OUP user menu

Long-term outcome in idiopathic granulomatous mastitis: a western multicentre study

A. Néel, M. Hello, A. Cottereau, J. Graveleau, P. De Faucal, N. Costedoat-Chalumeau, M. Rondeau-Lutz, C. Lavigne, L. Chiche, E. Hachulla, S. Seiberras, J. Cabane, N. Tournemaine, M. Hamidou
DOI: http://dx.doi.org/10.1093/qjmed/hct040 433-441 First published online: 13 February 2013

Abstract

Aim: To investigate the presentation, disease course and long-term outcome of a western cohort of idiopathic granulomatous mastitis (IGM) and to analyse the impact of different therapeutic strategies.

Methods: Multicentre retrospective study of 23 women followed over an extended period. Patients were recruited in nine French internal medicine departments.

Results: The median follow-up was 6 years. IGM presented commonly as a single inflammatory unilateral extra-areolar lump of varying size. Clinical course was heterogeneous and frequently remitting/relapsing. Most patients had at least one recurrence (18/23, 78%). The mean number of recurrences was 1.3 ± 1.5. Seven women had a bilateral evolution. Twelve women received steroids (corticosteroids). Only two of these did not respond to corticosteroids, whereas six relapsed when dose was tapered off. Nine patients received colchicine and/or hydroxychloroquine. First-line treatment consisted of excisional surgery in eight cases. At the date of last interview, 91% of the patients declared to be healed, 15 being free of treatment. However, 12/21 (57%) reported significant sequelae (unsightly scars: eight and/or lasting pain: six). Unsightly scars were not more prevalent in patients who had received steroids whereas they tended to be more frequent after breast excisional surgery. In addition, we found that excisional surgery did not prevent recurrences more successfully than a conservative approach.

Conclusions: Despite its retrospective nature, this Caucasian series provides novel information regarding long-term outcomes in IGM and argues in favour of conservative approaches. The value of immunomodulatory drugs such as colchicine or hydroxychloroquine deserves further investigation.

Introduction

Idiopathic granulomatous mastitis (IGM) is a rare chronic inflammatory disorder of the breast first described in 1972 by Kessler and Wolloch.1 This localized granulomatosis usually affects women of childbearing age and often mimics carcinoma or breast abscess.2,3 Histopathology evaluation then plays a crucial role in the positive and differential diagnosis of the disease.4 Its physiopathology remains elusive.

Several series of IGM have been published previously but have a limited follow-up duration.5–18 Moreover, they have been mainly reported by senological or surgical teams from non-western countries (India, Saudi Arabia, Turkey and China). Further, while various therapeutic approaches have been advocated, including excisional surgery on one side and systemic treatment (corticosteroid or immunosuppressant) on the other side, the impact of different treatment strategies has never been compared.19

The main purposes of this study were (i) to investigate the presentation, disease course and long-term outcome of a western cohort of IGM arising from internal medicine departments and (ii) to analyse the clinical impact of different therapeutic approaches.

Patients and methods

Study design and inclusion criteria

In 2011, a letter that proposed the enrolment of patients in the present study was sent to all members of the French Society of Internal Medicine (SNFMI). The study was conducted in compliance with Declaration of Helsinki and French laws. Inclusion criteria were: pathology-proven IGM with at least 12-months follow-up and informed consent. Pathological criteria for the diagnosis of IGM were the presence of non-caseating granulomatous inflammation. When no well-formed granuloma could be identified, the presence of an epithelioid histiocytic infiltrate was required.10 Tuberculous mastitis was excluded in all cases. Relevant data from medical records were collected and long-term outcome was determined by direct phone interview (M.H.) of patients (n = 19) or/and by the treating physician (n = 4) using a standardized questionnaire.

Statistical analysis

Continuous variables are expressed as means (±SD, standard deviation) or medians (range) and were compared using non-parametric Mann–Whitney U-tests. Categorical variables are expressed as numbers and were analysed using Fisher’s exact test. All P-values were two-sided; significance was set at P < 0.05. Data were analysed using GraphPad Prism version 4.02 software (GraphPad Software Inc., San Diego, CA, USA).

Results

Characteristics of the study population

Nine centres participated in the study and 23 patients were included. All patients were Caucasian women. Their characteristics are detailed in Table 1. The mean age at onset of the disease was 39 years ± 8 years (range: 25–61 years). Five patients (21%) had a previous history of autoimmune or inflammatory disease. There was no previous history of tuberculosis or systemic granulomatosis in any of them. Six women were nulliparous (26%). Among parous patients, less than a half had breastfed (8/17). No constant history of oral contraceptive use could be found.

View this table:
Table 1

Patients’ general characteristics and associated conditions

PatientAge at onset (years)Past pregnancyPast lactationContraception at onsetTobacco smokingMedical historyBiological autoimmunity
1333YesNoneNoNP
2314NoCOCNoSpondyloarthritisANA 1/160
3364NoNoneYesNP
4413NoNoneNoNP
5380NoIUDNoAntipsychotic drugNo
6371NoPOCYesNo
7432NoIUDNoNo
8373NoPOCNoNo
9432YesNoneNoThyroiditisAnti-TPO+
10340NoNoneNoAntipsychotic drugNo
11514YesIUDYesNo
12403YesSINoProlactinomaNP
13372YesIUDNoNo
14342NoNoneYesNP
15250NoNoneNoNo
16470NoSINoAntipsychotic drugNo
17381NoNoneNoDermatomyositis Ulcerative colitisAnti-JO1+
18391NoMechanicNoNP
19361YesNoneNoPolymyositisANA 1/320
20560NoNoneNoLupusNo
21612YesNoneNoNo
22342YesNoneYesEndometriosisANA 1/160
23300NoSINoAntipsychotic drugNo
  • ANA, antinuclear antibodies; COC, combined oral contraceptive; IUD, intrauterine diaphragm; NP, not performed; POC, progesterone oral contraceptive; SI, subdermal implant; TPO, thyroperoxydase.

Disease presentation

Clinical presentation is detailed in Table 2. IGM presented commonly as a single unilateral extra-areolar lump of varying size (from 1.5 to 6 cm) in 18 of 23 women (78%). Both breasts were equally involved, with no predilection for any particular site. The overlying skin was frequently inflammatory (19/23, 83%). Pain was frequent but inconstant (16/23, 70%). Cutaneous complications, such as fistulization and ulceration were described in 10 and 6 women, respectively. Mild grade fever was noticed in two cases.

View this table:
Table 2

Clinical presentation and long-term outcome

PresentationOutcome
PatientLocationLumps (n)Overlying skinPainComplicationsBilateral evolutionRecurrences (n)Unsightly scarsLasting painOutcomeFollow-up (month)
1Left1InflYesALNYes1YesNoHealed102
2Right1NormalYesALNYesOverlappingNoYesResistant78
3Left1Infl/Nip RetrNoNoneNo1NoYesHealed84
4Right1InflYesNoneNo1NoNoHealed117
5Right1NormalNoALN/Fis/UlcNo0YesNoHealed62
6Left1InflYesNoneNo0NoYesHealed122
7Right1Infl/Nip RetrNoALNYes4YesNoHealed336
8Left1NormalNoNoneNo0NoNoHealed70
9Right2Infl/Nip RetrNoFisNo0YesYesHealed68
10Right1InflYesFisYes3NoNoHealed70
11Left1InflYesNoneYes1NoNoHealed193
12Left1Infl/Nip RetrYesFisNo1NoNoHealed130
13Left3InflYesALN/Fis/UlcNo1NoNoHealed75
14Right2Infl/Nip RetrYesFisNoOverlappingNoNoHealed16
15Right1Infl/Nip RetrYesALNNo3YesYesResistant28
16Left3Infl/Nip RetrYesNoneNo4NoYesHealed56
17Left2Infl/Nip Retr/P OrNoALN/Fev/Fis/UlcNo1NoNoHealed14
18Left1InflYesUlcNo1NoNoHealed59
19Left1NormalNoNoneNo1YesNoHealed57
20Right1InflYesNoneYes5YesNoHealed97
21Left1InflYesFis/UlcNo1NoYesHealed107
22Right1Infl/Nip RetrYesFev/FisYes1YesYesHealed149
23Right1Infl/Nip Retr/P OrYesFis/UlcNo0NoNoHealed15
  • ALN, axillary lymph node; Fev, fever; Fis, fistulization; Infl, inflammation; Nip Retr, nipple retraction; P Or, peau d’orange; Ulc, ulceration.

Ultrasonography was available for 19 patients. The most common presentation was a non-specific irregular hypoechoic nodule (14/19, 74%), with duct ectasia in five cases. Two other patterns were described: heterogeneous lesion (3/19) and infiltration (2/19).

Pathological examination was obtained either from excisional biopsy (12/23, 52%) or from fine-needle aspiration biopsy (11/23, 48%). The main histological feature was non-caseating granuloma composed of epithelioid histiocytes and multinucleated giant cells (19/23, 83%). The other pattern consisted in a polymorphous inflammatory infiltrate with epithelioid histiocytes (4/23). Microabscesses were associated in seven cases (30%). No refractile or birefringent material was identified, and Ziehl–Nielsen and periodic acid-Schiff stains were consistently negative.

Bacteriological examination was available for 11 patients. More than half were sterile (7/11). Cultures identified Staphylococcus aureus in one case and Corynebacterium spp. in three cases.

Seventeen patients were searched for autoimmunity markers. Among them, four had antinuclear antibodies and one had anti-thyroperoxydase antibodies. Biological autoimmunity was associated with an overt autoimmune disease in four cases (inflammatory myopathy in two, spondyloarthropathy in one and thyroiditis in one).

Disease course and long-term outcome

The median duration of follow-up was 6 years and 3 months (range: 14 months to 28 years). Figure 1, which depicts individual data regarding disease activity and treatments, illustrates the remarkable heterogeneity of the disease. Most of the patients had at least one recurrence (18/23, 78%), two of whom had overlapping attacks. More than one quarter of the women (7/23) had a bilateral evolution. The overall mean number of recurrences was 1.3 ± 1.5 (range: 0–5). Among these women, the mean delay between the first and the last attack was 4 years ± 4 years and 9 months (range: 3 months–18 years and 6 months). At the date of last contact, most patients (21/23, 91%) declared to be healed, 15 free of treatment, with a mean time since last flare-up of 5 years and 7 months (range: 5 months to 10 years and 6 months).

Figure 1

Individual disease course and treatments of 23 patients with IGM.

Clinical impact of different therapeutic strategies

Impact on disease course

Therapeutic intervention was motivated solely by IGM in 21/23 patients. The two other patients receiving steroids and/or methotrexate (MTX) for an inflammatory myopathy were excluded from the analysis regarding both of these treatments.

A short course of antibiotics was used initially to relieve acute inflammation in most cases (18/23, 78%) and appeared to be effective. Abscess formation required surgical drainage in six women (26%). More than a half of the women received steroids (13/21), 10 of these as a first-line therapy. The median initial dosage was 40 mg/day (range: 20–60 mg) and the median duration was 15 months (range: 4–117 months). Steroids were effective in 11 cases (85%), but six patients (46%) relapsed when dose was tapered off, at a median dose of 10 mg/day (range: 2–15 mg). Only two patients did not respond to steroids. Colchicine and hydroxychloroquine were proposed in six and four patients, respectively, either to control a mild active disease, as a steroid sparing agent or as a maintenance treatment to avoid further attacks. Among the six patients receiving colchicine, four did not relapse under this treatment. The only patient treated with hydroxychloroquine as a monotherapy healed. Two women with antibiotics and steroids refractory IGM were treated with oral 15 mg weekly MTX: one did not respond and the other recovered. Interestingly, two other patients had developed IGM despite receiving MTX for an inflammatory myopathy.

Nine patients (39%) had at least one breast excisional surgery (excision or quadrantectomy: seven; total mastectomy: two). It was performed as a first-line therapy in eight cases. Surgical treatment did not prevent recurrences. Indeed, among the nine women who had breast surgery, seven relapsed after surgery (78%) (Figure 2).

Figure 2

Relapse-free survival of patients according to their first-line treatment [excisional surgery (n = 8) vs. conservative treatment (n = 15)] in Kaplan–Meier analysis.

Impact on long-term sequelae

At the date of last interview, 21/23 women had an inactive disease and none had developed a systemic inflammatory disease. However, a high proportion of them suffered unpleasant sequelae (n = 12, 57%). Eight (38%) displayed unsightly scars (Figure 3) and six (29%) felt lasting pain. Unlike the latter, unsightly scars tended to be associated with a greater number of recurrences (1.9 vs. 1.2, P = 0.5). Further, unsightly scars tended to be more frequent among women who had undergone breast surgery [4/8 (50%) vs. 4/13 (31%), P = 0.6], but not among those who had received steroids [5/12 (42%) vs. 3/9 (33%), P = 1]. Neither surgery nor steroids were associated with the occurrence of lasting pain.

Figure 3

Skin scars of IGM (sequelae).

Discussion

IGM is an uncommon and misunderstood inflammatory disease of the breast, which is poorly documented in literature. Herein, we report the largest Caucasian series of IGM, which provides a detailed overview of disease course and outcome over a prolonged follow-up period.

One of the originalities of our study is that all patients were investigated in internal medicine departments. Several arguments may suggest that IGM is an autoimmune reaction, such as its association with autoimmune diseases (Wegener’s granulomatosis, thyroiditis)20 or with extra-mammary manifestations (erythema nodosum, arthritis),21,22 the presence of markers of autoimmunity11 and the efficacy of steroids.23 In a previous limited series of eight patients, six had rheumatoid factor and two had antinuclear antibodies with anti-double-stranded DNA.11 In our study, IGM was associated with clinical and/or biological autoimmunity in five cases (22%). These findings may be seen as a clue to the autoimmune nature of IGM. However, our study arises from internal medicine departments and is therefore subjected to a recruitment bias. On the other hand, 3/11 bacterial cultures performed in our study isolated a Corynebacteria. Certainly, these organisms could be regarded as contaminants. However, this finding is reminiscent of those of two previous reports which pointed to a possible pathogenic role of Corynebacteria in IGM.24,25 Taylor24 reported a series of 34 mastitis from which a Corynebacterium was isolated. Of note, 14 patients also exhibited coryneform Gram-positive bacilli in histological sections. They found that the clinicopathological presentation of these cases was similar to that of IGM.25 More recently, a patient with Corynebacterium-associated IGM was found to have an impaired neutrophil function due to a variant of a pathogen sensor (NOD2) that predisposes to Crohn’s disease.26 Further studies are needed to confirm this association and to assess the efficacy of antibiotics. Of note, antibiotics did not change the course of the disease in our three cases. Importantly, initial evaluation of a granulomatous mastitis must include a complete microbiological workup. Nowadays, it should combine pathology, cultures and molecular biology techniques in order to search for lipophilic Corynebacterium, unexpected bacteria (such as Brucella spp.),27 fungi and Mycobacterium tuberculosis. Importantly, one should remember the low sensitivity of tuberculosis diagnostic tools in this setting.28,29 We assume that, given the length of study period, microbiological workup of our patients was not as thorough as it should be. Still, their prolonged follow-up and response to therapy rule out an infection.

Our results confirm the common clinical presentation of the disease. It usually appeared as a sensitive inflammatory lump in the breast in women in their third or fourth decade.2,5,8,10,16,30 What had never been emphasized previously is the great heterogeneity in the clinical course of IGM (Figure 1), which complicates the therapeutic management and scientific approach to this entity. Indeed, evolution appears to be unpredictable, sometimes spontaneously favourable as already observed,8,15,30 but more often characterized by attacks of varying intensity, alternating with remission periods. Our recurrence rate (78%) was higher than previously reported (17 and 23% with 31 and 15 months of follow-up, respectively).5,16 This result might be due to a recruitment biais and/or to a longer follow-up. A bilateral evolution was frequently noticed, as previously reported (25% in the series of 25 patients reported by Erzogen17). The long-term prognosis of IGM is unclear because follow-up was rather short in previous series. Further, the possibility of late recurrences makes it difficult to assess.5,9 Our global clinical impression is that the benign disease goes away with time. Moreover, none of our patients developed any systemic disease during follow-up. However, we noticed that IGM could significantly impair quality of life, due to numerous factors such as pain,10,30 skin complications (abscess formation, fistulization and ulceration),2,10 overlapping attacks, chronicity,5,6,9,15 apprehension of recurrences,5,9 poor response to treatment and sequelae. Hence, therapeutic management of these patients has three goals. First, controlling active inflammation; second, preventing recurrences and last, avoiding definite sequelae such as lasting pain and/or unsightly scars.

Unfortunately, little if any scientific evidence can guide the therapeutic management of IGM, which remains controversial. Indeed, some authors favour conservative treatment5,15,31 whereas others recommend surgery.7,16,32,33 Therefore, we aimed at evaluating the impact of different therapeutic approaches in our patients. As expected, steroids were often effective (85%) but relapses were frequent when the dose was tapered off (46%). This finding is in accordance with the data of a review of the literature in which Akbulut et al.19 found that 72% of 108 patients receiving steroids for IGM achieved full recovery, and 20% relapsed. The higher rate of relapse in our study may be explained by a longer duration of follow-up. However, Hugon et al.31 have recently reported a retrospective analysis of 14 cases where 78% of patients relapsed after initiation of steroid therapy.31 This difference may be due to a faster steroid tapering. Indeed, a prolonged course of low-dose steroids (<10 mg during 6–12 months) may be helpful to avoid recurrences, as in systemic granulomatous diseases.34,35 Besides, their patients had already exhibited a high number of relapses before steroid therapy and so appear to have a more severe disease than ours. These discrepancies may result from differences in disease-activity assessment or a recruitment bias. Interestingly, our finding that MTX had no obvious efficacy in three patients out of four contradicts the encouraging literature review reported by Akbulut et al.19 Importantly, our conclusions are corroborated by data from Hugon et al.,31 who reported that four patients out of five relapsed under treatment.

Unfortunately, the heterogeneity of disease course and the limited size of our cohort did not allow a clear quantification of the effect of immunomodulating agents such as colchicine and hydroxychloroquine. However, our clinical impression is that these molecules can be effective in some patients and may prevent the recourse to surgery. This is especially true for colchicine, which efficacy has already been noticed.36 Indeed, this drug has an anti-neutrophil activity and is routinely used in several inflammatory disorders (erythema nodosum, recurrent aphtosis, cutaneous vasculitis and neutrophilic diseases),37 whereas hydroxychloroquine is frequently used to treat cutaneous sarcoidosis.38 Pathological similarities between the aforementioned diseases and IGM (granulomatous and/or neutrophilic inflammation) and the favourable benefit/risk profile of these immunomodulating agents as well as their limited cost make them appealing drugs to be evaluated in IGM. Moreover, it may be tempting to try colchicine as an alternative to steroids given its rapid efficacy in some acute inflammatory disorders.

Importantly, we found that first-line surgical treatment did not prevent relapses, as suggested recently by others.7 However, in a review of the literature Wilson et al.32 stated that surgery was more effective than steroids. Indeed, they reported that partial or total mastectomy was successful in 79 and 100% of the cases, respectively, whereas steroids were considered effective in only 42%. In addition, 21% of the patients who had undergone breast surgery relapsed, vs. 58% of the patients on steroids.32 In contrast, we found that in the long term almost 80% of patients suffered a relapse after first-line surgery, and that unsightly scars tended to be more frequent among women who had undergone surgery. Therefore, we believe, like others,5,15,31,39 that a conservative treatment should be favoured. However, plastic surgery is certainly useful to remove unsightly remnant lesions.39

The results of our study should be interpreted with caution because of its retrospective nature and the biases associated with telephone interviews. Further, we may not capture the whole spectrum of IGM, since patients were recruited in internal medicine departments. Despite these limitations, it is the first Caucasian multicentre study that focuses on clinical course and long-term outcome of IGM in this setting. One should keep in mind that the good prognosis of this under recognized entity may be tempered by its impact on quality of life and the occurrence of sequelae. A close follow-up is warranted to detect early relapse and to avoid repeated and sometimes mutilating surgeries. Our findings argue in favour of an interdisciplinary approach to this disorder. While the efficacy of steroids is increasingly recognized, the effect of immunomodulatory drugs (colchicine, hydroxychloroquine) and antibiotics targeting Corynebacteria deserve further investigations.

Acknowledgement

Authors thank the French National Society of Internal Medicine (Société Nationale Française de Médecine Interne).

Conflict of interest: None declared.

Footnotes

  • *These authors contributed equally to this work.

References

View Abstract