Background

Mycobacterium abscessus (M. abscessus) comprises a group of non-tuberculous, rapidly growing mycobacteria (RGM) and consist of three subspecies: M. abscessus subsp. abscessus, M. abscessus subsp. bolletii, and Mycobacterium abscessus subsp. massiliense (MMA) [1]. Mycobacterium abscessus subsp. abscessus and bolletii exhibit macrolide resistance with a full-length functional erythromycin ribosome methyltransferase (erm)(41), whereas MMA exhibits macrolide susceptibility with a truncated, nonfunctional erm(41) [2]. Mycobacterium abscessus subsp. massiliense can be differentiated from the other two subspecies by a polymerase chain reaction (PCR) analysis of erm(41) as MMA has a shorter erm(41). The erm(41) DNA amplified from MMA are 397-bp long, whereas the erm(41) DNA amplified from the other 2 subspecies are 673-bp long [2]. The rrl mutation is associated with acquired resistance to macrolides in RGM [3]. Other mutations associated with macrolide resistance have also been reported in MMA, including mutations in the rpIV gene [4].

Although MMA can cause pulmonary diseases, surgical site infections (SSI), and disseminated diseases [1], no cases of aortic endograft infection have been reported. Here, we describe the first case of aortic endograft infection caused by MMA. During the clinical course, the bacteria acquired macrolide resistance, and mutations were observed in genes with no previous reports of associations with macrolide resistance.

Case presentation

An 85-year-old man was hospitalized with a 4-day history of fever and abdominal pain. He had undergone stent-graft insertion for an abdominal aortic aneurysm 2 months prior. Computed tomography (CT) revealed periaortic fluid collection consistent with stent-graft placement (Fig. 1A). Based on the clinical information, he was diagnosed with aortic endograft infection and administered infusion of meropenem (1 g twice daily) and vancomycin (0.4 g twice daily).

Fig. 1
figure 1

The results of computed tomography scan. A Before initiating antimicrobial therapy. B After 14 weeks of antibiotic therapy, periaortic fluid pooling became larger than that before antibiotics administration. C-D Three days after drainage surgery, the abscess expanded and spread along the drainage tube and formed a subcutaneous abscess around the drain insertion site. White arrowheads represent the abscess and white arrows represent the drainage tube

Full size image

Despite 14 weeks of antibiotic therapy, periaortic fluid accumulation increased as compared to that before antibiotic administration (Fig. 1B). Surgical abdominal drainage was performed, and an indwelling drain was inserted. Bacterial culture of the drainage fluid gave negative results on sheep blood agar, BTB lactose agar, and chocolate agar at 35 ℃ in 5% CO2 for 24–48 hours. Three days after the drainage surgery, the periaortic fluid increased again and spread along the drainage tube, forming a subcutaneous abscess around the drain insertion site (Fig. 1C, D). The abscess was drained, and fluorescent acid-fast staining of the abscess fluid revealed bacilli. The organism was identified as Mycobacterium abscessus by nucleic acid sequencing of housekeeping genes (16S rRNA, rpoB, hsp65, sodA) (Table 1). The PCR product length of erm(41) was only 397 bp, which indicates deletion of this gene, according to a previous study [2] (Fig. 2). Thus, the organism was identified as MMA. The same organism was isolated from the periaortic fluid specimens. In addition, the first drainage sample was positive on 7-day culture in thioglycolate broth, and the organism was identified as MMA. Based on these results, the patient was diagnosed with aortic endograft infection with MMA.

Table 1 Analysis results of 16S rRNA, rpoB, and hsp65 sequence
Full size table
Fig. 2
figure 2

Agarose gel electrophoresis of DNA extracted from drainage fluid. The white arrow shows amplification of the 397 bp-long erythromycin ribosome methyltransferase gene. Full-length gels are presented in Supplementary Figure 1

Full size image

The antibiotic regimen was changed to intravenous imipenem-cilastatin (IPM/CS) 500 mg four times a day, amikacin (AMK) 300 mg every other day (Ccr 26.3ml/min), and oral clarithromycin (CAM) 400 mg twice daily. Two months after this treatment, only oral CAM 400 mg twice daily and sitafloxacin (STFX) 100 mg once daily were administered because CT showed a decreased in the size of the abscess. Although the MIC of STFX is lower than that of other fluoroquinolones, the breakpoint of STFX is not well-defined. (Table 2). However, after 6 weeks of treatment, the abscess around the aortic aneurysm increased in size again, and a new abscess appeared around the bilateral iliac arteries and veins. Antimicrobial susceptibility testing of the drainage fluid from the abscess resulted in the isolation of an MMA strain that had acquired resistance to CAM (Table 2). Intravenous IPM/CS 500 mg four times a day, AMK 300 mg daily, and oral linezolid (LZD) 600 mg twice daily were added to the treatment regimen along with oral CAM and STFX. LZD was discontinued within 1 week because of thrombocytopenia. The four-antibiotic combination therapy and abscess drainage were continued for 4 months, but the size of the abscesses did not change. On day 286 of hospitalization, he was discharged to a nursing home with a drainage tube, and the treatment with oral CAM, STFX, and intravenous AMK was continued by a visiting physician. Six months after discharge from the hospital, the patient died of an unknown cause. Neither the full-length erm(41) gene nor the rrl gene mutation was found by Sanger sequencing in the pre- and post-treatment strains. The reserved pre- and post-treatment strains were cultured on sheep blood agar (Fig. 3) for bacterial recovery and DNA extraction, and whole-genome sequence analysis (WGS) was performed on a NextSeq550 sequencer (Illumina, San Diego, CA). WGS of the post-treatment strains revealed two indels and one single nucleotide polymorphism (SNP) (Table 3). There was one SNP between the pre- and post-treatment strains, and the average nucleotide identity was 99.99%; thus, both the strains were considered identical. Reinfection with another species of MMA was ruled out based on the WGS results.

Table 2 Susceptibility test results of Mycobacterium abscessus subspecies massiliense
Full size table
Fig. 3
figure 3

Colonies of post-treatment strains. Post-treatment strains cultured on sheep blood agar

Full size image
Table 3 Result of whole genome sequence analysis of two strains of Mycobacterium abscessus subspecies massiliense
Full size table

Discussion

This case is noteworthy because it presents three instructive points. First, aortic graft infection caused by MMA is extremely rare. Second, the identification of MMA from an aortic graft infection treated with prolonged antimicrobial therapy is clinically meaningful in terms of selection of antimicrobials. Third, the MMA strain acquired resistance to clarithromycin during the treatment course.

Aortic endograft infection is rare and has a poor prognosis, with an incidence rate of 0.43–2.0% [5,6,7,8,9,10] and a mortality rate of 11–50% [5, 6, 8,9,10,11,12]. The causative pathogens include Staphylococcus sp., Streptococcus spp., Propionibacterium spp., Escherichia coli, and Enterobacter cloacae, which are culture-negative in 20–55% of cases [5,6,7, 9,10,11]. Aortic endograft infection with nontuberculous mycobacteria (NTM) is extremely rare. Mizoguchi et al. [13] reported an endovascular device infection with an iliopsoas abscess caused by Mycobacterium tuberculosis var. bovis after intravesical bacillus Calmette–Guérin therapy; Shaikh et al. [14] reported an aortic graft infection with a Mycobacterium avium complex in a patient with a human immunodeficiency virus infection; and Plate et al. [15] reported an aortic endograft infection with Mycobacterium chimaera associated with an outbreak in the high-care units. To the best of our knowledge, this is the first report of aortic endograft infection with MMA. This case report highlights that despite the rarity, clinicians should still consider the possibility of mycobacterial infection, in cases of no improvement following treatment with broad-spectrum antibiotics.

In this case, the route of infection was unknown. Patients at risk for NTM infections are immunocompromised patients or those with pre-existing pulmonary conditions [16]. However, our patient was not immunocompromised and had no pulmonary abnormalities. Nontuberculous mycobacteria can be a causative agent of SSI, and outbreaks due to the contamination of medical devices have been reported. For example, ultrasound gel contaminated by MMA on plastic trays, [17] and Mycobacterium wolinskyi [18] and Mycobacterium chimaera [19, 20] transmission through heater-cooler-devices for extracorporeal circulation. Six months after this case, another patient who underwent ascending aorta replacement at our hospital, developed bacteremia due to MMA. We conducted microbial surveillance in the operating room environment, which included carts, shelfs, personal computers, ultrasonic probes, tube holders, and the hands of doctors and nurses. However, MMA was not detected anywhere, and no common devices were used between the two cases (extracorporeal circulation was not used in our case).

The MMA strain isolated in this case had acquired resistance to clarithromycin 2 months after the initiation of clarithromycin treatment. Two gene mutations related to macrolide susceptibility in RGM have been reported: erm(41) is involved in macrolide-induced resistance, and rrl is associated with acquired resistance to macrolides [3]. According to the WGS of the pre- and post-treatment strains, neither reinfection with another species of MMA with full-length erm(41) nor acquisition of the rrl mutation from the same species was found. Other mutations associated with macrolide resistance in MMA have been also reported, including mutations in the rpIV gene [4]. WGS of the post-treatment strains revealed a frameshift mutation in the gene encoding aminodeoxychorismate lyase and a missense mutation (CTG→CCG) in the gene encoding the luciferase-like monooxygenase class flavin-dependent oxidoreductase, which was not associated with erm(41), rrl, and rpIV mutations. Aminodeoxychorismate lyase catalyses the following chemical reaction, “4-amino-4-deoxychorismate = 4-aminobenzoate + pyruvate” [21]. Luciferase-like monooxygenase class flavin-dependent oxidoreductase transfers one oxygen atom of an oxygen molecule to a substrate while reducing the other oxygen atom to water [22]. The effect of these enzymatic changes on MMA is not known.

There were three mutation sites between the two strains shown by WGS, but these were in regions not associated with known macrolide susceptibility. It should be clearly stated that the cause of macrolide resistance in this instance remains uncertain. Inferring the mechanism of macrolide resistance solely from the function of the mutated genes in this case is challenging. Moreover, it cannot be definitively ruled out that these mutations may not be involved in drug resistance at all. In this case, the mechanism by which the MMA strain acquired macrolide resistance remains unknown. However, new gene mutations leading to clarithromycin resistance in MMA have been reported [23, 24], and further WGS results may clarify the association between the genetic variants in our case and clarithromycin resistance.

Conclusion

We report a case of infected aortic aneurysm caused by MMA. Aortic endograft infection caused by an MMA strain is extremely rare, but MMA should be suspected as the causative microorganism when broad-spectrum antimicrobials are ineffective. During the clinical course, the bacteria acquired macrolide resistance and mutations were observed in genes with no report of associations with macrolide resistance.