BJD

C L I N I C A L A N D L A B O R A T O R Y I N V E S TI G A T I O N S British Journal of Dermatology

Histopathology of drug rash with eosinophilia and systemic

symptoms syndrome: a morphological and phenotypical
study
N. Ortonne,1,2,3 L. Valeyrie-Allanore,3,4 S. Bastuji-Garin,3,5 J. Wechsler,1 S. de Feraudy,1 T.-A. Duong,3,4
M.-H. Delfau-Larue,2,3,6 O. Chosidow,3,4 P. Wolkenstein3,4 and J.-C. Roujeau3
Assistance Publique – H^opitaux de Paris (AP-HP), H^opital Henri-Mondor, 1Departement de Pathologie, 4Service de Dermatologie, 5Service de Sante-Publique,
6
Service d’Immunologie Biologique, 94010 Creteil Cedex, France
2
INSERM U955 equipe 9, H^opital Henri-Mondor, 94010 Creteil Cedex, France
3
Universite Paris Est Creteil (UPEC), Faculte de Medecine, LIC EA4393, 94010 Creteil Cedex, France

Summary

Correspondence Background The histopathological features of drug rash with eosinophilia and sys-
Nicolas Ortonne. temic symptoms (DRESS) syndrome remain poorly characterized.
E-mail: nicolas.ortonne@hmn.aphp.fr Objectives To better characterize the histopathological features of DRESS syndrome,
and define the phenotype of the effector cells in the skin and compare it with
Accepted for publication
25 January 2015
maculopapular rash (MPR).
Methods We conducted a retrospective study on 50 skin biopsies from patients
Funding sources with DRESS syndrome (n = 36). Histopathological and immunophenotypical fea-
None. tures were studied and compared with a series of MPRs (n = 20).
Results Foci of interface dermatitis, involving cutaneous adnexae, were frequently
Conflicts of interest seen in cases of DRESS. Eosinophils were seen in only 20% of cases and neu-
None declared.
trophils in 42%. Eczematous (40%), interface dermatitis (74%), acute general-
N.O., L.A., O.C. and P.W. contributed equally ized exanthematic pustulosis-like (20%) and erythema multiforme-like (24%)
to this paper. patterns were observed. The association of two or three of these patterns in a
single biopsy was significantly more frequent in cases of DRESS than in a series
DOI 10.1111/bjd.13683 of nondrug-induced dermatoses (P < 0
01), and appeared to be more marked
in DRESS syndrome with severe cutaneous lesions (P = 0
01) than in less severe
cases of DRESS and MPR. A higher proportion of CD8+ and granzyme B+ lym-
phocytes was observed in cases of DRESS with severe cutaneous eruptions (ery-
throderma and/or bullae). Atypical lymphocytes were found in 28% of
biopsies, and expressed CD8 in most cases; a cutaneous T-cell clone was rarely
found (6%).
Conclusions The histopathology of DRESS syndrome highlights various associated
inflammatory patterns in a single biopsy. Cutaneous effector lymphocytes com-
prise a high proportion of polyclonal CD8+ granzyme B+ T lymphocytes.

What’s already known about this topic?

• The histological features of drug rash with eosinophilia and systemic symptoms
(DRESS) vary from spongiotic dermatitis to an erythema multiforme-like aspect.
• In DRESS, skin biopsies often show eosinophils and apoptotic bodies.
• Atypical lymphocytes may be found in skin infiltrates.

What does this study add?

• The association of several inflammatory patterns in a single biopsy is suggestive for
diagnosis.
• Cutaneous infiltrates can compromise atypical lymphocytes resembling S
ezary cells.

50 British Journal of Dermatology (2015) 173, pp50–58 © 2015 British Association of Dermatologists

• DRESS shows a higher density of inflammatory infiltrates, more apoptosis, associ-
ated inflammatory patterns and granzyme B+ cells than maculapapular rash.
• Effector T cells are mainly polyclonal granzyme B+ CD8+ T cells.
The most common type of cutaneous drug reaction is the
morbilliform-type or maculopapular rash (MPR), but more
severe cutaneous drug eruption can occur, including drug rash
with eosinophilia and systemic symptoms (DRESS) syndrome.
In 1996, Bocquet and colleagues proposed this term in order
to reduce confusion with hypersensitivity syndrome. DRESS
syndrome is often characterized by a skin rash and multivis-
ceral involvement, usually associated with hypereosino-
philia.1,2 In some patients, DRESS syndrome can present with
more aggressive cutaneous lesions, including erythroderma, or
with bullae. The disease may be associated with reactivation
of herpesvirus replication, especially human herpesvirus
(HHV)-6 and HHV-7,3 but also Epstein–Barr virus (EBV) and
cytomegalovirus.4,5
The histopathological aspect of DRESS syndrome is not well
described in the literature, and is often only mentioned in
passing in dermatopathology textbooks.6 To the best of our
knowledge, three series of DRESS syndrome focusing on the
histology have previously been published, all highlighting the
polymorphous aspect of DRESS syndrome, with various
inflammatory patterns.7–9 Interestingly, Walsh et al.9 suggested
that the presence of apoptotic keratinocytes correlated with a
more aggressive phenotype, with liver injury and an erythema
multiforme (EM)-like cutaneous aspect, while most biopsies
showed a spongiotic dermatitis. Chi et al.7 also found that skin
biopsies of DRESS syndrome displayed various inflammatory
aspects, and showed that interface dermatitis with apoptotic
keratinocytes were more frequent in DRESS syndrome than in
MPR. Owing, in part, to the fact that the inflammatory infil-
trates of DRESS syndrome may comprise atypical lymphocytes,
in cases with diffuse skin eruption or erythroderma, differen-
tial diagnosis with lymphoma, especially S
ezary syndrome
(SS), may be histopathologically difficult. In addition, it has
been shown that T-cell clonality analyses may give positive
results in DRESS syndrome.10
In contrast to toxic epidermal necrolysis (TEN), in which it
has been shown that granulysin is the predominant cytokine
inducing the apoptosis of epithelial cells, little attention has
been paid to the cellular and molecular mechanisms involved
in DRESS syndrome.11 As in other severe cutaneous drug reac-
tions, understanding of the effectors involved in T-cell activa-
tion and organ cytotoxicity are important, and may represent
the first step in the development of new, targeted therapies. It
is known that, following drug exposure, DRESS syndrome is
characterized by an expansion of effector CD8+ T cells and
regulatory T cells (Tregs) in the blood.12,13
The aim of the present study was to better characterize
the histopathological features of DRESS syndrome, and to
© 2015 British Association of Dermatologists
Histopathology of DRESS syndrome, N. Ortonne et al. 51
make a comparison with various inflammatory dermatoses
and MPRs, which are less severe forms of drug reaction. We
also focused on the phenotype of the effector cells in the
skin.
Materials and methods
Patients and material selection
We retrospectively included 50 biopsy specimens over an 11-
year period from 36 patients [17 men and 19 women, med-
ian age 52
3 years (range 15
0–90
0)] who had a diagnosis of
DRESS syndrome. In the majority (39%) of cases the offending
drug was allopurinol, followed by carbamazepin (11%), mi-
nocycline, sulfamethoxazole + trimethoprim (5%) and sulfa-
salazine (5%). In each case, the diagnosis of DRESS syndrome
(at least probable) was established according to previously
published criteria.14 The mean delay between the onset of
symptoms and skin biopsy, available in 22 cases, was 14 days
(range 1–80). In all but three cases the delay ranged between
1 and 12 days and thus the biopsy was considered to have
been performed during the acute phase. We compared the
histopathology in patients with DRESS syndrome with severe
cutaneous lesions (n = 21), presenting with erythroderma
and/or bullae, with those with a less severe phenotype
(n = 17), presenting only with a maculopapular eruption. The
main clinical features of the patients are summarized in
Table 1.
We compared the biopsies from patients with DRESS syn-
drome with biopsies from patients with a less severe drug-
induced skin eruption (MPR). The latter group comprised 20
skin biopsies from 20 patients [six men and 14 women, med-
ian age 67 years (range 17–91)]. Patients were diagnosed
with MPR when they presented with a drug-induced rash and
met no criteria for DRESS syndrome according to Kardaun
et al.14
The following clinical and laboratory data were retrospec-
tively collected in all patients with DRESS syndrome: fever;
type of eruption (erythematous rash and MPR, or erythroder-
ma); presence of purpuric lesions on the lower limbs; presence
of skin necrosis symptoms (superficial erosions, blistering or
Nikolski sign); polyadenopathy; hypereosinophilia (> 700
eosinophils per mm3) and eosinophil blood count; presence
of circulating hyperbasophil or ‘atypical’ lymphocytes; liver
and renal dysfunction.
The study was approved by the Comit
e de Protection des
Personnes Ile de France IV (institutional review board no.
00003835).
British Journal of Dermatology (2015) 173, pp50–58
52 Histopathology of DRESS syndrome, N. Ortonne et al.

Table 1 Clinical characteristics of the 36 included patients with drug orthokeratosis or parakeratosis (continuous, psoriasiform or
rash with eosinophilia and systemic symptoms (DRESS) syndrome focal); (ii) aspect of granulous layer – normal or thickened;
(iii) aspect of the spinous layers – acanthosis [psoriasiform
Female sex 19 (52
8)
(regular hyperplasia) or irregular], epidermal atrophy, spongi-
Mean age, years (range) 52 (15–90)
Offending drug osis [grade 1 – diffusely enlarged intercellular spaces; grade 2
Allopurinol 14 (39) – marked confluent spongiosis (‘prevesicles’); grade 3 – con-
Carbamazepine 4 (11) stituted vesicles] or apoptotic keratinocytes, i.e. ‘Civatte
Sulfamethoxazole+ trimethoprim 2 (5) bodies’; (iv) aspect of dermoepidermal junction – normal,
Minocycline 2 (5) focal interface dermatitis (apoptotic keratinocytes or vacuo-
Sulfasalazine 2 (5)
lized basement membrane zone) or widespread interface
Other
dermatitis; (v) adnexal lichenoid interface dermatitis lesions –
Dermatological manifestations
Erythroderma 17 (47) follicular interface dermatitis or lichenoid interface dermatitis
Maculopapular rash 21 (58) at the acrosyringium; (vi) characterization of intraepidermal
Bullae/erosive lesions 6 (17) inflammatory cells – lymphocytes, eosinophils, neutrophils
Pustules 14 (39) and corneal/subcorneal pustules; (vii) presence or absence of
Purpuraa 5 (14) atypical lymphocytes (medium-sized lymphocytes with
General and extracutaneous manifestations
enlarged nuclei, lymphocytes with enlarged hyperconvoluted
Fever 32 (89)
nuclei suggesting S
ezary cells or large atypical lymphocytes);
Hyperbasophilic lymphocytes (n = 23) 15 (65)
Polyadenopathyb 24 (67) (viii) dermal infiltrate – localization (superficial, superficial
Liver dysfunctionc 26 (72) and deep, or deep dermis, hypodermis), architecture (band-
Renal dysfunctiond 16 (44) like, perivascular, periadnexal, interstitial, diffuse), and density
Hypereosinophiliae 32 (89) (low, i.e. few scattered lymphocytes not forming aggregates;
intermediate, i.e. cohesive cells forming some aggregates; or
Values are given as n (%) unless otherwise indicated. aThe pur-
pura was not infiltrated; bpatients with polyadenopathy had high, i.e. sheets of lymphocytes grouped in large aggregates)
enlarged lymph nodes (> 1 cm) in at least two different anatomi- and components (eosinophils, neutrophils, lymphocytes and
cal sites; cpatients with serum glutamic oxaloacetic transaminase plasma cells); (ix) dermal changes: vasculitis, oedema of pap-
and/or serum glutamic-pyruvic transaminase and/or gamma- illary dermis and leucocytoclastic nuclei.
glutamyl transferase and/or alkaline phosphatase abnormal The presence of particular inflammatory patterns was
level(s) [over twice the normal laboratory value(s)] were consid- recorded in each case. The eczematous pattern was defined as
ered to have liver dysfunction; drenal dysfunction was defined as a grade 2 or 3 spongiosis with lymphocytes exocytosis; inter-
an abnormal creatinine level; e700–15 000 eosinophils per mm3. face dermatitis as basal lymphocyte exocytosis with keratino-
cyte vacuolization and/or apoptosis; acute generalized
exanthematic pustulosis (AGEP)-like as a multilocular subcor-
Histopathology and immunohistochemistry: techniques
neal or intracorneal pustulosis; psoriasiform as an association
and parameters analysed
of psoriasiform hyperplasia with continuous parakeratosis;
Formalin-fixed, paraffin-embedded (FFPE) skin biopsies were pustular psoriasis as a psoriasis pattern with pustules; and EM-
retrieved from archive material of the Department of Pathol- like as lymphocytic exocytosis with aggregates of apoptotic
ogy, Assistance Publique–H^ opitaux de Paris. Haematoxylin, keratinocytes. We recorded in each case the number of inflam-
eosin and saffron (HES) staining and immunohistochemistry matory patterns observed in a single biopsy. As a comparative
were applied to 3-lm-thick sections. Immunostaining was group, we analysed the inflammatory patterns of 47 skin
done using monoclonal antibodies to CD2, CD3, CD4, CD5, biopsies of patients with a nondrug-induced dermatosis (21
CD7, CD8, CD20, CD56, CD123, granzyme B (Dako, Glost- men and 26 women, median age 50
4 years). This control
rup, Denmark) and FoxP3 (236/AE7; Abcam, Cambridge, group was established by selecting nondrug-induced dermato-
U.K.). We used a standard avidin–biotin–peroxidase method ses that are known to involve a dermal lymphocytic infiltrate
with diaminobenzidine (DAB) chromogen and the NexES im- and epidermal inflammatory changes: eczematous reaction
munostainer (Ventana, Tucson, AZ, U.S.A.), after antigen (n = 9), psoriasis (n = 11), lichen planus (n = 10), subacute/
retrieval by heat in the appropriate buffer. Immunostaining of chronic lupus (n = 9) and SS (n = 8).
FoxP3 was done manually using a biotin–avidin system conju- We performed an immunohistochemical study in 24 biop-
gated to horseradish peroxidase [HRP; VECTASTAINâ ABC–AP sies from the DRESS syndrome group (CD2, CD3, CD4, CD5,
kit (Vector Laboratories, Burlingame, CA, U.S.A.)]. Double- CD7, CD8, CD20, CD56, CD123, granzyme B, and FoxP3), as
staining experiments (CD3/CD8) were performed using the well as in biopsies from the SS and MPR groups (CD3, CD8,
Bond max device (Menarini Diagnostics, Rungis, France). granzyme B, FoxP3). The phenotype of atypical lymphocytes,
All HES slides were reviewed with a multiheaded micro- when present, was analysed. In three DRESS syndrome samples
scope and discussed by two dermatopathologists (N.O. and with atypical lymphocytes, we performed double stainings for
J.W.). The following morphological parameters were recorded CD3 and CD8. The presence of CD8+, CD56+, CD123+ and
systematically: (i) aspect of the stratum corneum – normal, granzyme B+ cells was assessed in both dermis and epidermis.

British Journal of Dermatology (2015) 173, pp50–58 © 2015 British Association of Dermatologists

The density of positive cells was categorized as negative (no
positive cell), low (0–5% positive lymphocytes), intermediate
(> 5–50%) or strong (> 50%).
Epstein Barr virus-specific in situ hybridization
The search for EBV in skin samples was done by in situ hybrid-
ization on deparaffinized slides, using consensus probes for
EBV-encoded small RNA (EBER) transcripts (Bond ISH probe;
Menarini Leica, Florence, Italy) coupled with fluorescein and
developed with secondary antifluorescein antibodies coupled
to HRP and DAB.
T-cell clonality analysis
For T-cell clonality studies, DNA was extracted from either a
snap frozen skin biopsy or from the FFPE samples, and analysed
by DNA amplification of TCRG using consensus primers and
separation of the amplimers by polymerase chain reaction dena-
turing gradient gel electrophoresis, as previously described.15
Statistical analysis
Qualitative variables were compared across different popula-
tions using the Fisher’s exact or Kruskal–Wallis tests. All tests
were two-tailed and P-values < 0
01 were considered statisti-
cally significant, using a Bonferroni correction. Statistical
analyses were performed using STATA (version 11.0; StataCorp,
College Station, TX, U.S.A.).
Results
Histopathological and phenotypic study of drug rash
with eosinophilia and systemic symptoms
The main histopathological findings are summarized in
Table 2. A diffuse parakeratotic layer (84%) and foci of liche-
noid interface dermatitis (76%) were frequent. In particular,
an interface dermatitis involving the adnexae was commonly
observed in the acrosyringium and the infundibular and isth-
mic portions of pilar units. Apoptotic keratinocytes were seen
in 60% of cases and neutrophil exocytosis in 12%, sometimes
with subcorneal pustules (18%). The dermal infiltrates
appeared to be polymorphous, with plasma cells in 18% of
cases and neutrophils in 42%. While most patients presented
with hypereosinophilia (89%), with an eosinophil count rang-
ing from 700 to 15 000 per mm3 (Table 1), eosinophils were
only significantly present in 20% of skin biopsies. Atypical
lymphocytes, sometimes resembling S
ezary cells, were present
in almost one-third of cases (14 of 50; 28%) (Fig. 1c).
Nuclear debris within the dermis was also often seen, but
leucocytoclastic vasculitis was not observed. In most cases, the
infiltrate was located only in the superficial dermis (88%),
and predominated around dermal capillaries (73%). The den-
sity of inflammatory infiltrates was low (42%) or intermediate
(50%) in the majority of cases. Overall, various inflammatory
© 2015 British Association of Dermatologists
Histopathology of DRESS syndrome, N. Ortonne et al. 53
patterns were observed. The most frequent was an interface
dermatitis (74%), followed by eczematous (40%), EM-like
(24%) and AGEP-like pustulosis (20%). Interestingly, more
than one pattern was frequently observed in a single biopsy.
This multiplicity of inflammatory pattern was significantly
more pronounced in DRESS than in nondrug-induced derma-
toses (56% vs. 25% of cases, respectively; P < 0
01). A repre-
sentative example of a DRESS biopsy with three different
patterns is shown in Figure 2.
The most frequently associated patterns were eczematous
and interface dermatitis (20%). The second most frequently
associated patterns were interface and EM-like dermatitis
(14%).
Lymphocyte phenotyping showed that most cases of DRESS
comprised a high proportion of CD8+ lymphocytes, with
numerous cytotoxic cells expressing granzyme B (Table 3;
Fig. 3). No CD56+ natural killer cells, and no or very few
CD123+ plasmacytoid dendritic cells and CD20+ B cells were
found (< 5% in all investigated cases). FoxP3+ lymphocytes
were identified in 58% of cases, but were scattered in most
samples.
Among 14 analysed skin biopsies from patients with DRESS,
EBER expression was only present in one, in a few scattered
lymphocytes. T-cell clonality results were available for 17
patients. A T-cell clone was detected in only one (6%) skin
sample. None of the seven patients with atypical lymphocytes
in the skin had a cutaneous T-cell clone. Two patients, includ-
ing one with a cutaneous T-cell clone, had a T-cell clone in
blood.
Clinical pathological correlations in drug rash with
eosinophilia and systemic systems
No correlation could be established between the most frequent
offending drug (allopurinol) and any of the morphological
parameters analysed. Among the 15 patients with circulating
atypical/hyperbasophilic lymphocytes, only three (20%) had
atypical lymphocytes identified in the skin. Cutaneous eosin-
ophils were identified in only nine (28%) of the 32 patients
with blood hypereosinophilia. A proportion (14%) of patients
had purpuric lesions, but leucoclastic or lymphocytic vasculitis
was not seen. Significant red blood cell extravasation was not
found, but skin biopsies were done on the lower limb in only
one patient with purpuric lesions.
Interestingly, biopsies from patients with a severe pheno-
type (erythroderma and/or bullae) more frequently showed
an EM-like pattern (P < 0
01), while the presence of other
inflammatory patterns, atypical lymphocytes, pustules and the
deepness of dermal infiltration did not significantly differ
between DRESS presenting with a maculopapular eruption or
MPR. Patients with apoptotic keratinocytes in skin biopsies
showed a tendency to have more liver (80% vs. 64%) and
renal (60% vs. 43%) dysfunctions. In addition, DRESS with
severe cutaneous lesions had significantly more associated pat-
terns (P = 0
01) and a higher proportion of granzyme B+
lymphocytes (P = 0
04). The proportion of CD8+ effector T
British Journal of Dermatology (2015) 173, pp50–58
54 Histopathology of DRESS syndrome, N. Ortonne et al.

Table 2 Comparison of the histological features of patients with drug rash with eosinophilia and systemic symptoms (DRESS) syndrome and
maculopapular rash (MPR)

Histological feature DRESS (n = 50) MPR (n = 20) P-value

Parakeratosis 42 (84) 8 (40) 0
20
Acanthosis 27 (54) 2 (10) < 0
01
Pustules 10 (20) 5 (25) 0
75
Focal ID 38 (76) 7 (35) < 0
01
Apoptotic keratinocytes 30 (60) 6 (30) 0
06
Lymphocyte exocytosis 32 (64) 7 (35) 0
03
Follicular interface dermatitis (n = 13/n = 3) 13 (100) 2 (67) –
Interface dermatitis of acrosyringium (n = 49/n = 6) 19 (39) 3 (59) –
Papillary oedema 24 (48) 7 (35) 0
43
Atypical lymphocytes 14 (28) 7 (35) 0
58
Dermal eosinophils 10 (20) 9 (45) 0
04
Dermal neutrophils 21 (42) 6 (30) 0
42
Leucocytoclasia 15 (30) 1 (5) 0
03
Dermal plasma cells 9 (18) 0 0
05
Mid and deep dermis infiltration 13 (26) 1 (5) 0
04
Infiltrate density < 0
01
Low 21 (42) 17 (85)
Intermediate 25 (50) 3 (15)
High 4 (8) 0
Inflammatory patterns
None (perivascular infiltrate only) 7 (14) 8 (40) –
Eczematous 20 (40) 7 (35) 0
79
ID 37 (74) 8 (40) 0
01
EM-like 12 (24) 0 0
01
AGEP-like 10 (20) 4 (20) 1
Mean no. of existing patterns < 0
01
None 6 (12) 8 (40)
1 16 (32) 7 (35)
≥2 28 (56) 5 (25)
Type of associated patterns
ID + eczematous 10 (20) 0 –
ID + eczematous + AGEP-like 6 (12) 2 (10) –
ID + EM-like 7 (14) 0 –

Values are given as n (%). Using a Bonferroni correction, P-values ≤ 0
01 were considered significant. ID, interface dermatitis; EM, erythema
multiforme; AGEP, acute generalized exanthematic pustulosis.

cells was higher in less severe DRESS, but the difference was
not significant. The number of cases studied for FoxP3 was
too limited to search for statistical differences between the
two groups.
Comparison with maculopapular rashes
Interface dermatitis was observed in seven (35%) cases of
MPR, which was significantly less than in DRESS syndrome
(P < 0
01). Apoptotic keratinocytes were seen only in 30% of
cases, while they were identified in 60% of instances of
DRESS syndrome, but the difference was not significant. Exo-
cytosis of neutrophils and pustulosis were seen in 25% of
cases, with no significant differences compared with DRESS
syndrome. Atypical activated lymphocytes were noted in
seven (35%) cases. As in DRESS syndrome, vasculitis was
never seen but nuclear debris was observed, although it
appeared to be rarer (P = 0
03). The density of the inflamma-
tory infiltrates was, in most cases, slight (17 of 20; 85%),
with only three biopsies showing an intermediate-density
infiltrate and none a high-density one. The density of dermal
infiltrate was significantly lower than in DRESS syndrome
(P < 0
01). By comparison, 58% of samples from the DRESS
syndrome group showed intermediate- or high-density infil-
trates (Table 2). In addition, patients with DRESS syndrome
had more mid-dermis infiltration (P = 0
04). In two cases of
MPR, there were minimal lesions, with a very subtle lympho-
cytic perivascular infiltrate and no epidermal injury. A signifi-
cant proportion of the MPRs had mainly dermal infiltrates,
with no significant inflammatory pattern (40% of MPR vs.
12% of DRESS; P < 0
01). The most represented patterns in
MPR were interface dermatitis and eczematous changes. How-
ever, the proportion of MPRs with interface dermatitis was
lower than in DRESS syndrome (P = 0
01), and no cases of
MPR vs. 24% of cases of DRESS syndrome had an EM-like
aspect (P = 0
01). Interestingly, fewer cases of MPR presented

British Journal of Dermatology (2015) 173, pp50–58 © 2015 British Association of Dermatologists
 Histopathology of DRESS syndrome, N. Ortonne et al. 55

(a) (b)

Fig 1. Various histopathological aspects of

cutaneous infiltrates of drug rash with (c) (d)
eosinophilia and systemic symptoms (DRESS)
syndrome. (a) Spongiotic dermatitis with
confluent areas of spongiosis within the
epidermis associated with lymphocyte
exocytosis. (b) This case of DRESS syndrome
shows a marked lichenoid interface dermatitis
with mild acanthosis and a heavy lymphocytic
infiltrate extending from the superficial dermis
to the epidermal basal layer, in which many
apoptotic keratinocytes are seen. (c) This case
closely resemble S
ezary syndrome, with a
mild perivascular infiltrate that comprises
atypical lymphocytes with enlarged
hyperchromatic nuclei (arrowheads and
inset). (d) A large multilocular pustule is
present, as usually seen in acute generalized
exanthematic pustulosis.

with two or more associated inflammatory patterns (25% vs.

56%; P < 0
01).
As in DRESS, MPRs showed infiltration of CD8+ T lympho-
cytes and comprised a proportion of granzyme B+ cytotoxic
effector cells (Table 3). No significant differences in the pro-
portion of FoxP3 lymphocytes were observed. Although the
proportion of CD8+ T cells was not different, the proportion
of granzyme B+ cells in both the epidermis (P < 0
01) and
the dermis (P < 0
01) was significantly higher in DRESS syn-
drome than in MPR.
Finally, we compared, for a selection of parameters, DRESS
syndrome with severe cutaneous lesions (erythroderma and/
or bullae) with DRESS syndrome presenting with a maculo-
papular eruption and with MPR. Interestingly, DRESS syn-
drome with severe cutaneous lesions more frequently showed
apoptotic keratinocytes and an EM-like pattern (P < 0
01),
which was never seen in MPR. In addition, DRESS syndrome
with severe cutaneous lesions had significantly more associated
patterns (mean 1
95 vs. 0
95; P = 0
01) and a higher propor-
tion of granzyme B+ lymphocytes (P = 0
04).
Discussion
DRESS syndrome is a severe systemic cutaneous drug reaction,
with a potentially fatal outcome due to visceral involvement,
and particularly to hepatic and cardiac injuries.16 Recently, the
European Severe Cutaneous Adverse Reactions to Drugs group
(EuroSCAR/RegiSCAR) proposed diagnostic criteria for the
disease, allowing constitution of homogeneous groups of
patients, as presented in this series. It is noteworthy that none
of these criteria rely on histopathology.
As for most drug reactions, except those with a particular
feature, such as TEN and AGEP, the histopathological aspect of
DRESS syndrome is not well described in the literature. In
textbooks, the syndrome is either not described on histopatho-
logical grounds,6 or it is only stated that it may present with
various histopathological features, including spongiotic derma-
titis, erythema multiformis or aspects of TEN.17 In Lever’s Histo-
pathology of the Skin, it is described as a variant of exanthemic
drug reactions, where eosinophils and scattered apoptotic
keratinocytes are commonly but not always seen.18 In their

© 2015 British Association of Dermatologists British Journal of Dermatology (2015) 173, pp50–58
56 Histopathology of DRESS syndrome, N. Ortonne et al.
(a) (b)
(c) (d)
clinicopathological study, Chiou et al.8 highlighted the
polymorphous histological expression of DRESS syndrome,
which may present with various inflammatory patterns. More
recently, in a series of 27 cases, Walsh et al.9 found that the
most frequent inflammatory changes were a spongiotic derma-
titis or basal cell vacuolization with apoptotic keratinocytes,
the former being more frequent in patients with MPR. We
found that the histopathological presentation of DRESS syn-
drome is highly variable, encompassing many inflammatory
patterns, from a slight perivascular lymphocytic infiltrate to a
pustular, AGEP-like, EM-like, eczematous or interface dermati-
tis, with the latter being the most frequent. What appeared to
be a special feature was the association of different inflamma-
tory patterns in a single specimen, a finding that was signifi-
cantly more frequent in DRESS syndrome than in nondrug-
induced dermatoses and MPR. The histopathological presenta-
tion did not seem to correlate with the culprit drug, although
relevant statistical analyses could only be obtained for allopuri-
nol. Interestingly, dermal eosinophils were present in only
20% of cases and apoptotic keratinocytes in only half of cases.
British Journal of Dermatology (2015) 173, pp50–58
Fig 2. Drug rash with eosinophilia and
systemic symptoms (DRESS) syndrome with
multiple different inflammatory patterns. (a)
In this unique section of a DRESS syndrome
skin biopsy, three distinct inflammatory
patterns can be observed: (b) a multilocular
pustule (arrowheads), (c) a spongiotic
dermatitis with a vesicle containing
Langerhans’ cells, and (d) foci of vacuolar or
lichenoid interface dermatitis at the dermal–
epidermal junction and within a follicular
adnexa.
Our results confirm that in a significant proportion of cases,
atypical lymphocytes may be found in dermal infiltrates, so
that differential diagnosis with cutaneous T-cell lymphoma
may be challenging, especially with SS. In this context, the
demonstration of a CD8+, granzyme B+ phenotype of the
atypical cells may be a helpful feature, as SS neoplastic cells
are CD4+. Taken together, our results further support a major
role for CD8+ effector T cells and the perforin/granzyme
pathway in drug reactions, especially in DRESS syndrome, as
previously shown in MPR.19 Interestingly, the histological
aspect of MPR, previously described in a large case series by
Gerson et al.,20 shares many features with DRESS syndrome.
Whether MPR and DRESS syndrome belong to the same spec-
trum of drug reactions, with MPR representing the less severe
end of the spectrum, is therefore questionable. The main dif-
ference in our series was the intensity of the inflammatory
changes, as the proportion of cases with epidermal lesions, as
well as the density of inflammatory infiltrates, appeared to be
lower in MPR than in DRESS syndrome, as previously shown.7
Whether the severity of skin inflammation and the extension
© 2015 British Association of Dermatologists
 Histopathology of DRESS syndrome, N. Ortonne et al. 57

Table 3 Phenotypical comparison between patients with drug rash with eosinophilia and systemic symptoms (DRESS) syndrome and
maculopapular rash (MPR)

DRESS (n = 24) MPR (n = 20) P-value

Proportion of CD8+ dermal cells (n = 22/n = 18)
≤ 5% 0 1 (5) 0
80
5–50% 4 (18) 3 (17)
> 50% 18 (82) 14 (78)
Presence of epidermotropic granzyme B+ cells (n = 20/17) 18 (90) 7 (41) < 0
01
Proportion of dermal granzyme B+ cells (n = 21/n = 11)
Score 1: ≤ 5% 4 (19) 4 (36) < 0
01
Score 2: 5–50% 13 (62) 3 (27)
Score 3: > 50% 4 (19) 4 (36)
Proportion of dermal FoxP3+ cells (n = 12/n = 14)
≤ 5% 8 (67) 6 (43) 0
60
5–50% 4 (33) 7 (50)
> 50% 0 1 (6)

Values are given as n (%).

(a) (b)

Fig 3. Phenotype of lymphocytes in drug rash

with eosinophilia and systemic symptoms
(DRESS) syndrome. (a) More than 50% of
lymphocytes, both in the epidermis and the
dermis, are CD8+ T cells. (b) Scattered (c) (d)
FoxP3+ lymphocytes are present within the
infiltrate (arrowheads). (c) DRESS syndrome
biopsy showing many granzyme B+
lymphocytes both in the dermis and in the
epidermis with typical cytoplasmic granular
staining. (d) Double-staining shows that the
atypical medium-sized lymphocytes seen in
the papillary dermis and around the capillaries
in this sample express CD8 (arrowheads).

to extracutaneous sites in DRESS syndrome, in contrast to
MPR, are due to differences in cytotoxic effector cells and/or
the infiltration of Tregs is therefore an interesting issue.
FoxP3+ Tregs were recently reported to be dramatically
expanded in blood during the acute phase of DRESS syn-
drome.13 Although both MPR and DRESS syndrome were
characterized by a significant infiltration of CD8+ and cyto-
toxic granzyme B+ cells, we found that DRESS syndrome with
a severe phenotype, presenting with an erythroderma and/or
bullae, had significantly more granzyme B+ cells. These cases
also presented with more inflammatory changes, as the num-
ber of associated inflammatory patterns was significantly
higher than in MPR and DRESS syndrome presenting with a
MPR. A potential role for skin-infiltrating Tregs in the limita-
tion of skin lesions in fixed drug reactions, in contrast to
TEN, has already been suggested, as sequential biopsies have
proven that the proportion of these cells in skin infiltrates
increases progressively.21 We found that FoxP3+ Tregs were
not as numerous as expected within the skin infiltrates during
the acute phase of DRESS syndrome. We speculate that DRESS
syndrome is actually characterized by a dramatic expansion of
Tregs in blood, while effector CD8+ T cells are recruited in
the skin, where they can exert cytotoxic functions because of
a lower proportion of Tregs recruited at the same time from
blood. The fact that different cytokinic environments may
coexist in blood and skin, leading to such differences, remains
speculative and requires further investigation. The major role
played by expanded activated T cells in DRESS syndrome has
already been demonstrated. In particular, it has been shown
recently that expanded CD8+ T-cell populations, found in all
the involved organs, are directed against herpesviruses, espe-
cially EBV. This suggests that DRESS syndrome may also be

© 2015 British Association of Dermatologists British Journal of Dermatology (2015) 173, pp50–58
58 Histopathology of DRESS syndrome, N. Ortonne et al.
regarded as a multiorgan antiviral T-cell response.12 For this
reason, we investigated whether EBV – a putative target of the
CD8+ T cell – was present in the skin. With the exception of
one case, we failed to demonstrate the recurrent presence of
EBV in situ. We also found no significant infiltration of CD56+
and CD123+ cells. This finding further suggests that the effec-
tor phase of DRESS mostly relies on adaptive immunity, with
the activation of cytotoxic T cells. In addition to lymphocyte
phenotyping, T-cell clonality probably represents an important
diagnostic criterion for differential diagnosis between DRESS
syndrome and cutaneous T-cell lymphomas. A T-cell clone in
skin or blood was only detected in two of 17 investigated
cases from this series. Interestingly, none of these cases
showed atypical T cells in the skin. This is in agreement with
a previous study in which T-cell clones were only detected in
blood of patients with DRESS syndrome.10 Of note, in most
cases, the search for a cutaneous T-cell clone was performed
on paraffin-embedded tissues, demonstrating that, in difficult
cases, this can be done easily by dermatopathologists.
In conclusion, various inflammatory patterns are observed
in DRESS syndrome, and these different patterns are often
seen to be associated in a single biopsy, which may repre-
sent a histopathological clue in diagnosis. Effector lympho-
cytes, at least in skin, comprise a high proportion of
polyclonal CD8+ granzyme B+ lymphocytes. The histopatho-
logical presentations and the cutaneous effector cells in
DRESS and MPR show some overlap, but DRESS is character-
ized by more inflammation and more granzyme B+ effector
cells, especially when patients present with an erythroderma
and/or bullae.
Acknowledgments
We thank Laetitia Gregoire, Unit
e de Recherche Clinique
(URC), Henri Mondor Hospital, and Audrey Colin from the
Department of Dermatology, Henri Mondor Hospital, for their
help in presenting the study to the ethics committee (Comit
e
de Protection des Personnes, Ile de France IV).
References
1 Ben m’rad M, Leclerc-Mercier S, Blanche P et al. Drug-induced
hypersensitivity syndrome: clinical and biologic disease patterns in
24 patients. Medicine (Baltimore) 2009; 88:131–40.
2 Kardaun SH, Sekula P, Valeyrie-Allanore L et al. Drug reaction with
eosinophilia and systemic symptoms (DRESS): an original multi-
system adverse drug reaction. Results from the prospective Regi-
SCAR study. Br J Dermatol 2013; 169:1071–80.
3 Tohyama M, Hashimoto K, Yasukawa M et al. Association of
human herpesvirus 6 reactivation with the flaring and severity of
British Journal of Dermatology (2015) 173, pp50–58
drug-induced hypersensitivity syndrome. Br J Dermatol 2007;
157:934–40.
4 Seishima M, Yamanaka S, Fujisawa T et al. Reactivation of human
herpesvirus (HHV) family members other than HHV-6 in drug-
induced hypersensitivity syndrome. Br J Dermatol 2006; 155:344–9.
5 Shiohara T, Inaoka M, Kano Y. Drug-induced hypersensitivity syn-
drome (DIHS): a reaction induced by a complex interplay among
herpesviruses and antiviral and antidrug immune responses. Allergol
Int 2006; 55:1–8.
6 Weedon D. Weedon’s Skin Pathology, 3rd edn. Edinburgh: Churchill
Livingstone, 2010.
7 Chi MH, Hui RC, Yang CH et al. Histopathological analysis and
clinical correlation of drug reaction with eosinophilia and systemic
symptoms (DRESS). Br J Dermatol 2014; 170:866–73.
8 Chiou CC, Yang LC, Hung SI et al. Clinicopathological features and
prognosis of drug rash with eosinophilia and systemic symptoms:
a study of 30 cases in Taiwan. J Eur Acad Dermatol Venereol 2008;
22:1044–9.
9 Walsh S, Diaz-Cano S, Higgins E et al. Drug reaction with eosino-
philia and systemic symptoms (DRESS): is cutaneous phenotype a
prognostic marker for outcome? A review of clinicopathological
features of 27 cases. Br J Dermatol 2013; 168:391–401.
10 Cordel N, Lenormand B, Courville P et al. Study of the clonality of
cutaneous and blood lymphocytes during drug-induced hypersen-
sitivity in 6 patients. Ann Dermatol Venereol 2004; 131:1059–61.
11 Chung WH, Hung SI, Yang JY et al. Granulysin is a key mediator
for disseminated keratinocyte death in Stevens-Johnson syndrome
and toxic epidermal necrolysis. Nat Med 2008; 14:1343–50.
12 Picard D, Janela B, Descamps V et al. Drug reaction with eosino-
philia and systemic symptoms (DRESS): a multiorgan antiviral T
cell response. Sci Transl Med 2010; 2:46ra62.
13 Takahashi R, Kano Y, Yamazaki Y et al. Defective regulatory T cells
in patients with severe drug eruptions: timing of the dysfunction
is associated with the pathological phenotype and outcome. J
Immunol 2009; 182:8071–9.
14 Kardaun SH, Sidoroff A, Valeyrie-Allanore L et al. Variability in the
clinical pattern of cutaneous side-effects of drugs with systemic
symptoms: does a DRESS syndrome really exist? Br J Dermatol 2007;
156:609–11.
15 Delfau-Larue MH, Petrella T, Lahet C et al. Value of clonality stud-
ies of cutaneous T lymphocytes in the diagnosis and follow-up of
patients with mycosis fungoides. J Pathol 1998; 184:185–90.
16 Bourgeois GP, Cafardi JA, Groysman V et al. A review of DRESS-
associated myocarditis. J Am Acad Dermatol 2011; 66:e229–36.
17 McKee PH, Calonje E, Granter SR. Pathology of the Skin with Clinical
Correlations, 3rd edn, vol. 1. London: Elsevier Mosby, 2005.
18 Elder DE. Lever’s Histopathology of the Skin, 10th edn. Philadelphia, PA:
Lippincott Williams and Wilkins, 2009.
19 Yawalkar N, Egli F, Hari Y et al. Infiltration of cytotoxic T cells in
drug-induced cutaneous eruptions. Clin Exp Allergy 2000; 30:847–55.
20 Gerson D, Sriganeshan V, Alexis JB. Cutaneous drug eruptions: a
5-year experience. J Am Acad Dermatol 2008; 59:995–9.
21 Mizukawa Y, Yamazaki Y, Shiohara T. In vivo dynamics of intraepi-
dermal CD8+ T cells and CD4+ T cells during the evolution of
fixed drug eruption. Br J Dermatol 2008; 158:1230–8.
© 2015 British Association of Dermatologists