The development of immune checkpoint inhibitors (ICBs) has radically changed the way that numerous cancers are treated. After the initial approval of ipilimumab (a monoclonal antibody blocking CTLA-4) for the treatment of metastatic melanoma, five more antibodies that target PD-1/PD-L1 pathways, including nivolumab and pembrolizumab (against PD-1) and atezolizumab, avelumab, and durvalumab (against PD-L1), have been licensed and introduced into the therapeutic algorithms for various malignancies in first and later lines of treatment, as well as in the neoadjuvant and adjuvant settings [1]. Cancer cells are able to evade host immune surveillance and escape tumor neutralization by inhibiting PD-1 targeted cancer-specific T cells via overexpression of PD-L1 [2]. The monoclonal antibodies prevents the binding of PD-1 to its ligand PD-L1, restores the T cell-mediated cytotoxicity and allows the immune natural defense to fight against cancer with significant clinical benefits [3]. However, the immune stimulation triggered by these drugs may lead to severe and even life-threatening, albeit infrequent, immune-related adverse events (irAEs) involving almost every organ [4, 5]. Current guidelines on the management of irAEs recommend the prompt administration of high-dose corticosteroids and if toxicity persists, further immunosuppression with steroid-sparing regimens (e.g. anti-tumor necrosis factor-alpha (TNF-a) agents or mycophenolate mofetil) [6].

In this context, the additional modulation of immune response due to the cancer itself, due to the administration of ICPIs or the supplementary medications (e.g. steroids or anti-TNF agents) for overcoming irAEs may unmask chronic underlying or opportunistic infections and rarely cause some serious infectious complications such as varicella-zoster virus infection, cytomegalovirus-associated enterocolitis, pulmonary aspergillosis, pneumocystis pneumonia and reactivation of latent tuberculosis with detrimental, in some cases, effects on cancer treatment outcome and patient’s survival [7, 8]. The overall incidence of these serious immune-related infections in 740 patients with metastatic melanoma receiving ICBs was estimated to be 7.3% in a recent review, where infectious complications were detected mainly in patients who required corticosteroids and/or TNF-a inhibitors [9].

Given the high incidence of Mycobacterium tuberculosis (MTB) infection worldwide and the poor prognosis of MTB reactivation, a renewed interest was developed to recognize individuals at high risk that should be screened for early detection of latent tuberculosis and treated to prevent active disease [10, 11]. The Center for Disease Control and Prevention (CDC), the World Health Organization (WHO), and the US Preventive Services Task Force (USPSTF) agree that the risk of exposure to MTB is higher: a) in patients living or working in endemic countries (e.g. East Asia and Central America) and b) in patients living in large group settings (e.g. homeless or military shelters and prisons). In most patients infected with MTB, the disease remains clinically asymptomatic and inactive, however, in 5–10% of them, the infection will reactivate at some point during their lifetime with a baseline risk between 6 and 20 per 100,000 person-years [12]. After that, the risk of reactivation depends on the specific type of immunosuppression [11, 13]. Compared to the general population, this risk is greater among solid organ transplant recipients (15xfold) [14] and stem cell transplant recipients (8-12xfold) [15], followed by patients treated with anti-TNF medications (5-7xfold) [16–19], while in patients with HIV infection, it reaches 50 times higher and causes up to 25% of deaths among patients [20]. Other host factors that may increase the susceptibility to develop active tuberculosis include older age (> 60 years), prior tuberculosis history, chronic obstructive pulmonary disease, heavy smoking or increased alcohol consumption, diabetes mellitus or end-stage renal disease and for these patients screening is also recommended [13, 21–23]. Cancer has been recognized as an independent risk factor for developing active MTB infection since the 1970s, however this risk widely varies among cancer types, is differentially affected by modern therapies (targeted agents and monoclonal antibodies) and remains to be precisely quantified.

In this study, we present two melanoma patients who developed active tuberculosis during their treatment with PD-1/PD-L1 blockade in our department. In view of these two cases, we review the literature from the preclinical data on the immune-mediated interactions of PD-1/PD-L1 inhibition and co-existent tuberculosis, and published clinical reports with ICB-associated tuberculosis. Integrating the current evidence with our institutional experience, we address questions about which cancer patients are at higher risk for MTB infection, whether ICB-treated cases should be still considered immunocompromised, and how they should be managed for latent or active tuberculosis.

A 76-year-old Greek woman was diagnosed with a cutaneous melanoma lesion of her left lower leg in August 2009 (Fig. 1). Her comorbidities included smoking of 45 pack*years, hypertension, dyslipidemia, coronary artery disease and osteopenia. She underwent a radical resection of the tumour, but the sentinel lymph node was grossly infiltrated (stage IIIb, T3aN1aM0), and she received interferon (IFN) 20,000 iu/m² every day during December 2009, according to contemporary recommendations. She remained disease free until July 2017 when she developed a new cutaneous lesion of her left calf (M1a, stage IV). PET/CT scanning did not show other distant metastasis. For her metastatic recurrent melanoma, the patient enrolled in a clinical trial (ClinicalTrials.gov ID: NCT03068455) and was randomized to receive monotherapy with nivolumab 240 mg every 2 weeks versus the combination of nivolumab with ipilimumab 1 mg/kg every 3 weeks. Due to her smoking history, she was under regular follow-up by pulmonologist and had a negative tuberculin skin test (TST) on March 2017 but the trial protocol did not require LTBC screening before the initiation of immunotherapy. In January 2018, after 8 doses of immunotherapy, she presented diarrhea grade 2 and started methylprednisolone 16 mg po twice daily with a slow taper (over 4–6 weeks). After a short-term improvement of her diarrhea to grade 1, her symptoms worsened again and a colonoscopy was performed. The endoscopic examination revealed grade 3 colitis with multiple ulcerative mucosal lesions. Therefore, immunotherapy was permanently discontinued and the dose of methylprednisolone was increased to 32 mg daily and intravenous (iv) infliximab administered at a dose of 5 mg/kg. After three doses of infliximab her colitis improved (to grade 1) and steroid taper was resumed. Two weeks later, the patient was admitted to our hospital with fever up to 38 °C, fatigue and weight loss. Physical examination suggested a lower respiratory tract infection. Laboratory tests revealed neutrophils: 6700/μL, hemoglobin: 10.5 g/dL, platelet count: 129,000/μL and elevated C reactive protein (CRP = 147.6 mg/L). The patient was empirically treated with iv tazobactam/piperacillin 4.5 mg 4 times daily. Computed tomography (CT) scanning of the chest showed interval development of ground glass and centrilobular nodular opacities predominantly in the right lung (Fig. 1). These imaging findings were not present in a prior CT scan 2 months before. On April 19, 2018, the patient underwent bronchoscopy and was started on anti-tuberculosis medication with rifampin 600 mg/day, isoniazid 300 mg/day and ethambutol 1200 mg/day and pyrazinamide 2000 mg/day as well as anti-pneumocystis treatment with trimethoprim-sulfamethoxazole 20 mg/kg/day. The polymerase chain reaction (PCR) of bronchoalveolar lavage (BAL) was positive for MTB complex. Despite immediate treatment, her respiratory function was progressively worsened and the patient was moved to intensive care unit. She was intubated but remained persistently febrile and hypotensive requiring vasopressor medications. The patient passed away 2 days later and subsequently the BAL culture grew MTB with no resistance to anti-tuberculosis treatment, according to susceptibility testing.Fig1Fig. 1

Development of active MTB in a patient treated with nivolumab+/-ipilimumab for metastatic melanoma in the setting of a clinical trial. a Timeline of therapy and disease status for both melanoma and TB. b Chest CT images of patient at the enrollment in the clinical trial before the initiation of PD-1 inhibition (15 October 2016, left) and ~20 weeks later (19 April 2018, right)

An 85-year-old Greek man was diagnosed with melanoma of the right parotid nodes confirmed by positive fine needle aspiration on December 2012 (Fig. 2). His medical history included hypertension, dyslipidemia, glaucoma and benign prostate hyperplasia. The patient underwent a total parotidectomy and regional lymph node dissection as well as a tonsil biopsy. Both right parotid gland and dissected lymph nodes were infiltrated by melanoma, while biopsy of tonsil was negative. No primary skin lesion was recognized and subsequent staging scans were also negative for residual disease (stage IIIb, TxN1bM0). Therefore, he received prophylaxis with high dose IFN 20000 iu/m² every day and he was followed-up until June 2018. At this time, he underwent a chest CT because of persistent cough, which revealed multiple lymph nodes and a mediastinal soft tissue mass (M1b, stage IV) (Fig. 2). In the context of a clinical trial (ClinicalTrials.gov ID: NCT03273153), he started a combination with atezolizumab 840 mg every 3 weeks and MEK inhibitor cobimetinib 60 mg every day for his metastatic BRAFV600 wild-type melanoma. The patient was not tested for LTBC before starting anti-melanoma treatment, because he had no known risk factors for MTB reactivation and the clinical trial protocol did not require it. After receiving the therapeutic combination for approximately 5 months, including a temporary interruption of cobimetinib because of grade 3 rash, the patient presented with symptoms of lower respiratory tract infection grade 3 on November 2018. No new imaging findings by X-ray were recognized at that time. Cobimetinib was interrupted again and the patient received a course of iv tazobactam/piperacillin 4.5 mg QID and levofloxacin 500 mg every day. During the next 3 months, the patient had two more episodes of fever grade 2, during which he was hospitalized and received broad-spectrum antibiotics for 1 week in total each time. Throughout this period, cobimetinib was temporarily interrupted, but immunotherapy with atezolizumab was continued without complications. Although acid-fast bacilli stain of sputum was negative, a sputum culture taken in his last hospitalization grew M. tuberculosis. Susceptibility testing showed susceptibility to all antimycobacterial agents. On February 2019, he initiated a 3-drug regimen, including isoniazid 300 mg/day, rifampin 600 mg/day and pyrazinamide 1500 mg/day. Currently, he continues anti-tuberculosis medication with good tolerability, whereas anti-melanoma combination is still withheld up to resolution of MTB imaging lesions.Fig2Fig. 2

Development of active MTB in a patient treated with atezolizumab and cobimetinib for metastatic melanoma in the setting of a clinical trial. a Timeline of therapy and disease status for both melanoma and MTB. b Chest CT images of patient at the enrollment in the clinical trial (July 2018, left) and 4 months later (November 2018, right)

Although T-helper type 1 CD4+ cells (Th1) are required for control of mycobacterial infections, the enhanced CD4 activity in the absence of PD-1 surveillance exacerbates tuberculosis in mouse models. This status of T cell exhaustion arises from the sustained activation with absence of inhibitory receptors and prevents optimal control of infection and tumors [24]. In fact, PD-1 knockout (PD-1−/−) mice are more susceptible to MTB mortality, developing large necrotic lesions with high bacterial loads and succumbing faster than even T cell-deficient mice [25–27]. The inability of PD-1−/− mice to control mycobacterial infection is attributed to increased Th1-mediated responses and overproduction of interferon-gamma (IFN-γ) [26]. Sakai et al. pointed out that activated PD-1/PD-L1 signaling suppresses the accumulation of parenchymal CD4+ T cells and limits IFN-γ production, protecting mice from fatal exacerbated pulmonary mycobacterial infection [28]. Cancer cells and infectious agents may evade early immune responses with other PD-1/PD-L1 mediated mechanisms: i) promotion of PD-L1 expression on dendritic cells and enhanced induction of Treg cells [29, 30], ii) overexpression of PD-1 on NK cells, as detected in patients with multiple myeloma [31] or infected with MTB [32] or HIV [33]. A recent study by Cao et al. [34] found that co-stimulation by MTB and lung cancer antigen in mice may partially reverse the loss of T-cell function via PD-1/PD-L1 pathway and prevent the rapid evolution of advanced lung cancer [34].

In human patients with active tuberculosis, PD-1 was increased on CD4+ T cells but not on CD8+ T cells compared to healthy controls [35, 36] while effective anti-tuberculosis treatment was associated with a down-regulation of PD-1 on CD4+ T cells [36]. Similarly, the expressions of PD-1 and PD-L1 on monocytes from patients with active MTB were much higher compared to healthy controls, while phagocytosis and intracellular killing activity of macrophages increased significantly with PD-1/PD-L1 blockade in vitro [35]. In a patient with Merkel cell carcinoma treated with nivolumab, IFN-γ-producing MTB-specific CD4+ T cells were detected in the blood months before the development of a tuberculoma [37]. Producing a scenario similar to immune reconstitution inflammatory syndrome (IRIS), the blockade of PD-1 axis boosts Th1-mediated inflammatory responses and causes a worsening of damage to the MTB-infected tissue [38]. Currently, the PD-1/PD-L1 pathway is also being studied as a novel host-directed target in multidrug-resistant tuberculosis [39, 40].

In order to identify other reported cases with ICB-associated MTB infection, we used the following terms for online search of PubMed: (1) terms suggestive of cancer (e.g., cancer, tumor, malignancy), (2) terms suggestive of immunotherapy (e.g., immune checkpoint inhibitors, PD-L1, PD-1, CTLA4, immunotherapy), (3) terms suggestive of tuberculosis (e.g., tuberculosis, TB, Mycobacterium tuberculosis). As restriction limits through the electronic search, we used English language and human-based studies. The full search strategy of the literature by the two independent reviewers, with the numbers of the records identified or excluded and the reasons of exclusions, is represented in Fig. 3, according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis). A secondary expanded search was conducted by using Medical Subject Headings (MeSH) subjects and by hand-searching of reference lists from previous reviews to identify additional publications. For the cumulative presentation of findings among the identified case-reports, the 95%CIs of proportions were estimated by modified Wald method.Fig3Fig. 3

Flow diagram of literature search strategy

Based on the synthesis of current evidence and our experience, we below address some emerging issues regarding the incidence and management of tuberculosis in oncological patients, and suggest clinical practice recommendations.

It is accepted that ICBs may have infectious complications, indirectly as a consequence of the need for corticosteroids or TNF-a inhibitors to control irAEs associated with ICB therapy. Tuberculosis may be an exception to this rule, as the majority of reported cases from the literature and our experience were receiving neither corticosteroids nor TNF-a inhibitors when their reactivation was documented. Therefore, MTB reactivation may represent a direct complication of immunotherapy, although more data are needed to unequivocally establish this. The exact mechanism of increased susceptibility to MTB following administration of ICBs is not yet known. Preclinical data recognize a crucial role of PD-1/PD-L1 blocking in T cell exhaustion, evasion of immune surveillance and development of active tuberculosis. However, in clinical practice, the management of M. tuberculosis among cancer patients receiving ICBs presents challenges. Cancer itself is an independent risk factor for developing active MTB infection. This generally occurs early in the course of disease, and cancer progression is the most common misdiagnosis when constitutional symptoms such as weight loss and fever, common with active MTB, are developed. Thus, before changing treatment for a supposed disease progression or initiating corticosteroids for a suspected irAE, all cancer patients with appropriate symptomatology should be tested for tuberculosis and checked for any previous exposure to MTB, and other risk factors. The prompt diagnosis of a mycobacterial infection, even in a subclinical stage, is essential to avoid later potentially morbid exacerbation. Given that inhibition of PD-1/PD-L1 pathway may favor tuberculosis reactivation, targeted screening for LTBC is suggested before initiation of an ICPI, especially in cancer subjects with additional independent risk factors (e.g. host comorbidities, exposure to MTB endemic regions, and immunosuppression). The preferred diagnostic modality (e.g. a single test or combination of TST and IGRA) for LTBC screening in these patients has not been clearly defined. In addition, no available data exist for the management of latent or active tuberculosis during PD-1/PD-L1 blockade; for this reason, therapeutic guidelines are adopted from the management of patients receiving TNF-a inhibition. In general, in case of active tuberculosis, ICPIs are temporarily withheld, any further immunosuppression is discontinued and anti-tuberculosis treatment is timely initiated. Also, in patients diagnosed with either active or latent tuberculosis, it is not clear how long after the corresponding anti-TB treatment ICPIs should be safely resumed or initiated, with a duration of 2–4 weeks to be suggested. The continuously expanded implementation of IPCIs in cancer treatment requires resolving of these challenges by upcoming research data in order to maximize the clinical benefits of immunotherapy uninterruptedly and safely.