Antimicrobial Treatment Options for Granulomatous Mastitis Caused by Corynebacterium Species

Abstract

Corynebacterium species are increasingly recognized as important pathogens in granulomatous mastitis. Currently, there are no published treatment protocols for Corynebacterium breast infections. This study describes antimicrobial treatment options in the context of other management strategies used for granulomatous mastitis. Corynebacterium spp. isolated from breast tissue and aspirate samples stored from 2002 to 2013 were identified and determined to the species level using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), 16S RNA sequencing, and rpoB gene targets. The MICs for 12 antimicrobials were performed using Etest for each isolate. Correlations of these with antimicrobial characteristics, choice of antimicrobial, and disease outcome were evaluated. Corynebacterium spp. from breast tissue and aspirate samples were confirmed in 17 isolates from 16 patients. Based on EUCAST breakpoints, Corynebacterium kroppenstedtii isolates (n = 11) were susceptible to seven antibiotic classes but resistant to β-lactam antibiotics. Corynebacterium tuberculostearicum isolates (n = 4) were multidrug resistant. Two nonlipophilic species were isolated, Corynebacterium glucuronolyticum and Corynebacterium freneyi, both of which have various susceptibilities to antimicrobial agents. Short-course antimicrobial therapy was common (median, 6 courses per subject; range, 1 to 9 courses). Patients with C. kroppenstedtii presented with a hot painful breast mass and underwent multiple surgical procedures (median, 4 procedures; range, 2 to 6 procedures). The management of Corynebacterium breast infections requires a multidisciplinary approach and includes culture and appropriate sensitivity testing to guide antimicrobial therapy. Established infections have a poor outcome, possibly because adequate concentrations of some drugs will be difficult to achieve in lipophilic granulomata. Lipophilic antimicrobial therapy may offer a therapeutic advantage. The role of immunotherapy has not been defined.

INTRODUCTION

Granulomatous mastitis (GM) is a rare inflammatory condition that typically occurs in parous women of reproductive age (1). Clinically, it can present as a breast mass with features similar to those of breast malignancy (2). Multiple infective and inflammatory conditions, including tuberculosis (TB), fungal infections, sarcoidosis, amyloidosis, and Wegener's granulomatosis, have been recognized, but recently, attention has been drawn to corynebacterial species as a specific pathogen in this disease. Typically, corynebacterial breast infections are characterized by abscess formation, granulomatous inflammation, and progression to sinus/fistula formation (3). This and other specific diagnoses should be excluded before a diagnosis of idiopathic granulomatous mastitis (IGM) is made (3–5).

Commonly, it is the lipophilic species of the Corynebacterium genus that cause mastitis (3, 6–8). As such, a microbiological diagnosis can be challenging, due to their fastidious growth requirements and prolonged incubation time (3, 7). It is likely that cases due to corynebacteria are underrecognized or its relevance underappreciated when isolated from breast samples (5). Furthermore, most antimicrobials are hydrophilic, with poor distribution to lipid environments, and this has implications for efficacy of treatment (see Table 1).

TABLE 1.

Antimicrobial lipophilicity and antimicrobial susceptibilities for C. kroppenstedtii, C. tuberculostearicum, C. glucuronolyticum, and C. freneyi

Antimicrobial agent	LogPa	EUCAST breakpointb	C. kroppenstedtii MICs (mg/liter) (n = 11)			C. tuberculostearicum MIC (mg/liter) for isolate (n = 4):				C. glucuronolyticum (n = 1)	C. freneyi (n = 1)
			MIC50	MIC90	Range (% resistant)	1	2	3	4
Penicillin	0.76	S, ≤ 0.12; R, >0.12	0.25	0.75	0.094 to 0.75 (91)	>32	0.125	>32	0.125	0.094	0.75
Amoxicillin-clavulanate	−2.3/−1.5	S, ≤2; R, >8c	0.19	0.38	0.064 to 0.5 (0)	>32	0.064	>32	0.094	0.047	0.38
Clarithromycin	3.24	NAd	0.023	0.023	<0.016 to >32	>32	>32	0.032	0.094	0.38	4.0
Vancomycin	−4.4	S, ≤2; R, >2	0.75	0.75	0.38 to 1.0 (0)	0.75	0.5	0.38	0.75	0.38	0.38
Tetracycline	−3.5	S, ≤2; R, >2	0.38	0.5	0.25 to 0.5 (0)	0.75	>32	>32	0.75	>32	0.75
Ceftriaxone	−1.8	S, ≤1; R, >2c	0.75	4	0.125 to >32 (27)	>32	0.75	>32	0.75	1.0	1.0
Ciprofloxacin	−0.81	S, ≤1; R, >1	0.19	>32	0.094 to >32 (18)	>32	0.125	>32	0.094	0.25	0.094
Clindamycin	1.04	S, ≤0.5; R, >0.5	0.125	0.19	0.064 to >32 (9)	>32	>32	0.5	0.75	0.125	0.047
Rifampin	2.77	S, ≤0.06; R, >0.5	<0.006	0.006	<0.006 to 0.008 (0)	0.125	0.047	0.032	0.064	<0.002	0.004
Trimethoprim-sulfamethoxazole	1.28/0.79	S, ≤1; R, >2	0.094	0.094	0.032 to 0.094 (0)	>32	0.75	0.5	0.5	0.25	0.19
Linezolid	0.64	S, ≤2; R, >2	0.38	0.5	0.25 to 0.5 (0)	0.5	0.5	0.75	1.5	0.19	0.38
Moxifloxacin	−0.5	S, ≤0.5; R, >0.5	0.064	2.0	0.047 to 3.0 (18)	1	0.094	>32	0.064	0.094	0.023

^aFor LogP values (26), the partition coefficient is the ratio of the concentration of the compound in octanol to its concentration in water and is a measure of lipophilicity. Antimicrobials with high LogP coefficients (>1) are preferentially distributed to lipophilic compartments.

^bS, susceptible; R, resistant.

^cNon-species related EUCAST breakpoints.

^dNA, not available.

Current knowledge about antimicrobial options to treat Corynebacterium mastitis is lacking. Antimicrobial choices for these infections have largely been driven by nonstandardized disk testing, indicating broad susceptibility to a range of antimicrobial classes. Many patients undergo multiple surgical procedures and repeated short courses of antimicrobials before symptoms abate months to years later, suggesting that this approach has little effect on the natural history of this disease (3, 9–11).

The aims of this study were first to describe the Corynebacterium species isolated from cases of granulomatous mastitis, and second to describe antimicrobial treatment options for Corynebacterium breast infections with reference to their likely concentration at the site of infection. This is placed in the context of other suggested management strategies for granulomatous mastitis, such as surgery and immunomodulator therapy.

MATERIALS AND METHODS

Case definition.

Cases were included if a Corynebacterium species was isolated from a specimen obtained by a sterile technique from a breast mass. Cases were excluded if there was any other cause found for the condition on histology or culture.

A database search of samples processed at Canterbury Health Laboratories (Christchurch, New Zealand) found a record of 27 isolates identified as Corynebacterium spp. from breast specimens from 2002 to 2013. Twenty isolates (two from the same patient) were from a sterile site (surgical tissue specimen or percutaneous aspirate) and able to be grown in culture.

Microbiological investigations.

Each isolate was cultured on 5% sheep blood agar (Fort Richard Laboratories, Auckland, New Zealand) and incubated at 5% CO₂ and 37°C. Colonies were identified by phenotypic characteristics and matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) using the Bruker Biotyper (Bremen, Germany). Acceptable identification by MALDI-TOF MS was made on scores of ≥2.0; otherwise, samples were reanalyzed after ethanol-formic acid extraction (12). The identification of all 20 isolates was confirmed by 16S RNA sequencing (13). Isolates established to be Corynebacterium spp. had rpoB gene PCR performed (14). The DNA sequences were compared to reference strains deposited in the GenBank database using the BLAST algorithm (15).

Susceptibility testing.

The MICs for 12 antimicrobials (listed in Table 1) were performed by Etest (bioMérieux, Marcy l'Etoile, France). The antimicrobial selection was based on published susceptibility data, empirical use in breast infections, and suitability for prolonged courses or in combination therapy (7, 16, 17).

Mueller-Hinton agar (Fort Richard Laboratories, Auckland, New Zealand) was used for the Etests and incubated at 37°C and 5% CO₂. The plates were inoculated with a bacterial suspension in brain heart infusion (BHI) broth (Fort Richard Laboratories) of turbidity equivalent to 1 McFarland standard, as per the Etest manufacturer recommendations for Corynebacterium. European Committee on Antimicrobial Susceptibility Testing (EUCAST) Corynebacterium breakpoints were used for the interpretation of MICs (www.eucast.org). Nonspecies-related breakpoints were used for amoxicillin-clavulanate and ceftriaxone.

Clinical correlation.

The clinical notes of patients with confirmed Corynebacterium breast infection were reviewed to confirm the presence of a breast mass, a granulomatous reaction on histology, and to exclude any with alternative pathology. Basic demographic information, antimicrobial therapy, duration of treatment, and outcome were recorded.

RESULTS

Identification of isolates.

From the initial 20 isolates, three were reclassified to the genus Actinomyces, leaving 17 Corynebacterium isolates from 16 patients. Prior to the availability of MALDI-TOF MS and molecular techniques in our laboratory, identification combined phenotypic characteristics and biochemical profile (API Coryne system; bioMérieux, France).

Ten patients had Corynebacterium kroppenstedtii isolated from breast specimens (one patient had C. kroppenstedtii isolated twice), four had Corynebacterium tuberculostearicum isolated, one had Corynebacterium glucuronolyticum, and one had Corynebacterium freneyi. Morphologically, C. kroppenstedtii and C. tuberculostearicum had round, nonpigmented, and nonhemolytic colonies (<1 mm). C. glucuronolyticum colonies were large, white, and mucoid, whereas C. freneyi had large, yellow, and wrinkled colonies.

Of the 17 isolates, six reliably scored ≥2.0 on MALDI-TOF MS, and a further five scored ≥1.8. The remainder had values below the accepted score for reliable identification (<1.7). MALDI-TOF MS was unable to differentiate C. freneyi from Corynebacterium xerosis.

16S rRNA gene sequencing was able to confirm species identification in 16 isolates with ≥98% similarity to the type strain of the species. The exception was C. freneyi, which needed rpoB gene sequencing to be differentiated from C. xerosis, with 98% gene similarity to the type strain.

Antimicrobial susceptibility.

Etest MICs for the lipophilic C. kroppenstedtii and C. tuberculostearicum were read at 24 and 48 h, due to their slow growth. C. glucuronolyticum and C. freneyi Etest MICs were read at 24 h.

The susceptibilities of C. kroppenstedtii, C. tuberculostearicum, C. glucuronolyticum, and C. freneyi isolates are shown in Table 1. Based on the EUCAST breakpoints for this genus, isolates of C. kroppenstedtii were susceptible to rifampin, tetracycline, trimethoprim-sulfamethoxazole, linezolid, and vancomycin and resistant to the β-lactams (www.eucast.org). One isolate was multidrug resistant. Isolates of C. tuberculostearicum were multidrug resistant, consistently being sensitive to linezolid and vancomycin only.

Clinical correlation.

Pure growth of a Corynebacterium species was isolated from the initial breast specimen in 13 subjects. Of these, 12 presented with a new breast lesion (10 C. kroppenstedtii, 1 C. tuberculostearicum, and 1 C. freneyi), and one grew C. tuberculostearicum from a collection 8 months after surgery. The one male patient in this series grew C. glucuronolyticum and an Actinomyces sp. from an acutely infected breast cyst. In two patients who had previous surgical intervention, other bacteria were identified from the same sample (C. tuberculostearicum and Staphylococcus lugdunensis in one patient and C. tuberculostearicum, Finegoldia magna, and Staphylococcus aureus in the other). Gram-positive bacilli were seen on microscopy of a direct Gram stain with 8/17 specimens.

In most cases, growth of corynebacteria was observed early in the clinical presentation, but repeated isolation was uncommon in successive specimens. Notably, only C. kroppenstedtii was isolated from specimens taken at a later date in two of the 10 patients. Mycobacterium culture had been requested in six of the 12 patients who presented with a new breast mass; these were all negative. No patient had fungal testing requested. Ten of the 16 patients, including the one male patient, were of New Zealand (NZ) European or European ethnicity. Other ethnicities included Pacific Islander (n = 2), Maori (n = 2), African, and East Asian. The median age was 42 years (range, 21 to 71 years).

All 16 patients were referred to general surgery for management, and 14 underwent fine-needle or open biopsy to exclude malignancy. Three patients had been seen by the infectious diseases team for advice on antimicrobial therapy once a microbe had been isolated. Empirical antimicrobial choice, treatment, and outcome are shown in Table 2. Repeated short courses (5 to 7 days) of antimicrobial therapy were common (median, 6 courses per subject; range, 1 to 9). No patient had been treated with immunomodulator or antituberculous therapy.

TABLE 2.

Clinical presentation, antibiotic selection, and outcome

No.	Age (yr)/sexa	Ethnicity	Presentation	Histology	Empirical antimicrobialb	Culture	Treatment	Susceptibility by Etest	Outcome
1	47/F	NZ European	Right breast abscess for 4 mo	Granulomatous mastitis	Ciprofloxacin,c clindamycin,c metronidazole	C. kroppenstedtii; isolated twice, 17 days apart	Doxycycline,c ciprofloxacinc	R to penicillind	Right breast mass for 24 mo, left breast IGM 5 yr later
2	35/F	Pacific Islander	Right breast mass with pain for 2 wk	Acute and chronic inflammatory cells	Nil	C. kroppenstedtii	Doxycyclinec	R to penicillin	Recurrence with wound discharge for 25 mo
3	21/F	Maori	Recurrent bilateral breast lumps with 24-h history of acute right breast abscess	Inflammatory cell infiltrate with granulomata	Flucloxacillin, roxithromycin, ciprofloxacinc	C. kroppenstedtii; isolated twice, 3 wk apart	Flucloxacillin, ciprofloxacin,c amoxicillin- clavulanate,c cefaclor, doxycyclinec	2 isolates with same susceptibility profile; R to penicillin	Recurrent bilateral breast abscesses treated for 4 mo then loss to follow-up
4	47/F	NZ European	Left breast mass for 1 mo	Multifocal granulomatous inflammation and microabscess formation (3 mo later)	Nil	C. kroppenstedtii	Amoxicillin	R to penicillin	Diagnosed as IGM; left breast mass resolved 6 mo later; recurrent multifocal IGM 3 yr later
5	42/F	British European	Right breast abscess for 1 wk	Acute inflammatory infiltrate	Nil	C. kroppenstedtii	Clindamycinc	R to penicillin, ceftriaxone	Developed fistula, resolved 10 mo later
6	57/F	NZ European	Left acute breast abscess	Consistent with abscess	Flucloxacillin	C. kroppenstedtii	Flucloxacillin, ciprofloxacin, amoxicillin-clavulanate,c metronidazole	R to penicillin, ciprofloxacin, moxifloxacin	Repeated left breast drainage, left breast central duct excision 23 mo later
7	39/F	NZ European	NAe	NA	NA	C. kroppenstedtii	NA	R to penicillin	NA
8	38/F	Pacific Islander	Large right breast abscess for 2 mo	Diffuse acute inflammation	Flucloxacillin,c amoxicillin- clavulanatec	C. kroppenstedtii	Flucloxacillin,c amoxicillin- clavulanate,c doxycyclinec	No resistance	Large draining cavity packed for 1 mo; resolved after 4 mo
9	34/F	African	Large left abscess for 1 wk with fever	Ill-defined granulomata with giant cells	Amoxicillin- clavulanate, flucloxacillin	C. kroppenstedtii	Doxycycline,c ciprofloxacin,c amoxicillin	R to penicillin	Fistulotomy, partial mastectomy, resolution after 6 mo
10	41/F	Southeast Asian	Left breast abscess for 8 days	Acute inflammation	Flucloxacillin, cefazolin, gentamicin, clindamycin	C. kroppenstedtii	Doxycyclinec	R to penicillin, clarithromycin, ceftriaxone, ciprofloxacin, clindamycin, moxifloxacin	Resolved after 3 mo
11	42/F	Pacific Islander	Large right breast abscess for 2 mo	Granulomata surrounding lipid spaces with central microabscess formation	Flucloxacillin, Amoxicillin- clavulanatec	C. tuberculostearicum	Flucloxacillin, amoxicillin-clavulanatec	R to penicillin, clindamycin	Chronic multifocal right breast mastitis for 17 mo
12	51/F	NZ European	Right breast abscess for 1 wk	No evidence of malignancy (previous left breast cancer)	Flucloxacillin	C. freneyi	Flucloxacillin, doxycyclinec	R to penicillin	Separate abscess right breast 1 mo later (same species isolated).
13	51/F	NZ European	Postoperative wound infection	Not performed	Amoxicillin- clavulanate, flucloxacillin, cefazolin	C. tuberculostearicum, F. magna, S. aureus	Penicillin, metronidazole, cefazolin	Multiresistant strain	4 mo of wound care
14	71/F	NZ European	Seroma postsurgery	Not performed	Nil	C. tuberculostearicum, S. lugdunensis	Flucloxacillin	Multiresistant strain	Resolved with drainage
15	53/F	NZ European	Collection 8 mo postsurgery	No evidence of recurrence (malignancy)	Flucloxacillin, ciprofloxacinc	C. tuberculostearicum	Amoxicillin-clavulanatec	R to penicillin, clarithromycin, tetracycline, clindamycin	Collection with chronic nipple discharge resolved after 12 mo
16	43/M	NZ European	Right chest wall abscess for 10 wk	Consistent with abscess	Flucloxacillin	C. glucuronolyticum, Actinomyces neuii	Ciprofloxacinc	R to tetracycline	Recurrence 2 mo later with right breast abscess and chronic sinus

^aF, female; M, male.

^bEmpirical antimicrobial therapy defined as antimicrobials given prior to a positive Gram stain or microorganism isolated from specimen.

^cStrain susceptible to antimicrobial based on in vitro Etest. Strain considered sensitive to doxycycline if susceptible to tetracycline. R = resistant.

^dR, resistant (EUCAST breakpoint).

^eNA, no clinical information available for case 7.

Clinical information was available for nine of the 10 patients from whom C. kroppenstedtii had been isolated. All presented with a single peripheral breast mass of up to 4 months duration that was hot, painful, and swollen. Three subjects went on to develop bilateral breast disease with multiple breast abscesses on both sides. Two patients had an aspirate cultured from a secondary breast mass, and no microbe was isolated. Granulomatous inflammation was reported on histology in four cases, but no bacteria were seen on the pathology samples. The median number of surgical interventions per patient was 4 (range, 2 to 6 interventions). These included incision and drainage, fistulotomy, and partial mastectomy.

DISCUSSION

The role of Corynebacterium in invasive infections is often debated, as most of these species are part of the endogenous skin flora. However, the link between Corynebacterium and granulomatous mastitis has periodically been reported (6, 9, 10, 18) since the publication of a case series by Taylor et al. (3) in 2003. Growth in pure cultures, isolation from sterile sites, and distinct histological findings from breast tissue samples support the causative link (3, 19). In addition, Corynebacterium is a well-known cause of granulomatous mastitis in animals (20). In our series, all 12 patients who presented with a new breast mass had Corynebacterium sp. as a single isolate from a tissue or aspirate sample, further supporting the pathogenic role of these species in breast disease.

Corynebacterium identification and genus characterization have become more reliable with the availability of MALDI-TOF MS and molecular techniques (7, 21, 22). To determine species-level identification of all isolates in this study, 16S rRNA and rpoB gene sequencing were used. MALDI-TOF MS reliably identified 6/17 isolates, with reliability likely to improve as more strains are added to the Biotyper database.

As found in this study, strains of C. tuberculostearicum are often multiresistant, demonstrating the macrolide-lincosamide-streptogramin B (MLSB) mechanism conferred by the ermX gene (23, 24). While data are limited, antimicrobial susceptibility by nonstandardized disk diffusion has been performed on isolates of C. kroppenstedtii, reporting broad susceptibility to most classes of antimicrobials (9, 11). Isolates in this study, however, were largely resistant to penicillin by Etest, although this might represent a selection bias with the penicillin-sensitive strains treated. The variable Corynebacterium strain resistance suggests that correct species identification and antimicrobial susceptibility testing would ideally be performed for all isolates.

The presence of granulomatous inflammation should lead to a consideration of further therapy. Corynebacteria survive in lipid-filled vacuoles surrounded by a reactive neutrophilic granulomatous infiltrate rather than in the inflamed tissue (3, 19). This inflammatory reaction can be rapid; in our series, the development of granulomata was seen as early as a week from the start of clinical symptoms. In this environment, adequate tissue concentrations for bactericidal activity may be achieved by agents that are highly lipophilic and have a high volume of distribution, which include rifampin, clarithromycin, trimethoprim-sulfamethoxazole, and clindamycin (24, 25). In contrast, β-lactams and fluoroquinolones have low lipid solubility and therefore are expected to be less effective. However, fluoroquinolones are used in other active granulomatous infections, such as tuberculosis.

Short-course antimicrobial treatment did not appear to help clinical outcomes, particularly once granulomatous inflammation was present. However, there were insufficient data to demonstrate that a long duration of treatment was beneficial, as in TB. Furthermore, as the selection of antimicrobial therapy was varied, outcome cannot be inferred from the MIC results. For selection of an antimicrobial agent, the results of sensitivity testing, MICs, and the pharmacological properties of an agent should be considered and evaluated in prospective studies. Immunomodulator therapy may be a useful therapeutic adjunct by modifying the initial immune response and granuloma formation (22). Corticosteroid therapy in IGM is increasingly used (26–28). Further studies looking into the impact of immunomodulator treatment in cases of granulomatous breast infections are needed, with particular caution if combined with an antimicrobial regime that includes rifampin (29).

Surgical intervention, particularly for sampling of breast tissue and management of abscesses and fistulae, is necessary. Success when combining these procedures with antimicrobial therapy has been reported (9, 11). Ideally, prompt diagnosis will avoid consequent disfiguring complications.

In summary, the primary goal in any patient who presents with a nonpuerperal breast mass is accurate diagnosis, including the exclusion of malignancy. This study suggests that a failure of empirical treatment of mastitis with β-lactam antimicrobials should trigger further microbiological investigations. Given the poor outcomes seen with Corynebacterium mastitis treated with β-lactam therapy, we suggest other antimicrobials be used as empirical therapy for atypical breast infections, for example, doxycycline or trimethoprim-sulfamethoxazole. If granulomatous disease is present, it seems prudent to choose agents that are both active against Corynebacterium spp. and have physicochemical properties that promote activity within the lipid-filled spaces. Preferred choices include clarithromycin and rifampin, which are also active in other granulomatous infections, such as mycobacteria. Further clinical studies of antibiotic choice and duration of therapy are necessary. Oral corticosteroids or other immunosuppressive agents might be of benefit to slow the host immune response and therefore curtail the development of granulomatous disease once a microbiological diagnosis has been established; again, further research is required.