ARTICLE
Auteur(s) : Anastasia Papagrigoraki, Paolo Gisondi, Paolo Rosina,
Manuela Cannone, Giampiero Girolomoni
Section of Dermatology and Venereology, Department
of Medicine, University of Verona, Piazzale A. Stefani 1,
37126 Verona, Italy
accepté le 1 Ao�t 2010
Erythema nodosum (EN) is the most frequent variant of
panniculitis, with an incidence of approximately 1-5 cases per
100,000 persons/year, with a male to female ratio of 1:6 and with
the peak incidence between 20 and 30 years of age.
Clinically, it is characterized by the sudden onset of multiple
tender painful bluish-purple erythematous nodules, measuring one to
six cm in diameter. It is most commonly located bilaterally on the
extensor surface of the lower limbs, particularly on the pretibial
area. Nodules are self-limited and undergo typical colour changes
from bright red to brownish yellow bruise-like discoloration, then
subside after one to six weeks without signs of scarring,
ulceration or atrophy. Histologically, EN represents a
stereotypical example of septal panniculitis with no
vasculitis [1]. Histology, however, is rarely required to confirm
the diagnosis. EN is thought to result from an exaggerated
cell-mediated immune response to various antigens, although some
data indicate a possible involvement of the deposition
of immune complexes on subcutaneous venules [2]. Recently, it
has also been shown that radical oxygen intermediates might play a
role in the pathogenesis of EN [3]. EN can be associated with
a vast variety of disease processes, including infections, drugs,
malignant diseases, sarcoidosis, chronic inflammatory bowel
diseases, rheumatological diseases, autoimmune disorders and
pregnancy [4]. The aetiology of EN may vary considerably according
to the geographical area and the medical specialization of the
department collecting the cases [5-10]. In about one sixth to one
third of patients, EN is a recurrent disease. In general, relapsing
EN presents with smaller and less painful nodules which last for a
shorter time [6, 7, 9]. It is unclear whether relapsing EN has an
aetiology distinct from the sporadic cases, and whether the
recurrent course can be predicted.
In this study, we reviewed the cases of EN observed in our
clinic over the past 10 years, focusing on the cases in which
the disease was recurrent and trying to correlate the relapsing
course with a specific aetiology.
Patients and methods
One hundred and twenty-four patients who received the diagnosis of
EN between 1997 and 2007 were enrolled in the study. In most
patients the diagnosis was clinical, and it was confirmed by
histopathology in only 16 patients. EN was defined as the
acute onset of tender, painful, erythematous, subcutaneous nodules,
located bilaterally on the pretibial areas. All patients were
carefully questioned for drug usage, as well as preceding and
concurrent systemic symptoms. A thorough physical examination
was also performed in all patients. Laboratory investigations
included a complete blood count, erythrocyte sedimentation rate,
C-reactive protein, liver function tests, serum lipids, urea and
glucose, as well as rheumatoid factor, anti-nuclear antibodies
(ANA), and angiotensin converting enzyme (ACE). Hormonal research
included serum levels of estrogens, progesterone, luteinizing
hormone (LH), follicle stimulating hormone (FSH) and free and total
testosterone levels. Bacteria and parasite screening was performed
in all patients by nose and throat swab cultures and faeces
analysis. Serum anti-streptolysin O (ASO) and anti-DNAse titers,
screening for Epstein-Barr, hepatitis B and hepatitis C infections,
and common respiratory viruses were performed. A tuberculin
skin test and chest X-ray were routinely done. After the first
manifestation of EN, patients were re-contacted by phone and asked
to return for a follow-up visit. In particular, physical
examination and laboratory investigations were repeated. The
criteria for defining EN as post-infective was the temporal (<
2 months) [1], association between the appearance of EN and
systemic symptoms, such as sore throat, fever, malaise, diarrhoea
and arthralgias; positive infective sample of the rhino-pharynges,
urinary track, faeces and skin; elevated ASO and anti-DNAse titers;
elevated white blood cell levels; a positive intradermal reaction
to the tuberculin skin test and pulmonary parenchyma involvement on
the chest X-Ray examination. Diagnosis of drug induced EN was based
on temporal correlation (< 2 months) [11-15], the relapse of EN
after drug re-introduction and the absence of relapsing EN with a
continuous treatment with the imputed drug [13, 14]. When the above
criteria were excluded and EN was not associated with an underlying
systemic disease or pregnancy, it was considered as idiopathic. The
time from the primary EN and the re-evaluation ranged from one to
ten years (mean ± SD 5 ± 4; median 3 years). Recurrence of EN
was defined as the re-emergence of the typical erythematous nodules
after a disease-free period of at least one month. In order to
determine which factors could predict recurrences of EN,
aetiological factors deemed responsible for the first manifestation
of EN were compared to those imputed as causing the relapse(s).
Statistical analysis
All analyses were performed using the STATA (version 10.0
Stata-Corp LP, College Station, TX) and Graph-Pad (version 4.0, El
Camino Real, San Diego, CA) software packages. Categorical
variables were expressed as a percentage and continuous variables
in mean ± standard deviation (SD). Data is expressed as means ± SD
or percentages. Statistical analyses included the paired t test and
logistic regression analysis. The independence of the association
of EN relapse was assessed by multivariate logistic regression
analysis. In the fully adjusted regression model, age, gender,
re-assumption of the imputed drug, infections, systemic disease and
pregnancy were included as baseline covariates. Values at p <
0.05 were considered statistically significant.
Results
The study included 124 patients; nearly all were females (113,
91.1%) and the male to female ratio was one to ten. The mean age of
the patients was 39.5 ± 16 (SD) years, ranging from 4 to
90 years, with a median age of 37 years. A specific
aetiology for the first manifestation of EN was attributed in 73
(58.8%) patients. The leading incidental factor suspected to be the
cause of EN in 32 patients was found to be infection (25.8%
(32/124) of the total number of patients; 43.8% (32/73) of patients
with an attributed aetiology). Seventy-one of 124 patients
presented infective foci at the pharynges (50%), teeth (27%),
gastrointestinal track (9%), urinary track (7%), lungs (4.2%) or
skin (3%). One patient had a positive faeces sample for
Blastocystis Hominis. It was possible to associate EN with an
infection in only 32 (25.8%) of the 71 patients. In
30 patients (24.1%, 30/124; 41%, 30/73) diagnosis of
post-streptococcal EN was made based on a positive culture sample
of the pharynges for Streptococcus pyogenes beta-haemolytic group
A and/or elevated ASO and anti-DNAse titers. In two patients,
EN was secondary to pulmonary tuberculosis. Both patients were
young (16 and 19 years old) African females and each
manifested a strong positive intradermal reaction to the tuberculin
skin test; one patient presented a chest X-ray for a miliary
tuberculosis and the other for a primary tuberculoma localized on
the apical region of the left lung.
Drug exposure was considered responsible in 19 cases
(15.3%, 19/124; 26%, 19/73). Drugs responsible for inducing EN were
found to be sexual hormones (mostly estrogen-progestin
combinations; (5.6%, 7/124; 9.6%, 7/73; 36.8%, 7/19), followed by
antibiotics (3.2%, 4/124; 5.5%, 4/73; 21%, 4/19) and NSAIDs (3.2%;
5.5%; 21%) and antidepressants (1.6%, 2/124; 2.7%, 2/73; 10.5%,
2/19) (table 1). In
14 patients (11.2%, 14/124; 19.2%, 14/73) a systemic disease
was associated, mostly sarcoidosis (7 patients 5.6%; 9.6%),
followed by malignancies (2 cases of Hodgkin's lymphoma, one of
non-Hodgkin's lymphoma, and one of astrocytoma), Crohn's disease (2
cases), and Behçet's syndrome (1 case). Eight of the
113 female patients were pregnant.
Forty-two of the 124 patients reported having prodromal
symptoms, such as fever, malaise and arthralgias, present in
particular in cases associated with infection (24 out of 42).
The duration of the nodules was 14 ± 6 days (mean ± SD), with
persistence for more than 10 days in 92% of the patients. EN
was treated in 28.2% with only topical application of
corticosteroids and heparinoids, in 24.2% with systemic antibiotics
(macrolides), in 8% with analgesics (tramadol), in 6.45% with
systemic corticosteroids and in 33% with a combined treatment
(antibiotics, steroids, analgesics and topical).
EN relapsed in 33 patients (26.6%) with only one relapse in
17 patients (13.7%; 51.5%), two relapses in 6 (4.8%; 18.2%),
and three or more relapses in 8 patients (6.45%; 24.2%).
Relapse duration was 9 ± 2 days (p < 0.05 vs. primary EN), with
less than a ten-day duration in 90.9% of the patients. EN relapsed
after one year in 21 patients (63.6%), after two years in
9 patients (27.2%) and after three or more years in
3 patients (9%). Moreover, nodules were smaller in number and
size, and less painful. Relapses were never associated with
prodromal symptoms. An aetiological factor was identified in 25
(75.8%) of the 33 cases of relapse. Aetiologies of the first
manifestation and of the relapses causing EN are presented in figure 1. The
leading factor causing EN relapse was infection (33.3% of the total
number of relapsing EN; 44% of relapsing EN with an attributed
aetiology), all post-streptococcal. However, relapsing
post-streptococcal EN occurred in patients in whom the first
manifestation of EN was classified as idiopathic (5/11), as
associated with pregnancy (2/11) or as drug-induced (1/11). The
second most common cause of relapsing EN was drugs (24.2% and 32%),
followed by systemic diseases (9% and 12%) and pregnancy (9% and
12%). Relapsing drug-induced EN was always associated with the use
of the same drug, particularly sex hormones (table 2). Patient 7 had 10 relapses
of EN under progesterone. This patient was taking progesterone for
recurring uterine fibromas, but the gynaecologist did not agree
with withdrawing the medicine from the therapy, and therefore EN
continued to recur. After multiple regression analysis, the only
significant factor associated with the relapse was use of the same
drug (odds ratio 4.09; p = 0.025). In contrast, EN caused by other
aetiologies relapsed due to different factors which could not be
predicted. No significant correlation was found between the
relapses and sex, age, infections, pregnancy, and systemic
disorders such as sarcoidosis, inflammatory bowel diseases and
lymphomas (table 3). In most
patients, relapsing EN was treated only with topical treatment
(39.4%) and systemic treatment was added in only 18.2% of the
recurrent cases. The remaining cases did not receive therapy.
Table 1 Drugs associated with erythema nodosum
|
DRUG
|
Number and percentage of patients out
of the total number of patients (n = 124)
|
Percentage of patients out of the number
of patients with an attributed aetiology (n =
73)
|
|
SEXUAL HORMONES
|
7 (5.6%)
|
9.6%
|
|
Ethinil estradiol/desogestrel
|
3
|
|
|
Ethinil estradiol/levonogestrel
|
1
|
|
|
Ethinil estradiol/gestodene
|
1
|
|
|
Progesterone
|
1
|
|
|
Estradiol hemihydrate
|
1
|
|
|
NSAIDs
|
4 (3.2%)
|
5.5%
|
|
Nimesulide Ibuprofen
|
2 2
|
|
|
ANTIBIOTICS
|
4 (3.2%)
|
5.5%
|
|
Penicillins Macrolides Sulfamides Cephalosporines
|
1 1 1 1
|
|
|
ANTIDEPRESSANTS
|
2 (1.6%)
|
2.7%
|
|
Benzodiazepines Barbiturates
|
1 1
|
|
|
OTHER DRUGS
|
2 (1.6%)
|
2.7%
|
|
Montelukast Mesterolone
|
1 1
|
|
Table 2 Drugs associated with relapsing erythema
nodosum
|
Patient
|
Age
|
Sex
|
N° of relapses
|
Drug
|
|
1
|
25
|
F
|
1
|
Nimesulide
|
|
2
|
29
|
F
|
6
|
Estradiol/desogestrel
|
|
3
|
31
|
F
|
3
|
Estradiol/gestodene
|
|
4
|
61
|
F
|
4
|
Benzodiazepine
|
|
5
|
16
|
F
|
2
|
Estradiol/desogestrel
|
|
6
|
24
|
F
|
1
|
Estradiol/levonogestrel
|
|
7
|
27
|
F
|
10
|
Progesterone
|
|
8
|
18
|
F
|
6
|
Ibuprofen
|
Table 3 Independent predictors of erythema nodosum
relapse
|
Independent variables
|
Odds Ratio
|
95% Confidence Interval
|
p
|
|
Relapse
|
Age (years)
|
1.00
|
0.97-1.04
|
0.63
|
|
Gender (male vs. female)
|
1.61
|
0.26-9.97
|
0.06
|
|
Re-introduction of the imputed drug (yes vs. no)
|
4.09
|
1.19-14.0
|
0.025
|
|
Infections (yes vs. no)
|
0.4
|
0.14-1.33
|
0.14
|
|
Systemic disease (yes vs. no)
|
1.2
|
0.37-4.06
|
0.72
|
|
Pregnancy (yes vs. no)
|
0.97
|
0.94-1.00
|
0.11
|
Discussion
EN occurs mostly in young females, women in the first and second
trimester of pregnancy and in patients using oral contraceptives
containing high doses of estrogens [16-19]. An aetiology can be
attributed in nearly half of the cases, generally with infections
as the leading cause (in particular streptococcal pharyngitis and
primary tuberculosis), followed by drugs and associated diseases
such as sarcoidosis, Behçet's syndrome and inflammatory bowel
diseases [5, 10, 16-20]. In our series we were able to attribute a
cause in 58.8% of patients, most commonly infections, followed by
drugs, systemic diseases (sarcoidosis and Crohn's disease) and
pregnancy. The high rate of attributed aetiology in this study is
based on a temporal correlation between the appearance of EN and
the aetiology, which allowed for distinguishing between association
and causality. The high percentage of drug-induced EN (15.3%;
26%) in our series is also explained by the fact that some
patients, during hospitalization, reported in their medical history
that there had been a recent change in treatment medication for a
recurrent streptococcal infection of the pharynges (with systemic
symptoms, elevated serum ASO and/or anti-DNAse titers, and/or
positive culture sample of the pharynges for Streptococcus
pyogenes). In these cases EN was not correlated to the infection,
but to the change in medication and therefore regarded as drug
induced. In addition, the same group of patients reported during
their follow up visits that they never had a relapse of EN after
stopping usage of the new drug, though infective pharyngitis
continued to recur, thus indicating a correlation to the drug and
not to the disease.
The aetiology of EN may vary considerably according the
geographical area. Streptococcal pharyngitis is the most frequent
aetiology of EN throughout the world, both in developed and
developing countries [5, 6, 8, 9, 20]. Tuberculosis is a common
aetiology of EN in the countries where tuberculosis is endemic,
whereas it is an uncommon cause of EN in developed countries [5, 6,
9, 10, 21]. Sarcoidosis is commonly associated with EN, with a high
prevalence in the USA, a present but less persistent prevalence in
Europe, a low prevalence in the Middle East and Southern Asia, and
a very low prevalence in Africa [22]. In our study, 7 of
124 patients were affected by sarcoidosis and in each patient
EN observation led to the diagnosis of sarcoidosis. One of these
patients had Löefgren's syndrome, whereas the others had a
multi-organ disease. It has been found that patients affected by
sarcoidosis who develop EN have a nucleotide exchange at position
-308 in the human TNF-α gene promoter, whereas patients with EN
without sarcoidosis displayed a similar allele frequency to
controls. These results support the notion that EN associated to
sarcoidosis may be pathogenetically linked to altered TNF-α
production [16, 23]. In our study Behçet's syndrome was observed in
only one male patient. This is a rather low incidence in comparison
with other studies performed in countries such as Turkey, where
Behçet's syndrome is more prevalent [6, 8]. Moreover, only two
patients had Crohn's disease, although inflammatory bowel diseases
(such as ulcerative colitis and Crohn's disease) are relatively
common in our country. This is explained by the fact that the
diagnosis of the inflammatory bowel disease precedes the
manifestation of EN in the majority of the patients [24], and these
patients are usually followed-up on and treated in gastrointestinal
units rather than dermatological units.
Previous studies could not determine an association between
relapses and aetiology, nor could they predict relapses,
classifying most relapses as idiopathic. Some cases of recurrent EN
were associated with a systemic disease such as sarcoidosis,
Crohn's disease or Behçet's syndrome; a few were considered as post
infective and none were associated with drugs [6, 7, 9] (table 4). In our study, however, an aetiology
for the relapses could be attributed in 25 of 33 (75.8%)
patients, attributed in particular to infections, followed by
drugs, systemic diseases and pregnancy. Factors responsible for the
first manifestation of EN almost always differed from those causing
relapses in the same patients. On the other hand drug-induced
relapsing EN could be predicted. Drug-induced EN relapsed in
8 of 19 patients and 5 of 8 drug-induced
relapsing cases of EN were associated with the use of sexual
hormones (table 2). It was also
observed that when a patient continued to be treated with the same
drug, EN constantly relapsed.
Our study has several limits, including the fact that patients
were recruited in a tertiary referral centre, and thus they may not
be completely representative of patients with EN visited in other
settings. Moreover, the attribution of an aetiology is not based on
validated criteria, but this is a common concern in several
multi-factorial diseases.
In conclusion, many factors may lead to EN, and the prevalence
of these disorders varies according to geographical area and the
medical department collecting the cases. Therefore it appears
advisable to establish local guidelines in order to determine the
most effective diagnostic approach to EN, as well as to prevent
relapses.
Table 4 Medical departments and the related
geographical areas where relapsing EN has previously been studied
|
Reference
|
Medical department
|
Country
|
Patients
|
Relapse (%)
|
Aetiology of relapsing EN (N, %)
|
|
Mert A et al. [6]
|
Infectious Disease Microbiology Internal Medicine Dermatology
|
Turkey
|
100
|
34 (34%)
|
Idiopathic 33 (97%) Behçet 1 (3%)
|
|
Cribier et al. [7]
|
Dermatology
|
France
|
129
|
8 (6.2%)
|
Idiopathic 5 (62.5%) Sarcoidosis 1(12.5%) Infections 2 (25%)
|
|
Garcia-Porrua et al. [9]
|
Rheumatology
|
Spain
|
106
|
17 (16%)
|
Idiopathic 13 (76.5%) Crohn's disease 1 (5.9%) Sarcoidosis 1 (5.9%)
Infections 2 (11.8%)
|
|
Our study
|
Dermatology
|
Italy
|
124
|
33 (26.6%)
|
Infections 11 (33.3%) Drugs 8 (24.2%) Idiopathic 8 (24.2%) Systemic
diseases 3 (9%) Pregnancy 3 (9%)
|
Disclosure
Funding sources: none. Conflict of interest: none.
References
1 Requena L, Sánchez Yus E. Erythema nodosum. Semin Cutan
Med Surg 2007; 26: 114-25.
2 Jones JV, Cumming RH, Asplin CM, Laszlo G,
White RJ. Letter: Circulating immune complexes in erythema
nodosum and early sarcoidosis. Lancet 1976; 1: 153.
3 Kunz M, Beutel S, Bröcker E. Leucocyte
activation in erythema nodosum. Clin Exp Dermatol 1999; 24:
396-401.
4 Schwartz RA, Nervi SJ. Erythema nodosum: a sign of
systemic disease. Am Fam Physician 2007; 75: 695-700.
5 Erez A, Horowitz J, Sukenik S. Erythema nodosum
in the negev area-a survey of 50 patients. Isr J Med Sci 1987;
23: 1228-31.
6 Mert A, Kumbasar H, Ozaras R, et al.
Erythema nodosum: an evaluation of 100 cases. Clin Exp
Rheumatol 2007; 25: 563-70.
7 Cribier B, Caille A, Heid E, Grosshans E.
Erythema nodosum and associated diseases. A study of
129 cases. Int J Dermatol 1998; 37: 667-72.
8 Mert A, Ozaras R, Tabak F, Pekmezci S,
Demirkesen C, Ozturk R. Erythema nodonum: an experience
of ten years. Scand J Infect dis 2004; 36: 424-7.
9 García-Porrúa C, González-Gay MA,
Vázquez-Caruncho M, et al. Erythema nodosum: etiologic
and predictive factors in a defined population. Arthritis Rheum
2000; 43: 584-92.
10 Psychos DN, Voulgari PV, Skopouli FN,
Drosos AA, Moutsopoulos HM. Erythema nodosum: the
underlying conditions. Clin Rheumatol 2000; 19: 212-6.
11 Valeyrie-Allanore L, Sassolas B, Roujeau JC.
Drug-induced skin, nail and hair disorders. Drug Saf 2007; 30:
1011-30.
12 Fiszenson-Albala F, Auzerie V, Mahe E,
et al. A 6-month prospective survey of cutaneous drug
reactions in a hospital setting. Br J Dermatol 2003; 149:
1018-22.
13 Benahmed S, Picot MC, Hillaire-Buys D,
Blayac JP, Dujols P, Demoly P. Comparison of
pharmacovigilance algorithms in drug hypersensitivity reactions.
Eur J Clin Pharmacol 2005; 61: 537-41.
14 Aberer W, Bircher A, Romano A, et al.
Drug provocation testing in the diagnosis of drug hypersensitivity
reactions: general considerations. Allergy 2003; 58: 854-63.
15 Vedove CD, Del Giglio M, Schena D,
Girolomoni G. Drug-induced lupus erythematous. Arch Dermatol
Res 2009; 301: 99-105.
16 Labunski S, Posern G, Ludwig S, Kundt G,
Bröker EB, Kunz M. Tumour necrosis factor-alpha promoter
polymorphism in erythema nodosum. Acta Derm Venereol 2001; 81:
18-21.
17 Mana J, Marcoval J. Erythema nodosum. Clin Dermatol
2007; 25: 288-94.
18 Amoli MM, Thomson W, Hajeer AH, et al.
HLA-DRB1 associations in biopsy proven erythema nodosum. J
Rheumatol 2001; 28: 2660-2.
19 Jhaveri K, Halperin P, Shin SJ, Vahdat L.
Erythema nodosum secondary to aromatase inhibitor use in breast
cancer patients: case reports and review of the literature. Breast
Cancer Res Treat 2007; 106: 315-8.
20 Soderstrom RM, Krull EA. Erythema nodosum.
A review. Cutis 1978; 21: 806-10.
21 Vashisht P, Sahoo B, Khurana N, Reddy BS.
Cutaneous tuberculosis in children and adolescents: a
clinico-histological study. J Eur Acad Dermatol Venereol 2007; 21:
40-7.
22 Crystal RG. Sarcoidosis. In: Braunwald E,
Fauci AS, Kasper DL, Hauser SL, Longo DL,
Jameson JL, eds. Harrison's principles of internal medicine.
New York: McGrow-Hill Ltd, 2001: 1969-74.
23 McDougal KE, Fallin MD, Moller DR, Song Z, Culter DJ, Steiner
LJ, Cutting GR and ACCESS Research group. Variationi in the
lynphotoxin-α/Tumor necrosis factor locus modifies risk of erythema
nodosum in sarcoidosis. J Invest Dermatol 2009; 129: 1921-26.
24 Farhi D, Cosnes J, Zizi N, et al.
Significance of erythema nodosum and pyoderma gangrenosum in
inflammatory bowel diseases: a cohort study of 2402 patients.
Medicine 2008; 87: 281-93.
|
Printable version
Version PDF
