Mycobacterium abscessus soft tissue infections associated with subcutaneous injection of lipolytic agents

Abstract

Background: Management of Mycobacterium abscessus (M. abscessus) skin and soft tissue infections outbreaks require collaboration between clinicians, public health authorities and reference laboratories providing bacterial molecular identification and genotyping.

Objective: This study reports on a M. abscessus skin and soft tissue outbreak linked to mesotherapy treatments in Montréal, Canada. We present an innovative approach combining Nanopore long read and Illumina short read whole-genome sequencing data with a novel open access bioinformatic molecular epidemiology pipeline to support public health investigations by identifying genomically related isolates.

Methods: Public health investigations and physician questionnaires were used for outbreak identification and investigation. The complete genomes of six isolates from four individuals were sequenced. Nanopore and Illumina data were combined to assemble genomes de novo using a hybrid approach. Single nucleotide polymorphisms (SNPs) were identified within each genome by mapping the short reads to a reference genome. Single nucleotide polymorphisms distances among isolates, and between isolates and the reference genome were used alongside core SNP alignment analyses to evaluate genetic similarity between isolates and other contemporary M. abscessus genomes.

Results: Identified individuals had received mesotherapy injections from the same esthetician within a one-month period. Outbreak isolates were genomically nearly identical, differing by only 0–2 SNPs when using the earliest collected isolate as a reference. A phylogenetic tree revealed genetic distinctness between outbreak isolates and other contemporary M. abscessus isolates from Montréal.

Conclusion: Genomic sequencing and the presented bioinformatic approach can identify M. abscessus genomic relatedness suggestive of clinical outbreaks. This approach can support public health investigations, particularly when epidemiological links are uncertain.

Introduction

Nontuberculous mycobacteria are ubiquitous organisms that can be found in healthcare and other environments and cause outbreaks of skin and soft tissue, pulmonary, lymph nodes, and joint infections ^{Footnote 1Footnote 2Footnote 3Footnote 4}. Diagnosis and treatment of Mycobacterium abscessus (M. abscessus) soft tissue infections is particularly challenging due their frequently atypical nodular and mildly inflammatory presentation, the inability of conventional bacterial cultures to identify this rapidly growing mycobacteria and its innate and acquired resistance to available antimicrobial agents ^{Footnote 3Footnote 5}. Outbreaks of M. abscessus skin and soft tissue infections have been previously reported in immunosuppressed individuals, following exposure to contaminated water, as post-operative complication of plastic surgery or following tattooing, acupuncture and mesotherapy treatments—a procedure in which multiple injections of pharmaceuticals or vitamins are delivered into the mesodermal layer of skin tissues to promote the loss of fat or cellulitis ^{Footnote 5Footnote 6Footnote 7}. Past outbreak investigations required coordinated efforts between public health, clinical professionals and reference laboratories providing bacterial molecular identification and genotyping ^{Footnote 8Footnote 9}.

The increased availability of next-generation sequencing platforms and the development of semi-automated data analysis pipelines support the increased uptake of next-generation sequencing-based molecular typing for outbreak investigation by public health and clinical microbiology laboratories ^{Footnote 10}. Bacterial whole-genome sequencing (WGS) allows much greater discrimination of epidemiologically clustered isolates and therefore helps support or refute the results of standard public health investigations ^{Footnote 11Footnote 12Footnote 13}. By combining WGS and epidemiological investigations, Olawoye et al. recently showed that health care-associated human-to-human transmission of M. abscessus was rare on the island of Montréal ^{Footnote 14}. In a multi-country study comparing strains from distinct global M. abscessus outbreaks, Tettelin et al. highlighted that while strains from the same outbreak were clustered together, high genomic relatedness could also be observed between epidemiologically unrelated isolates ^{Footnote 15}. Next-generation sequencing-based molecular typing systems must be well adapted to pathogens’ intrinsic genomic diversity and evolution to best complement public health investigations and accurately support or refute the relatedness of isolates.

We report on the investigation of a M. abscessus skin infections outbreak associated with mesotherapy procedures in Québec. We also describe a unique innovative approach combining long read and short read sequence data to improve the resolution of our molecular epidemiology analysis. This novel bioinformatic pipeline for the phylogenetic analysis of M. abscessus isolates is openly accessible.

Methods

Public health investigation

In 2021, an infectious disease clinician reported a case of disseminated panniculitis lesions associated with mesotherapy injections to the Montréal public health department. Although M. abscessus infections are not a mandatory reportable disease in Québec, the attending physician had perceived a potential risk for the public since aesthetic injections were the only identifiable risk factor. An investigation was initiated to identify and attempt to eliminate the infection source. An interview with the esthetician was conducted to catalog the injection products that were used and to review infection control practices. A detailed list of past clients having underwent mesotherapy procedures using the same injectable product and procedures was obtained. Putative outbreak cases were defined as individuals with microbiologically confirmed M. abscessus skin and soft tissue infections having been injected with the same product by the same esthetician. Cases were identified by reviewing the clients’ medical charts, including skin infections and positive culture results for M. abscessus. Attending physicians of all newly identified infected individuals were asked to complete a questionnaire collecting basic epidemiology, clinical presentation and microbiology data (Appendix, Supplementary material).

Genomic sequencing and bioinformatic analysis

Mycobacterial isolates had initially been cultured by hospital-based clinical laboratories and referred to the Laboratoire de santé publique du Québec for M. abscessus speciation using 16s RNA sequencing. All available isolates from all individuals were retrieved and included in the molecular analysis. One randomly selected contemporary M. abscessus clinical isolate (control X) from Montréal was included as control for laboratory procedures and bioinformatic analyses. Mycobacterial DNA was extracted from pure culture and sequenced using Illumina and Oxford Nanopore next-generation sequencing platforms (Appendix, Supplementary material). All isolates passed sequencing quality controls including pre-alignment genomic coverage and absence of contaminating data (Appendix, Supplementary material).

Sequencing data was analyzed following previously described methods from Waglechner et al. ^{Footnote 13}. Both Nanopore and Illumina sequencing data were analyzed together using the Snakemake workflow engine (version 7.32.4) to execute the following pipeline using the ”hybrid” branch ^{Footnote 16}. Kraken2 v2.1.3, using a prebuilt MiniKraken database (version k2_plusfp_20220908_16) containing bacterial, viral, human, archaeal, vector, plasmid, protozoa and fungal sequences, was used for taxonomic assignment and contamination checking of each raw sequence data set (Nanopore and Illumina) for each sample ^{Footnote 17Footnote 18}. Using short read data, de novo assembly was performed and assigned to the most appropriate subspecies using mashtree along with reference sequences of Mycobacterium tuberculosis H37Rv (outgroup) (NC_000962) and the three M. abscessus subspecies: M. abscessus subspecies abscessus ATCC 19977 (NCBI GCF_000069185.1), M. abscessus subspecies massiliense CCUG 48898/JCM 15300 (NCBI GCF_000497265.2) and M. abscessus subspecies bolletii BD (NCBI GCF_003609715.1) ^{Footnote 18}. A hybrid de novo assembly was also prepared for each sample with long and short read data using dragonfly v1.2.1 with three rounds of short read polishing with polypolish v0.6.0 ^{Footnote 19Footnote 20}.

First, to assess outbreak individuals’ isolates relatedness, samples’ short read de novo assemblies for cases A through D were mapped against 1) the M. abscessus subspecies massiliense CCUG 48898 and 2) the hybrid short-read polished assembly of the putative outbreak earliest case (case B) using bwa-mem 2 v2.2.1 with default parameters ^{Footnote 19}. Single nucleotide polymorphisms (SNPs) were identified using samtools mpileup v1.14 ^{Footnote 15}. Variants were extracted into an alignment and subjected to recombination correction using gubbins v3.3.0 and counted using snp-dists v0.8.2 to assess relatedness of isolates ^{Footnote 21}. Core SNP alignments were analyzed by iqtree v2.3.6 with ultrafast bootstraps to examine the relatedness of outbreak isolates in the context of using either an outbreak isolate (case B hybrid assembly) or a subspecies type (M. abscessus subspecies massiliense CCUG 48898) as the reference genome. ETE (v3.1.1) was used to visualize maximum likelihood phylogenetic trees ^{Footnote 22}.

Second, to confirm outbreak isolates’ clustering within local circulating M. abscessus strains, a core genome was estimated from predicted open reading frames present in 99% or more of all assemblies from the outbreak isolates and 220 additional recent Montréal M. abscessus genomes from Olawoye et al. using the pangenome workflow from bactopia v3.1.0 using default parameters ^{Footnote 14Footnote 23}.

Data availability

Whole genome sequencing data is available on the National Center for Biotechnology Information GenBank database (bioproject ID 1221978).

Public health investigation

The initial investigation included only the first two clients and revealed that an esthetician had performed mesotherapy injections in a rented commercial space. Esthetic services were advertised on social media as part of a spa package and were performed in a total of five clients during a one-month period. Among those, the public health investigation and medical chart review concluded that four clients were infected (Figure 1). Within days following the injection procedures, all four individuals had presented with subcutaneous, erythematous nodules at various sites where injections occurred, including the waist, lower back and thighs, accompanied by systemic symptoms (fever, chills, fatigue). Infected individuals had no other risk factors for mycobacterial infection, such as immunosuppression, implanted foreign material or other surgeries. Skin biopsy cultures had confirmed M. abscessus as the etiologic agent in all four individuals, including two with two distinct cultured isolates each (C–1 and C–2, D–1 and D–2). Each individual’s clinical management and treatment were retrieved (Appendix, Supplementary material).

Figure 1 - Text description

This figure presents a horizontal timeline of a Mycobacterium abscessus outbreak in Montréal, showing clinical events for four patients (A, B, C, and D) over a 14-month period. The horizontal axis represents time in months, with month 0 defined as the time of the first patient’s injections. The vertical axis lists patients A through D.

For each patient, multiple time intervals and events are displayed:

• Incubation period is shown as a short bar near the beginning of each patient’s timeline, representing the time between injection and onset of symptoms. This period occurs within approximately the first 1 to 2 months for all patients.
• Definitive treatment duration is represented by longer horizontal bars starting after diagnosis and extending for varying durations. Treatment lasts approximately:
  • Patient A: about 1 month (around months 5 to 6)
  • Patient B: about 5 months (approximately months 3 to 8)
  • Patient C: about 6 months (approximately months 5 to 11)
  • Patient D: about 1 to 1.5 months (approximately months 4 to 5.5)
• Time of sample collection is indicated by circular markers. For all patients, at least one sample is collected around month 3. Patients C and D have repeated samples collected over time, reflecting persistent lesions.
• Time of surgery is indicated by “X” markers. Surgical interventions occur:
  • Patient A: around month 5
  • Patient B: around month 13
  • Patient C: around months 5 and 11
  • Patient D: multiple procedures between approximately months 4 and 6
• A vertical line at approximately month 2.5 marks the end of at-risk injection activities by the esthetician. All initial exposures and incubation periods occur before this point, while most diagnoses, treatments, and interventions occur afterward.

Overall, the figure illustrates that symptom onset occurred within a short time after exposure, while treatment and clinical management extended over several months, with variability in duration and need for repeated interventions among patients.

A registered letter based on regulations in Québec’s public health law ordered the esthetician to assure a cessation of all activities. The esthetician had discarded all injection products after clients reported adverse reactions, so these products were unavailable for culture. The mesotherapy product lacked a Drug Identification Number in Health Canada’s Drug Product Database and was not a licensed natural product on Health Canada’s Licensed Natural Health Product Database ^{Footnote 24Footnote 25}. Health Canada’s Central Triage Unit, Regulatory Operations and Enforcement Branch was alerted to the possible M. abscessus contamination of the product. The case was referred to the Health Product Compliance East Unit of Health Canada which performed a compliance verification for the product; however, the results of the investigation, including additional information obtained from direct communication with the manufacturer, were not shared in accordance with the Privacy Act. As of February 2025, the Health Product Compliance East Unit reported “no more noncompliance” for the product. Also, no recalls or safety alerts have ever been issued for the product according to the Government of Canada’s Recalls and Safety Alerts database ^{Footnote 26}.

Outbreak isolates relatedness molecular analysis

All outbreak-related isolates were identified as M. abscessus subspecies massiliense. The contemporary clinical control isolate was identified as M. abscessus subspecies abscessus and excluded from further pairwise SNP comparisons using the outbreak-specific hybrid assembly reference. Short read data from the outbreak isolates mapped against a polished assembly of an early outbreak isolate (Case B) showed 100% mapping coverage. The outbreak isolates were nearly identical with each other by pairwise SNP distance, ranging from 0 to 2 SNPs. The same procedure using M. abscessus subspecies massiliense CCUG 48898 as a reference yielded pairwise SNP distances ranging from 32 to 50 SNPs between outbreak isolates, while outbreak isolates ranged from 26,612 to 26,644 pairwise SNPs apart from the reference (Figure 2, Figure 3). Sequentially cultured clinical isolates from the same individual (e.g., C–1 vs. C–2, D–1 vs. D–2) sampled up to eight months apart were not more closely related to the earliest isolate in each pair than to isolates from other individuals.

Figure 2 - Text description

This figure contains two panels (a and b), each showing a maximum likelihood phylogenetic tree of Mycobacterium abscessus outbreak isolates from Montréal. Both trees depict genetic relatedness among isolates from six clinical cases (cases A, B, C-1, C-2, D-1, and D-2), with branch lengths representing genetic distance.

Panel a:

The phylogenetic tree is rooted using a reference genome labeled M. abscessus subsp. massiliense CCUG 48898 (indicated in bold). All outbreak isolates cluster closely together on a single branch, distinct from the reference genome, indicating high genetic similarity among cases relative to the external reference. The scale bar indicates a genetic distance of 0.002 substitutions per site. Within the outbreak cluster, isolates from cases C-1, D-1, A, C-2, B, and D-2 are nearly indistinguishable, with very short branch lengths separating them.

Panel b:

The phylogenetic tree is reconstructed using a de novo assembled genome from case C-1 as the reference (indicated in bold as “Case C-1 de novo”). In this tree, all outbreak isolates again cluster tightly, with minimal genetic distances between them. The scale bar indicates a much smaller genetic distance (8.54 × 10^-8 substitutions per site), reflecting the high resolution of comparison within the outbreak. Minor branching differences are visible between isolates, but all cases remain closely related, consistent with a common source.

Across both panels, the trees demonstrate that isolates from all patients are highly genetically similar, supporting their inclusion in a single outbreak cluster, with only minimal variation observed between samples from the same or different individuals.

Figure 3 - Text description

This figure presents two pairwise single nucleotide polymorphism (SNP) distance matrices comparing Mycobacterium abscessus isolates from six outbreak cases (A, B, C-1, C-2, D-1, and D-2), along with reference genomes.

Panel a:

Distances are calculated using Illumina sequencing data mapped to the M. abscessus subsp. Massiliense CCUG 48898 reference genome. The reference strain shows very large genetic distances from all outbreak isolates (on the order of tens of thousands of SNPs), indicating it is highly divergent. In contrast, all outbreak isolates are closely related to each other, with pairwise distances generally in the range of a few dozen SNPs. Variation within this group is modest and relatively uniform, with no clear separation into subgroups by patient. Distances between samples from the same individual are similar in magnitude to those observed between different individuals.

Panel b:

Distances are recalculated using a de novo long-read assembly from an early outbreak case (case C-1) as the reference. In this panel, pairwise distances among outbreak isolates are minimal, typically ranging from zero to two SNPs. Several isolate pairs are identical (zero SNP differences), while the largest observed differences are limited to two SNPs. This pattern is consistent across all cases, with no distinct clustering or separation between patients.

Overall, the two panels show that while absolute SNP distances vary depending on the reference genome used, the relative relationship remains consistent: all outbreak isolates form a highly homogeneous group with minimal genetic variation and no evident stratification by case.

Outbreak isolates clustering amongst locally circulating strains

The pangenome was constructed using assemblies from the six outbreak isolates as well as the Control X isolate along with 220 additional M. abscessus isolates collected from 2010 to 2018 recently published from the island of Montréal and other Québec laboratories ^{Footnote 5}. The core genome (sequences found in 99% or more of genomes) consisted of 3,403 open reading frames totaling 3,951,644 bp when aligned. After recombination detection and masking with ClonalFrameML, pairwise SNPs were counted from this alignment and used to produce a phylogenetic tree of 227 Montréal M. abscessus isolates from all three subspecies (Figure 4). This tree shows that the outbreak isolates are distinct from other subspecies massiliense isolates previously sequenced except for one. This isolate had been cultured from the respiratory track of a cystic fibrosis patient who had not received mesotherapy treatments and shared no epidemiological links with the outbreak individuals.

Figure 4 - Text description

This figure shows a phylogenetic tree based on core-genome single nucleotide polymorphisms (SNPs) for Mycobacterium abscessus isolates collected in Montréal in 2025. The tree includes outbreak isolates from six clinical cases (A, B, C-1, C-2, D-1, and D-2) and a larger background set of approximately 220 additional isolates from the same geographic region.

Branches represent genetic relationships, with branch length corresponding to genetic distance. A scale bar indicates the level of divergence.

Isolates are annotated along the right side with a vertical color strip indicating subspecies: blue for M. abscessus subsp. abscessus, red for M. abscessus subsp. massiliense, and pink for M. abscessus subsp. bolletii. A labeled control isolate (“Control X”) is positioned within the broader tree outside the outbreak cluster.

The outbreak isolates are grouped together within the M. abscessus subsp. massiliense portion of the tree, forming a tight cluster with very short branch lengths, indicating high genetic similarity. This cluster is clearly separated from other M. abscessus isolates in the Montréal dataset, which show greater genetic diversity and are distributed across multiple branches.

Overall, the figure shows that outbreak-associated isolates are closely related to each other and distinct from the broader population of M. abscessus circulating in Montréal.

Discussion

During this M. abscessus infection outbreak investigation, we confirmed the ability of a conventional epidemiological investigation to comprehensively identify individuals involved in a point-source outbreak. We also developed an improved molecular approach leveraging both short (Illumina) and long (Nanopore) read sequence data to confirm the relatedness of epidemiologically linked cases and assess their bacterial isolates’ molecular clustering within an extended catalog of locally circulating strains.

Esthetician services and easy access to injection products on-line do not fall under Health Canada regulations. In Québec, anyone who wishes to receive regulated product injections for cosmetic purposes must undergo a prior medical evaluation so that a doctor can establish an individualized treatment plan. A nurse or licenced practical nurse, if prescribed to do so, may undertake such treatments. Since esthetician services are not regulated, it is important for the public to be aware of the risks associated with invasive treatments which should only be provided under medical supervision. In this context, we highlight the importance of case signaling by health professionals to Public Health departments. This outbreak investigation was initiated following physicians’ report of a post-injection infection. Public health authorities have a responsibility to investigate real and perceived threats to the public’s health; as seen in this investigation, they use their authority under the law to order the cessation of activities or use of suspected products, until a public health investigation can be completed and the threat removed or diminished.

A strength of our investigation is the combination of traditional epidemiological investigation and multi-modal sequencing molecular epidemiology. Using clinically related isolates from this outbreak investigation, we extended our previously published M. abscessus bioinformatic pipeline to make use of long read data from Oxford Nanopore instruments ^{Footnote 15}. Long read data enabled the production of essentially closed and complete genome assemblies of the outbreak isolates and enabled the use a more relevant reference for pairwise SNP analysis. Before the common availability of long read sequencing, mapping relied on using the closest publicly available genome sequence that, in practice, could result in bias in SNP calling. This bias can be seen when mapping our outbreak isolates against the M. abscessus subspecies massiliense CCUG 48898 strain where the clinical isolates were collectively approximately 28,000 SNPs apart from the reference. More importantly, bias was introduced at the mapping and SNP calling stages that exaggerated the differences between outbreak isolates. Using the complete de novo Case C–1 assembly as a reference 100% of the reference was covered revealing limited genomic variability between outbreak isolates where each was within 0–2 SNPs of each other.

In this study, all outbreak isolates were found to be almost genetically identical. Sequentially sampled isolates from the same infected individuals did not exhibit higher or lower genomic relatedness (e.g., C–1 and C–2, D–1 and D–2). Previous studies suggested that M. abscessus subspecies massiliense has a lower mutation rate than other M. abscessus ^{Footnote 27Footnote 28Footnote 29}. In a previous M. abscessus molecular epidemiology study conducted over nine years, clinical isolates collected from the same individual did preferentially clustered together ^{Footnote 14}. Bryant et al., suggested that SNP distances of less than 25 were associated with related M. abscessus subspecies massiliense isolates and that SNP distances of 50 to 200 correlated with distinct clusters ^{Footnote 27}. Doyle et al. found median SNP distances of 2,084 within the same sequence cluster and advocated for variant calling against more genetically similar reference sequences or when unavailable, the first isolated sample, to better highlight differences between strains and decluster genetically similar sequences ^{Footnote 30}. We believe the differences between within-host bacterial evolution in lungs and soft tissue infections, the bioinformatic analytic approach and the choice of alignment reference sequence explain these differences between previously published results and ours. Among other previously reported M. abscessus skin and soft tissue outbreaks, one included short-read sequencing-based WGS analysis and also found related isolates to be within three SNPs of each other and source environmental isolates ^{Footnote 6}.

Conclusion

Nontuberculous mycobacteria skin infections outbreaks can be associated with aesthetic injection products. Bacterial WGS and the presented bioinformatic pipeline can support M. abscessus outbreak investigations. Combining Nanopore and Illumina sequencing data for molecular clustering analyses represents a novel approach to improve M. abscessus clinical isolates clustering analysis. Genomic sequencing is of important value when molecular clustering further supports a clinically suspected iatrogenic infectious.

Authors' statement

XQ-N — Methodology, validation, formal analysis, investigation, writing–original draft, visualization
NW — Methodology, software, validation, formal analysis, data curation, writing–original draft, visualization
MV — Investigation, formal analysis, data curation, writing–original draft
P-AV — Investigation, writing–review & editing
FP — Software, formal analysis, data curation, writing–review & editing
BT — Investigation, writing–review & editing
MZ-N — Investigation, writing–review & editing
CT — Investigation, writing–review & editing
NP — Formal analysis, investigation, writing–review & editing
MT — Investigation, writing–review & editing
AU — Investigation, writing–review & editing
P-MA — Formal analysis, investigation, writing–review & editing
JC — Formal analysis, investigation, writing–review & editing
RL — Software, validation, formal analysis, resources, supervision, writing–review & editing
SGL — Conceptualization, methodology, validation, investigation, resources, data curation, writing–review & editing, supervision, project administration

Xavier Quan-Nguyen, Nicholas Waglechner and Maxime Veillette contributed equally to this article.

The content and view expressed in this article are those of the authors and do not necessarily reflect those of the Government of Canada.

Competing interests

The authors declare no competing interests.

ORCID numbers

Xavier Quan-Nguyen — 0009-0005-1448-4418
Maxime Veillette — 0000-0002-0493-3428
Melissa Zarandi-Nowroozi — 0000-0003-3635-4292
Joseph Cox — 0000-0002-7041-1556
Robyn Lee — 0000-0001-7120-9053
Simon Grandjean Lapierre — 0000-0003-3646-1573

Acknowledgements

None.

Funding

No dedicated funding was obtained for this study. SGL is supported by a clinical-scholar award from the Fonds de Recherche Québec – Santé (Award #367743).

Appendix

Supplemental material is available upon request to the author: simon.grandjean.lapierre@umontreal.ca

Figure S1: Data collection questionnaire

Figure S2: Mycobacterial DNA extraction and sequencing

Figure S3: Sequencing quality metrics

Figure S4: Clinical management and treatment

Table S1: Mycobacterium abscessus outbreak patients’ treatment and clinical management

• Previous Page
• Table of contents
• Next page