Article Text

Download PDFPDF
Novel selective medium for the isolation of corynebacterium kroppenstedtii from heavily colonised clinical specimens
  1. Sally Cheuk Ying Wong1,2,
  2. Rosana W S Poon1,2,
  3. Chuen-Hing Foo1,
  4. Antonio H Y Ngan1,2,
  5. Herman Tse1,2,
  6. Vivian C M Lam1,2,
  7. Tiffany H Y Leung1,2,
  8. Chun-Pong Wong1,2,
  9. Vincent C C Cheng1,2,
  10. Jonathan H K Chen1,2,
  11. Kwok-Yung Yuen1,2
  1. 1 Department of Microbiology, Queen Mary Hospital, Hong Kong, China
  2. 2 Department of Microbiology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong, China
  1. Correspondence to Professor Kwok-Yung Yuen, Carol Yu Centre for Infection, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China; kyyuen{at}


Aims Granulomatous mastitis due to Corynebacterium kroppenstedtii is an increasingly recognised cause of an indolent and distressing mastitis in non-lactating females. This slow-growing lipophilic organism is not reliably isolated using routine culture methods. A novel selective culture medium (CKSM) is designed to optimise the isolation of this organism from clinical specimens.

Methods CKSM contains 10% galactose and Tween 80 (10%) to enhance the growth of C. kroppenstedtii, fosfomycin (100 µg/mL) to suppress the other bacteria, and differentiate C. kroppenstedtii from non-kroppenstedtii lipophilic corynebacteria by esculin hydrolysis. The medium was evaluated for its ability to support the growth of C. kroppenstedtii, selection and differentiation of C. kroppenstedtii from other bacteria in non-sterile clinical specimens.

Results C. kroppenstedtii grew as 1–2 mm colonies with black halo on CKSM within 72 hours of incubation, compared with barely visible pinpoint colonies on routine blood agars. During the four-month period of evaluation with 8896 respiratory specimens, 103 breast specimens, 1903 female genital tract specimens, 617 newborn surface swabs and 10 011 miscellaneous specimens, 186 C. kroppenstedtii were isolated, including 127 (1.4%) respiratory and 59 (0.5%) miscellaneous specimens, 184 of them were found only on CKSM. Besides the three (2.9%) positive breast specimens, 27 (1.4%) high vaginal and endocervical swabs, and 11 (1.8%) surface swabs of newborns were positive for C. kroppenstedtii.

Conclusions CKSM is a useful addition to routine agar media for the isolation of C. kroppenstedtii, and will be helpful for studying the epidemiology and transmission of this unusual Corynebacterium causing granulomatous mastitis.

  • corynebacteria
  • diagnostics
  • breast
  • infections

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Granulomatous mastitis (GM) is a relapsing, distressing and disfiguring condition affecting non-lactating women. In recent years, reports have illustrated the role of Corynebacterium kroppenstedtii, which are lipophilic Gram-positive bacilli, in a subgroup of patients with GM through Gram stain, culture and molecular methods.1–4 Our recent report has also found C. kroppenstedtii in a group of non-lactating females at reproductive age, with many suffering hyperprolactinemia associated with anti-psychotic medications or pituitary microadenoma.3 We suspect that there has been substantial under-reporting of C. kroppenstedtii from breast specimens due to the slow-growing nature of this lipophilic organism.5 The usual procedure using blood agar and chocolate agar, with incubation at 35°C in 5% CO2 for up to 48 hours, is inadequate for consistent isolation of this organism. Here, a novel selective and differentiating medium (CKSM) is designed for the isolation of C. kroppenstedtii from non-sterile specimens taken from heavily colonised anatomical sites. The medium was evaluated with C. kroppenstedtii, non-kroppenstedtii corynebacteria and other bacteria commonly isolated from clinical specimens, and its performance with clinical specimens received in a microbiology laboratory.


Agar component

One litre of the C. kroppenstedtii selective medium, CKSM, contains 25 g brain-heart-infusion base (CM1135, oxoid), 15 g agar no.1 (Difco agar, BD), 10 g glucose (VWR chemicals, US pharmacopoeia), 10 g galactose (G0750, Sigma-Aldrich), 10 mL 10% Tween 80 (P1754, Sigma-Aldrich), 1 g esculin (E8250, Sigma-Aldrich), 0.5 g ferric (III) citrate (Merck, Germany) and 100 mg fosfomycin (Infectopharm, Germany). A pilot study was performed to determine the optimal concentration of glucose and galactose that best support the growth of C. kroppenstedtii; Tween 80 was included to facilitate lipophilic growth; esculin and ferric citrate were added for the detection of esculin hydrolysis by C. kroppenstedtii; 5 and fosfomycin was used for the suppression of potential coexisting flora, at a concentration of 100 µg/mL chosen according to previous reported breakpoints of other bacteria.6 7 All ingredients, except fosfomycin, were dissolved in deionised water and autoclaved at 121°C for 15 min. When the mixture was cooled to 50°C, fosfomycin was added aseptically. The resulted agar has a pH of 6.4, with a tinge of green due to the presence of esculin and ferric citrate. On CKSM, C. kroppenstedtii typically appears as 1–2 mm off-white colonies, with a surrounding blackened halo due to esculin hydrolysis (figure 1A). Although sniffing of plates is discouraged in the microbiology laboratory, the growth of C. kroppenstedtii on CKSM was associated with a pungent vinegar odour after 48–72 hours of incubation at 35°C with 5% CO2, apparent as soon as culture plates were opened for examination.

Figure 1

Growth of lipophilic corynebacteria on Corynebacterium kroppenstedtii selective agar (CKSM, left) and blood agar (BA, right) after 72 hours of incubation (24 hours in 35°C with 5% CO2 and 48 hours at 35°C in air). Pure bacterial culture from 20 µL of 105 CFU/mL bacterial suspension in (A–F). (A) clinical isolate of Corynebacterium kroppenstedtii, with blackening of agar on CKSM; (B) clinical isolate of C. tuberculostearicum; (C) clinical isolate of C. macginleyi; (D) clinical isolate of C. jeikeium; (E) clinical isolate of C. resistens; (F) clinical isolate of C. bovis; and (G) non-sterile specimen with C. kroppenstedtii identified on CKSM. White arrow lines: C. kroppenstedtii; black arrowheads: non-C. kroppenstedtii bacteria.

C. kroppenstedtii and incubation conditions using CKSM

Five clinical strains of C. kroppenstedtii from our recent study,3 identity confirmed using 16 s rRNA gene sequencing (using primers 1F 5′-AGTTTGATCMTGGCTCAG-3′ and 2R 5′-GGACTACHAGGGTATCTAAT-3′) and nanI gene as previously described,3 8 were inoculated onto CKSM from 0.5 McFarland suspension using 10 µL standard loops. As some laboratories may have limited CO2 incubation facilities, we tested the isolation of C. kroppenstedtii on CKSM under three different conditions: three days at 35°C with 5% CO2 (condition 1), two days at 35°C with 5% CO2 followed by 1 day at 35°C in air (condition 2) and one day at 35°C with 5% CO2followed by two days at 35°C in air (condition 3). For comparison, the C. kroppenstedtii suspensions were inoculated onto blood agar (BA, CM0331 Columbia agar base, Oxoid, with sterile horse blood) and chocolate agar (CHO, CM0331 Columbia agar base, Oxoid, with sterile horse blood) for 72 hours. All agars were observed daily for bacterial growth.

C. kroppenstedtii and non-kroppenstedtii lipophilic corynebacteria on CKSM in pure culture

Six strains of C. kroppenstedtii (one C. kroppenstedtii DSM 44385), five C. tuberculostearicum, two C. macginleyi, four C. urealyticum (two strains from external quality assessment programme, EQAP), three C. jeikeium (two strains from EQAP), three C. resistens, two C. bovis and three C. accolens were used to evaluate the ability of CKSM to select and differentiate C. kroppenstedtii from non-kroppenstedtii lipophilic corynebacteria (online supplementary appendix table 1). All C. kroppenstedtii strains were identified using both MALDI-TOF MS and 16 s rRNA gene sequencing as previously described,8 while clinical isolates of non-kroppenstedtii Corynebacterium were identified using matrix-assisted laser desorption/ionisation-time of flight mass spectrometry (MALDI-TOF MS) only, with a score >2.0. Each strain was made into suspension at four different concentrations (105, 106, 107 and 108 CFU/mL), 20 µL of each suspension was inoculated in triplicates onto CKSM, BA and CHO. All inoculated CKSM were incubated in condition 3 as it was found to provide non-inferior growth of C. kroppenstedtii with minimum CO2 incubation requirement (see Results), and there are limited CO2 incubators in our laboratory. Inoculated BA and CHO were incubated in condition 1 to simulate the usual bacterial culture condition for respiratory and miscellaneous specimens in our laboratory.

Supplemental material

Non-lipophilic corynebacteria and clinically encountered bacteria on CKSM in pure culture

Twenty-three ATCC bacterial isolates, one NCTC strain, 17 EQAP strains and four clinical strains of bacteria were tested on CKSM (table 1). These isolates were inoculated onto CKSM using a standard 10 µL loop from a 0.5 McFarland bacterial suspension. The plates were examined after 72 hours of incubation under condition 3.

Table 1

Growth of commonly encountered bacteria on CKSM after three days of incubation (35°C at 5% CO2 for 24 hours, followed by 48 hours at 35°C in air)

Validation of selective agar with non-sterile clinical specimens

Respiratory specimens and miscellaneous specimens received in our laboratory from 26 October 2015 to 29 February 2016 were inoculated onto CKSM, in additionto routine media according to our laboratory standard operating procedures (SOPs) (online supplementary appendix table 2). Respiratory specimens included nasal, oral and throat swabs, sputum, tracheal aspirates, bronchoalveolar lavage and bronchial aspirates. Miscellaneous specimens included swabs, fluid and tissues obtained from any sites of the body, excluding blood, urine, faeces or rectal swabs, or specimens from the central nervous system. CKSM agars were incubated in condition 3. All plates were examined daily for bacterial growth. Potential C. kroppenstedtii colonies, i.e. 1–2 mm off-white colonies with or without a blackened halo, were subjected to identification using MALDI-TOF MS (Bruker Daltonics, Germany) MALDI Biotyper version 3.1 with reference library version DB_5989 and in-house enhanced database with an extra 46 strains of C. kroppenstedtii identified in our recent study.3 Enhancement of database was done as previously described,9 and was used in this study as MALDI-TOF MS score for C. kroppenstedtii is often low (<1.8) when only the Biotyper database was used (data not shown). Formic-ethanol extraction was used if the initial MALDI-TOF MS score was <1.7.10 Isolates identified to be C. kroppenstedtii with no blackened halo on CKSM were inoculated on esculin agar slant (47 g brain-heart-infusion agar (Oxoid CM1136), 0.5 g ferric (III) citrate (Merck, Germany), 1 g esculin hydrate (E8250, Sigma-Aldrich), per litre of deionised water) for incubation in 5% CO2 at 35°C for 48 hours to test for esculin hydrolysis.

Statistical analysis

χtest, Student’s t-test and McNemer’s test with continuity correction were used for statistical analysis where appropriate, using SPSS statistics version 20 in the former two and R 3.4.1 for the latter. A P value of less than 0.5 is considered statistically significant.


C. kroppenstedtii and incubation conditions using CKSM

CKSM consistently yielded larger C. kroppenstedtii colonies compared with BA (figure 1A). Significant difference of colony morphology between the three media was noted from day 2 of incubation, and the poorest growth was observed on CHO. Off-white 1–2 mm colonies with distinct vinegar odour were observed on CKSM in all three conditions regardless of duration of CO2 enrichment (figure 2). Incidentally, one of the five C. kroppenstedtii strains did not demonstrate the blackened halo on CKSM, despite that esculin hydrolysis was demonstrated on esculin agar slant by that strain. All four other strains demonstrated blackened halo on CKSM within 72 hours of incubation in all three conditions. Nonetheless, larger colonial size and blackening were observed earlier (on day 1 or 2) on CKSM when incubated in an CO2-enriched environment for ≥48 hours. In comparison, C. kroppenstedtii were seen as pinpoint colonies, odourless, on BA and CHO after three days (figure 1A).

Figure 2

Colony morphology of C. kroppenstedtii after incubation in different atmospheric conditions at 35°C. Condition 1: 72 hours in CO2-enriched environment; condition 2: 48 hours in CO2-enriched environment and 24 hours in air; Condition 3: 24 hours in CO2-enriched environment and 48 hours in air.

C. kroppenstedtii and non-kroppenstedtii lipophilic corynebacteria on CKSM in pure culture

CKSM supported growth of C. kroppenstedtii from suspension at concentrations from 105 to 108 CFU/mL after three days with an average of 166 CFU from 105 CFU/mL suspensions. Visible colonies were observed from day 2 at all concentrations. Distinct vinegar odour was detected in all, while esculin hydrolysis was observed in five of the six strains at all concentration. In comparison, significantly smaller and fewer colonies were seen on BA and CHO after three days, where hazy pinpoint colonies were seen on both, with a mean of 147 and 143 CFU from 105 CFU/mL suspensions (P value=0.04 and 0.01 compared with CKSM), respectively. MALDI-TOF MS spectra of C. kroppenstedtii obtained from colonies on CKSM, BA and CHO showed no significant difference, and a score of >1.7 was obtained (data not shown).

For non-kroppenstedtii lipophilic corynebacteria, 17 out of 18 strains grew on CKSM. C. macginleyi strains grew poorly on CKSM, one strain failed to grow altogether and the other with scanty colonies after three days (figure 1C). In contrast, all other non-kroppenstedtii lipophilic corynebacteria demonstrated good growth on CKSM after three days, with larger colonies noted on CKSM when compared with BA and CHO (figure 1B–F). As expected, none of the non-kroppenstedtii Corynebacterium species tested demonstrated blackened halo.

Non-lipophilic corynebacteria and clinically encountered bacteria on CKSM in pure culture

The results of non-lipophilic Corynebacterium species and other bacteria tested are included in table 1. For non-kroppenstedtii non-lipophilic corynebacteria, C. sundsvallence and C. renale failed to grow on CKSM (growth detected on BA and CHO), while all others grew on both CKSM and BA. Better growth was noted on BA compared with CKSM for C. ulcerans and C. striatum. Other non-corynebacterial bacteria that grew on CKSM demonstrated colony morphology entirely different from C. kroppenstedtii and can be easily differentiated from C. kroppenstedtii (figure 1G). Some Lactobacillus species were described to have esculin hydrolysing activities. Of the three Lactobacillus species included, only L. acidophilus demonstrated darkening of agar on CKSM, with more translucent colonies compared with C. kroppenstedtii (online supplementary appendix figure 1).

Supplemental material

Selective agar and non-sterile clinical specimens

Over the four-month period, 8896 respiratory and 12 633 miscellaneous specimens were screened. C. kroppenstedtii were identified from 186 specimens on CKSM, including 127 (1.4%) respiratory and 59 (0.5%) miscellaneous specimens (table 2), from 114 and 51 unique patients respectively. Compared with the SOPs, an additional 184 C. kroppenstedtii, from 127 respiratory and 57 and miscellaneous specimens, were identified with the use of CKSM. The use of CKSM increased the isolation of C. kroppenstedtii from 0% and 0.015% to 1.4% and 0.5% (P value <0.001 in both) in respiratory and miscellaneous specimens compared with the SOPs. The most common co-existing bacteria included E. coli, Staphylococcus aureus, coagulase-negative Staphylococcus species, Streptococcus species, Pseudomonas aeruginosa, Acinetobacter baumannii and Candida species. Seven specimens (two breast-related, two newborn ear swabs and three wound specimens) yielded no bacterial growth on routine culture media, but growth of C. kroppenstedtii was detected on CKSM. The two specimens where C. kroppenstedtii was also detected through routine culture methods included a perianal pus swab from a neonate with perianal abscess (with other two strains of E. coli), and tissue from a necrotic area of the knee (pure culture of C. kroppenstedtii on BA as pinpoint colonies after two days).

Table 2

Number of clinical specimens with isolation of C. kroppenstedtii

During the identification process, non-kroppenstedtii lipophilic corynebacteria, and Lactobacillus species were the most common bacteria that could mimic C. kroppenstedtii on CKSM, i.e. 1–2 mm off-white colonies. None of them, however, displayed the blackened halo or the vinegar odour. While sniffing of plates is not encouraged in the microbiology laboratory, the presence of vinegar odour on CKSM was highly predictive of the presence of C. kroppenstedtii in our experience. Notably, 13 (7.0%) of the 186 clinically isolated C. kroppenstedtii lacked blackened halo on CKSM, but esculin hydrolysis activity was subsequently demonstrated in 12 of these 13 on esculin agar slant. MALDI-TOF MS with enhanced database against C. kroppenstedtii was useful for final identification.

Incidentally, C. kroppenstedtii were noted in 24 out of 1903 (1.4%) specimens collected from the female genital tract (predominately high vaginal swabs), and 11 out of 617 (1.8%) surface swabs of newborns, all of whom were within one week of birth by normal vaginal delivery. Eleven (6.7%) of the 110 swabs of Tenckhoff catheter exit sites or percutaneous endoscopic gastrostomy site were also positive with C. kroppenstedtii. Isolation of C. kroppenstedtii was noted in 126 out of 8895 respiratory specimens, where the isolation rate of C. kroppenstedtii in males was significantly higher than those in females (1.9% vs 0.7%, P=0.0001).


Granulomatous mastitis is a recurrent, disfiguring condition often affecting non-lactating women of reproductive age. Other than the well-known infective agents such as Mycobacterium tuberculosis, dimorphic fungi and Brucella species, recent publications have identified a subset of granulomatous mastitis due to Corynebacterium kroppenstedtii, often associated with cystic neutrophilic granulomatous inflammation.11 12 Many existing reports identified C. kroppenstedtii from Gram stain, culture, 16 s rRNA gene sequencing or metagenomic analysis.1–4 However, significant under-reporting of C. kroppenstedtii infections are likely as the routine standard culture procedures in clinical laboratories are unreliable in isolating C. kroppenstedtii from clinical specimens. This could impede understanding of the organism and its role in human infection, leading to delayed treatment. Selective media have previously been devised for C. equi, C. urealyticum, oral Corynebacterium species to facilitate further study of similar organisms.13–15

In our study, the novel CKSM selective and differential media have significantly increased the culture yield of C. kroppenstedtii. This was achieved through inhibition of many clinically encountered Gram-positive and Gram-negative bacteria by fosfomycin, and growth enrichment of lipophilic C. kroppenstedtii with Tween 80.5 16 Visible colonies of C. kroppenstedtii were observed on CKSM within three days, with features allowing easy differentiation from other non-corynebacterial bacteria by means of colony morphology and the non-kroppenstedtii lipophilic corynebacterial colonies by esculin hydrolysis. An incidental observation of the vinegar odour from C. kroppenstedtii on CKSM was found to be an additional characteristic. The larger colonies, blackened halo, presence of vinegar odour and suppressed concurrent non-corynebacterial growth enhanced detection of C. kroppenstedtii even in mixed bacterial growth as shown by a substantial increase in detection rate. We believe that overgrowth by other bacteria, and the small pinpoint colonies of C. kroppenstedtii on BA and CHO have likely prohibited identification of C. kroppenstedtii colonies among other co-existing bacteria from clinical specimens on these non-selective agars, despite it being known that C. kroppenstedtii can grow on these media. It was noted in our study that 7% of C. kroppenstedtii failed to demonstrate esculin hydrolysis on CKSM, therefore 1–2 mm off-white colonies without blackened halo on CKSM should also be further characterised to exclude the possibility of C. kroppenstedtii. In addition, only 24 hours of CO2 enrichment is needed for the isolation of C. kroppenstedtii on CKSM. This could be important in laboratories with limited CO2 incubation facilities. Overall, the use of CKSM increases the detection of C. kroppenstedtii from clinical specimens, shorten time-to-visible growth, with low requirement of CO2 enrichment.

Interestingly, C. kroppenstedtii were isolated from female genital tracts and surfaces of newborns. These sites have not previously been reported to have C. kroppenstedtii. While C. kroppenstedtii, like many other Corynebacterium species, were considered part of normal human microbiota,17 the exact niche of this organism is unknown. Our data suggested that C. kroppenstedtii could be part of the normal commensal flora of the female genital tract, with newborns acquiring the bacteria during passage through the maternal genital tract.

There are several limitations in our study. First, we compared routine culture methods with the addition of CKSM in the isolation of C. kroppenstedtii from non-sterile specimens. While a superior sensitivity is demonstrated with the addition of CKSM, there is no current ‘gold standard’ for the isolation of C. kroppenstedtii from clinical specimens. Therefore, true sensitivity and specificity of CKSM cannot be ascertained. Second, only respiratory and miscellaneous specimens were tested because previously reported C. kroppenstedtii were isolated from these sites.1–3 5 CKSM performance on other specimen types, especially more contaminated specimens such as stools cannot be extrapolated from our data. However, C. kroppenstedtii in these other sites are unlikely to be of clinical significance. Third, some non-lipophilic esculin-hydrolysing corynebacteria such as C. aurimucosum, C. canis, C. durum, C. freiburgense, C. glucuronolyticum, C. matruchotiiand C. timonense have not been included in the study, thus it is uncertain if their colony morphology, including the blackened halo due to their esculin-hydrolyzing ability, could mimic those of C. kroppenstedtii. However, as they are non-lipophilic, these bacteria could be differentiated from C. kroppenstedtii by the lack of lipophilicity. Finally, clinical strains of non-kroppenstedtii lipophilic corynebacteria used for initial evaluation of CKSM were only identified using MALDI-TOF MS, thus the possibility of misidentification cannot be excluded. However, only isolates with a good MALDI-TOF MS score of >2.0 were included, and a recent study has demonstrated good species correlation between MALDI-TOF MS and rpoB gene sequencing for Corynebacterium species.18

We believe that the CKSM can substantially improve the culture yield of C. kroppenstedtii from clinical specimens, which will be particularly useful for patients with potential C. kroppenstedtii infections such as granulomatous mastitis, Tenckhoff catheter exit site infection, or culture-negative skin and soft tissue infections.3 The isolation of the organisms not only facilitate diagnosis, but allow antibiotic susceptibility testing to be performed as antibiotics-resistant C. kroppenstedtii have been reported.16

Take home messages

  • Corynebacterium kroppenstedtii has an increasingly recognised role in patients with granulomatous mastitis, but is not reliably isolated from clinical specimens using standard culture methods.

  • The novel selective-differentiating culture medium, CKSM, improved culture yield from clinical specimens, through the use of antibiotic suppression of co-existing bacteria, optimisation of growth through Tween 80 and differentiation through esculin hydrolysis. This use of CKSM can facilitate diagnosis, as well as the study of the epidemiology and transmission of this organism.

  • C. kroppenstedtii was found in female genital tract specimens and surfaces of newborns, suggesting that the organism could be part of the normal flora of the female genital tract. This finding, to the best of our knowledge, has not been previously described.

Abstract translation

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


We thank the technical staff in the Department of Microbiology, Queen Mary Hospital, University of Hong Kong, for assistance with the microbiological lnvestigations. This study was approved by the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (UW 15-643).



  • Handling editor Tony Mazzulli.

  • Contributors SCW designed, performed analyses and wrote the manuscript. RWP designed, performed analyses and reviewed the manuscript. CHF, AHN, VCL, THL, CPW conducted experiments, collected data and reviewed the manuscript. HT, VCC performed analysis and reviewed the manuscript. KYY designed experiments, performed analysis and reviewed the manuscript.

  • Funding This study was supported by the Hong Kong University Foundation.

  • Competing interests None declared.

  • Patient consent Detail has been removed from this case description/these case descriptions to ensure anonymity.

  • Ethics approval Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster.

  • Provenance and peer review Not commissioned; externally peer reviewed.