Increased Tuberculosis Reactivation Risk in Patients Receiving Immune Checkpoint Inhibitor-Based Therapy

Abstract

Objective

Reports of tuberculosis (TB) during anticancer treatment with immune checkpoint inhibitors (ICIs) are increasing. However, it is not clear whether the use of ICIs is a significant risk factor for TB, including reactivation or latent TB infection (LTBI).

Methods

To determine the risk of TB reactivation in patients with lung cancer who use ICIs or tyrosine kinase inhibitors (TKIs), we conducted a retrospective study using a hospital-based cancer registry. In addition, we monitored patients with cancer using ICI or TKI in a multicenter prospective study to check the incidence of LTBI.

Results

In the retrospective study, several demographic factors were imbalanced between the ICI and TKI groups: the ICI group was younger, had more males, exhibited more squamous cell carcinoma in histology rather than adenocarcinoma, had fewer EGFR mutations, and received more chemotherapy. Propensity score matching was used to control for confounding factors, and we found that the incidence of TB was higher among patients with lung cancer who received ICIs than among those who received TKIs (2298 vs 412 per 100 000 person-years, P = .0165). Through multivariable analysis, group (ICI vs TKI) was the independent risk factor for TB development (adjusted hazard ratio (aHR): 6.29, 95% CI, 1.23-32.09, P = .0269). In the prospective cohort, which included 72 patients receiving ICIs and 50 receiving TKIs, we found that the incidence of positive seroconversion of LTBI by interferon gamma release assay (IGRA) was significantly higher in patients receiving ICIs (18% vs 0%, aHR: 9.88, P = 0.035) under multivariable Cox regression.

Conclusion

The use of ICIs may be linked to a higher likelihood of TB reactivation and LTBI than individuals solely receiving TKIs as anticancer therapy. Consequently, the implementation of a screening program for TB reactivation and LTBI among patients undergoing ICI treatment could prove advantageous by enabling early detection and prompt treatment of the infection.

Reports of tuberculosis (TB) during anticancer treatment with immune checkpoint inhibitors (ICIs) are increasing. This article reports on whether the use of ICIs is a significant risk factor for TB, including reactivation or latent TB infection.

Graphical Abstract

Graphical Abstract.

Implications for Practice.

Immune checkpoint inhibitors increase the risk of tuberculosis in patients with lung cancer. Immune checkpoint inhibitors increase the likelihood of positive seroconversion for latent tuberculosis infection. Routine screening for tuberculosis should be considered before initiating immune checkpoint inhibitors.

Introduction

Tuberculosis (TB) is a leading cause of death from infectious diseases, with an estimated 10 million people suffering from active infection and 1.2 million deaths in 2020, according to the World Health Organization (WHO).¹ After the success of “The Global Plan to Stop TB 2006-2015,”² WHO proposed the END TB Strategy for the post-2015 era, and various frameworks have contributed to progress toward TB elimination.³ Strategies include optimizing current treatment protocols and screening high-risk groups for latent TB infection (LTBI).

Patients with malignancy, particularly those with lung cancer, are at high risk of developing active TB because of their immunocompromised state.^4-6 With advancements in cancer treatments, the average life span after a diagnosis of cancer has been extended, resulting in a growing population of susceptible patients.^7-9 WHO estimates that there were 19.3 million new cancer cases in 2020, with a projected 47% increase in new cases over the next 20 years, particularly in transitioning countries where TB is more prevalent.⁸ Therefore, it is crucial to target this high-risk population for TB prevention and screening efforts.

Immune checkpoint inhibitors (ICIs) have become the standard of care for multiple types of cancer, including lung cancer, and have moved from last resort treatment for refractory disease to front-line therapy for metastatic disease and even in early-stage disease. ICIs have been approved for neoadjuvant therapy in non–small cell lung cancer and breast cancer and as adjuvant therapy in bladder cancer, esophageal cancer, melanoma, non–small cell lung cancer, and renal cell carcinoma. Additionally, ICIs are increasingly being used in combination with chemotherapies, other ICIs, or targeted therapies, leading to a significant shift in cancer treatment.¹⁰

Recently, there have been increasing reports of TB reactivation in patients receiving ICIs.^11-16 A previous study suggested screening for TB before initiating ICI treatment,¹¹ but there were conflicting opinions because the number of reported TB cases remains small.^11,12 In addition, the incidence of active TB and LTBI in patients with cancer receiving ICIs is currently not well understood. Notably, LTBI seroconversion, which represents recent Mycobacterium tuberculosis (Mtb) infection, is associated with an approximately 8-fold higher risk of progression to active TB disease and is more predictive than a single positive interferon-gamma release assay (IGRA) result.¹⁷ In this study, our objective was to examine the risk of tuberculosis development using a 2-fold approach. First, we conducted a retrospective cohort study to investigate the risk of active tuberculosis development among lung cancer patients receiving ICIs. Second, we conducted a prospective cohort study to assess the risk of LTBI seroconversion in pan-cancer patients. Patients receiving tyrosine kinase inhibitors (TKIs), as they do not undergo immune modulating therapy such as ICIs, demonstrate a relatively similar profile to that of the general population. Consequently, they were selected as the control group for comparison.

Materials and Methods

Tuberculosis Reactivation Cohort

We conducted a retrospective cohort study at the National Taiwan University Hospital and its Yun-Lin branch from January 2015 to June 2020. We used the hospitals’ integrated medical database to review patients aged 20 years or older with advanced non–small cell lung cancer (inoperable stage IIIb or IV disease) and no history of tuberculosis before their lung cancer diagnosis. The study was approved by the Research Ethics Committee (IRB No: 202002118RINC), and written consent was not required because of the retrospective nature of the study.

Demographic information, such as age, sex, smoking status, and body mass index (BMI), was collected. Cancer histology, EGFR mutation, and duration from cancer diagnosis to ICI/TKI were recorded. Co-morbidities, including cancer, asthma, chronic obstructive pulmonary disease (COPD), bronchiectasis, pneumoconiosis, idiopathic pulmonary fibrosis, cirrhosis, end-stage renal disease, cerebrovascular disease, congestive heart failure, diabetes mellitus (DM), gastroesophageal reflux disease (GERD), autoimmune diseases (systemic lupus erythema, rheumatoid arthritis, Sjőgren’s syndrome, polymyositis, and dermatomyositis), and solid-organ transplantations were identified using diagnostic codes in the medical records.

Grouping and Outcome Measurement for TB Reactivation

The study tracked the use of all pharmacological treatments for lung cancer, including ICI, TKI, chemotherapy, and corticosteroids (systemic or inhaled), from diagnosis until the end of the study. Details of these treatments are provided in Supplementary Tables S1-S3. Participants were divided into 2 groups based on their treatment regimens: the ICI group and the TKI group. The index date was defined as the first dose of ICI or TKI. The patients were monitored until the study endpoint of TB reactivation, death, loss of follow-up, or the end of the study period (with a maximum follow-up of 18 months).

TB diagnosis was made through a positive culture for M. tuberculosis or a clinical diagnosis (based on one inpatient diagnosis or a minimum of 2 outpatient diagnoses, made by clinical symptoms, radiological findings, and laboratory test results such as typical TB pathology, nucleic acid amplification test (NAAT), or pleural effusion adenosine deaminase level).¹⁸ To decrease between-group differences, we further selected propensity score-matched cohorts using a Greedy nearest neighbor matching method and performed the outcome analysis with a standardized mean difference, SMD < 0.1).¹⁹

LTBI Seroconversion Cohort

To augment our findings, we conducted a prospective study in 5 teaching hospitals in Taiwan between April 2020 and October 2022 with the approval of the hospital’s Research Ethics Committee (IRB No: 202002118RINC, 201910079RINC, 109043-F, FIRB091). The study group consisted of patients with cancer scheduled to receive FDA-approved ICI therapy, whereas patients receiving TKI therapy were enrolled as the control group. Patients who received recent chemotherapy were not included because of lymphocyte depletion and not good candidate for LTBI testing.²⁰ Patients were excluded from the study if they or their family refused to participate, if they could not comply with the study protocol, or if their life expectancy was less than 3 months. To evaluate LTBI status, we examined their blood using QuantiFeron-TB Gold In-tube (QFT) testing, an interferon gamma release assay (IGRA), during anticancer therapy. IGRA testing was performed every 2 months during the first 6 months of treatment. We compared the prevalence at the initial stage and incidence of LTBI (seroconversion) within 6 months in the ICI group to the TKI group.

Statistical Analysis

Statistical analysis for the study was performed using SAS 9.4 software. To compare categorical and continuous variables, Chi-squared and Student’s t tests were used, respectively. Kaplan-Meier survival curves were analyzed using log-rank tests. The study calculated univariable and multivariable hazard ratios using Cox proportional hazard regression. Factors that were found to be statistically significant in the univariable analysis were included in the multivariable analysis. A P-value of <.05 was considered statistically significant.

Results

Patient Enrollment in the TB Reactivation Cohort

We reviewed the cancer registry and selected 14 196 patients with lung cancer diagnosed between 2015 and 2020. We excluded patients who did not receive ICI or TKI treatment (n = 10 669), those who had used drugs before a formal diagnosis of cancer (n = 97), those who had a TB diagnosis before the diagnosis of lung cancer (n = 34), those who were diagnosed with lung cancer before 2015 when ICIs were first introduced (n = 810), and those with non-advanced stage cancer at initial diagnosis. A total of 2049 patients were included in the study, 442 of whom received ICIs and 1607 received TKIs. The study flowchart was shown in Figure 1.

Figure 1.

Study flowchart of the tuberculosis (TB) reactivation cohort. Abbreviations: ICI, immune checkpoint inhibitor user; TKI, tyrosine kinase inhibitor user.

Demographics of the TB Reactivation Cohort

Table 1 shows the demographics and clinical characteristics of the original cohort. The ICI group were younger (63.2 vs 66.1 years), had a higher proportion of male patients (60.41% vs 38.33%), a lower proportion of adenocarcinoma (57.47% vs 92.78%), a longer duration from cancer diagnosis to ICI/TKI use (288 ± 337 vs 176 ± 308 days), a higher percentage of patients who received chemotherapy (85.52% vs 40.2%), more systemic (44.3% vs 25.1%), and inhaled (16.97% vs 7.65%) corticosteroid use than the TKI group. There were more underlying comorbidities, including COPD (11.54% vs 3.11%), GERD (6.79% vs 3.92%), DM (18.10% vs 12.51%), Sjőgren’s syndrome (1.58% vs 0.37%), and sinusitis (0.68% vs 0.12%) in the ICI group.

Table 1.

Demographics of tuberculosis reactivation cohorts according to using immune checkpoint inhibitor (ICI) or tyrosine kinase inhibitor. The matched cohorts are based on propensity matched method.

	Original cohort				Matched cohort
Table	ALL (N = 2049)	ICI group (N = 442)	TKI group (N = 1607)	P-value	ALL (N = 862)	ICI group (N = 431)	TKI group (N = 431)	P-value
Age, years	65.48 (11.89)	63.23 (11.67)	66.1 (11.87)	6.52E−06	62.8 (11.58)	63.03 (11.68)	62.56 (11.49)	.4789
Sex, male	883 (43.09%)	267 (60.41%)	616 (38.33%)	1.04E−16	505 (58.58%)	256 (59.4%)	249 (57.77%)	.6783
Cancer histology
Squamous cell carcinoma	120 (5.86%)	88 (19.91%)	32 (1.99%)	2.18E−78	99 (11.48%)	84 (19.49%)	15 (3.48%)	2.81E−28
Adenocarcinomas	1745 (85.16%)	254 (57.47%)	1491 (92.78%)		642 (74.48%)	249 (57.77%)	393 (91.18%)
Others	184 (8.98%)	100 (22.62%)	84 (5.23%)		121 (14.04%)	98 (22.74%)	23 (5.34%)
EGFR mutation, N = 1697	1362 (80.26%)	71 (25.63%)	1291 (90.92%)	1.29E−137	396 (61.40%)	70 (25.83%)	326 (87.17%)	6.34E−58
Duration from diagnosis to using TKI or ICI, days	200  ± 317	288 ± 337	176 ± 308	3.87E−11	247 ± 334	284 ± 333	210 ± 332	.0012
Chemotherapy, presence	1024 (49.98%)	378 (85.52%)	646 (40.2%)	6.71E−64	738 (85.61%)	367 (85.15%)	371 (86.08%)	.7710
Systemic steroid, presence	599 (29.2%)	196 (44.3%)	403 (25.1%)	3.11E−15	340 (39.44%)	187 (43.39%)	153 (35.5%)	.0178
Inhaled steroid, presence	198 (9.66%)	75 (16.97%)	123 (7.65%)	4.37E−09	125 (14.5%)	67 (15.55%)	58 (13.46%)	.4391
Comorbidities
COPD	101 (4.93%)	51 (11.54%)	50 (3.11%)	4.23E−13	70 (8.12%)	42 (9.74%)	28 (6.5%)	.1044
Asthma	44 (2.15%)	9 (2.04%)	35 (2.18%)	0.8555	21 (2.44%)	9 (2.09%)	12 (2.78%)	.6597
Bronchiectasis	7 (0.34%)	2 (0.45%)	5 (0.31%)	.6483	2 (0.23%)	2 (0.46%)	0 (0.00%)	.4994
IPF	5 (0.24%)	1 (0.23%)	4 (0.25%)	1	3 (0.35%)	1 (0.23%)	2 (0.46%)	1
GERD	93 (4.54%)	30 (6.79%)	63 (3.92%)	.0103	51 (5.92%)	28 (6.5%)	23 (5.34%)	.5640
Obesity	1 (0.05%)	1 (0.23%)	0 (0.00%)	.2157	1 (0.12%)	1 (0.23%)	0 (0.00%)	1
Liver cirrhosis	8 (0.39%)	2 (0.45%)	6 (0.37%)	.6848	5 (0.58%)	2 (0.46%)	3 (0.7%)	1
ESRD	8 (0.39%)	1 (0.23%)	7 (0.44%)	1	1 (0.12%)	1 (0.23%)	0 (0.00%)	1
CHF	42 (2.05%)	8 (1.81%)	34 (2.12%)	.6878	10 (1.16%)	7 (1.62%)	3 (0.7%)	.3409
Stroke	18 (0.88%)	4 (0.9%)	14 (0.87%)	1	6 (0.7%)	4 (0.93%)	2 (0.46%)	.6864
DM	281 (13.71%)	80 (18.1%)	201 (12.51%)	.0025	151 (17.52%)	78 (18.1%)	73 (16.94%)	.7201
SLE	4 (0.2%)	2 (0.45%)	2 (0.12%)	.2053	2 (0.23%)	2 (0.46%)	0 (0.00%)	.4994
RA	10 (0.49%)	4 (0.9%)	6 (0.37%)	.2372	6 (0.7%)	3 (0.7%)	3 (0.7%)	1
Sjőgren’s syndrome	13 (0.63%)	7 (1.58%)	6 (0.37%)	.0105	11 (1.28%)	7 (1.62%)	4 (0.93%)	.5462
Autoimmune	23 (1.12%)	9 (2.04%)	14 (0.87%)	.0395	15 (1.74%)	8 (1.86%)	7 (1.62%)	1
Transplant	5 (0.24%)	0 (0.00%)	5 (0.31%)	.5915	2 (0.23%)	0 (0.00%)	2 (0.46%)	.4994
Sinusitis	5 (0.24%)	3 (0.68%)	2 (0.12%)	.0705	4 (0.46%)	3 (0.7%)	1 (0.23%)	.6241
Development of TB	18 (0.88%)	7 (1.58%)	11 (0.68%)	.0728	9 (1.04%)	7 (1.62%)	2 (0.46%)	.0938

Abbreviations: CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; ESRD, end-stage renal disease; GERD, gastroesophageal reflux; ICI, immune checkpoint inhibitor; IPF, idiopathic pulmonary fibrosis; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; TKI, tyrosine kinase inhibitor. The bold value indicates P-value <.05.

To reduce bias, a propensity score-matched cohort was generated. After matching, all demographics and clinical characteristics of the ICI and TKI groups were well balanced except for a higher percentage of systemic steroid use (43.39% vs 35.5%, P = .0178), a lower percentage of adenocarcinoma (57.77% vs 91.18%, P < .0001) and a longer duration from diagnosis to TKI or ICI use (284 vs 210 days, P = .0012) in the ICI group (Table 1) to maintain a standardized mean difference <0.1.

Analysis of Active TB Reactivation

The original cohort had TB development rates of 1.58% and 0.68% in the ICI and TKI groups, respectively (P = .0728; Table 1). The Kaplan-Meier method, for time-to-event analysis, was used to calculate the event-free rate within 18 months, which was found to be higher in the TKI group than in the ICI group (P = .0138 by log-rank test; Fig. 2A). After propensity score matching, TB occurred in 7 (1.62%) patients in the ICI group and 2 (0.46%) patients in the TKI group (P = .0938). The difference between the groups was significant (P = .0165 by log-rank test; Fig. 2B). In the original cohort, 7 of 442 patients (cumulative time at risk: 311.78 person-years) in the ICI group and 11 of 1607 patients (cumulative time at risk: 1745.37 person-years) in the TKI group developed active tuberculosis, corresponding to incidence rates of 2245 and 630 cases per 100 000 person-years, respectively. In the matched cohort, 7 of 431 patients (cumulative time at risk: 304.56 person-years) in the ICI group and 11 of 1607 patients (cumulative time at risk: 485.34 person-years) in the TKI group developed active tuberculosis, with corresponding incidence rates of 2298 and 412 cases per 100 000 person-years, respectively.

Figure 2.

Kaplan-Meier curves for the incidence of tuberculosis (TB) reactivation in the retrospective study cohort before and after propensity score matching. (A) In the original cohort, the incidence rate of TB development was higher in the ICI group than in the TKI group (2245 vs 630 per 100 000 person-year, P = 0.0138 by log-rank test). (B) After matching, the significance remained (2298 vs 412 per 100 000 person-year, P = .0165).

Univariable and Multivariable Analyses of TB Reactivation

The risk of TB development was assessed using the Cox proportional hazards model. Results from the univariable analysis revealed that being in the ICI group (crude HR: 5.62, 95% CI, 1.15-27.34), adenocarcinoma (crude HR: 0.08, 95% CI, 0.01-0.48), comorbidity of chronic obstructive pulmonary disease (crude HR: 5.99, 95% CI, 1.23-29.25), and congestive heart failure (crude HR: 19.21, 95% CI, 2.38-154.82) were all significantly associated with TB compared with the TKI group (P = .0326, P = .0268, P = .0055, respectively). The factors significant (P < .05) in the univariable analysis were included into the multivariable analysis. Because the ICI/TKI group was significantly correlated with cancer type, we entered one of the 2 factors into the multivariable analysis separately and analyzed the model of multivariable analysis. In the multivariable analysis (model 1), only the ICI group was found to remain statistically significant as a risk factor for TB (adjusted HR: 6.29, 95% CI, 1.23-32.09, P = .0269; Supplementary Table S4). In model 2, being adenocarcinoma was significantly associated with a lower risk of TB (adjusted HR: 0.11, 95% CI, 0.01-0.79, P = .0284).

Prospective LTBI Seroconversion Cohort

A total of 122 patients with cancer were enrolled in the study from May 2020 to November 2022 in 5 teaching hospitals in Taiwan (National Taiwan University Hospital, National Yang-Ming Chiao-Tung University Hospital, Far-East Medical Hospital, National Taiwan University Hospital Yunlin branch, E-Da Hospital), with 72 receiving immune checkpoint inhibitor (ICI) treatment and 50 receiving tyrosine kinase inhibitor (TKI) treatment. The age and sex distribution of the patients was similar between the 2 groups. The baseline latent tuberculosis infection (LTBI) status, as determined by the QuantiFERON-TB Gold In-tube test (QFT), was 20% and 15.1% in the TKI and ICI groups, respectively (no statistically significant difference, P > .05; Table 2).

Table 2.

Demographics of the prospective cohort for latent tuberculosis infection (LTBI).

	ICI group (n = 77)	TKI group (n = 50)	P value
Age, year	60.7 ± 11.9	63.9 ± 9.9	.122
Male sex	59 (77)	19 (38)	<.001
Smoking			.001
Current smoker	12 (16)	0
Ex-smoker	34 (44)	15 (30)
Body weight, kg	60.0 ± 12.1	62.8 ± 13.4	.236
Cancer types			<.001
Lung cancer	35 (46)	50 (100)
Head and neck cancer	21 (27)	0
Others	21 (27)	0
ICI agents
Anti-PD-1 ab	63 (82)	—
Anti-PD-L1 ab	14 (18)	—
Duration from cancer diagnosis to ICI/TKI use, days	248 ± 356	312 ± 510	.447
Underlying disease	1 (1)	1 (1)	.368
Diabetes mellitus	16 (21)	6 (12)	.451
Liver cirrhosis	2 (3)	0	.251
ESRD	1 (1)	0	.419
COPD	10 (13)	13 (26)	.071
Presence of symptoms*	46 (60)	26 (52)	.390
Hemoglobin, g/dL	11.1 ± 2.4	12.7 ± 1.9	.002
Serum creatinine, mg/dL	0.99 ± 0.67	0.98 ± 0.51	.228
Baseline QFT-GIT result			.158
Positive	10 (14)	11 (22)
Negative	48 (65)	34 (68)
Indeterminate	16 (22)	5 (10)

Data are mean ± standard deviation or number (%).

^*Respiratory symptoms include chronic cough, sputum production, and shortness of breath.

Abbreviations: COPD, chronic obstructive pulmonary disease; ESRD, end-stage renal disease; ICI, immune checkpoint inhibitor; PD-1, programmed death-1; PD-L1, programmed death ligand 1; QFT-GIT, QuantiFeron Gold In-tube; TKI, tyrosine kinase inhibitor.

Of the enrolled patients with cancer who underwent QFT examination, 88 participants (51 ICI users and 37 TKI users) were followed up, and 66 of them (39 ICI users and 27 TKI users) had a negative baseline LTBI status. During the follow-up period, the LTBI status changed to positive in 7 patients receiving ICI, which was a higher incidence than that in the TKI group (18% vs 0%, P = .036 as determined by log-rank test; Fig. 3).

Figure 3.

Kaplan-Meier curve of the probability of free from interferon gamma release assay seroconversion in the prospective latent tuberculosis infection (LTBI) study cohort. The risk of positive seroconversion of the interferon gamma release assay (IGRA) was significantly higher in patients receiving ICIs (P = .036).

Univariable analysis, taking into consideration potential confounding factors such as age, sex, BMI, smoking, cancer type, history of chronic obstructive pulmonary disease, white blood count, and hemoglobin level, revealed that group (being in the ICI group rather than in the TKI group) was a significant risk factor for positive LTBI seroconversion (Supplementary Table S5 in the supplement file). Factors found to be significant or borderline significant (group, age, hemoglobin level) were further included in the multivariable analysis. The results of the multivariable analysis showed that group (being in the ICI vs TKI group; adjusted HR 9.88, 95% CI, 1.13-1297.16, P = .035) and age (adjusted HR 1.07, 95% CI, 1.00-1.16 per 1-year increment, P = .039) were statistically significant factors for LTBI positive seroconversion (Supplementary Table S5).

Discussion

In a systematic review and meta-analysis of 36 phase II or III clinical trials including over 20 000 patients, ICI use alone was found to be safer and associated with a lower risk of bacterial infection compared to chemotherapy. However, the study mainly focused on pyogenic infections. Other infections, such as viral, fungal, and mycobacterial infections, were not discussed.²¹ Our study revealed that patients with lung cancer treated with ICIs had a higher incidence of TB reactivation (2298 per 100 000 person-years) than those treated with TKIs (412 per 100 000 person-years) in the TB reactivation cohort after adjusting for potential confounders via propensity score matching and multivariable analysis. The multivariable Cox proportional hazards analysis showed that the ICI group was an independent risk factor for TB development (adjusted HR: 6.29, P = .0269) when compared to the TKI group. In contrast, adenocarcinoma was associated with a lower risk of TB than squamous cell carcinoma and other cancer cell types. In addition, this finding was reinforced by a prospective LTBI seroconversion cohort study, which demonstrated that patients who received ICIs had a greater HR of positive IGRA seroconversion compared with the control group (patients who received TKIs). The incidence rate of TB in both groups significantly exceeds that in the Taiwanese population (30 per 10⁵ person-years in 2020, as reported by the Taiwan CDC). These comparisons suggest that the lung cancer population, regardless of treatment, represents a distinct high-risk group.²²

A recent prospective study in Japan conducted serial IGRA testing on patients with lung cancer undergoing immunotherapy. This study enrolled 178 patients between 2017 and 2020, of whom 123 completed serial IGRA tests. Among these patients, 15% initially tested positive for IGRA, whereas 4% of those who initially tested negative experienced seroconversion during their treatment course. Notably, one patient with positive baseline IGRA result and another patient who experienced seroconversion developed active tuberculosis.²³ In contrast, a recent cohort study conducted in Korea did not identify an association between ICI use and active TB incidence. However, it is worth noting that their analysis was initiated from the date of cancer diagnosis rather than the date of drug use, potentially introducing bias in evaluating the relationship between ICIs and TB incidence.²⁴

The findings of this study agree with those of previous studies.^11-15,23 However, our study extends upon previous findings by incorporating one of the largest retrospective cohorts to investigate TB in ICI users. The use of a Cox proportional hazards model in our cohort facilitated a sequential analysis, offering deeper insights beyond mere data description or association analysis in cross-sectional studies.^13-15 Furthermore, we used propensity score matching to account for between-group differences, and the risk of TB reactivation (HR) remained 6 times higher in ICI users than in the control group. These methodological strategies bolster the credibility of the association between ICIs and TB. To further validate the findings from our retrospective study, we conducted a prospective study in which we continuously monitored patients’ latent tuberculosis infection (LTBI) status throughout their treatment. This allowed for a comprehensive analysis of Mycobacterium tuberculosis (Mtb) infection characteristics and the timeline of reactivation following the initiation of ICI treatment. In addition to the previous Japanese study, we included patients with diverse tumor types in our cohort. Furthermore, we established a control group consisting of patients receiving TKI therapy to assess the rate of seroconversion among individuals with a lower risk of TB infection.

In the natural history of Mtb infection, approximately 30% of close contacts will be infected following exposure to Mtb bacilli in the patient’s airway.²⁵ Subsequently, the LTBI status is defined by the absence of clinical or radiological evidence of active TB but a positive immune test for Mtb, such as IGRAs.²⁶ For patients with cancer, the reactivation risk is higher than that in the general population.^4-6 Studies have demonstrated that the new positive seroconversion of QuantiFERON-TB Gold In-tube (QFT) testing, a type of IGRA, is associated with a higher risk of TB reactivation and serves as a more accurate marker than a single positive test of IGRA.¹⁷ Given the relatively low incidence of TB, we used positive LTBI seroconversion as a surrogate marker for the elevated risk of TB development in our prospective study because of its ease of measurement and follow-up. The findings of our prospective study were consistent with those obtained from our retrospective cohort on TB reactivation.

It may seem counterintuitive that anticancer immunotherapy, designed to enhance the immune system’s function, can increase the risk of tuberculosis. Several mechanisms may explain the link between ICI therapy and TB. For example, PD-1 expression is increased in immune cells such as CD4 + T cells, NK cells, neutrophils, and monocytes in patients with tuberculosis, suggesting a regulatory role.^27,28 Anti-PD1 therapy may disrupt the homeostasis of anti-Mtb-specific T cells, leading to the dysregulation of several cytokines, including TNF-α, which can favor the growth of Mtb.²⁸ In addition, the immune checkpoint plays a role in maintaining immune system balance within TB granulomas, preventing tissue destruction and excessive inflammation.²⁹ However, rapid T-cell activation by ICIs may lead to more destruction of the extracellular matrix and increased recruitment of monocytes or neutrophils, thereby promoting Mtb growth.^12,30 The rise in active TB cases following ICI use suggests that an excess of immunity can be just as damaging as insufficient immunity, resembling the mechanism seen in immune reconstitution inflammatory syndrome (IRIS) among human immunodeficiency virus-infected patients receiving antiviral therapy, as demonstrated by pathological examination of a patient with TB pleurisy after anti-PD1 therapy.¹⁶ Studies have also shown that PD-1-deficient mice have fatal inflammation after Mtb infection,^11,31,32 and similar impaired defense against Mtb under PD-1 blockade has been observed in rhesus macaques.³³ This may be due to the role of the PD-1 pathway in avoiding harmful inflammation and promoting normal CD4 T-cell responses. ICI may also have immune-related adverse effects. Clinicians may use corticosteroids and other immunosuppressants to suppress dysregulated immune function.³⁴ These immunosuppressants may also increase the incidence of TB reactivation.

Our study has several strengths. First, the first retrospective part is a large-scale cohort study comprising 14 196 lung cancer patients, with 2049 ultimately included in the analysis. A matched control group was used for comparison, eliminating the potential for bias due to demographic factors. In addition, a prospective multi-center cohort was collected to reinforce the results and show compatible findings. To date, this is the only known prospective cohort study evaluating the association between ICI and LTBI seroconversion. Furthermore, the study was conducted in Taiwan, where the incidence of tuberculosis is intermediate,^1,35 making it easier to evaluate outcomes related to TB.

This study has some limitations to consider. The retrospective part of our study relies on claimed data, which could lead to potential confounding factors and missing information. The baseline IGRA test was not mandatory before starting treatment; therefore, the baseline LTBI condition was not available. Patients with mutation-addicted lung cancer may have different demographics and disease characteristics compared with those without. In addition, patients in the ICI group may have had less oncogene-dependent disease compared with those in the TKI group, which is significantly correlated with the cancer type of adenocarcinoma. In the prospective cohort study, IGRA-positive seroconversion was used as a surrogate. The overall TB risk remained low even in patients with IGRA-positive seroconversion.¹⁷ The case number and outcome number in the studies was small and did not include all types of cancer sensitive to ICI. The study was conducted in Taiwan, and caution should be exercised when applying the findings to other populations and ethnicities without further validation. The number of patients in the prospective cohort is small, which limits the ability to further analyze confounding factors. In addition, our study lacks a control group consisting of individuals from the normal population, thereby requiring us to rely on the estimated incidence of tuberculosis in the general population obtained from governmental reports.

Conclusion

This study found that patients with lung cancer who received ICIs had a higher incidence of TB reactivation, with a rate of 2298 per 100 000 person-years, which was significantly higher than that of patients who received TKI. Analysis using Cox proportional hazards found an adjusted HR of approximately 6.29 for ICI users. A prospective LTBI seroconversion cohort showed results similar to those of the retrospective analysis. Although we have adjusted for other differences between the 2 groups, we acknowledge that, due to the trial design, we cannot directly attribute the higher risk of LTBI or TB reactivation to immune checkpoint inhibitor immunotherapy. However, further large-scale prospective studies are required to confirm these findings. Until then, it might be recommended that routine screening for TB be conducted for patients with lung cancer considering ICI treatment because chest radiography and sputum examinations are easily accessible and non-invasive procedures. Early detection of TB can reduce mortality in patients with cancer.

Supplementary Material

Supplementary material is available at The Oncologist online.

Funding

This study was partially supported by a research grant from the National Taiwan University Hospital (111-T0014, 112-E0017, and 113-E0005), the National Taiwan University Hospital Yunlin branch (NTUHYL109.N001) and the Far Eastern Memorial Hospital National Taiwan University Hospital Joint Research Program (109-FTN05 and 111-FTN0017). The funders played no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflict of Interest

The authors indicated no financial relationships.

Author Contributions

Conception/design: H.W.C., Y.W.K., C.C.S. Provision of study material or patients: H.W.C., Y.W.K., C.Y.C., W.C.L., J.P.C., C.Y.C., Y.F.W., S.C.C., C.C.S., J.Y.W., W.Y.L., H.C.W., C.J.Y. Collection and/or assembly of data: H.W.C., Y.W.K., C.C.S. Data analysis and interpretation: C.H.C., S.M.W., Y.C.C., C.C.S. Manuscript writing: H.W.C., Y.W.K., C.Y.C., W.C.L., J.P.C., C.Y.C., Y.F.W., S.C.C., C.C.S., J.Y.W., W.Y.L., H.C.W., C.J.Y. Final approval of manuscript: All authors.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.