We analyzed 27 biopsies of cutaneous/mucous NTMD and 27 biopsies of TB diagnosed between December 2, 2013 and May 17, 2023. 26 biopsies of cutaneous NTMD and 26 biopsies of TB were from the First Affiliated Hospital of Sun Yat-sen University, 1 biopsy of cutaneous NTMD and 1 biopsy of TB were from the Affiliated Panyu Central Hospital of Guangzhou Medical University. The inclusion criteria comprised cases of cutaneous/mucous NTMD and TB. The diagnosis was based on clinical presentation, imaging, pathogenesis, pathological examination, metagenomics next-generation sequencing (mNGS), and tuberculosis polymerase chain reaction (TB-PCR) testing (Sardiña et al., 2020; Epperson et al., 2024). Cutaneous/mucous tissue, cutaneous pus, blood, and bronchoalveolar lavage fluid (BALF) were collected for acid-fast bacilli (AFB) culture, mNGS, and TB-PCR testing. mNGS was used as a high-resolution reference method, mNGS plays a critical role in the identification and diagnosis of infectious diseases. However, mNGS still presents several limitations, such as FFPE contamination, detection threshold, and semiquantitative interpretation challenges (Appak et al., 2018; Huang et al., 2019; Zhou et al., 2019; Dulanto Chiang and Dekker, 2020; Qian et al., 2020). In this cohort, all 27 biopsies of NTMD were diagnosed by mNGS, among which 12 biopsies underwent AFB culture, with a positive rate of approximately 58.3% (7/12). All 27 biopsies of tuberculosis (TB) were confirmed by mNGS and TB-PCR, among which 10 biopsies underwent AFB culture, with a positive rate of approximately 60.0% (6/10).

Specimens were fixed in 4% neutral formaldehyde, routinely dehydrated, clarified, paraffin-embedded, and cut into 3 μm thick sections. These sections were stained with hematoxylin and eosin (HE) and subjected to acid-fast staining. Two dermatopathologists with subspecialty expertise conducted independent morphological analyses to assess differences between the groups.

Immunohistochemical staining was performed on 4 μm sections cut from formalin-fixed, paraffin-embedded tissue using an automated immunostainer. The primary monoclonal antibodies used included CD68 (1:100, mouse monoclonal, clone KP1, catalog No. ZM-0060; OriGene), CD86 (1:100, rabbit monoclonal, clone E2G8P; Cell Signaling), CD163 (RTU, mouse monoclonal, clone 10D6, catalog No. ZM-0428; OriGene), CD206 (1:1000, rabbit monoclonal, catalog No. 18704-1-AP; Proteintech), and PD-L1 (1:50, mouse monoclonal, clone 22C3, catalog No. M3653; Dako). The cells were counted manually by two individuals (X-Y Z and Z W). In cases of discrepancies between the two evaluators, the slides were re-examined and recounted to reach a consensus.

The percentage of positive macrophages was calculated for CD68, CD86, CD163, and CD206, based on the total number of macrophages. PD-L1 expression was assessed by counting PD-L1 stained cells (lymphocytes and macrophages/histiocytes) relative to the total number of live mononuclear inflammatory cells, expressed as a percentage. Membrane staining intensity was considered PD-L1 expression, including in multinucleated giant cells. In PD-L1 assessment, there is currently no standardized criteria for inflammatory lesions. Therefore, we adopted the ≥10% cut-off in immune cells based on the study by Selma Emre (Emre et al., 2023). Samples were categorized into low PD-L1(+) (less than 10% positivity) and high PD-L1(+) (10% or greater positivity) groups.

When selecting body fluids or fresh tissue from the infection site for mNGS testing, strict aseptic techniques must be followed during the collection of sterile body fluid/tissue samples to avoid false positives caused by environmental mycobacterial contamination. Samples were aseptically sealed and stored at −20 °C or transported on dry ice to the molecular laboratory for mNGS detection. The samples underwent grinding and cell wall disruption procedures followed by host DNA depletion. Upon completion of pre-processing, nucleic acid extraction was performed using magnetic bead-based methods. RNA and DNA libraries constructed from patient-derived samples were sequenced on an Illumina HiSeq platform in rapid-run mode, achieving a depth of 5 to 10 million single-end reads, each 140 base pairs in length. The raw sequencing data were quality-controlled using fastp (v0.23.2), which included adapter trimming, low-quality read filtering, low-complexity sequence removal, and elimination of excessively short reads. A human reference database was constructed based on GRCh38.p13 and the NCBI nt database to identify and remove human-derived sequences through alignment with bowtie2 (v2.3.5.1). A pathogen database was built using the NCBI RefSeq database (v20221231), FDA-ARGOS v1.0, the Genome Taxonomy Database release 206, and the NCBI GenBank database (v20221231). Non-human sequences were aligned to the pathogen database using the Burrows-Wheeler Aligner (BWA, v0.7.17-r1198-dirty). Contaminating sequences were filtered using a laboratory background database, followed by taxonomic classification and statistical analysis (in-house script) to obtain species-specific read counts, relative abundance, and genome coverage. All detected read counts were normalized to 20 M reads. For specific pathogens such as NTM and MTB, the detection of one unique species-specific read was considered positive (Naccache et al., 2014; Schlaberg et al., 2017; Han et al., 2022).

Paraffin-embedded tissues were sectioned at 5 μm thickness, and DNA was extracted using the Paraffin-Embedded Tissue Genomic DNA Extraction Kit (Qiagen). The concentration and purity of extracted DNA were assessed by UV spectrophotometry. Subsequently, samples underwent an MTB assay according to the instructions of the Mycobacterium tuberculosis DNA fluorescence diagnostic kit (PCR-fluorescence probing, Sansure Biotech Inc.). Positive and negative controls were established to ensure the accuracy and reliability of results (Choi et al., 2012; Lim et al., 2019; Sawatpanich et al., 2022).

Statistical analysis was performed using SPSS 27.0 software. Categorical variables were presented as frequencies and percentages. The Fisher’s exact test was used to compare pathological features, immune cell infiltration, granuloma types, and PD-L1 expression on immune cells. P-values were adjusted for multiple testing using the Benjamini-Hochberg (BH) method, with a significance threshold set at P < 0.05. The correlations among the positive rates of macrophage markers (CD68, CD86, CD163, and CD206) were analyzed as follows. The Shapiro-Wilk test (prioritized for sample sizes n ≤ 50) was used to assess normality. Variables with p > 0.05 were considered normally distributed, and Pearson correlation coefficients were calculated. For variables failing to meet the normality criterion (p ≤ 0.05), nonparametric Spearman correlation was employed. Correlation strength was quantified using R-values ranging from –1 to 1, where values closer to ±1 denote a stronger linear association between variables. A p-value less than 0.05 was considered statistically significant.

A total of 27 biopsies of NTMD were analyzed, with cutaneous/mucosal biopsy specimens obtained from each case ( Table 1 ). The biopsies included 8 of M. marinum, 6 of M. abscessus, 5 of the M. avium-intracellulare complex (MAC), 3 of M. haemophilum, 3 of M. avium, 1 of M. colombiense, and 1 of M. mantenii. Specimens were collected from various sites: 11 from the limbs, 4 from the face, 2 from the back, 2 from the abdomen, 1 from the neck, 5 from unspecified cutaneous locations, and 2 from the nasal mucosa.

The 27 biopsies represented 20 patients with an average age range of 41–72 years (mean age 57.9, median age 59). Most patients were middle-aged or elderly, with a male-to-female ratio of 1:1. Five patients had multiple biopsies from different sites either concurrently or at different times. One patient had successive infections with two NTM species: initially with M. avium-intracellulare complex in 2020, which was clinically resolved, followed by M. mantenii in 2022.

Among the 20 patients, 6 had disseminated NTMD, all of whom had compromised or deficient immune function, including 4 with adult anti-IFN-γ autoantibody-related immunodeficiency syndrome, one with a history of lymphoma, and one with drug-induced liver damage. Among the 14 biopsies of NTM cutaneous infections, 2 had adult anti-IFN-γ autoantibody-associated immunodeficiency syndrome, and one had diabetes. All 6 patients with anti-IFN-γ syndrome received prednisone for anti-inflammatory treatment, and pathological biopsies were performed prior to the initiation of therapy. The NTMD lesions were primarily painless papules, nodules, or plaques, some merging into larger lesions that exhibited bleeding, crusting, and scarring in chronic cases. A few patients had co-infections with human herpesvirus type 4, human herpesvirus type 5, fine cyclic virus, Pseudomonas aeruginosa, and Candida albicans.

For tuberculosis (TB), 27 cutaneous/mucosal biopsy specimens were obtained from 24 patients, with ages ranging from 8 to 83 years (mean age 40.8, median age 36). This group displayed a wide age distribution, from children to the elderly, also with a male-to-female ratio of 1:1. Three patients underwent cutaneous biopsies from different sites simultaneously. The biopsies from patients with cutaneous manifestations of tuberculosis were collected from the limbs (12), neck (4), chest (2), buttocks (2) and face (1), as well as from other sites including facial cutaneous, the vocal cords, the floor of the mouth, nasal mucosa, and the inner canthus of the eye. The TB cutaneous lesions typically appeared as plaques or nodules, dark red in color, and were often accompanied by desquamation, vesiculobullous eruptions, oozing, ulceration, and crusting.

A comparative analysis of the pathomorphological characteristics of the epidermis, dermis, and subcutaneous adipose tissue was conducted (Supplemental Digital Content Supplementary Figure S1 , which demonstrates gross pathological morphological features of NTMD and TB), focusing on subtle features ( Figures 1 , 2 ). Epidermal characteristics included hyperplasia, ulceration, parakeratosis/hyperkeratosis, granulocytic exudates, plasma exudates, and spongy edema. Dermal and subcutaneous features included interstitial edema, hemorrhage, microvessels hyperplasia, microvessels dilation and congestion, vasculitis, necrosis, and destruction of appendages. Four NTMD biopsies and eight TB biopsies exhibited loss of the epidermis due to ulceration or crusting, while two NTMD biopsies and two TB biopsies from nasal mucosa. Consequently, fewer biopsies were available for evaluation of epidermal-related features. A comparison of the features revealed that NTMD exhibited more microvessels hyperplasia, dilation, congestion, and interstitial edema than TB, although these differences did not reach statistical significance ( Table 2 ). Pseudoepitheliomatous hyperplasia was observed in NTMD (3/27, 11.1%) and TB (2/27, 7.4%), with no significant difference. Epidermal hyperplasia and ulceration were present in both NTMD and TB biopsies, as were parakeratosis/hyperkeratosis, granulocytic exudation, plasma exudation, spongiotic edema, hemorrhage, vasculitis, and destruction of appendages, with no significant differences noted.

A comparison of the pathomorphological features of NTMD from seven different strains revealed distinct findings ( Figure 1 , Supplemental Digital Content Supplementary Table S2 , which demonstrates NTMD pathomorphological characterization of diverse bacterial strains): (i) M. marinum exhibited two distinct histopathological patterns: The first was characterized by well-formed granulomas (300 -500 μm) composed of abundant epithelioid cells, often with central fibrinous necrosis. The second pattern featured marked interstitial edema, hemorrhage, and loosely distributed epithelioid cells without distinct nodular formation, accompanied by peripheral granulation tissue proliferation. (ii) M. abscessus infection consistently demonstrated neutrophilic infiltration. In severe cases, aggregates of neutrophils within newly formed blood vessels were observed. A distinctive feature was the presence of foamy histiocytes with abundant pale cytoplasm, these foamy histiocytes exhibit the distinctive feature of containing faintly basophilic granules and filamentous cellular debris within their cytoplasm. (iii) MAC infection typically demonstrate relatively sparse inflammatory cell infiltration. Histopathological examination reveals only mild neutrophilic infiltration. (iv) A distinctive histopathological feature of M.haemophilum infection is the presence of interface dermatitis. (v) M. avium infection is characterized by prominent, coarse collagen proliferation, resulting in characteristic keloid-like fibrosis. (vi) M. colombiense demonstrated focal interstitial edema, loosely arranged epithelioid cells, hemorrhage, granulation tissue proliferation, and granular necrosis. (vii) M.mantenii presented with small clusters of epithelioid cells (100 μm), diffuse neutrophilic infiltration without distinct abscess formation, and minimal intravascular neutrophil aggregation.

Both NTMD and TB primarily present with granulomatous lesions and nonspecific chronic suppurative inflammation. Immune cell types present include lymphocytes, plasma cells, eosinophils, neutrophils, histiocytes, and multinucleated giant cells. Immune cell infiltration was graded on a scale of 0 to 3, where grade 0 indicates absence, grade 1 denotes focal or scattered infiltration, grade 2 signifies multifocal aggregates and grade 3 indicates small patchy or diffuse infiltration. The histopathological grading of neutrophilic infiltration is categorized as: grade 1 indicates neutrophils are sparsely distributed with low cellular density, grade 2 indicates moderate neutrophil density with formation of small neutrophil foci, grade 3 indicates dense, confluent neutrophilic infiltration with microabscess or abscess formation. A comparative analysis revealed significant differences in the infiltration of lymphocytes, plasma cells, neutrophils, and histiocytes between NTMD and TB, with no significant differences in eosinophils or multinucleated giant cell distribution ( Table 3 ). The number of lymphocytes, plasma cells, and neutrophils was significantly higher in NTMD compared to TB. Neutrophil infiltration was observed in all NTMD biopsies, mostly in small focal distributions (grade 1, 14/27, 51.85%). A few neutrophilic microabscesses (grade 3) were seen in 4 biopsies (4/27, 14.8%), while neutrophils were absent in 29.63% of TB biopsies. Only one case of TB exhibited neutrophilic microabscesses (grade 3). In addition to forming granulomas as epithelioid cells, histiocytes in NTM biopsies could also be seen as foamy histiocytes scattered or patchily distributed in the dermis and subcutaneous tissue (excluding the ulcerated granulation tissue), predominantly in focal distributions (grade 1, 14/27, 51.85%). In TB biopsies, histiocytes were mainly involved in granuloma formation as epithelioid cells, with only a few biopsies (5/27, 18.52%) showing scattered or patchily distributed foamy histiocytes (grade 1) in the stroma (excluding ulcerated granulation tissue).

Granulomatous lesions can be classified as necrotizing or non-necrotizing, based on necrosis presence. Necrotizing granulomas are further categorized into caseous, coagulative, and pyogenic necrosis. NTMD presented with granulomas in 19 of 27 biopsies (70.4%), while TB exhibited granulomas in 26 of 27 biopsies (96.3%), indicating a significantly higher frequency in TB(Supplemental Digital Content Supplementary Table S1 , which demonstrates comparison of granuloma types between NTMD and TB). Both NTMD and TB can have non-necrotizing or necrotizing granulomas; however, NTMD predominantly exhibited non-necrotizing granulomas (11/27, 40.7%) ( Figures 3A–G ), whereas TB predominantly showed caseous necrotizing granulomas (12/27, 44.4%) ( Figures 3H, I ), with NTMD presenting caseous necrotizing granulomas in 14.8% (4/27). Statistically significant differences were found in both the proportion and classification of granulomas between NTMD and TB.

CD68 is a pan-macrophage marker, M1-type macrophages is marked by the surface marker CD86, while M2-type macrophages are characterized by CD163 and CD206. Macrophages were found distributed in the interstitium, present as scattered histiocytes and granuloma epithelioid macrophages. The mean positivity of macrophages for CD68, CD86, CD163, and CD206 in NTMD was 42.5%, 3.0%, 55.4%, and 43.3%, respectively. In TB, the positivity rates were 41.7%, 2.7%, 44.2%, and 39.0%. Both NTMD and TB exhibited high expression of CD68, CD163, and CD206, with CD163 being the most prominently expressed marker. CD86 was either negative or weakly positive (less than 5% positivity).

In NTMD and TB, CD68 and CD206 were primarily expressed by granuloma epithelioid macrophages, while CD163 was expressed by interstitial scattered macrophages ( Figure 4 ). A positive correlation was noted between CD68 and CD206 positivity, as well as between CD86 and CD163 in NTMD, reaching statistical significance ( Figure 5 ). No significant correlations were found in TB macrophages (Supplemental Digital Content Supplementary Figure S2 , which demonstrates statistical analysis of the correlation between the positive rates of CD68, CD86, CD163, and CD206 in macrophages in TB).

In both NTMD and TB, PD-L1 was expressed by granuloma epithelial cells, lymphocytes, and multinucleated giant cells. Granuloma epithelial cells predominantly expressed PD-L1, whereas lymphocytes and multinucleated giant cells showed low or absent PD-L1 expression ( Figure 6 ). In NTMD, mononuclear inflammatory cells exhibited a PD-L1 positivity rate of 0% to 40%, with a mean positivity of 6.4%, which was higher in biopsies of M. avium and M. marinum. In TB, PD-L1 positivity ranged from 0% to 60% in mononuclear inflammatory cells, with a mean positivity of 14.3%. A higher percentage of mononuclear inflammatory cells in TB (45.8%) demonstrated high PD-L1 expression (≥10%) compared to NTMD (25%), though no significant difference was noted between the two groups (Supplemental Digital Content Supplementary Table S3 , which demonstrates PD-L1 expression of immune cells in NTMD and TB).