Talaromyces marneffei and nontuberculous mycobacteria co-infection in HIV-negative patients

To describe the clinical features and the risk factors for nontuberculous mycobacteria (NTM) and Talaromyces marneffei (TM) co-infections in HIV-negative patients. A multicenter retrospective study in 13 hospitals, and a systematic literature review were performed of original articles published in English related to TM/NTM co-infections. HIV-negative patients with TM and NTM co-infections comprised Group 1; TM-only infection Group 2; NTM-only infection Group 3; and healthy volunteers Group 4. Univariate logistic analysis was used to estimate the potential risk factors of TM/NTM co-infections. A total of 22 cases of TM and NTM co-infections were enrolled. Of these, 17 patients (77.3%) had a missed diagnosis of one of the TM or NTM pathogens. The anti-IFN-γ autoantibodies (AIGAs) titer, white blood cell (WBC), neutrophil counts (N), erythrocyte sedimentation rate (ESR), C reactive protein (CRP), globulin, and immunoglobulin G (IgG) levels of Group 1 were higher than those of the other groups, whereas the levels of CD4+T cells was lower than those of other groups. There was a significant negative correlation between the AIGA titers and the number of CD4+T cells (P < 0.05). Factors including the ratio of the actual values to the cut-off values of AIGAs, WBC, N, HGB, CD4+T cells, IgG, IgM, IgA, serum globulin, ESR, and CRP were taken as potential risk factors for TM and NTM co-infection. Most patients with TM and NTM co-infection had a missed diagnosis of one of the TM or NTM pathogens. The levels of AIGAs, WBC, N, ESR, and CRP in TM and NTM co-infections were remarkably higher than in mono-infection. High-titer AIGAs may be a potential risk factor and susceptibility factor for co-infection of TM and NTM in HIV-negative hosts.

Laboratory findings and clinical features. Laboratory findings are shown in Table 1 and Fig. 1. Routine bloodwork including, erythrocyte sedimentation rate (ESR), C reactive protein (CRP) lymphocyte phenotyping, Table 1. Baseline demographics and clinical characteristics of the 106 participants. Bold values indicate significant difference between groups or in univariate logistic regression analysis. a Indicates statistical significance between Groups 1 and 2. b Indicates statistical significance between Groups 1 and 3. c Indicates statistical significance between Groups 1 and 4. d Indicates statistical significance between Groups 2 and 3. e Indicates statistical significance between Groups 2 and 4. f Indicates statistical significance between Groups 3 and 4. g A total number of 14 patients in Group 1 had AIGAs titer, WBC, N, L, HGB, ESR, CRP, globulin, immunoglobulins (IgG, IgA, IgM), and lymphocytes subpopulations (CD4 + T cell, CD8 + T cell, CD3 + T cell) data. Data are expressed as median ± interquartile range. Kruskal-Wallis H test was used to determine statistical significance among the 3 or 4 groups, followed by a 2 by 2 comparison across groups through a Fisher's exact test. P < 0.05 indicates statistical significance. Group 1 = patients with TM and NTM co-infections; Group 2 = patients with TM mono-infection; Group 3 = patients with NTM mono-infection; Group 4 = healthy control volunteers. *Indicates the nature of the underlying disease in three groups. Group 1: 5 cases with Sweet's syndrome, 1 case with malignant tumor, 1 case with cystic fibrosis, 1 case with Behcet's syndrome, and 1 case with diabetes; Group 2: 1 case with thalassemia, 1 case with Sjogren's syndrome, 1 case with ankylosing spondylitis, 1 case with major trauma or surgery, 1 case with hyperthyroidism, 2 cases with glucocorticoids and or immunosuppressive agents, 1 case with hypertension, and 1 case of diabetes. Group 3: 3 cases with major trauma or surgery, 3 cases with hypertension, and 1 case with diabetes. **Serums from 14 participants in Group 1, all patients in Groups 2 and 3, and 40 health volunteers were tested for anti-IFN-γ autoantibodies. Six of eight patients in the literature review cohort in Group 1 were defined as AIGA-positive, while the last 2 patients were not assessed. Thus, a total of 20 patients were tested for AIGAs in Group 1. BMI body mass index, AIGAs anti-IFN-γ auto-antibodies, ND no data, WBC white blood cell, N neutrophil counts, L lymphocyte counts, HGB haemoglobin, ESR erythrocyte sedimentation rate, CRP C-reactive protein, Ig immunoglobulin. Normal range: IgG, 8-18 g/L; IgA, 2.01-2.69 g/L; IgM, 0.84-1.32 g/L; CD4 + T cell, 410-1590 cells/μL; CD8 + T cell, 190-1140 cells/μL; CD3 + T cell, 690-2540 cells/μL. www.nature.com/scientificreports/ and serum immunoglobulin G (IgG)] were performed for 14 patients from the retrospective study, and were not available in patients from the literature review cohort. White blood cell (WBC), neutrophil counts (N), ESR, and CRP in Groups 1 and 2 were significantly higher than in Group 3 (P < 0.001). Hemoglobin (HGB) in Groups 1 and 2 were lower than in Group 3 (P < 0.05). Globulin, IgG, and IgM levels of Group 1 were higher than those of the other groups. CD4 + T and CD3 + T lymphocyte counts in Group 1 were lower than normal reference values and that in Groups 2 and 3, respectively (P < 0.05) ( Table 1, Fig. 1A-C). Significant differences in clinical manifestations were found (P = 0.002) through Chi-square statistical tests for all the clinical manifestations between Groups 1, 2, and 3 in the three groups ( Table 2). The most common clinical features in Group 1 were lymphadenopathy, fever, and cutaneous lesions, followed by cough, weight loss, and ostealgia. Fever, lymphadenopathy, and ostealgia were more common in Groups 1 and 2. However, weight loss and cough were more common in Group 3. Chest high resolution computed tomography (HRCT) was also conducted in the three groups (Table 2), showing significant differences in the prevalence of mediastinal lymphadenopathy, fibrous cords, pleural effusion and/or pleural thickening, bronchiectasis, and cavitary lesions. Mediastinal lymphadenopathy was more common in Groups 1 and 2, whereas fibrous cord, cavitary lesions, and bronchiectasis were more common in Group 3.
These indexes were also compared between Groups 1A and 1B, which showed no significant differences (Supplementary Table 1). However, the CD4 + T lymphocytes and CD3 + T lymphocytes in Group 1A were lower than in Group 1B. In addition, patients in Group 1A receiving combined treatment (anti-fungal with anti-NTM treatment) showed a significant decrease in the inflammatory indexes (WBC, N, ESR, and CRP). By contrast, in Group 1B, the inflammatory indexes did not decrease, but rather increased following a single regimen therapy (anti-fungal or anti-NTM treatment). However, upon identifying the second pathogen and providing combined treatment, these inflammatory indexes, symptoms, and signs in patients improved (Fig. 1D, E, Supplementary Table 2).  Table 3, Fig. 2A), lymph nodes, skin, and bone/joints were the most commonly infected sites in Groups 1 and 2. Lung involvement was more common in Groups 2 and 3, with pleural as the most commonly involved site in Group 2. Microbiology and pathology in patients with concomitant or sequential infections by TM and NTM. TM was most commonly isolated from respiratory specimens (14 cases), including bronchoalveolar lavage fluid (BALF) (7 cases), sputum (5 cases), and lung tissue (2 cases), followed by blood (5 cases), purulent Table 2. Symptoms and imaging findings of Chest HRCT in three groups. Bold values indicate significant difference between groups or in univariate logistic regression analysis. a Indicates statistical significance between Groups 1 and 2. b Indicates statistical significance between Groups 1 and 3. c Indicates statistical significance between Groups 2 and 3. Data are presented as n (%). Fisher's exact test and Kruskal-Wallis H test were used to calculate P-values. P < 0.05. Group 1 = patients with TM and NTM co-infections, Group 2 = patients with TM infections only, Group 3 = patients with NTM infections only. HRCT high resolution computed tomography. *Two patients from the systematic literature review did not undergo HRCT. Thus, a total of 20 patients received HRCT.  www.nature.com/scientificreports/ secretion (4 cases), and lymph nodes (3 cases) in Group 1. By contrast, NTM was most commonly isolated from lymph nodes (7 cases), sputum (4 cases), and blood (4 cases) in Group 1. Granulomatous lesions (12 cases), followed by non-specific inflammation (11 cases) and suppurative lesions (7 cases), were the most common histological findings in 27 pathological specimens from Group 1. Further, positive PAS staining (40.7%) of tissues and secretions were more frequent than acid-fast (AFB) staining (14.8%) in Group 1.
The distribution of rapid-and slow-growing nontuberculous mycobacterial species was similar in Group 1 (Supplementary Table 4). In these patients, the most commonly isolated species was Mycobacterium abscessus (4/11, 36.4%), followed by Mycobacterium chelonae and Mycobacterium kansasii (3/11, 27.6%). In addition to TM and NTM, other common co-infecting pathogens in Group 1 were Staphylococcus aureus, Aspergillus, Salmonella, and Burkholderia. Moreover, one patient was infected by up to six pathogens during the course of disease.

Increased AIGA levels in TM and NTM co-infection. Serums obtained from 14 participants in Group
1, all patients in Groups 2 and 3 (n = 22 in each group), and 40 health volunteers were tested for AIGAs. Furthermore, six of the eight patients in the literature review cohort in Group 1 were defined as AIGA-positive, the other 2 patients were not assessed. The positivity rate of AIGAs was significantly different across groups, specifically 100% (20/20), 81.8%, and 63.6% in Groups 1, 2, and 3, respectively (P = 0.000). When comparing AIGA titers between the groups, Groups 1, 2, and 3 were remarkably higher than the healthy volunteer group, with Group 1 showing the highest AIGA titer (Table 1, Fig. 2B). Meanwhile, there was a significant negative correlation between AIGA titers and the number of CD4 + T cells (P < 0.05, Fig. 2C).

Univariate analysis logistic regression analyses for risk factors of TM and NTM co-infections.
We analyzed risk factors for developing TM/NTM coinfections in group 1 compared to groups 2 and 3. We found that factors including the ratio of the actual values to the cut-off values of AIGAs, WBC, N, HGB, CD4 + T cells, IgG, IgM, IgA, serum globulin, ESR, and CRP were taken as potential risk factors for TM and NTM co-infection (Table 3).
Treatment and outcome. The prognosis and outcomes of patients in Group 1 was worse than that of patients in Groups 2 and 3, especially in cases of persistent and/or relapsed infections (P < 0.001) ( Table 4).
Treatment outcomes are presented in Table 3 among 22 patients: 19 received anti-NTM medical treatment and 22 received anti-fungal treatment. Furthermore, 1 case was lost to follow-up, 1 died from multiple organ failure, 7 were effectively cured of both TM and NTM), 9 relapsed, and 6 had persistent infection. Of the 13 patients with positive AIGA, only 1 patient (P17) received AIGA treatment. Upon receiving combined methylprednisolone and rituximab treatment, the AIGA titer of P17 decreased from more than 1: 10,000 to 1: 5000 after 2 courses of therapy. The total treatment time, including anti-fungal and anti-NTM, was 40 months (6-114 months) ( Table 5).

Discussion
To our knowledge, this is the first report showing the differences between TM and NTM co-infection and their respective mono-infections. Some clinical differences were noticed across groups. The severity of inflammation (WBC, N, ESR, CRP), inflammatory anemia, and prevalence of involved sites in TM and NTM co-infection were more evident than in TM or NTM mono-infection, especially when compared. Noteworthy, when patients received single active antifungal or single anti-NTM treatment, some symptoms improved while others worsened. Inflammatory markers (WBC, N, CRP, ESR) did not significantly decline or increase, but did not maintain normal levels, indicating the presence of double or multiple infections, especially in patients with high-titer AIGAs. Univariate analysis for risk factors of TM and NTM co-infection found that high level of AIGAs, WBC, N, HGB, IgG, IgM, IgA, serum globulin, ESR, and CRP and low level of CD4 + T cells were taken as potential risk factors for TM and NTM co-infection. Most importantly, the titer of AIGAs was significantly positively correlated with the number of sites involved, which suggested that the titer of AIGAs was associated with disseminated infection. Thus, high-titer AIGAs may represent a potential risk factor and susceptibility factor for co-infection of TM and NTM in HIV-negative hosts. Monitoring the AIGA titer is the most important step in screening for co-infections or disseminated infections.
IFN-γ is produced principally by T lymphocytes and natural killer cells after stimulation with microbial products and interleukin (IL)-12 20 . Patients with positive AIGAs often suffer from recurrent infections, especially due to NTM 8,9,11 . Because IFN-γ is an activator of macrophage differentiation and a pro-inflammatory activator of innate immunity, the blockade effects of the AIGAs on IFN-γ present in the serum of patients with NTM are hypothesized to regulate the antimicrobial function of macrophages 20 . Recently, a study showed that AIGAs can neutralize IFN-γ, affect the activation of the IFN-γ receptor (IFN-γR), and downregulate the production of its downstream factors, such as TNF-α and IL-12, and inhibit IFN γ-STAT-1 phosphorylation 11 . IFN-γ is also an essential activator of CD4 + T cell differentiation into Th1 cells 21 . In the present study, the AIGA titers and positive rates of patients with co-infection were significantly higher than those of other groups, while their CD4 + T and CD3 + T cell levels were significantly lower than those of other groups. Meanwhile, there was a significant negative correlation between AIGA titers and the number of CD4 + T cells. Thus, the neutralizing and blockade effects of the AIGAs may be related to the low level of CD4 + T cells, which may be the reason for patients susceptible to opportunistic pathogens, especially intracellular pathogens.
TM and NTM showed very similar clinical manifestations such as fever, anemia, weight loss, cough, expectoration, and skin lesions. They both can involve skin lesions, respiratory system, and bone, leading to local or disseminated infections. High recurrence and/or persistent infection rates (59.1%) was found in TM and NTM co-infected patients, primarily due to misdiagnosis and/or missed diagnoses as each other or TB. In HIV-negative individuals with TM and NTM co-infection, only one pathogen (TM or NTM) was discovered in the early stages of disease in most patients (77.3%). Moreover, inflammatory markers in TM and NTM co-infection were higher than in NTM mono-infection, though no significant difference was found between simultaneous and successive TM and NTM. These suggest that most patients found to have sequential TM and NTM infections were in fact infected with both TM and NTM simultaneously; however, one pathogen was missed at diagnosis, resulting in poor prognosis.
Furthermore, TM histopathology often manifests as granuloma, but caseous granuloma is rare, which was characteristic of positive TM cultures in this study. Second, it is more difficult to make a differential diagnosis of NTM from TB because of its similar histopathology and acid-fast staining. Thus, even if it is positive for acid-fast staining, metagenomic next-generation sequencing and culture of mycobacteria is essential to detect NTM, especially when anti-tuberculosis treatment is not effective. Third, TM and NTM co-infection has a higher inflammatory index and dissemination than NTM infection, which may be related to AIGAs and TM.

Conclusion
High-titer AIGAs represent an independent risk factor for TM and NTM co-infection in HIV-negative hosts. AIGA may be a major susceptibility factor for intracellular pathogens such as TM and NTM. Further, poor prognosis of TM and NTM co-infection may be due to misdiagnosis and/or missed diagnoses. Therefore, AIGA patients the outcome assessment was performed at their last outpatient follow-up. For 8 patients that were part of the systematic literature review, their outcome assessment was extracted from the literature. For the following patients, the duration between the time the treatment was stopped, and the outcome assessment was respectively: 12 months for P15; 19 months for P16; 10 months for P18; 12 months for P19; and 6 months for P22. For the following patients, the outcome assessment time was performed when they were discharged: P17, P21, and P21. Systematic literature review cohort. For a systematic review of articles related to TM/NTM co-infection, original articles published in English from Jan 2004 to July 2019 were reviewed using the following electronic databases: PubMed, Web of Science, Embase, and BIOSIS. Screening of relevant studies was based on combinations of keywords, such as "non-tuberculosis", "non-tuberculous", "non tuberculosis", "nontuberculous", "nontuberculous mycobacterium", "nontuberculosis mycobacteria", "NTM", "MOTT", "atypical mycobacterium", "penicilliosis", "Penicillium marneffei", "Talaromycosis", "Talaromyces marneffei", "T. marneffei", and "P. The data presented in this study result from a merge of these 2 cohorts (Guangxi cohort and literature review cohort). Clinical outcomes definitions: (1) Cured (no recurrence of TM and/or NTM infection for at least six months after discontinuation of antifungal/anti-NTM therapy); (2) persistent or relapsed infection (persistent infection: no improvement of clinical symptoms after antifungal/anti-NTM treatment, relapsed infection: improvement of clinical symptoms, negative pathogen detection after antifungal/anti-NTM effective treatment, followed by the reappearance of pathogen-associated infectious signs and/or positive pathogen testing); and (3) death. A disseminated disease was defined as an infection in at least two noncontiguous and sterile sites.
Diagnostic criteria for NTM and TM. Each patient fulfilled the diagnostic criteria of each disease. NTM was diagnosed following the 2007 American Thoracic Society (ATS)/Infectious Disease Society of America guidelines 22,23 . TM infection was diagnosed as follows: (1) positive cultures for TM, characterized by dimorphic fungi that grew either as a mold at 25 °C or as yeast at 37 °C; (2) characteristic morphology of the yeast form of TM, confirmed by cytology and histopathology from tissues and secretions using Periodic Acid-Schiff (PAS) staining or Wright's stain, including a transverse septum 23 ; or (3) TM and/or NTM isolated by metagenomic next-generation sequencing from clinical specimens.
Anti-IFN-γ autoantibody assay. Serum samples obtained under sterile conditions before the patient received antimicrobial therapy treatment and during the active stage of the infection. Serum samples were retrieved from a serum bank and stored at − 80 °C. AIGAs were detected in all participants. All serum samples were tested at the first thaw. The detection of AIGAs was performed using an enzyme-linked immunosorbent assay kit (Cloud-Clone Corp. Wuhan, China) whose detection range is 12-200 ng/ml. According to the manufacturer's protocols: the serum samples from patients were 1:1500 diluted, and serum samples from a healthy control were 1:600 diluted by phosphate-buffered saline (PBS). The normal range for the anti-IFN-γ-autoantibody concentration was defined by the 99th percentile for the healthy controls and was estimated using the log-normal distribution. Outlying concentrations were classified as positive for anti-IFN-γ autoantibodies 1,6 . www.nature.com/scientificreports/ IFN-γ, IL-4, IL-6, IL-8, TNF-α assay. Serum samples obtained under sterile conditions before the patient received antimicrobial therapy treatment and during the active stage of the infection. Serum samples were retrieved from a serum bank and stored at − 80 °C. IFN-γ, IL-4, IL-6, IL-8, TNF-α were detected in all participants. All serum samples were tested at the first thaw. The detection of IFN-γ, IL-4, IL-6, IL-8, TNF-α was performed using a human enzyme-linked immunosorbent assay kit (Cloud-Clone Corp. Wuhan, China) according to the manufacturer's instructions.
Statistical analysis. Continuous variables were expressed as median ± interquartile range. Differences between groups were compared using Kruskal-Wallis H or Mann-Whitney U tests. Dunn-Bonferroni test was used for post-hoc comparisons. Chi-square or Fisher's exact tests were used to compare categorical variables. Spearman's correlation coefficient was used for ranked data to measure the dependence of two nonparametric variables. Univariate logistic analysis was used to estimate risk factors of co-infection. We used SPSS (version 25.0), and GraphPad Prism (version 7) for statistical analysis and graph illustrations, and a two-sided P-value of 0.05 or less was considered significant.
Ethical approval. This study was approved by the Ethical Review Committee of the First Affiliated Hospital of Guangxi Medical University (2018.KY-E-094). The clinical trial was registered on www. clini caltr ials. gov (NCT03819348). Written informed consent was provided by all participants in the prospective cohort study.
Consent to participate. All study participants provided informed consent, and the study design was approved by the appropriate ethics review board.

Consent for publication.
Written informed consent for publication was obtained from all participants.

Data availability
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.