Disseminated Mycobacterium marinum in Human Immunodeficiency Virus Unmasked by Immune Reconstitution Inflammatory Syndrome

Abstract

Distinguishing disseminated Mycobacterium marinum from multifocal cutaneous disease in persons with human immunodeficiency virus/AIDS can present a diagnostic challenge, especially in the context of immune reconstitution inflammatory syndrome (IRIS). In this work, we demonstrate the utility of flow cytometry and whole genome sequencing (WGS) to diagnose disseminated M. marinum unmasked by IRIS following initiation of antiretroviral therapy. Flow cytometry demonstrated robust cytokine production by CD4 T cells in response to stimulation with M. marinum lysate. WGS of isolates from distinct lesions was consistent with clonal dissemination, supporting that preexisting disseminated M. marinum disease was uncovered by inflammatory manifestations, consistent with unmasking mycobacterial IRIS.

Mycobacterium marinum typically presents with skin lesions localized in areas of pertinent exposure. Disseminated disease is uncommon because of the organism’s preference for lower temperatures but has been reported in immunocompromised hosts. Immune reconstitution inflammatory syndrome (IRIS) is the worsening or unmasking of an underlying infection following antiretroviral therapy (ART) initiation and has only been reported once with cutaneous M. marinum infection [1]. IRIS can present with a marked inflammatory response to a preexisting infection and may require corticosteroids for improvement once other etiologies have been excluded. We describe a case report highlighting the use of sequencing and flow cytometry to diagnose disseminated M. marinum IRIS in a patient living with human immunodeficiency virus (HIV)/AIDS.

METHODS

Patient

The patient described in this report was a participant of a National Institutes of Health institutional review board–approved protocol (NCT0214705) and had given informed consent.

Flow Cytometry

Cryopreserved peripheral blood mononuclear cells (PBMCs) were collected before initiation of ART and at IRIS timepoints. For in vitro stimulation experiments, PBMCs were incubated in the presence of heat-inactivated, sonicated M. marinum coincubated with anti-CD28/CD49d, as previously described [2]. Data were acquired on BD LSRFortessa and analyzed using FlowJo version 10.6.

Microbiologic Workup and Molecular Identification

Skin biopsy specimens were used for touch preparation and ground preparation acid-fast bacilli (AFB) smears (auramine-rhodamine staining). Ground material was treated with N-acetyl-l-cysteine -sodium hydroxide to reduce growth of commensal flora. Culture was performed in mycobacteria growth indicator tube media (Becton Dickinson, Franklin Lakes, New Jersey) at 30ºC and 35ºC, 7H11 Middlebrook agar plates (Remel/Thermo Fisher Scientific, Lenexa, Kansas) at 35ºC, and Tripticase soy agar with 5% sheep blood (sheep blood agar, Remel) at 30ºC.

Following positive AFB smear from the first biopsy, an approximately 3-mm piece was subjected to pan-mycobacterial real-time polymerase chain reaction (PCR) targeting the secA1 gene followed by Sanger sequencing of the amplicon and comparison to publicly available DNA sequences in GenBank and institutional secA database for species-level identification (data not shown) [3]. Isolate identification was also performed on mycobacterial colonies by matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometry (Bruker, Inc).

Whole Genome Sequencing

Colonies from 7-day-old subcultures grown on 7H11 Middlebrook agar at 30ºC from each clinical isolate were used for DNA extraction and library preparation. For long-read sequencing, DNA extraction of MMA1 was performed using phenol-chloroform method as previously described [4]. The Pacific Biosciences protocol “Preparing multiplexed microbial SMRTbell libraries for the PacBio Sequel System” was used to create libraries from 3 μg of DNA. Sequencing was performed using a Sequel sequencer (Pacific Biosciences) using version 3 SMRT cells and sequencing reagents with 10-hour movies. Demultiplexed PacBio Sequel subreads were assembled with the Hierarchical Genome Assembly Process (HGAP4) pipeline within the PacBio SMRT Link version 6.0.0 package. The assembled contigs were circularized using Circlator 1.5.3 [5] and corrected reads generated from HGAP4 and Canu version 1.6 [6]. The circularized sequence was processed with 3 rounds of consensus polishing with the PacBio SMRTLink version 6.0.0 resequencing pipeline. The FASTA assembly was annotated using the Prokka (version 1.13) pipeline [7].

For short-read Illumina sequencing, colonies from MMA2 and MMA3 underwent bead beating (Powerlyzer 24, MoBio) for 40 seconds at 4000 rpm, followed by automated extraction (EasyMag, bioMérieux); 250 ng of the genomic DNA was sheared to approximately 500 bp with a Covaris m220 instrument (Covaris, Inc) and used for library preparation with NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs). Short-read Illumina sequencing was performed on an Illumina MiSeq instrument (300 bp PE) and assembled with SPAdes version 3.11 (--careful option) [8]. Genome assemblies were filtered to retain contigs >5000 bases. Variant calling in MMA2 and MMA3 isolates was performed using the BreSeq pipeline (version 0.33) [9] using MMA1 PacBio assembly as a reference.

Genomic Analysis

The patient’s genome assemblies were compared with all 22 nonrepeated available M. marinum genome assemblies downloaded from the National Center for Biotechnology Information (NCBI) RefSeq database (accessed September 2019; Supplementary Table 1). We also included a representative of each of the 2 closest relatives: Mycobacterium ulcerans (Agy99) (RefSeq accession number NC_008611.1) and Mycobacterium liflandii (128FXT) (RefSeq accession number NC_020133.1). A whole genome similarity matrix was generated using the Mash algorithm (version 2.1) [10], which was clustered and visualized using the heatmap.2 function in R (version 5.3) [11].

Data Availability

The genome sequences analyzed in this study have been deposited under NCBI BioProject accession number PRJNA630296.

RESULTS

Case Presentation

A 35-year-old Hispanic man initially presented to a community hospital with fevers, fatigue, weight loss, diarrhea, and a nonhealing arm wound after an injury at work. During his stay he was diagnosed with HIV/AIDS (CD4 T-cell count of 66 cells/µL; HIV-1 RNA viral load of 187 312 copies/mL), hepatitis C, Cryptosporidium enteritis, Staphylococcus lugdunensis bacteremia, and disseminated histoplasmosis. He was treated with ampicillin-sulbactam, amphotericin, and nitazoxanide with overall improvement, except for a persistent left arm wound with nodular lymphangitis, and was referred to our institution (Figure 1A).

Two months before his presentation he had rubbed and abraded his left elbow on a rough surface at his pet grooming salon. The lesion never healed, and a month later, it had started expressing semisolid white yellow material from 1 of the nodular lesions with persistent ulcerations on his elbow. In addition to his occupation as a pet groomer, he had cats, dogs, rabbits, birds, and a fish in a large aquarium at home.

A skin biopsy of the left elbow wound demonstrated granulomatous tissue and AFB (Supplementary Figures 1 and 2) confirmed as M. marinum by MALDI-TOF mass spectrometry and culture, as well as mycobacterial real-time PCR targeting the secA1 gene (not shown). He was placed on therapy with azithromycin, ethambutol, and rifabutin, and later transitioned to azithromycin, ethambutol, and moxifloxacin because of a pruritic rash with rifabutin. He received antimycobacterial therapy for a week before ART with dolutegravir/abacavir/lamivudine was started.

After 10 days of ART, he presented with worsening erythema and swelling of the left elbow with hypermetabolic nodular lymphangitis and new left thumb and right foot nodules, showing abnormally increased uptake on research ¹⁸F-fluorodeoxyglucose positron emission tomography–computed tomography (PET-CT) (Figure 1B–D and Figure 2A). The foot (heel) lesion was biopsied, and M. marinum was identified. His CD4 count had risen to 216 cells/µL and the viral load had decreased to 1068 copies/mL. A QuantiFERON TB Gold test, which was previously negative prior to ART, was repeated and was now positive.

Clinical Isolate Whole Genome Sequencing Supports Disseminated Infection

De novo assembly of the long PacBio reads from MMA1 generated a single closed contig, with a genome size of 6.3 Mb. Short-read sequencing of MMA2 and MMA3 generated 458 872 and 795 364 paired reads, respectively, which generated assemblies of approximately 6.3 Mb in length (N50 of 291 258 bp and 243 448 bp, respectively). The average GC content was 65.79% for all assemblies.

Variant calling of the second (MMA2) and third (MMA3) clinical isolates using the earliest isolate (MMA1) as reference revealed near identity of the 3 genomes, with 1 single-nucleotide variant (SNV) relative to MMA1 shared between the MMA2 and MMA3 isolates, and 3 additional SNVs and a frameshifting indel present in the MMA3 isolate (Supplementary Table 2). This high degree of similarity is consistent with the interpretation that the 3 isolates constitute a single clonal lineage.

We next sought to place the sequenced genomes from the patient in the context of previously sequenced M. marinum genomes. It has been suggested that M. marinum may be divided in “M” and “Aronson” lineages [12]. We compared whole genome nucleotide divergence between the patient’s isolates and 22 publicly available M. marinum genomes using Mash. Compared to the M. marinum genome sequences from GenBank, the patient’s isolates were most closely related to M. marinum MB2 strain with a Mash difference score of 0.005 (~0.5% divergence), and group within the “M” lineage (Figure 2B).

Enhanced In Vitro Response to M. marinum

In vitro stimulations of PBMCs with M. marinum was performed from this timepoint and demonstrated emerging polyfunctional cytokine response to M. marinum after ART compared to before ART, corroborating the QuantiFERON results (Figure 2C). Based on the absence of any new, alternative diagnoses, the emergence of cell-mediated responses to M. marinum, and confirmation that the new lesions were also M. marinum, a clinical diagnosis of disseminated M. marinum IRIS was made. The patient improved with addition of corticosteroids and a prolonged corticosteroid taper with prednisone for 10 weeks along with drainage of fluctuant lesions. Antimycobacterial therapy was continued for 15 months, with complete resolution of clinical findings.

DISCUSSION

To our knowledge, this is the first case of disseminated M. marinum complicated by IRIS described in the literature. An additional unique feature was that the patient initially presented with cutaneous disease, and disseminated disease was unmasked following immune reconstitution after initiation of ART and confirmed by molecular techniques.

The use of sequencing data strongly supported the hypothesis of disseminated disease as opposed to multiple sites of localized infection. Although highly similar independent inoculations cannot be definitively ruled out by this methodology, in addition to the anatomical distance and lack of water exposure of the heel, the degree of relatedness observed strongly supports the hypothesis of dissemination of a single-strain inoculum. This fact was significant as it demonstrated that disseminated M. marinum may initially present as only cutaneous disease in an immunocompromised host and the true extent of disease may only appear with immune reconstitution. Additionally, disseminated disease is a strong indication for prolonged antimicrobial course. The patient was treated with a longer course of antimycobacterial medications for disseminated disease [13, 14]. Evidence of a clonal pathogen also supported the diagnosis of a paradoxical inflammatory immune response to the pathogen rather than a new infection.

Production of multiple cytokines in response to underlying infections by CD4 T cells have been demonstrated in patients with IRIS [2, 15], and in vitro stimulations of PBMCs can be used to corroborate a clinical diagnosis of IRIS by demonstrating emergence of T-cell responses to underlying antigens. The hypermetabolic activity seen on research PET-CT has also been reported in IRIS, and changes in glycolytic activity following ART have been found to correlate with levels of inflammatory cytokines [16]. The use of flow cytometry is not always easily accessible, but the QuantiFERON Gold test detects production of interferon-γ in response to Mycobacterium tuberculosis–specific antigens, specifically ESAT-6 and CFP-10, which are shared by a few nontuberculous mycobacteria, namely M. marinum, M. szulgai, M. flavescens, and M. kansasii.

In summary, this clinical presentation and diagnostic workup gave us insight into the pathogenesis of M. marinum, especially in the immunocompromised host. The use of molecular techniques and flow cytometry were instrumental in the diagnostic evaluation and helped direct the optimum management.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. This research was supported by the Intramural Research Program of the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH). The patient signed informed consent for participation in the NIAID protocol (PET Imaging and Lymph Node Assessment of IRIS in Persons with AIDS [PANDORA], NCT02147405). This work utilized the computational resources of the NIH High Performance Computing Biowulf cluster (http://hpc.nih.gov). We acknowledge Morgan Park for the PacBio assembly and long-read Sequence Read Archive submission and Pam Thomas for the GenBank submission.

Financial support. This research was supported by the Intramural Research Program of the NIAID, NIH.

Potential conflicts of interest. All authors: No reported conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes