Introduction

Granuloma formation is a pathological hallmark of mycobacterial infection. Acid-fast bacillus staining methods, such as Ziehl–Neelsen staining, are useful for diagnosing mycobacterial lymphadenitis, but unfortunately have low detection rates [1, 2]. In the case of mycobacterial lymphadenitis negative for acid-fast bacillus staining, the presence of granulomatous lesions on pathological examination becomes even more important. However, insufficient granuloma formation may be observed in an immunodeficient state [3].

Anti-interferon-gamma (IFN-γ) autoantibodies have been recognized as a cause of adult-onset immunodeficiency for mycobacterial infection, especially nontuberculous mycobacterial infections [4, 5]. Here, we report a case of disseminated Mycobacterium abscessus infection with underlying adult-onset immunodeficiency caused by anti-IFN-γ antibodies. Mycobacterial lymphadenitis without granuloma formation was observed, which occurred due to the immunodeficiency associated with anti-IFN-γ antibodies. The discrepancy in mitogen response between QuantiFERON (QFT) and T-SPOT assays aided the diagnosis of anti-IFN-γ antibodies in this case.

Case presentation

A previously healthy 49-year-old Japanese woman presented with a 3-month history of swelling and tenderness of the left cervical lymph nodes. Associated symptoms included low-grade fever and fatigue. Laboratory investigations revealed elevated inflammatory markers, including C-reactive protein (4.27 mg/dL) and serum soluble interleukin-2 receptor (1564 U/mL). Liver and renal function tests were unremarkable, except for albumin levels, which were lower than normal (1.9 g/dL). Human immunodeficiency virus (HIV) antibody test was negative. As we initially suspected infectious inflammatory lymphadenitis, we began administering antibiotics, including garenoxacin, minocycline, and clarithromycin, but the patient’s clinical symptoms worsened. Positron emission tomography-computed tomography (PET-CT) revealed pathological 18F-fluorodeoxyglucose (18F-FDG) uptake in the left cervical, left supraclavicular, bilateral axillary, left hilar, mediastinal, and left cardiac lymph nodes. It also detected bone lesions, mainly in the sternum (Fig. 1). Therefore, we performed cervical lymph node biopsy to exclude malignant lymphoma. The biopsy tissue was evaluated by pathological analysis of frozen sections, immunohistochemistry, flow cytometry, fluorescence in situ hybridization (FISH), and chromosomal analysis. Staining of sections with hematoxylin and eosin showed reactive lymphadenitis with abscess formation. No granulomatous lesions were present in the lymph nodes (Fig. 2) and Ziehl–Neelsen staining revealed no acid-fast bacilli. No evidence of malignant lymphoma was found on immunohistochemical staining, flow cytometry, FISH, or chromosomal analysis. Blood, urine, and sputum cultures were all negative for bacteria, fungi, and mycobacteria. Based on these initial results, we made a diagnosis of nonspecific inflammatory lymphadenitis. However, despite continued intravenous administration of antibiotics, including levofloxacin and cefepime, the lymphadenopathy and fever worsened.

Fig. 1
figure 1

a Positron emission tomography-computed tomography demonstrated 18F-fluorodeoxyglucose uptake by multiple lymph nodes and bone lesions. b Cervical lymph node. c Axillary lymph node and bone lesion (in the sternum). d Hilar lymph nodes and bone lesion (in a vertebra)

Full size image
Fig. 2
figure 2

Lymph node biopsy showed lymphadenitis with abscess formation. No granuloma was found in the lymph node (hematoxylin and eosin staining)

Full size image

As PET-CT showed 18F-FDG uptake in the sternum (among other bones) rather than the lymph nodes, we attempted sternal bone marrow aspiration at the location of 18F-FDG accumulation. However, despite several attempts, we failed to obtain bone marrow cells. Such an outcome is referred to as a “dry tap” and suggests a pathological process. The clinical findings were consistent with osteomyelitis, and we considered this to be the cause of 18F-FDG uptake in the bone [6]. Osteomyelitis is usually caused by bacterial infection, although in rare cases it may be caused by fungal or mycobacterial infection in immunosuppressed patients [7,8,9]. Osteomyelitis is known to be an occasional complication of tuberculosis [10]. Recently, multifocal osteomyelitis after BCG vaccination has been reported in people with Mendelian susceptibility to mycobacterial diseases (MSMD), especially in cases of IFN-γR1 deficiency [11, 12]. To test for Mycobacterium tuberculosis (MTB) infection in our patient, we performed IFN-γ release assays (IGRAs) with QFT-TB Gold Plus (QFT-Plus) and T-SPOT. The IFN-γ level in the QFT-Plus mitogen control was extremely low (enzyme-linked immunosorbent assay [ELISA] value = 0.09 IU/ml), but high in the T-SPOT mitogen control (> 20 spots) (Table 1). This discrepancy strongly suggested that IFN-γ was being inhibited by factors in the patient’s plasma. Therefore, we performed a repeat cervical lymph node biopsy, followed by bacterial and mycobacterial culture test on the tissue. The histopathological finding was identical to the initial lymph node biopsy, and no granuloma formation was evident. Polymerase chain reaction tests of the lymph node for Mycobacterium tuberculosis, M. avium, and M. intracellulare were all negative, but Mycobacterium abscessus was isolated from the mycobacterial culture. Based on these findings, we made a diagnosis of disseminated nontuberculous mycobacterial infection (DNTM) caused by M. abscessus infection. The presence of anti-IFN-γ antibodies was confirmed by analysis of IFN-γ-neutralizing antibody titer and IFN-γ-neutralizing activity at Niigata University Medical and Dental Hospital [13]. A high titer of neutralizing autoantibodies to IFN-γ (concentration of anti-IFN-γ: 141.5 E.U., control 22.9 E.U.; STAT1 phosphorylation index: − 6.9, control: 396) was identified in the patient’s serum by ELISA and a flow cytometry-based method. The final diagnosis was disseminated M. abscessus infection with underlying adult-onset immunodeficiency caused by anti-IFN-γ antibodies.

Table 1 The discrepancy in mitogen response between QuantiFERON-TB Gold Plus (QFT-Plus) and T-SPOT.TB (T-SPOT) in a female patient with disseminated Mycobacterium abscessus infection
Full size table

The patient was treated for the disseminated M. abscessus infection with combination chemotherapy consisting of meropenem (1000 mg/day), amikacin (600 mg/day), and clarithromycin (800 mg/day). One month after initiation of antimycobacterial treatment, the patient was afebrile and her lymphadenopathy had shrunk markedly. Treatment was continued on an outpatient basis with daily clarithromycin and an infusion of meropenem and amikacin three times a week. Three months later, PET-CT showed no abnormal 18F-FDG accumulation. Antimicrobial treatment was continued for 1 year. At the time of writing, no regression was observed and the patient’s condition remained stable for at least 6 months after the antimicrobial treatment was discontinued.

Discussion

DNTM mostly occurs as an opportunistic infection in immunosuppressed individuals, including those with advanced HIV infection [14]. Anti-IFN-γ antibodies have recently been reported to affect the IFN-γ immune pathway and cause adult-onset immunodeficiency syndrome in DNTM and other opportunistic infections, especially in East Asia [4, 5, 15,16,17]. The course of such infections in patients with this syndrome can be severe and progressive despite prolonged antimicrobial therapy, and treatment may fail [18, 19]. Anti-IFN-γ antibodies are detected in about 80% of DNTM patients without obvious immunodeficiency, such as HIV infection [4, 5]. Therefore, if a previously healthy person develops DNTM, the majority of cases are considered to have acquired immunodeficiency caused by anti-IFN-γ antibodies. The commonly involved organs include the lymph nodes (79%), bone/joints lesions (34%), and lungs (32%) [20]. As DNTM in HIV infection rarely causes bone lesions, the presence of bone lesions may be suggestive of an association with anti-IFN-γ antibodies [20,21,22]. For patients with both lymphadenopathy and bone involvement, DNTM-associated anti-IFN-γ antibodies is one of the most important differential diagnoses. In our case, focusing on the presence of bone lesions provided insight that allowed us to reach the correct diagnosis.

Granulomas are a pathological hallmark of mycobacterial infection. Their main function is to localize and contain mycobacteria, while limiting the immune response to a relatively small area. Therefore, granulomas formation may be insufficient in an immunodeficient state [3]. As anti-IFN-γ antibodies cause an immunodeficient state specific to mycobacteria, it is possible that granuloma formation may be insufficient in this disease. In fact, a previous case report described mycobacterial infections in which granuloma formation was not observed [23]. However, this previous case report was published before anti-IFN-γ antibodies became widely known, and this case may have been associated with anti-IFN-γ antibodies. It is very difficult to diagnose mycobacterial infection without granuloma formation on pathological examination without performing mycobacterial culture. A number of DNTM cases with insufficient granuloma formation may have been overlooked.

If the presence of anti-IFN-γ antibodies is confirmed, mycobacterial infection should be suspected even in the absence of granuloma formation. Recently, the QFT IGRA, which has been used extensively for in vitro diagnosis of MTB infection, has been repurposed for screening of patients with suspected anti-IFN-γ antibodies [24, 25]. IGRAs are blood tests that detect the secretion of IFN-γ ex vivo by assessing lymphocytes after stimulation with two or three proteins that are fairly specific for MTB. IGRAs using the M. tuberculosis-specific antigens ESAT-6 and CFP-10 have high specificity for the diagnosis of tuberculosis and rarely yield false positive for most nontuberculous mycobacterial infections, including M. abscessus [26, 27]. However a small number of nontuberculous mycobacterial infections, such as M.kansasii, M. marinum, and M. szulgai, which express ESAT-6 and CFP-10, can cause positive IGRA results [28]. Currently, two types of FDA-approved IGRA are available: the QFT assay (Qiagen) and T-SPOT assay (Oxford Immunotec). The QFT assay measures the concentration of IFN-γ by ELISA, whereas T-SPOT measures the number of IFN-γ-secreting lymphocytes by enzyme-linked immunospot (ELISPOT) assay [29, 30]. In addition to measuring IFN-γ in the antigen-simulated sample, IGRAs also measure IFN-γ levels in positive and negative controls. The negative control contains no additives and is used to adjust for background IFN-γ. In the positive control, mitogen, which is a strong, nonspecific T cell stimulant, is used to stimulate lymphocytes, thus controlling for IFN-γ release that is unrelated to the tuberculosis antigen. A lack of IFN-γ in the positive control usually implies that the patient’s T cells do not respond to nonspecific IFN-γ-releasing antigens. As IGRAs assess IFN-γ release, a common explanation for a negative result in the positive control is an immunocompromised state, for example due to corticosteroid treatment, HIV infection, or cancer [31]. The QFT assay quantifies the IFN-γ released from sensitized lymphocytes in whole blood containing the patient’s plasma incubated with MTB antigens. Therefore, in addition to an immunocompromised state, a lack of IFN-γ in the positive control in the QFT assay may also indicate the presence of anti-IFN-γ antibodies. This cannot be the case in the T-SPOT assay, as it counts the number of IFN-γ-secreting cells, which are peripheral blood mononuclear cells separated from whole blood by centrifugation, a process that removes the patient’s plasma. Immunocompromised patients who do not have a positive control in the QFT assay are likely to not have a positive control in the T-SPOT assay. Therefore, it is recommended that the QFT and T-SPOT assays be used together to test for anti-IFN-γ antibodies. T-SPOT assay alone does not indicate the presence of anti-IFN-γ antibodies, and even if the QFT alone suggests the possibility of anti-IFN-γ antibodies, it may be erroneously considered to be due to an immunodeficient state. Thus, a low mitogen response in the QFT assay together with an adequate mitogen response in the T-SPOT assay, as in the case presented here, strongly suggests the presence of anti-IFN-γ antibodies. In our patient, we confirmed the diagnosis based on IFN-γ-neutralizing autoantibody titer and analysis of IFN-γ-neutralizing activity, which are specialized tests that can only be performed at specialist facilities [13]. The combination of QFT and T-SPOT allows the presence of anti-IFN-γ antibodies to be confirmed using these widely available tests.

In conclusion, granulomas are a pathological hallmark of mycobacterial infection, but may not fully form in an immunodeficient state. Clinicians should be aware of the possibility of mycobacterial infections without granuloma formation due to anti-IFN-γ antibodies. Diagnosis of these cases is very challenging but the clinical findings of lymphadenitis with bone involvement, and the discrepancy in mitogen response between QFT and T-SPOT assays, are helpful.