Vous êtes sur la page 1sur 14

Contact Dermatitis 2010: 62: 18–31 © 2010 John Wiley & Sons A/S

Printed in Singapore. All rights reserved


CONTACT DERMATITIS

Review Article

Formaldehyde-releasers in cosmetics: relationship


to formaldehyde contact allergy
Part 2. Patch test relationship to formaldehyde contact allergy,
experimental provocation tests, amount of formaldehyde released,
and assessment of risk to consumers allergic to formaldehyde

Anton de Groot1 , Ian R. White2 , Mari-Ann Flyvholm3 , Gerda Lensen1 AND Pieter-Jan Coenraads1
1 Department of Dermatology, University Medical Center Groningen, University of Groningen,

The Netherlands,
2
Department of Cutaneous Allergy, St John’s Institute of Dermatology, St. Thomas’ Hospital, London, UK, and
3
National Research Centre for the Working Environment, Copenhagen, Denmark

This is the second part of an article on formaldehyde-releasers in cosmetics. The patch test relationship
between the releasers in cosmetics to formaldehyde contact allergy is reviewed and it is assessed
whether products preserved with formaldehyde-releasers may contain enough free formaldehyde to pose
a threat to individuals with contact allergy to formaldehyde. There is a clear relationship between
positive patch test reactions to formaldehyde-releasers and formaldehyde contact allergy: 15% of
all reactions to 2-bromo-2-nitropropane-1,3-diol and 40–60% of the reactions to the other releasers
are caused by a reaction to the formaldehyde in the test material. There is only fragmented data
on the amount of free formaldehyde in cosmetics preserved with formaldehyde donors. However,
all releasers (with the exception of 2-bromo-2-nitropropane-1,3-diol, for which adequate data are
lacking) can, in the right circumstances of concentration and product composition, release >200 p.p.m.
formaldehyde, which may result in allergic contact dermatitis. Whether this is actually the case in
any particular product cannot be determined from the ingredient labelling. Therefore, we recommend
advising patients allergic to formaldehyde to avoid leave-on cosmetics preserved with quaternium-15,
diazolidinyl urea, DMDM hydantoin, or imidazolidinyl urea, acknowledging that many would tolerate
some products.

Key words: benzylhemiformal; 5-bromo-5-nitro-1,3-dioxane; 2-bromo-2-nitropropane-1,3-diol;


cosmetics; diazolidinyl urea; DMDM hydantoin; formaldehyde; imidazolidinyl urea; preservative;
quaternium-15; sodium hydroxymethylglycinate. © John Wiley & Sons A/S, 2010.
Accepted for publication 22 July 2009

In the first part of this article (1), key data on relevance of patch test reactions, and their frequency
formaldehyde-releasers used in cosmetics (benzyl- of use in cosmetics. In this second part, the patch
hemiformal, 5-bromo-5-nitro-1,3-dioxane, 2-bromo- test relationship of these chemicals to formalde-
2-nitropropane-1,3-diol, diazolidinyl urea, DMDM hyde are reviewed and assessed whether products
hydantoin, imidazolidinyl urea, quaternium-15 and preserved with formaldehyde-releasers may con-
sodium hydroxymethylglycinate) were presented, tain enough free formaldehyde to pose a threat to
including applications, frequency of sensitization, individuals who are contact allergic to formaldehyde.
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 19

Relationship between Formaldehyde-Releasers indicate that at least in a number of patients co-


and Contact Allergy to Formaldehyde reactions between formaldehyde and releasers are
When patients are tested with formaldehyde and due to formaldehyde sensitivity. On the other hand,
one or more formaldehyde-releasers, concomitant the investigation of Perret and Happle also indi-
positive reactions may be observed to formalde- cates that sensitivity to diazolidinyl urea can exist
hyde and (one or more) donors. It is often assumed independent of formaldehyde release, as none of the
that this is due to ‘cross-allergy’ to formalde- seven patients allergic to diazolidinyl urea but not
hyde (more correctly termed ‘pseudo-cross-allergy’ to formaldehyde reacted to any of other releasers.
in our opinion), i.e. that the patch test reac- The existence of allergy to the releaser 2-
tion to the donor is caused by the formalde- bromo-2-nitropropane-1,3-diol per se (independent
of formaldehyde contact allergy) was demonstrated
hyde present in the test material of the releaser.
25 years ago by Storrs and Bell (6). These inves-
In favour of this assumption is that often more
tigators described seven patients reacting to patch
than one releaser reacts in formaldehyde-sensitive
tests with 2-bromo-2-nitropropane-1,3-diol 0.25%
patients, even when the preservatives are not struc-
and/or 0.5% and/or 1% aqua or pet. They had
turally related (cross-reactions unlikely). In a recent
all been sensitized by Eucerin® cream, which was
study, for example, among 219 patients aller-
widely used in their geographic area as an emollient
gic to formaldehyde and/or formaldehyde-releasers
for dry and damaged skin, containing 0.05% 2-
(quaternium-15, diazolidinyl urea, imidazolidinyl
bromo-2-nitropropane-1,3-diol. The Eucerin® cream
urea, 2-bromo-2-nitropropane-1,3-diol, DMDM
was also patch test positive in all patients. None of
hydantoin, and ethylene urea/melamine-formaldehy-
them were allergic to formaldehyde and they did not
de resin), 76 (35%) were positive to only one of react to either quaternium-15 or imidazolidinyl urea.
these compounds, 53 (24%) reacted to two, 27 All patients performing a ROAT for a maximum
(12%) reacted to three, 31 (14%) to four, 25 (11%) of 2 weeks on normal skin with Eucerin® cream
to five, and 7 (3%) patients reacted to six of the developed dermatitis. This indicated that 2-bromo-
tested formaldehyde(releasing) materials (2). Of 59 2-nitropropane-1,3-diol allergy per se exists and that
patients, patch test positive to quaternium-15 1% 0.05% 2-bromo-2-nitropropane-1,3-diol in a cream
petrolatum (pet.), 31 (53%) were also allergic to is sufficient to give a positive patch test and a
formaldehyde (tested 2% aqua). However, of seven positive ROAT in 2-bromo-2-nitropropane-1,3-diol
patients allergic to quaternium AND to 2-bromo-2- allergic patients who are not sensitive to formalde-
nitropropane-1,3-diol, imidazolidinyl urea, or both, hyde. In addition, the investigators had seen another
ALL were also allergic to formaldehyde, attest- four patients reacting to 2-bromo-2-nitropropane-
ing to the role of formaldehyde in multiple reac- 1,3-diol but NOT to Eucerin® cream. These patients
tions to releasers (3). In a USA study with 454 DID co-react to formaldehyde and to quaternium-15
patients allergic to formaldehyde, 38 patients (8.4%) and/or imidazolidinyl urea. They concluded that the
reacted to both formaldehyde and two releasers amount of formaldehyde in the cream containing
and 4 reacted to all three donors tested (2-bromo- 0.05% 2-bromo-2-nitropropane-1,3-diol was insuffi-
2-nitropropane-1,3-diol, imidazolidinyl urea, and cient to cause a positive patch test (6).
quaternium-15). Of these patients reacting to two
or three releasers (suggesting formaldehyde to be
the cause), quaternium-15 (which probably releases How often are patients reacting to a releaser also
the most formaldehyde) was positive in ALL (4). allergic to formaldehyde?
Perret and Happle investigated 13 patients with pos- In many studies, patients have been patch tested with
itive patch test reactions to diazolidinyl urea 2% formaldehyde and one or more releasers, especially
aqua. Six of them were also allergic to formalde- with quaternium-15, which is included in the base-
hyde. All 13 patients were additionally tested with line series in both Europe and the USA. This gives
the formaldehyde-releasers urea formaldehyde (8% an opportunity to study the possible relationship
aqua), glyoxal urea (10% aqua), and quaternium-15 between formaldehyde and the donors, at least in
(1% pet.). All six patients reacting to diazolidinyl this diagnostic setting. In Table 1, data are presented
urea AND to formaldehyde also reacted to the on the frequency of co-reactions to formaldehyde in
other releasers, whereas of those patients not aller- patients allergic to formaldehyde-releasers.
gic to formaldehyde, NONE had a positive reaction There is a consistent picture for 2-bromo-2-
to any of the formaldehyde-releasers. It was con- nitropropane-1,3-diol. Less than 25% (and usually
cluded that in the six patients allergic to diazolidinyl far less) of 2-bromo-2-nitropropane-1,3-diol-sen-
and formaldehyde, the reactions to diazolidinyl urea sitive patients co-react to formaldehyde. This
were probably caused by formaldehyde in the patch suggests that most patch test reactions to 2-bromo-
test (5). Such studies, of which there are many, 2-nitropropane-1,3-diol indicate sensitivity to this
20
DE GROOT ET AL.

Table 1. Percentages positive to formaldehyde in patients reacting to formaldehyde-releasersa


2-Bromo-2-nitro-propane-1,3-diol Diazolidinyl urea DMDM hydantoin Imidazolidinyl urea Quaternium-15 Reference
Country and Number of %+ Number of %+ Number of %+ Number of %+ Number of %+
period of study patients formaldehyde patients formaldehyde patients formaldehyde patients formaldehyde patients formaldehyde
Finland 2001–2007 15 83% (7)
UK 2004–2005 87 17% 78 55% 64 53% 129 59% (8)
USA 1992–2004 788b 69%b 515b 83%b 756b 63%b (9)
Germany 1994–2004 180 37% (?) (10)
IVDK 1992–1995 160 5% 129 12% 85 15% 64 47% (11)
Austria 1992–1993 43 2% 36 11% 70 27% (12)
UK 1982–1993 89 25% 91 25% 156 24% 299 55% (13)
USA 1989–1991 58 81% (14)
USA 1982–1989 244 20% 87 46% 181 65% (15)
The Netherlands 1984–1988 13 46% (5)
UK 1985 16 38% (3)
UK 1983–1984 59 53% (3)
USA 1983–1987 89 55% (15)
USA <1980 24 83% (16)
a Only studies with at least 10 positive reactions to the releasers are included. Data on <10 patients are presented in Refs (2, 6, 7, 17–21).
b Mean numbers/percentages in two patient groups reacting to the releaser in water and to the releaser in pet. vehicle.
Contact Dermatitis 2010: 62: 18–31
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 21

Table 2. Percentages positive to formaldehyde-releasers in patients allergic to formaldehydea


Percentage of patients allergic to formaldehyde with positive reactions to:
Number of 2-Bromo-2- DMDM
Country and patients allergic nitropropane- Diazolidinyl hydantoin Imidazolidinyl Quaternium-15
period of study to formaldehyde 1,3-diol (%) urea (%) (%) urea (%) (%) Reference
Finland 2001–2007 73–79 4 11 5 21 (7)
UK 2004–2005 142 10 30 23 52 (8)
USA 1992–2004 2225 24.50 19 17 (9)
UK 1992–1993 629 4 4 6 47 (13)
Austria 1992–1993 105 1 4 18 (12)
USA 1982–1989 454 10 10 31 (4)
USA <1980 30 30 (16)

a
Only studies with at least 20 positive reactions to formaldehyde are included. Data on <20 patients are presented in Refs (2) and (6).

Table 3. Testing with DMDM hydantoin and its parent compound in patients allergic to formaldehyde (22)
Group I. 14 patients tested with DMDM hydantoin
Concomitant allergy to DMDM hydantoin
Patients allergic to formaldehyde 3.0% (only) 1% (also) 0.3% (also) Negative
1% only =7 3 4
1% and 0.3% =4 2 2
1% and 0.3% and 0.1% =3 2 1
Group II. 21 patients tested with MDM hydantoin
Concomitant allergy to MDM hydantoin
Patients allergic to formaldehyde 3.0% (only) 1% (also) 0.3% (also) Negative
1% only = 12 1 11
1% and 0.3% =4 1 1 2
1% and 0.3% and 0.1% =5 2 2 1

preservative per se and are not related to formalde- showed that about one out of five formaldehyde-
hyde allergy. The picture for quaternium-15 is also sensitive individuals was also patch test positive to
fairly consistent. In all but two studies, half the DMDM hydantoin. The relationship between posi-
number of patients or more reacted to formalde- tive patch tests to formaldehyde allergy and patch
hyde, indicating that formaldehyde is in >50% of test reactions to DMDM hydantoin was investigated
the cases responsible for the quaternium-15 patch in detail by De Groot et al (22). Thirty-five patients
tests. Less consistent are the figures for diazo- allergic to formaldehyde were patch tested with
lidinyl urea (12–81% formaldehyde co-reactions), formaldehyde and dimethylhydantoin (DM hydan-
DMDM hydantoin (37–83%), and imidazolidinyl toin; the parent compound). Twenty-one of these
urea (11–63% co-reactions to formaldehyde). formaldehyde-sensitive subjects were also tested
with MDM hydantoin (DM hydantoin +1 molecule
formaldehyde), the other 14 also with DMDM
hydantoin (two molecules formaldehyde). Test con-
How often are patients reacting to formaldehyde centrations were 0.1–0.3–1% for formaldehyde,
also allergic to one or more 0.3–1–3% for DMDM hydantoin and MDM hydan-
formaldehyde-releasers? toin, and 1–3–10% for DM hydantoin, all in water
The reverse situation, i.e. what percentage of (w/w). Control tests in 97 patients had excluded irri-
patients allergic to formaldehyde reacts to the tancy. The results are shown in Table 3. No patient
releasers, can also be deducted from many studies. reacted to the parent compound DM hydantoin. Of
The available data are shown in Table 2. the 14 patients in Group I, 8 (57%) reacted to
For 2-bromo-2-nitropropane-1,3-diol, the data are DMDM hydantoin. Of the 21 patients in Group II, 7
consistent: less than 10% of patients allergic to (33%) reacted to MDM hydantoin. In both groups,
formaldehyde react to 2-bromo-2-nitropropane-1,3- most negative reactions to (D)MDM hydantoin were
diol. On the other side of the spectrum, in all but observed in patients reacting only to the 1% solution
two studies at least 30% of the patients reacting to of formaldehyde. Patients with ‘stronger’ allergies to
formaldehyde also reacted to quaternium-15. Less formaldehyde (reacting to 0.1% and/or 0.3% also)
patients (<30%) will react to diazolidinyl urea, and tended to show more positive reactions even to the
still fewer (<23%) to imidazolidinyl urea. One study lower concentrations of (D)MDM hydantoin.
22 DE GROOT ET AL. Contact Dermatitis 2010: 62: 18–31

Table 4. Percentages positive to other formaldehyde-releasers in patients allergic to a particular formaldehyde-releaser


Percentage of patients allergic to releaser with positive reactions to:
Number of 2-Bromo-2-
Country and patients allergic nitropropane- Diazolidinyl Imidazolidinyl
period of study to formaldehyde 1,3-diol (%) urea (%) urea (%) Quaternium-15 (%) Reference
2-Bromo-2-nitropropane-1,3-diol
UK 2004–2005 87 (100) 10 8 13 (8)
UK 1982–1993 89 (100) 2 2 13 (13)
USA 1982–1989 244 (100) NT 5 14 (4)
Diazolidinyl urea
UK 2004–2005 78 11 (100) 60 49 (8)
UK 1982–1993 91 2 (100) 75 21 (13)
USA 1981–1991 58 5 (100) 29 41 (14)
Imidazolidinyl urea
UK 2004–2005 64 12 75 (100) 47 (8)
UK 1982–1993 156 1 44 (100) 22 (13)
USA 1982–1989 87 10 NT (100) 40 (4)
Quaternium-15
UK 2004–2005 129 9 30 23 (100) (8)
UK 1982–1993 154 2 5 6 (100) (13)
USA 1982–1989 181 15 NT 17 (100) (4)
NT, not tested.

Using the data from Rosen and McFarland (23), a consistent pattern: few reactions to and from other
it was estimated that 3% DMDM hydantoin con- formaldehyde-releasers, formaldehyde plays a minor
tains 0.3% (3000 p.p.m.) and the 0.3% solution role. There is in two of three studies a very high
0.02% (200 p.p.m.) free formaldehyde. Thus, 3 of co-reactivity between imidazolidinyl urea and dia-
14 patients reacted to 200 p.p.m. upon patch testing. zolidinyl urea, higher than with formaldehyde, indi-
This may be comfounded somewhat by the fact that cating (true) cross-allergy or a reaction to a common
the patients were also concurrently tested with the (non-formaldehyde) constituent (see under imidazo-
higher concentrations and with formaldehyde (addi- lidinyl urea and under diazolidinyl urea, Ref. (1)).
tive effect). These data demonstrate that aqueous Of the patients reactive to diazolidinyl urea or imi-
preparations of DMDM hydantoin, in concentrations dazolidinyl urea, up to nearly half have reactions to
comparable to those used in cosmetic products, con- quaternium-15, indicating a role for formaldehyde
tain enough free formaldehyde to cause dermatitis in because these chemicals are structurally unrelated.
some patients allergic to formaldehyde when patch Conversely, however, of the patients reacting to
tested (22). quaternium-15, only up to 23% react to imidazo-
lidinyl urea and up to 30% to diazolidinyl urea.
In such cases, the patient may well be allergic to
How often are patients reacting to a formaldehyde
formaldehyde, but the amount of formaldehyde in
donor also allergic to one or more other the diazolidinyl urea and imidazolidinyl urea test
formaldehyde-releasers? substances is not enough to cause a positive patch
In four large studies, the dynamics of patch test reac- test reaction.
tivity between the various formaldehyde-releasers The recent experience of the British Contact
was studied. The results are shown in Table 4. Dermatitis Society with formaldehyde and formal-
Again, for 2-bromo-2-nitropropane-1,3-diol, there is dehyde-releasers is shown in Table 5 (8). This table

Table 5. Relationship between formaldehyde and formaldehyde-releasers: UK experience 2004–2005 (8))


2-Bromo-2-nitro- Imidazolidinyl
Formaldehyde (%) Quaternium-15 (%) propane-1,3,diol (%) urea (%) Diazolidinyl urea (%)
Formaldehyde (100) 52 10 23 30
Quaternium-15 59 (100) 9 23 30
2-Bromo-2-nitropropane- 17 13 (100) 8 10
1,3-diol
Imidazolidinyl urea 53 47 12 (100) 75
Diazolidinyl urea 55 49 11 60 (100)
This table may be read as follows: horizontal row, 52% of patients allergic to formaldehyde are also allergic to quaternium-15 and 23%
are also allergic to imidazolidinyl urea. Conversely, vertical row, 59% of patients allergic to quaternium-15 are allergic to formaldehyde,
and 53% of the patients allergic to imidazolidinyl urea (55% for diazolidinyl urea) are also allergic to formaldehyde.
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 23

may be read as follows: horizontal row, 52% of ROAT on the arm (25), so false-negative reactions
patients allergic to formaldehyde are also allergic may have occurred.
to quaternium-15 and 23% are also allergic to Twelve patients allergic to formaldehyde per-
imidazolidinyl urea. Conversely, vertical row, 59% formed a ROAT with a cream containing 1%
of patients allergic to quaternium-15 are allergic to DMDM hydantoin (twice daily for a week in the
formaldehyde, and 53% of the patients allergic to flexor aspect of the lower arm) and in four patients
imidazolidinyl urea (55% for diazolidinyl urea) are (33%), dermatitis was observed after 3–6 days.
also allergic to formaldehyde. These four subsequently tested the same cream but
The associations between formaldehyde and the now containing 0.25% of the preservative (estimated
releasers were highly significant (Kendall’s τ - to contain approximately 200 p.p.m. free formalde-
b correlation score [P < 0.001]); this was also hyde) in a ROAT; the test was negative in two,
the case for the associations between the vari- produced only itching but no visible changes in one,
ous releasers (8). As the chemical structures of and provoked mild dermatitis in one (cave at: only
these compounds–with the exception of diazo- 1 week) (22). It was concluded that cosmetics pre-
lidinyl urea and imidazolidinyl urea–are dissimi- served with 0.25% DMDM hydantoin may pose a
lar, there is probably a major role for formalde- threat to certain patients allergic to formaldehyde.
hyde in these co-reactivities (which may then be Zachariae et al. (25) performed an experimen-
called ‘pseudo-cross-reactions’ to formaldehyde). tal use test exposure study with a cream pre-
Indeed, the strongest association is seen between served with diazolidinyl urea in patients allergic
formaldehyde and quaternium-15, known to release to formaldehyde. They developed a dose-escalating
the most formaldehyde of the releasers. 2-Bromo-2- design, including four diazolidinyl urea concen-
nitropropane-1,3-diol, on the other hand, releases lit- trations, and exposure to three different anatomi-
tle formaldehyde, and the association with formalde- cal regions (upper arm, neck, and face) each of
hyde, though statistically significant, is far weaker. 2-week duration. The test population consisted of
Of course, concomitant (‘dual’) sensitization to 30 patients allergic to formaldehyde, 10 patients
formaldehyde and the parent molecule may also allergic to diazolidinyl urea (of which 7 had also
contribute to these associations in a number of reacted to formaldehyde) and 10 non-allergic con-
cases (11, 24). trols. The four versions of the test cream were
preserved with 0.05%, 0.15%, 0.30%, and 0.60%
diazolidinyl urea (maximum allowed in the EU:
Experimental Provocation Tests with Cosmetics 0.5%), which were found to contain 130, 370, 730
Containing Formaldehyde-Releasers in and 1500 p.p.m. formaldehyde, respectively. They
Formaldehyde-Sensitive Subjects were applied twice daily for 2 weeks on a 5 × 5 cm
Fifteen patients with contact allergy to formalde- area of the flexor aspect of the upper arm (ROAT). If
hyde (n = 7) and/or the formaldehyde-releasers no reaction occurred after 2 weeks, the cream appli-
quaternium-15 (n = 8), DMDM hydantoin (n = cation continued on the neck for another 2 weeks,
1), imidazolidinyl urea (n = 2), and/or 2-bromo-2- and on the face, in a similar fashion, for further
nitropropane-1,3-diol (n = 6) but patch test negative 2 weeks. If a positive reaction was observed at any
to diazolidinyl urea underwent a ROAT (twice daily time before the end of the 6-week period, the patient
application to the antecubital fossa for 2 weeks) was examined and withdrawn from the study.
with a moisturizer preserved with 0.2% diazolidinyl The results were as follows: of the 10 formalde-
urea. In only one patient, the ROAT was clearly hyde allergic patients tested with the cream contain-
positive. Afterwards, this patient was retested with ing 0.05% diazolidinyl urea, none had a positive
diazolidinyl urea 1% pet. and 1% aqua and (now) reaction. The cream preserved with 0.15% diazo-
had a positive reaction to 1% aqua but not to diazo- lidinyl urea elicited two reactions in 10 patients,
lidinyl urea 1% pet. The amount of free formalde- 0.3% induced seven positive reactions and the
hyde was not investigated. It was concluded that ROAT with 0.6% resulted in seven positive reac-
in patients allergic to formaldehyde and/or one tions in 10 patients. Thus, with an increasing con-
or more formaldehyde-releasers, complete avoid- centration of diazolidinyl urea from 0.15% to 0.6%
ance of other formaldehyde-releasers to which the (corresponding to formaldehyde concentrations of
patient is patch test negative is not always necessary, 370–1500 p.p.m., respectively), a dose dependent
‘notably if the amount of formaldehyde released by increase in patients’ reactivity was seen, not only
that compound is below the threshold of reactivity in the number of reactors but also in the strength
for virtually all formaldehyde-sensitive individuals of the reaction. On the basis of this investigation,
(around 250 p.p.m.)’ (2). It should be realized, how- the amount of formaldehyde that does not elicit der-
ever, that a ROAT on the neck or on the face may be matitis in formaldehyde-sensitive subjects should be
positive in the absence of a positive reaction to the between 130 and 370 p.p.m.
24 DE GROOT ET AL. Contact Dermatitis 2010: 62: 18–31

Of the 10 patients allergic to diazolidinyl urea (7 corticosteroid cream. From this study, it may be
of which also reacted to formaldehyde), 9 had a pos- concluded that the presence of 175 p.p.m. formalde-
itive ROAT to the cream with 0.15% diazolidinyl hyde in a corticosteroid cream has a negative effect
urea, 6 on the arm, and 3 in the neck (compared on the therapeutic efficacy of the corticosteroid on
with 2 of 10 in patients reacting to formaldehyde experimentally induced allergic contact dermatitis in
but patch test negative to diazolidinyl urea) (25). patients allergic to formaldehyde and such creams
There are several possible explanations for this: should therefore not be used by formaldehyde aller-
(i) Patients were not only allergic to formaldehyde gic individuals (26).
but also to another compound in diazolidinyl urea Flyvholm et al. (27) saw a positive reaction to
(see also under chemical structure). The threshold a stay-on cosmetic preserved with 0.3% imidazo-
for a patch test response is lowered by simultane- lidinyl urea (containing approximately 300 p.p.m.
ous presence of more than one allergen (25). (ii) free formaldehyde) in a ROAT lasting 1 week in
These patients were more sensitive to formaldehyde one of three patients allergic to both formaldehyde
(and thus reacted to formaldehyde and diazolidinyl and imidazolidinyl urea. In 20 patients with allergy
urea) than those allergic to formaldehyde, for whom to formaldehyde but negative reactions to imida-
the amount of formaldehyde in the diazolidinyl urea zolidinyl urea, the ROAT produced some follicular
test preparation was insufficient to elicit a positive papules in 5 of 20. These were considered to be a
reaction. (iii) In at least two patients with a posi- mild allergic reaction, as none of the controls sub-
tive ROAT, formaldehyde was negative. This may jects had any reaction at all (27).
have been false negative. But if they were truly
non-allergic, the allergy to the (other) compound in
diazolidinyl urea that caused the reaction was strong The Amount of Formaldehyde Released by
enough to induce a positive ROAT. From these Formaldehyde-Releasers
data, it may (tentatively) be concluded that levels There is only fragmented data available on the
of 370 p.p.m. formaldehyde (0.03–0.04%) released amounts of formaldehyde that will be released by
by a formaldehyde-releaser may–under certain con- formaldehyde donors. These amounts depend on the
ditions of use–induce allergic contact dermatitis nature of the releaser, its concentration, the pH of
on normal skin in a number of patients sensitive the product, the temperature (the higher the temper-
to formaldehyde. Some patients may react to even ature the more formaldehyde is present in solution
lower formaldehyde concentrations. When patients after constant time) (28), the age of the product
react to both formaldehyde and the releaser itself, (upon storage increased levels of formaldehyde will
the risk of developing allergic contact dermatitis be released), the level of microbial contamination,
increases. and the other constituents of the products containing
Isaksson et al. (26) have investigated whether the releaser (11, 28–30).
the presence of 175 p.p.m. formaldehyde released The pH may greatly influence the release of
from the preservative imidazolidinyl urea in a cor- formaldehyde. Diazolidinyl urea, for example,
ticosteroid cream influences (i.e. diminishes) the releases formaldehyde foremost in an alkaline prod-
cream’s therapeutic efficiency in treating experimen- uct, i.e. with high pH value (31, 32). Imidazo-
tally induced allergic contact dermatitis. In seven lidinyl urea in lower concentrations is stable in
patients allergic to nickel and formaldehyde, patches a low pH (acidic) environment, but with increas-
of 8 × 9 cm of allergic contact dermatitis were pro- ing pH the amount of free formaldehyde also
voked on the outer part of both upper arms. In increases (32, 33). The same holds true for DMDM
a double-blind fashion, these were treated with a hydantoin, whereas sodium hydroxymethylglycinate
corticosteroid cream containing approximately 175 tends to release formaldehyde more in acidic solu-
p.p.m. formaldehyde and (the other arm) with a cor- tions (32). In a 0.1% quaternium-15 solution,
ticosteroid cream of similar strength not containing the amount of formaldehyde is strongly depen-
formaldehyde. As controls served 17 patients aller- dent on the pH: <200 p.p.m. formaldehyde at pH
gic to nickel but not to formaldehyde. In two of >9, 250–300 p.p.m. at pH 7 and 600 p.p.m. at
seven patients allergic to formaldehyde (29%), the pH 3 (34).
eczema treated with the formaldehyde-containing Other ingredients, i.e. the nature and composi-
cream healed completely within 4 weeks, compared tion of a product, may also influence the amount of
with 12 of 17 (71%) in the control group; the dif- free formaldehyde. In a protein-free shampoo, 0.1%
ference was statistically significant. There was no quaternium-15 released 482 p.p.m. formaldehyde,
correlation between the formaldehyde reactivity, as but in a shampoo with protein only 122 p.p.m. Pre-
measured by serial dilution testing, and the ten- sumably, the protein forms complexes with released
dency to healing in the group allergic to formalde- formaldehyde (23). In mascara preserved with 0.4%
hyde and treated with the formaldehyde-containing imidazolidinyl urea, the levels of formaldehyde were
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 25

<10 p.p.m., but in creams or milks containing released upon complete hydrolysis (see diazolidinyl
0.3–0.5% the concentration of formaldehyde varied urea, chemical structure (1)).
between 100 and 130 p.p.m. (28). DMDM hydantoin is produced by reacting
Several methods for determining formaldehyde formaldehyde with DM hydantoin (5,5-dimethyl-
have been used (28, 32–38) which make compar- hydantoin). The composition of DMDM hydantoin
isons difficult. None of the methods is absolutely as determined by gas chromatography is as fol-
reliable or does always give reproducible results, lows: 94–98% DMDM hydantoin, 2.5–3% MDM
which may in part account for (sometimes greatly) hydantoin (monomethyloldimethylhydantoin), other
varying results. With some methods (including the dimethylhydantoin formaldehyde products comprise
official EU adopted method), the releasers deliver in the balance. One mole of DMDM hydantoin can
the presence of the reagent free formaldehyde con- release a total of 2 mol of formaldehyde. The con-
tinuously during its determination, which may result centrations of free formaldehyde depend on the
in too high and non-reproducible values (11, 28). pH; in alkaline environment more formaldehyde
Several tests are now available to overcome this is released. DMDM hydantoin is said to contain
problem (28, 32, 34, 36). It should be realized 0.5–2% free formaldehyde and concentrations in
that free formaldehyde present in products preserved cosmetics range from 0.1% to 0.6% (42, 43). It
with a formaldehyde-releaser may not only originate has been suggested that most formulations will,
from the releaser itself but may also comprise excess therefore, contain between 20 and 120 p.p.m. free
formaldehyde used to synthesize the preservative, formaldehyde (31, 44). However, these estimates
formaldehyde from other formaldehyde-based raw may be too low, as a cream preserved with 0.15
materials used to prepare the cosmetic product, wt% DMDM hydantoin contained 0.013 wt% (130
formaldehyde used as a preservative for the raw p.p.m.) free formaldehyde (32).
materials themselves, and from degradation of non- Imidazolidinyl urea is prepared from 3 mol of
formaldehyde ingredients such as polyethylenegly- formaldehyde and 2 mol of allantoin (compare: dia-
col ethers (39). zolidinyl urea from 4 mol of formaldehyde and
1 mol of allantoin). From each molecule of imida-
Benzylhemiformal is the reaction product of ben-
zolidinyl urea, two formaldehyde molecules can be
zyl alcohol and formaldehyde, and is rarely used in
released (2 mol formaldehyde/mole imidazolidinyl
cosmetic products. It degrades rapidly in an aqueous
urea) (44). Release is hastened by increased tem-
solution, and at pH 5 the maximum possible amount
perature and pH. Only about 75% of this theoretical
of 300 p.p.m. formaldehyde of an 0.1% benzylhemi- amount of formaldehyde in imidazolidinyl urea will
formal solution can be demonstrated (36). Emeis be released upon complete hydrolysis (44).
et al. (32) found that benzylhemiformal at 2% w/w Sodium hydroxymethylglycinate. Emeis et al.
aqueous solution mixes spontaneously, with decom- investigated the release of formaldehyde from
position of 97% of the benzylhemiformal. Higher sodium hydroxymethylglycinate with 13 C NMR
dilution leads to instantaneous, complete decompo- spectroscopy (32). At a concentration of 1% and a
sition of the releaser into formaldehyde and benzyl pH of 5.5, sodium hydroxymethylglycinate decom-
alcohol. At the maximum permissible concentration poses completely into formaldehyde and sodium
of 0.15% benzylhemiformal in cosmetic rinse-off glycinate. In a basic environment, decomposition is
products, the concentration of free formaldehyde incomplete. It is not until further dilution to 0.5%
should according to these authors be just under 0.05 w/w–the maximum concentration allowed in cos-
wt% (500 p.p.m.) (32). metic products–that it decomposes completely at a
2-Bromo-2-nitropropane-1,3-diol only releases pH of 8.5. The concentration of free formaldehyde
formaldehyde at higher pH values. Breakdown of 2- then is 0.12% w/w (1200 p.p.m.) (32).
bromo-2-nitropropane-1,3-diol results in formation Rosen and McFarland (23) investigated protein
of formaldehyde and bromonitroethanol, a process and non-protein shampoos with the preservatives
which is accelerated by increased pH and/or tem- DMDM hydantoin, imidazolidinyl urea, diazolidinyl
perature. The half life is more than 5 years at pH 4 urea, and quaternium-15 added in concentrations
but drops to 2 months at pH 8 (40, 41). of 0.1%, 0.2%, 0.4%, and 0.8%, and determined
Diazolidinyl urea is produced from 4 mol of the amounts of free formaldehyde in the sham-
formaldehyde and 1 mol of allantoin (compare: poos. In the first experiment, the shampoos were
imidazolidinyl urea from 3 mol of formaldehyde heated to steambath temperatures. It was found that
and 2 mol of allantoin). From each molecule of DMDM hydantoin had released all its formalde-
diazolidinyl urea, four formaldehyde molecules can hyde (2 mol/mole DMDM hydantoin) after 15 min,
be released (4 mol formaldehyde/mole diazolidinyl quaternium-15 had released its total of 6 mol/mole
urea). However, only about 50% of the theoretical after 3 min and imidazolidinyl urea its 2 mol
amount of formaldehyde in diazolidinyl urea can be after 5 min. Diazolidinyl urea, theoretically able to
26 DE GROOT ET AL. Contact Dermatitis 2010: 62: 18–31

Table 6. Formaldehyde recoveries from NON-PROTEIN shampoo containing various preservatives (23)
Free formaldehyde
Preservative DMDM hydantoin Imidazolidinyl urea Diazolidinyl urea Quaternium-15
Concentration (%) mg/kg (p.p.m.) %a mg/kg (p.p.m.) %a mg/kg (p.p.m.) %a mg/kg (p.p.m.) %a
0.10 108 61 92 60 150 69 482 67
0.20 183 52 140 45 262 61 546 38
0.40 294 42 230 37 452 52 731 25
0.80 505 36 270 22 740 43 777 14
a
Percentage of theoretical total releasable formaldehyde.

release 4 mol/mole, released 2.1 mol after 5 min, quaternium-15 is analogous to DMDM hydantoin
and 3.3 mol/mole formaldehyde after 60 min at and imidazolidinyl urea, even though quaternium-15
steam bath temperatures. has six bound formaldehyde moieties. At 0.1% con-
In the second experiment, free formaldehyde was centration, four of the six formaldehyde moieties are
determined in the protein and non-protein shampoos released, which is 66% of the total available (23).
at 23◦ C by a microdiffusion method. The results Karlberg et al. developed a method for quantifi-
are shown in Tables 6 (non-protein shampoo) and 7 cation of formaldehyde in the presence of formal-
(protein-containing shampoo). dehyde-releasers in skin-care products (45). They
As the concentration of each preservative is used this method to determine the amount of
increased in the shampoo, the amount of free formaldehyde in cream and lotion preserved with
formaldehyde generated by hydrolysis increases. formaldehyde-releasers by adding formaldehyde-
With increasing concentrations, the percentage of releasers in concentrations ranging from 200 to 2400
(theoretical) total formaldehyde released, however, μg/g (0.02–0.24%) to diluted Essex® cream and
decreases. The amount of free formaldehyde Essex® lotion. Diazolidinyl urea and imidazolidinyl
released changes when protein is incorporated into urea were added to the cream and 2-bromo-2-
the shampoo, which is now depressed for all four
nitropropane-1,3-diol to the lotion. A linear correla-
preservatives when compared with the non-protein
tion between the concentration of the formaldehyde
shampoos. This indicates that free formaldehyde has
donor and the amount of formaldehyde detected
been complexed by the protein introduced.
The order of formaldehyde release is imidazo- was found. The amounts (approximate, read from
lidinyl urea <DMDM hydantoin <diazolidinyl urea a graphic in the publication) results are shown in
<quaternium-15 (23). Table 8.
DMDM hydantoin and imidazolidinyl urea each Using quantitative 13 C NMR spectroscopy, the
contain two methylol groups while diazolidinyl urea concentration of free formaldehyde was determined,
has four. The molar percentages of free formalde- released from diazolidinyl urea and imidazolidinyl
hyde found for these three materials are about the urea compared with the maximum releasable con-
same over the concentration range studied. An aver- centrations of formaldehyde as a function of dilution
age of 1.3 mol of free formaldehyde per mole of and pH (Table 9) (32).
preservative was found at 0.1%. This represents In 8 of 161 rinse-off products and 124 leave-
65% of the formaldehyde in DMDM hydantoin and on products produced in various European coun-
imidazolidinyl urea, but only 33% for diazolidinyl tries and the USA, free formaldehyde levels of
urea. This means that–under the condition of the >0.05% (500 p.p.m.) were found with a maximum
investigation–only two of the four methylol groups of 0.111% (1110 p.p.m.). Three of these products
are formaldehyde donors. The relationship of free were labelled with quaternium-15 and one with
molecular formaldehyde versus concentration for diazolidinyl urea (46). In Switzerland, 19 of 34

Table 7. Formaldehyde recoveries from PROTEIN-CONTAINING shampoo containing various preservatives (23)
Free formaldehyde
Preservative DMDM hydantoin Imidazolidinyl urea Diazolidinyl urea Quaternium-15
Concentration (%) mg/kg (p.p.m.) %a mg/kg (p.p.m.) %a mg/kg (p.p.m.) %a mg/kg (p.p.m.) %a
0.10 36 21 17 11 58 27 122 17
0.20 84 24 36 12 125 29 187 13
0.40 160 23 74 12 235 27 237 8
0.80 292 21 117 9 384 22 278 5
a
Percentage of theoretical total releasable formaldehyde.
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 27

Table 8. Amount of formaldehyde released with varying concentrations of formaldehyde-releasers (45)


Concentration of formaldehyde-releaser
0.02% 0.08% 0.16% 0.24%
Formaldehyde-releaser Free formaldehyde in the sample
Diazolidinyl urea (in dil. cream) (p.p.m.) 40 140 240 350
Imidazolidinyl urea (in dil. cream) (p.p.m.) 30 80 120 160
2-Bromo-2-nitropropane-1,3-diol (in lotion) (p.p.m.) 20 30 50 70

Table 9. Free formaldehyde released from diazolidinyl urea and imidazolidinyl urea as a function of dilution and pH (32)
Diazolidinyl urea Imidazolidinyl urea
0.5% 0.2% (w%) 0.6% 0.2% (w%)
Free formaldehyde (w%)
pH 9 0.084 (840 p.p.m.) 0.047 (470 p.p.m.) 0.024 (240 p.p.m.) 0.015 (150 p.p.m.)
pH 6 0.036 (360 p.p.m.) 0.019 (190 p.p.m.) 0.012 (120 p.p.m.) 0.011 (110 p.p.m.)
pH 4 0.032 (320 p.p.m.) 0.024 (240 p.p.m.) 0.015 (150 p.p.m.) 0.010 (100 p.p.m.)
Maximum releasable (w%) 0.21 (2100 p.p.m.) 0.086 (860 p.p.m.) 0.14 (1400 p.p.m.) 0.046 (460 p.p.m.)

cosmetic products (56%) investigated for the pres- be noted that in the cases of diazolidinyl urea,
ence of formaldehyde using three analytical methods DMDM hydantoin, and imidazolidinyl urea, these
including high-performance liquid chromatography percentages are heavily influenced by one major
were found to contain free formaldehyde in amounts USA study with by far the largest numbers of
ranging from 1 to 169 p.p.m.. Nine products con- allergic patients and high percentages co-reacting
tained more than 50 p.p.m.. In stay-on cosmetics, the to formaldehyde (9). Nevertheless, these data do
highest concentration proved to be 145 p.p.m. (35). clearly show that, with the exception of 2-bromo-2-
Other reports of free formaldehyde released nitropropane-1,3-diol, about half the reactions or so
from products containing formaldehyde-releasers (40–60%) to the releasers are caused by formalde-
are summarized in Table 10. hyde sensitivity. Of course, dual sensitization to
The maximum levels permitted in the EU and formaldehyde and a releaser per se is also possible
the highest values of free formaldehyde found (11). Obviously, the fact that 85% of positive patch
for these or lower concentrations are shown in test reactions to 2-bromo-2-nitropropane-1,3-diol
Table 11. These data indicate that–with the excep- are seen in patients negative to formaldehyde, means
tion of 2-bromo-2-nitropropane-1,3-diol, for which not only that allergy to 2-bromo-2-nitropropane-1,3-
adequate data are lacking–in the right circum- diol per se (i.e. independent of formaldehyde) does
stances the amount of free formaldehyde released exist, but that this is the case in the great majority
by all donors may be sufficient to induce dermatitis for this releaser. For the others, about half of the
with regular use (48). Benzylhemiformal is per- cases are contact allergies independent of formalde-
mitted in rinse-off products only and may conse- hyde sensitivity. It should be appreciated that in a
quently pose less threat to formaldehyde-sensitive number of these cases, the reaction to formaldehyde
subjects. may have been false negative, which would make
the percentage reacting to formaldehyde in the test
substance even greater.
Discussion Conversely, not all patients allergic to formalde-
hyde react to one or more releasers, in fact it is
Relationship between formaldehyde and releasers
a small minority (Table 2). The mean percentages
and between various releasers (adjusted for sample size) of patients allergic to
There is ample evidence that patch test reactions formaldehyde who also react to a releaser in the var-
to the various releasers in patients co-reacting to ious studies summarized in Table 2 were 16 for 2-
formaldehyde may be caused by formaldehyde in bromo-2-nitropropane-1,3-diol, 20 for diazolidinyl
the patch test preparation. The mean percentages urea, 19 for DMDM hydantoin, 14 for imidazo-
(adjusted for sample size) of patients allergic to lidinyl urea, and 40 for quaternium-15. Whether or
a releaser that also react to formaldehyde in the not a formaldehyde-sensitive individual will react to
various studies summarized in Table 1 were 15 one or more releasers is dependent on the strength of
for 2-bromo-2-nitropropane-1,3-diol, 49 for imida- the formaldehyde sensitization and on the amount of
zolidinyl urea, 55 for quaternium-15, 57 for DMDM formaldehyde present in (aqueous test substances)
hydantoin, and 59 for diazolidinyl urea. It should or releasable by the test preparation (pet.). Thus,
28 DE GROOT ET AL. Contact Dermatitis 2010: 62: 18–31

Table 10. Reports of free formaldehyde released from products containing formaldehyde-releasers
Amount of free formaldehyde Reference
Formaldehyde-releaser Present in (type of product) Concentration (%) % p.p.m.
Benzyl hemiformal Preservative solution 0.1 300 (28)
Shampoo 0.2 100 (28)
2-Bromo-2-nitropropane-1,3-diol Shampoo 0.02 10 (6)
Emulsion 0.02 15 (6)
DMDM hydantoin Preservative solution 0.25–pH 8.5–9 0.047 470 (32)
–pH 6–6.5 0.027 270 (32)
–pH 4–4.5 0.021 210 (32)
Cream 0.15 0.013 wt% 130 (32)
Preservative buffer solution 0.1 300 (33)
Imidazolidinyl urea Shampoo 0.1 8 (28)
Milky cleanser 0.3 100 (28)
Cleansing cream 0.4 92 (28)
Mascara 0.4 7 (28)
Night cream 0.5 131 (28)
Preservative saline solution 1.0 0.14 1400 (47)
Preservative buffer solution 0.1 100 (33)
Oil-in-water emulsion 0.3 300 (27)
Methenamineb Preservative solution 0.1–pH 3 480a (28)
–pH 5 220a (28)
–pH 8 10a (28)
Quaternium-15 Bath foam 0.2 398 (34)
Cleansing milk 0.2 482 (34)
Lotion 0.2 525 (34)
Shampoo 0.2 310 (34)
Sunscreen emulsion 0.2 406 (34)
product not specified 0.1 100 (31)
Shampoo 0.2 265 (28)
Eyeliner 0.2 286 (28)
Eye cream 0.2 266 (28)
Preservative solution 0.1 –pH>9 <200 (34)
–pH 7 250–300 (34)
–pH 3 600 (34)
Preservative buffer solution 0.1 700 (33)
Sodium hydroxymethylglycinate Preservative solution pH 8.5 0.5 (wt) 0.12 wt% 1200 (32)
Triazinetriethanolb Hand cleaner 0.2 344 (28)
a
Approximate values, read from a graphic.
b These releasers will be discussed in part 4.

one in about six formaldehyde allergic patients will to formaldehyde (reacting to 0.1% and/or 0.3%) had
also react to 2-bromo-2-nitropropane-1,3-diol and positive DMDM hydantoin patch test reactions to
imidazolidinyl urea, one in five to DMDM hydan- test preparations with lower formaldehyde content
toin and diazolidinyl urea, and about two in five (Table 3) (22).
to quaternium-15, which indeed (Tables 6, 7, and These findings are corroborated by the pattern of
10) releases most formaldehyde of all donors. The co-reactivity between the various releasers (Tables 4
relationship between strength of the reaction to and 5). 2-Bromo-2-nitropropane-1,3-diol patch test
formaldehyde and patch test reactions to donors reactivity, for instance, is–as mentioned above–in
was already clearly shown by de Groot et al., who only 15% attributable to formaldehyde. Indeed,
demonstrated that patients with stronger sensitivities co-reactivity to the other releasers was not even

Table 11. Permitted EU use levels for formaldehyde-releasers and levels of free formaldehyde demonstrated at these levels
Formaldehyde-releasers Permitted concentration (%) Product and concentration (%) Free formaldehyde (p.p.m.) Reference
Benzylhemiformal 0.15 Preservative solution: 0.1 300 (28)
2-Bromo-2-nitropropane-1,3-diol 0.1 Lotion: 0.08 30 (45)
Diazolidinyl urea 0.5 Non-protein shampoo: 0.4 452 (23)
Preservative solution: 0.5 840 (32)
Cosmetic cream: 0.3 730 (25)
DMDM hydantoin 0.6 Non-protein shampoo: 0.4 294 (23)
Preservative solution: 0.25 470 (32)
Imidazolidinyl urea 0.6 Oil-in-water emulsion: 0.3 300 (27)
Quaternium-15 0.2 Non-protein shampoo: 0.2 546 (23)
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 29

in a single study in more than 14% of these weak and poses little threat. However, formalde-
patients. The highest percentages of co-reactivity hyde being a ‘difficult’ allergen with risk of both
were seen in quaternium-15, which is readily false-positive and false-negative reactions, such con-
explained by the fact that this releaser has the clusions may not always be valid.
highest formaldehyde content. In patients aller- Although we do not know what level of formalde-
gic to imidazolidinyl urea and diazolidinyl urea, hyde in products is safe, levels >200 p.p.m.
many patients co-react to quaternium-15, for the probably are not (at least for some formalde-
same reason. However, the highest degree of co- hyde allergic subjects) (22, 48). The amount of
reactivity is between diazolidinyl urea and imida- formaldehyde in cosmetic products will depend
zolidinyl urea (up to 75% in several studies). This on the releaser used for their preservation, its
is not due to formaldehyde but to either cross- concentration, the age of the product, the pH,
reactivity or contact allergy to common ingredi- and other ingredients. Quaternium-15 releases the
ents in the test preparation such as compound HS largest amounts of formaldehyde, followed by
[(4-hydroxymethyl-2,5-dioxo-imidazolidin-4-yl)- diazolidinyl urea, DMDM hydantoin, and imi-
urea] or allantoin-formaldehyde condensation prod- dazolidinyl urea. 2-Bromo-2-nitropropane-1,3-diol
ucts (44, 49). releases little formaldehyde (Tables 8 and 10). As
Table 11 has shown, all releasers (with the excep-
tion of 2-bromo-2-nitropropane-1,3-diol, for which
Should cosmetics containing adequate data are lacking) can–in the right cir-
formaldehyde-releasers be avoided by patients cumstances of concentration and product compo-
allergic to formaldehyde and/or the releasers? sition–release >200 p.p.m. formaldehyde, which
may result in a number of patients in allergic con-
When a patient reacts to formaldehyde and to, for tact dermatitis. Whether this is actually the case
example, diazolidinyl urea, it would be common in any particular product cannot be decided on the
sense to avoid cosmetic products containing diazo- basis of ingredient labelling. We thus recommend
lidinyl urea, at least the stay-on varieties. The same to advise patients allergic to formaldehyde to avoid
holds true for a patient reacting to the releaser itself stay-on cosmetics preserved with quaternium-15,
but not to formaldehyde: avoidance is logical. But diazolidinyl urea, DMDM hydantoin, or imidazo-
how to decide when a patient has a positive patch lidinyl urea, especially when to be used regularly
test to formaldehyde but not to any of the releasers and/or on ‘sensitive’ or damaged skin. When there
or to one or two of the releasers (or when only is no alternative, products containing 2-bromo-2-
formaldehyde and quaternium-15 are tested in the nitropropane-1,3-diol may be tried. With products
standard series)? Should the ‘non-reactive’ releasers that the patient already possesses, provocation tests
also be avoided (50) or may this be unneces- under normal usage conditions may demonstrate
sary (2)? Unfortunately, there are insufficient data whether they are safe (or not) in this particular
to give firm and decisive recommendations, only patient.
tentative advice can be given. Whether or not a cos- The dose–response phenomenon has sparsely
metic product preserved with a releaser will elicitate been investigated with regard to formaldehyde and
an allergic reaction in patients sensitive to formalde- its releasers, and the clinical relevance of low
hyde depends on: (i) the type of product; (ii) the values of formaldehyde in consumer products and
mode of usage (frequency and location of appli- cosmetics clearly needs further investigation.
cation); (iii) the skin status; (iv) the strength of
the hypersensitivity to formaldehyde; and (v) the
amount of free and releasable formaldehyde in the References
product. 1. de Groot A C, White I R, Flyvholm M-A, Lensen G J,
Rinse-off products, in general, will have a low Coenraads P J. Formaldehyde-releasers in cosmetics: rela-
risk of causing allergic contact dermatitis. The tionship to formaldehyde contact allergy. Part 1. Characteriza-
tion, frequency and relevance of sensitization, and frequency
daily use of stay-on products increases the risk of use in cosmetics. Contact Dermatitis 2010: 62: 2–17.
(compared with products that are used infrequently 2. Herbert C, Rietschel R L. Formaldehyde and formaldehyde
or intermittently), especially when applied more releasers: how much avoidance of cross-reacting agents is
than once per day, when used on ‘sensitive’ skin required? Contact Dermatitis 2004: 50: 371–373.
3. Ford G P, Beck M H. Reactions to quaternium 15, Bronopol
such as the eyelids, the face, the neck, or the and Germall 115 in a standard series. Contact Dermatitis
axillae, and when used on damaged skin, e.g. dry 1986: 14: 271–274.
(dermatitis) skin in atopic individuals. 4. Fransway A F, Schmitz N A. The problem of preservation
in the 1990s: II. Formaldehyde and formaldehyde-releasing
It could be argued that, when there is a single biocides: incidences of cross-reactivity and the significance
reaction to formaldehyde but not to any of the of the positive reponse to formaldehyde. Am J Contact
releasers, the allergy to formaldehyde must be Dermatitis 1991: 2: 78–88.
30 DE GROOT ET AL. Contact Dermatitis 2010: 62: 18–31

5. Perrett C M, Happle R. Contact sensitivity to diazolidinyl product containing formaldehyde releaser. Contact Dermatitis
urea (Germall II). Arch Dermatol Res 1989: 281: 57–59. 1997: 36: 26–33.
6. Storrs F J, Bell D E. Allergic contact dermatitis to 2-bromo- 28. Engelhardt H, Klinkner R. Determination of free formalde-
2-nitropropane-1,3-diol in a hydrophilic ointment. J Am Acad hyde in the presence of donators in cosmetics by HPLC
Dermatol 1983: 8: 157–170. and post-column derivation. Chromatographia 1985: 20:
7. Aalto-Korte K, Kuuliala O, Suuronen K, Alanko K. Occu- 559–565.
pational contact allergy to formaldehyde and formaldehyde 29. Geier J, Lessmann H, Schnuch A, Uter W. Contact sensitiza-
releasers. Contact Dermatitis 2008: 59: 280–289. tions in metalworkers with occupational dermatitis exposed
8. Statham B N, Bodger O. Cross reactivity of isothiazolinone to water-based metalworking fluids: results of the research
and formaldehyde releasing preservatives. Presented at the project ’FaSt’. Int Arch Occup Environ Health 2004: 77:
Annual Scientific Meeting of the British Contact Dermatitis 543–551.
Society, July 2006, Birmingham. 30. Geier J, Lessmann H, Becker D et al. Formaldehydabspalter.
9. Rietschel R L, Warshaw E M, Sasseville D et al. Sen- Dermatologie in Beruf und Umwelt 2008: 56: 34–36.
sitivity of petrolatum and aqueous vehicles for detect- 31. Fransway A F. The problem of preservation in the 1990s: I.
ing allergy to imidazolidinylurea, diazolidinylurea, and Statement of the problem. Solution(s) of the industry, and
DMDM hydantoin: a retrospective analysis from the North the current use of formaldehyde and formaldehyde-releasing
American Contact Dermatitis Group. Dermatitis 2007: 18: biocides. Am J Contact Dermat 1991: 2: 6–23.
155–162. 32. Emeis D, Anker W, Wittern K-P. Quantitative 13 C NMR
10. Uter W, Frosch P J. Contact allergy from DMDM hydantoin, spectroscopic studies on the equilibrium of formaldehyde
1994–2000. Contact Dermatitis 2002: 47: 57–58. with its releasing preservatives. Anal Chem 2007: 79:
11. Geier J, Lessmann H, Schuch A, Fuchs Th. Kontaktallergien 2096–2100.
durch formaldehydabspaltende Biozide. Allergologie 1997: 33. Kijami K, Takeda M, Okaya Y et al. A study on release
20: 215–224. of formaldehyde from its donor-type preservatives. Anal Sci
12. Kränke B, Szolar-Platzer C, Aberer W. Reactions to 1991: 7(Suppl): 913–916.
formaldehyde and formaldehyde releasers in a standard series. 34. Benassi C A, Semenzato A, Zaccaria F, Bettero A. High-
Contact Dermatitis 1996: 35: 192–193. performance liquid chromatographic determination of free
13. Jacobs M-C, White I R, Rycroft R J G, Taub N. Patch test- formaldehyde in cosmetics preserved with Dowicil 200.
ing with preservatives at St John’s from 1982 to 1993. Con- J Chromatogr 1990: 502: 193–200.
tact Dermatitis 1995: 33: 247–254. 35. Gryllaki-Berger M, Mugny Ch, Perrenoud D, Pannatier A,
14. Hectorne K J, Fransway A F. Diazolidinyl urea: incidence of Frenk E. A comparative study of formaldehyde detection
sensitivity, patterns of cross-reactivity and clinical relevance. using chromotropic acid, acetylacetone and HPLC in cos-
Contact Dermatitis 1994: 30: 16–19. metics and household products. Contact Dermatitis 1992: 26:
149–154.
15. Parker L U, Taylor J S. A 5-year study of contact allergy to
36. Rastogi S C. Analytical control of preservative labelling on
quaternium-15. Am J Contact Dermat 1991: 2: 231–234.
skin creams. Contact Dermatitis 2000: 43: 339–343.
16. Fisher A A. Cosmetic dermatitis. Part II. Reactions to some
37. Flyvholm M-A, Tiedemann E, Menné T. Comparison of
commonly used preservatives. Cutis 1980: 26: 136.
2 tests for clinical assessment of formaldehyde exposure.
17. de Groot A C, Bos J D, Jagtman B A, Bruynzeel D P, van
Contact Dermatitis 1996: 34: 35–38.
Joost Th, Weyland J W. Contact allergy to preservatives–II.
38. Fregert S, Dahlquist I, Gruvberger B. A simple method for
Contact Dermatitis 1986: 15: 218–222.
the detection of formaldehyde. Contact Dermatitis 1984: 10:
18. Frosch P J, Weickel R. Contact allergy to the preservative 132–134.
bronopol. Hautarzt 1987: 38: 267–270. 39. Dahlquist I, Fregert S, Gruvberger B. Detection of formalde-
19. Peters M S, Connolly S M, Schroeter A L. Bronopol allergic hyde in corticoid creams. Contact Dermatitis 1980: 6: 494.
contact dermatitis. Contact Dermatitis 1983: 9: 397–401. 40. Croshaw B. Preservatives for cosmetics and toiletries. J Soc
20. de Groot A C, Bruynzeel D P, Jagtman B A, Weyland J W. Cosmet Chem 1977: 28: 3–17.
Contact allergy to diazolidinyl urea (Germall II® ). Contact 41. Bryce D M, Croshaw B, Hall J E, Holland V R, Lessel B.
Dermatitis 1988: 18: 202–205. The activity and safety of the antimicrobial agent bronopol
21. Dooms-Goossens A, de Boulle K, Dooms M, Degreef H. (2-bromo-2-nitropropane-1,3-diol). J Soc Cosmet Chem 1978:
Imidazolidinyl urea dermatitis. Contact Dermatitis 1986: 14: 29: 3–24.
322–324. 42. Rosen M. Glydant and MDMH as cosmetic preservative. In:
22. de Groot A C, van Joost T, Bos J D, Van der Meeren H L M, Cosmetic and Drug Preservation. Principles and Practice,
Weyland J W. Patch test reactivity to DMDM hydantoin. Kabara J J (ed.): New York, Marcel Dekker Inc., 1984:
Relationship to formaldehyde. Contact Dermatitis 1988: 18: pp. 165–190.
197–201. 43. Expert Panel of the Cosmetic Ingredient Review. Final report
23. Rosen M, McFarland A. Free formaldehyde in anionic sham- of the Safety Assessment for DMDM Hydantoin, Washington,
poos. J Soc Cosmet Chem 1984: 35: 157–169. CTFA, 1987.
24. Schnuch A, Geier J, Uter W, Frosch P J. Patch testing 44. Johansen M, Bundgaard H. Kinetics of formaldehyde release
with preservatives, antimicrobials and industrial biocides. from the cosmetic preservative Germall 115. Arch Pharm
Results from a multicentre study. Br J Dermatol 1998: 138: Chem Sci Ed 1981: 9: 117–122.
467–476. 45. Karlberg A-T, Skare L, Lindberg I, Nyhammer E. A method
25. Zachariae C, Hall B, Cottin M, Cupperman S, Ander- for quantification of formaldehyde in the presence of
sen K E, Menné T. Experimental elicitation of contact allergy formaldehyde donors in skin-care products. Contact Dermati-
from a diazolidinyl urea-preserved cream in relation to tis 1998: 38: 20–28.
anatomical region, exposure time and concentration. Contact 46. Rastogi S C. A survey of formaldehyde in shampoos and skin
Dermatitis 2005: 53: 268–277. creams on the Danish market. Contact Dermatitis 1992: 27:
26. Isaksson M, Gruvberger B, Goon AT-J, Bruze M. Can an 235–240.
imidazolidinyl urea-preserved corticosteroid cream be safely 47. Andersen K E, Boman A, Hamann K, Wahlberg J E. Guinea
used in individuals hypersensitive to formaldehyde? Contact pig maximization tests with formaldehyde releasers.
Dermatitis 2006: 54: 29–34. Results of two laboratories. Contact Dermatitis 1984: 10:
27. Flyvholm M A, Hall B M, Agner T et al. Threshold for 257–266.
occluded formaldehyde patch test in formaldehyde sensitive 48. de Groot A C, Flyvholm M-A, Lensen G J, Menné T,
patients. Relationship to repeated open application test with a Coenraads P J. Formaldehyde releasers: relationship to
Contact Dermatitis 2010: 62: 18–31 FORMALDEHYDE-RELEASERS IN COSMETICS: PART 2 31

formaldehyde contact allergy. Contact allergy to formalde- Address:


hyde and inventory of formaldehyde-releasers. Contact Der- Anton de Groot
matitis 2009: 61: 63–85. Department of Dermatology
49. Lehmann S V, Hoeck U, Breinholdt J, Olsen C E, Kreil- University Medical Center Groningen
gaard B. Characterization and chemistry of imidazolidinyl University of Groningen
urea and diazolidinyl urea. Contact Dermatitis 2006: 54: PO Box 30.001
50–58. 9700 RB Groningen
50. Flyvholm M-A, Menné T. Allergic contact dermatitis from the Netherlands
formaldehyde. A case study focussing on sources of formalde- Tel: +31(0)521320332
hyde exposure. Contact Dermatitis 1992: 27: 27–36. e-mail: antondegroot@planet.nl

Vous aimerez peut-être aussi