
Cinnamic aldehyde36.4
Cinnamic aldehyde4/10
Compositae mix3/9
Oak moss3/10
Cinnamic alcohol3/10
Benzyl alcohol728.6Balsam of Peru4/7
α-Amyl-cinnamic aldehyde475.0All substances of the FMAt least 1/4


The aim of this study was to evaluate the importance of 26 fragrance compounds qualified as allergens by the EU, with the consequence that they have to be labelled if contained in a product (10), by patch testing consecutive, unselected patients with suspected allergic contact dermatitis with these compounds. The main result of this study is that these compounds are highly heterogeneous with regard to their impact as contact allergens. One group of compounds can undisputedly be regarded as important allergens, namely, our group I (Table 2). Another group of compounds is clearly allergenic but less important in terms of sensitization frequency (group II). In contrast, a third group comprises 10 compounds which have turned out to be (extremely) rare sensitizers in our analysis, or which in other instances may even be considered as nonsensitizers.

The allergens of group I: Since a long time, oak moss is recognized as an important sensitizer (18–20), particularly confirmed by testing the single constituents in FM I-positive patients (2, 21). In contrast, tree moss is not contained in cosmetic patch test series, and this is the first study in which tree moss was tested in a larger population. With a sensitization prevalence of 2.4% in our patients it turned out to be the most frequent allergen. Among the >100 constituents identified in oak moss and tree moss (18), atranol and chloroatranol (degradation products of atranorin and chloroatranorin) figure as the most potent allergens (22, 23). The EC3 values in the Local Lymph Node Assay (LLNA) (0.6% and 0.4%) classify them as strong sensitizers (cited in SCCP/0847/04). These substances were found in many perfumes in considerable concentrations (24). In view of the extreme potencies of both substances, which are able to elicit in use tests with concentrations on the ppm level (0.0005%) and in patch testing in the ppb level (0.000015%), the Scientific Committee on Cosmetics (SCCP) came to the conclusion that both substances should not be present at all in cosmetic products (SCCP/0847/04). In a separate data analysis (unpublished) we found that only 53.8% (35/65) tree moss allergic patients reacted to the FM I (containing oak moss), indicating that oak moss alone is probably not suitable for diagnosing tree moss allergy.

Furthermore, tree moss may contain abietic acid and dehydroabietic acid, important allergens of colophony after oxidation. This may explain the high proportion of concomitant reactions to colophony (Table 3). In contrast, oak moss patch test preparations shown to be not contaminated by tree moss or its resins, do not react considerably together with colophony (25, 26).

The notable CA risk associated with exposure to the synthetic fragrance HMPCC has recently gained interest (4, 8, 9, 27–29). In a large European study, 50 out of 1855 consecutive patients (2.7%) tested with a screening series for fragrance allergy had a positive reaction to HMPCC (8). In only 60% of these HMPCC-hypersensitive individuals, there was a simultaneous CA to the FM (8). HMPCC was tested by the DKG in 3245 consecutive patients of which 62 (1.9%) had positive reactions (9). A group of European dermatologists developed a mix of 6 fragrances (FM II), including HMPCC, to be tested in consecutive patients (28). In this study, 2.9% reacted to the medium concentration (14%) of this mix, and >1/3 were shown to be allergic to its compound HMPCC, making it the dominating allergen of the FM II (28). In view of a sensitization frequency of 2.3% observed in our study, HMPCC must be regarded as one of the most important single fragrance allergens. In addition to the evident impact of HMPCC derived from clinical epidemiology, a use test (repeated open application test, ROAT) with 2 different concentrations (0.5% and, if negative, 3% with ethanol as vehicle) was conducted in 18 patients allergic to HMPCC as diagnosed by prior patch testing (29). In 16 of 18 cases (89%) a positive use test developed, 11 reacting to the low and 5 only to the high concentration.

Isoeugenol was, for a long time, the second most important fragrance allergen of the FM I (19, 21, 30). According to our data hydroxycitronellal (1.3% positive, Table 2), although less potent but probably more often used, is an allergen as important as isoeugenol (1.1%). Marked concomitant reactivity may be explained by the presence of isoeugenol in balsam of Peru or coexposure together with oak moss (Table 3). According to human and animal predictive tests isoeugenol is classified as a sensitizer of moderate potency (Human class 2, LLNA EC 3 of 1.3%) (31, 32). The biochemical mechanism of sensitization through isoeugenol may rely on the formation of an orthoquinone, whereas eugenol may react via a phenolic radical mechanism, explaining the relative rarity of concomitant reactions between the 2 substances (33) (Table 3). Interestingly, cases of sensitization in Japanese populations seem significantly less frequent than in Caucasian populations (34). Quantitative aspects of isoeugenol CA were assessed by use and patch tests (35–37).

Hydroxycitronellal is a fragrance widely used in perfumes and products of daily life (38, 39), and, if contained in higher concentrations, capable of causing CA (40). In experimental sensitization tests it was shown to be a weak to moderate sensitizer (18, 41), with an EC3 value of 20% (31), and a weak experimental elicitor (42). In a Human Repeated Insult Patch Test it was shown that humans can be sensitized by 5% and that challenge with concentrations as low as 1% can elicit reactions (41). Reviewing the literature, the Research Institute on Fragrance Materials (RIFM) expert panel concluded that 1% hydroxycitronellal is not likely to induce sensitization in humans with repeated exposure (43). If applied in the course of a ROAT in the area of former exposure and eczema (axillae), very low concentrations (0.032–0.32%) were sufficient to elicit reactions (44). In contrast, in an experimental model (hand immersion study) simulating real life exposure to diluted dish washing liquids containing the fragrance in higher concentrations (250 ppm), the development of visible eczema was not increased in sensitized subjects (45), although the combined exposure to an allergen and a detergent may enhance the allergic (patch test) reaction (46).

Testing with 4% yielded 13.8% allergic reactions (34), a very high rate caused by the selection of patients, and additionally by the high patch test concentration. Frosch et al. (27)used patch test concentrations of 1% in pet., which resulted in a lower frequency of 0.75% allergic reactions. In our study, hydroxycitronellal turned out to be an important sensitizer, with 1.3% allergic reactions, of which a considerable number were strong (++/+++), expressed by a low PR (Table 2). A relatively high rate of concomitant reactions to oak moss and HMPCC is very probably due to coexposure in cosmetics.

In numerous experimental studies (with and without adjuvant) in animals and humans the skin sensitizing properties of cinnamic aldehyde (cinnamal) have been demonstrated (47). Sensitization in humans was induced by concentrations in the range of 0.5–1% (48, 49). The EC3 values derived from the LLNA range from 1.4% and 3.1%, classifying the substance as a moderate to strong sensitizer (32, 50, 51). In an exposure-based risk assessment it was shown that the use of 1000 ppm (0.1%) in a leave-on cosmetic would pose an unacceptably high risk of sensitization, whereas the same concentration in a shampoo would pose an acceptable risk (52). In a ROAT it has been shown that deodorants containing cinnamic aldehyde can elicit axillary eczema in sensitized individuals, with concentrations ranging from 0.01% (1 reacting) to 0.1%, the majority reacting to 0.032% (53), complementing former results where 8/22 individuals reacted to 0.1% (54). In a larger European study in unselected patients cinnamic aldehyde was one of the more frequently diagnosed contact allergens (0.9%) (27), which is in line with the frequency found in our study (1.0%). A higher frequency was found in a North American study (1.7%) which can probably be explained by preselection of patients (55). In different test periods 10%, 13%, and 20% of FM I-positive patients reacted to cinnamic aldehyde (21, 30, 56), also reflecting the decreasing sensitization to cinnamic aldehyde (19, 57), which is probably due to a restricted use of this fragrance (58).

Cinnamic aldehyde is partly transformed into cinnamic alcohol and mainly to cinnamic acid, whereas the transformation of cinnamic alcohol to the aldehyde is minimal (59, 60). Concomitant reactions between aldehyde and alcohol (Table 3) may therefore be due to metabolism, cross-reactions, and coexposure, but isolated reactions to either compounds do occur quite often (∼50%) (61).

Farnesol, a fragrance with some antimicrobial activity (62), is often used in deodorants, exploiting this very additional property. The first larger patch test study with 5% and 10% in petrolatum (pet.) was done in 1985 in Japan. The suitable concentration was found to be 10%, and the proportion of positive reactions was 1.1% (63). In 2 European multicentre studies 1855 and 1703 consecutive patients were patch tested with farnesol (5% in pet) (4, 28). 0.5% and 0.35% reacted allergic, but the number of doubtful reactions was high (1.1% and 1.76%). With a frequency of 0.9% (CI: 0.6–1.2), 30 + and 8 ++/+++ reactions in our study (Table 2) and with results from a 20-year-old study reporting again 1.1% positive reactions (64) together with several case reports (65–67) there is no doubt that farnesol must be considered an important sensitizer, although its potency was classified as only weak (68) to moderate (69) in animal experiments.

The distinct pattern of low frequencies of cosensitizations with other fragrances, but a surprisingly high association with nickel (Table 3), may be cautiously interpreted as a hint on more specific exposure pathways to this fragrance. The fact that female clerks of younger age (7) and nickel sensitization were somehow overrepresented in farnesol-positive patients may give rise to further thoughts on a social class-specific consumer behaviour, having in mind, that nickel-allergy could be regarded as a ‘socially guided allergy’(70).

The allergens of group II according to our list (see Table 2), although frequently used (38, 39, 71–73), are clearly less important with regard to frequency of sensitization – partly (in contrast to our results) with no reactions at all in a number of previous studies (4, 27, 28, 34, 63, 64, 74–76)– as well as limited sensitizing potency (18, 31, 32, 49, 77, 78). However, the reaction profiles (RI and PR) indicate that a few unequivocal (stronger) allergic reactions did occur, leaving no doubt that these compounds do have sensitizing properties.

In this group, the ‘unusual’ characteristics of those patients sensitized to citral and coumarin stand out. In both subgroups, men prevailed, and occupational dermatoses was more often suspected, going along with the hands as most commonly affected anatomical site. As this pattern had been noted before, it was suggested to include coumarin and citral in a special ‘hand eczema series’(79).

Interestingly, concomitant reactions between citral and geraniol occurred frequently (83%; Table 3). This may be due to coexposure, but probably also to cross-reactions, as both compounds are structurally closely related. In larger European studies, citral ranked third and second among several fragrances (4, 28). In a selected group of patients with hand eczema, who were tested with 14 fragrances found in household products, citral was found to be an important allergen (28/658; 4.3%) (79). In a second study the authors further analysed the patients reacting to citral (80). Beside the 28 positive cases, there were 82 cases with irritant patch test reactions, showing citral as an allergen and an irritant, at least under patch test conditions.

In contrast to our results and a former study (4), no cases of coumarin allergy were observed in a recent study (28). Kunkeler et al. (81) reviewed all the cases tested in their department in Amsterdam between 1978 and 1997 (n ∼ 14 000). They identified 58 patients with at least a + reaction (0.4%). Given the missing sensitization to coumarin in predictive tests, the authors speculate that the coumarin positive cases may be sensitized by (alkyl-substituted) coumarin derivatives.

Compounds of group III (Table 2) can be regarded as very rare allergens or apparently turn into allergens only after substantial oxidation (e.g. limonene and linalool) (82). In the case of some compounds the alleged sensitizing properties can even be doubted, considering the possibility of false positive reactions in view of mainly doubtful or irritant reactions (see RI and PR Table 2). Even if there may indeed be single cases of CA to these compounds reported in the world literature, these may be more indicative of an increased individual susceptibility, than of the substance-specific sensitizing properties, as in the case of highly purified white pet. Ph. Eur. (83). One example may be AHCA, with one unequivocal case of CA to AHCA, displaying multiple sensitization to other fragrances seen by one of the authors (A.S.). Although the substance is used as a positive control, as a calibrant for comparing the consistency of LLNA responses (84), the number of documented allergic cases in humans is very low. We observed 3+ and 11 irritant reactions (Table 2). In a recently published European study on 1701 consecutive patients, 2 allergic and 16 doubtful reactions to 10% AHCA were reported (28).

Doubts are even more justified with regard to the ‘allergen’ BB. It was shown to be a ‘weak sensitizer’(69), and rare cases had been observed (85–89). However, in view of its frequent use as an acaricide with 25% concentration (90), and the virtual lack of allergic reactions to this topical drug, it seems not plausible to regard BB as a ‘significant contact allergen’(91).


This study emphasizes again the need for a ‘differentiated look’ on fragrances as contact allergens (21). The 26 fragrances were allocated to 3 different classes according to their importance in terms of frequency of sensitization, only. Nevertheless, a differentiated evaluation of compounds of each class may be needed for overall evaluation, considering not only frequency of sensitization, but also the amount of exposure or use, as well as allergenic potency, eventually together with the real extent of exposure to (highly) oxidized materials. For some substances regulation in terms of use concentration restriction, labelling, or even ban is needed (group I), for others labelling alone may be an adequate instrument of secondary prevention (group II). For at least some of the group III compounds neither restrictions nor labelling seems justified. Based on very low frequencies of sensitization despite a widespread use and (very) low potencies in predictive tests, these are probably not significant allergens at all. Further studies in other European countries on large test populations like ours should be performed as there might be regional differences in sensitization. The decision of the EU on labelling 26 compounds (because they were considered as allergens) should be revised. Some manufacturers of cosmetics have decided to use none of the ‘26 annex compounds’ but other compounds instead which do not need to be labelled. However, these alternative fragrance compounds may be less well studied from a toxicological point of view, and as they are mostly unknown to dermatologists, they are not patch tested, and possible CA remains undetected. In summary, prudent labelling must take into account both the risk profile of the respective compound, and subsequent replacement policies of manufacturers which may, in turn, have serious implications for consumer safety.