Risk factors and mechanisms of immune checkpoint inhibitor-related pneumonitis

ABSTRACT

The use of immune checkpoint inhibitors (ICIs) – including ipilimumab, nivolumab, and atezolizumab has demonstrated remarkable clinical efficacy in cancer therapy. However, due to the activation of the immune system, ICIs can also precipitate immune-related adverse events (irAEs) across multiple organ systems. Among these toxicities, checkpoint inhibitor-related pneumonitis (CIP) is uncommon but carries substantial mortality and poses notable diagnostic and therapeutic challenges. In this review, we summarize the current advances regarding the risk factors and mechanisms of CIP and provide an overview of its incidence and mortality, imaging characteristics, diagnostic approaches, and treatment strategies.

Plain Language Summary

With the expanding body of research on immune checkpoint inhibitor-associated pneumonitis, an increasing number of risk factors have been identified, many of which align with proposed pathogenic mechanisms. This article synthesizes the current knowledge on immune checkpoint inhibitor associated pneumonitis and provides a systematic overview of its risk factors and underlying mechanisms. In addition, the future research direction of immune checkpoint inhibitor associated pneumonitis also discussed in this article, emphasizing the need for continued investigation to elucidate disease mechanisms and to develop effective strategies for prevention and treatment.

Introduction

Immune checkpoint inhibitors (ICIs) primarily target inhibitory receptors such as cytotoxic T lymphocyte associated antigen 4 (CTLA-4), programmed cell death protein 1 (PD-1), and programmed cell death-ligand 1 (PD-L1), with several novel ICIs targeting LAG3, TIM3, TIGIT, and BTLA currently undergoing clinical development.^1–3 CTLA‑4 is a coinhibitory receptor on T cells that attenuates T‑cell activation by competing with CD28 for B7 ligands. The earliest study blocking CTLA-4 in murine models of colon cancer and fibrosarcoma demonstrated tumor regression and durable antitumor immunity,⁴ paving the way for therapeutic strategies that relieve tumor‑induced immunosuppression alongside surgery, radiotherapy, chemotherapy, and targeted therapy. PD‑1 was initially identified in studies of programmed cell death.⁵ Subsequent research revealed two ligands of PD-1, PD-L1 and PD-L2 had been discovered, and PD-1/PD-L1 axis was found closely related to tumor immune evasion. A seminal study found that enforced PD‑L1 expression in tumor cells increased tumorigenicity and invasiveness, and the authors proposed that blockade of the PD‑1/PD‑L1 axis could restore tumor‑specific immunity, laying the groundwork for modern cancer immunotherapy.⁶ With the development and approval of ICIs, however, a broad spectrum of immune‑related adverse events has emerged, including dermatologic toxicity, gastrointestinal toxicity, endocrinopathies disease, pneumonitis, rheumatologic toxicity, etc.⁷ Among irAEs, CIP is notable for its variable clinical severity and potentially fulminant course, rendering it a life‑threatening complication. To date, the underlying mechanism of CIP remains incompletely understood, and its management and treatment are usually based on clinical experience and existing practice guidelines.⁸ Challenges in prediction and diagnosis contribute to persistently high rates of severe pneumonitis and mortality. Accordingly, future research should focus on exploring molecular mechanism to find relevant predictive biomarkers as well as refine patient risk stratification to improve the early diagnosis rate of CIP and reduce its mortality.

Methods

A literature search of the PubMed database was performed using the following keywords: “immune checkpoint inhibitors” and “pneumonitis” and “immune checkpoint inhibitor-related adverse reactions.” Only research articles written in English were considered, and no predefined restrictions were set based on study type. Data sources were screened independently by three authors. Data analysis was performed by three authors.

Incidence rate and mortality of CIP

The incidence of CIP varies by cancer type. For instance, among patients receiving nivolumab or pembrolizumab monotherapy, the incidence of any-grade pneumonitis was 4.1% in non – small cell lung cancer (NSCLC), significantly higher than the 1.6% observed in melanoma. A similar incidence (4.1%) was reported in renal cell carcinoma, again exceeding that in melanoma. Nevertheless, most studies do not stratify CIP incidence by tumor type, and robust cross-cancer comparisons remain limited.⁹ Prior studies have reported that the overall incidence rate of CIP in clinical is about 3%–5%,^10–12 which is significantly lower than the incidence rate in the real world (about 9%–19%).^13,14 This disparity may reflect broader ICI use in routine practice and the restrictive inclusion criteria of clinical trials.¹⁵ However, a recent study on the incidence of CIP in NSCLC in the real world pointed out that it was only 2.49%, attributing earlier higher estimates to small sample sizes and overemphasis on baseline risk factors (e.g., cancer type, prior radiotherapy or chemotherapy, and pre‑existing lung disease).¹⁶ In addition, since the diagnosis of CIP is a diagnosis of exclusion, medical workers lack a unified diagnostic consensus during the diagnosis process, which leads to false positives or false negatives, thereby underestimating or overestimating incidence of CIP. Regarding mortality, CIP accounts for a substantial proportion of fatal irAEs, especially in the single drug treatment of anti-PD-1/PD-L1 (35%) and mortality rate of 15% or even higher.¹⁷ It is generally hypothesized that the occurrence of CIP indicates a more robust activation of the immune system against the tumor, potentially correlating with greater clinical benefit. However, recent studies have also shown that the median survival period of CIP is shorter than that of the cohort without CIP (428 d vs 1240 d).¹⁶ The beneficial effect of tumor inhibition from immune system activation and its detrimental effect from excessive inflammation affect the survival of patients treated with ICIs (Figure 1). More unified and accurate statistics should perform for better understanding the significance of CIP.

Figure 1.

The beneficial effect of tumor inhibition from immune system activation and its detrimental effect from excessive inflammation affect the survival of patients treated with ICIs.

Imaging modes and diagnosis of CIP

CIP manifests through several distinct radiographic patterns, primarily categorized into four types: 1) organizing pneumonia (OP); 2) nonspecific interstitial pneumonia (NSIP); 3) hypersensitivity pneumonia (HP); 4) acute interstitial pneumonia (AIP)/acute respiratory distress syndrome (ARDS). Research by Mizuki Nishino et al. identified the OP pattern as the most prevalent among these imaging manifestations.¹⁸ This finding is corroborated by a retrospective study of 64 CIP patients conducted by Myriam Delaunay, which reported a similar distribution of pattern frequency.¹⁹ The most common manifestations of OP are bilateral ground-glass or reticular changes, cumulative bilateral lung consolidation. A relatively specific feature of this pattern is the “reversed halo sign” (RHS), where a focal rim of consolidation surrounds a central area of ground-glass opacity.^20–22 In contrast, the NSIP pattern typically exhibits bilateral, peribronchovascular ground-glass or reticular opacities that demonstrate temporal and spatial homogeneity.²³ HP manifests as diffuse ground-glass changes and centrilobular nodules.¹⁸ AIP/ARDS manifests as diffuse ground-glass changes or parenchymal changes that can involve the entire lung.²⁴ A study on chronic CIP indicated that imaging changes in recurrent patients often appear in the same location, which may be related to organic changes in lung tissue.²⁵ In patients suffered from ICIs with CT imaging of pneumonitis, CIP can be diagnosed after excluding pulmonary infection, tumor progression, other causes of nonpulmonary interstitial disease like pulmonary vasculitis, pulmonary embolism and pulmonary edema. At the same time, it can be divided into four grades according to clinical symptoms, grade 1: asymptomatic, only clinical findings; grade 2: new onset of dyspnea, cough, chest pain, etc. or aggravation of original symptoms, affecting instrumental daily activities; grade 3: Severe symptoms, limited self-care ability; grade 4: life-threatening respiratory symptoms.^8,24,26,27 In addition to imaging, laboratory evaluation is essential and should include complete blood counts, C-reactive protein (CRP), procalcitonin (PCT), and other inflammatory markers. To facilitate differential diagnosis and exclude infectious etiologies, microbiological testing and culture are recommended, including sputum and nasopharyngeal specimens, among others. For suspected CIP cases that remain diagnostically challenging, bronchoscopy and, when feasible, lung biopsy may be considered. Further evaluation should be undertaken within a multidisciplinary team (MDT).

Management and treatment of CIP

CIP is treated differently according to its grade. Mild (Grade 1): Postpone the use of ICIs and resume them after improvement; Moderate (Grade 2): Suspend the use of ICIs and reassess the pros and cons before deciding whether to continue using them after recovery; Severe (≥Grade 3): Permanently discontinue ICIs.^8,24,26–28 For the treatment of CIP, corticosteroids are currently the main treatment drugs, which alleviate CIP through immunosuppressive effects. In cases of steroid-refractory CIP (defined as a lack of clinical improvement after 48–72 h of high-dose corticosteroid therapy) or recurrent disease, the addition of secondary immunosuppressive agents is recommended. Guideline-recommended options include infliximab (an anti-TNF-α monoclonal antibody), mycophenolate mofetil, and cyclophosphamide, among others.^27,29 Supporting this approach, a study by Jian Gao et al. successfully constructed a humanized mouse model of CIP, and proved that anti-TNF-α treatment significantly alleviated CIP.³⁰ Additional case reports have also documented favorable outcomes with anti-TNF-α therapy in steroid-refractory scenarios.³¹ However, another retrospective study pointed out that patients treated with anti-TNF-α combined with steroids showed worse survival outcomes compared with patients treated with steroids alone (median OS: 17 months vs 27 months).³² The effectiveness of anti-TNF-α drugs still needs to be verified by more basic and clinical experiments. The latest point of view is that, in addition to the prophylactic use of TNF-α blockade, targeting the PI3K-Akt signaling pathway—a crucial axis downstream of PD-1 engagement that regulates T-cell survival and activation – has been proposed as a novel strategy to ameliorate CIP by directly modulating dysregulated immune responses.³³ Furthermore, intravenous immunoglobulin also slows down the irAEs through immunosuppression, and related cases have also begun to be reported on CIP.³⁴ In addition to these commonly used immunosuppressants, some programs that are still in early research, such as anti-IL-6 receptor antibodies, anti-VEGF antibodies and soluble CTLA-4 mutants can also theoretically relieve CIP.^35–37 Meanwhile, a recent study by Yanlin Li et al. pointed out that the application of these immunomodulators (such as tumor necrosis factor alpha inhibitors, intravenous Ig, mycophenolate mofetil) has a considerable improvement effect in the treatment of grade 3–4 steroid-refractory CIP.³⁸ Looking forward, the treatment and management of CIP are expected to evolve in an increasingly personalized and individualized direction, leveraging biomarker-driven insights to tailor therapeutic strategies based on the specific immune pathology and severity profile in each patient.

Risk factors of CIP

The incidence and severity of CIP are significantly influenced by a variety of patient-specific clinical characteristics and treatment-related factors A primary risk factor is the type of primary malignancy (Table 1). Compared with other malignant tumors, the incidence of CIP in lung cancer is higher, and the onset time is shorter.^19,39 Several patient demographics and comorbidities also modulate risk. A history of heavy smoking (evaluated in pack-years) is a significant risk factor for developing all-grade CIP.^14,40 Interestingly, some studies show that women may also be an influential factor.^{13,14,41–44} For the use of ICIs, compared with PD-L1 inhibitors, PD-1 inhibitors have a higher risk of CIP.⁹ Besides, the use of CTLA-4 inhibitors does not increase the incidence of CIP, but the combination of CTLA-4 and PD-1 showed the greatest incidence of CIP.¹² Radiotherapy, as an important factors affecting both radiation pneumonitis and CIP, may conduct some common channels of mechanism such as common cytokines (IL-4, IL6, IL10 and IL17) release⁴⁵ and activation of common molecule (TGF-β, cGAS-STING, etc.).^46,47 A recent study pointed out that the simultaneous use of ICIs and radiotherapy will increase the incidence of whole-grade CIP,⁴⁸ and radiotherapy for curative purposes has a higher risk, but not correlated with radiotherapy space and time.⁴⁹ However, when combined with chemotherapy, the incidence and severity of CIP were reduced, which may be related to the damage to the immune system caused by chemotherapy (Figure 2).^50–53

Table 1.

The clinical characteristics, treatment, and outcomes of CIP.

Clinical characteristics	Treatment	Type of cancer	Total people	Incidence of CIP	Incidence of severe CIP	Relative risk	Relative risk of severe CIP	Other instructions
lung cancer18	PD-1/PD-L/CTLA-4/combination therapy	NSCLC/melanomas/ other types of cancer	1826	–	43.8	2.7	0.8	median (range) time: NSCLC vs melanomas(2.1 months vs 5.2 months)
squamous histologic type13	PD-1/PD-L/CTLA-4/combination therapy	NSCLC	205	32.6	–	2.29	–	relative risk is the ratio of squamous carcinoma to adenocarcinoma
smoking history38	PD-1/PD-L/CTLA-4/combination therapy	NSCLC/lung cancer/other types of cancer	142703	–	–	1.39	1.78	relative risk of severe CIP refers to the history of heavy smoking, i.e., smoking no less than 50 pack-years
previous lung diseases12	PD-1/PD-L/CTLA-4/combination therapy	NSCLC/SCLC	315	–	–	3.13	–	previous lung diseases refer to COPD by history, obstruction on spirometry, or emphysema on baseline chest CT scan.
Women13	PD-1/PD-L/CTLA-4/combination therapy	NSCLC	205	24.9	–	1.34	–	–
combination of multiple ICIs11	combination of PD-1 and CTLA-4/	NSCLC/lung cancer/ other types of cancer	12876	–	–	3.47	3.48	CTLA4 inhibitor ipilimumab treatment alone could not increase the risk of CIP
simultaneous use with radiotherapy42	PD-1/PD-L/combination therapy	NSCLC/SCLC	164	8.2	4.1	1.49	1.2	relative risk of curative-intent treatment and palliative radiotherapy:3.42
combination of chemotherapy46	PD-1/PD-L/combination therapy	advanced lung cancer	22178	–	–	0.27	–	Relative risk is the ratio to a single ICI

Figure 2.

Risk factors and mechanisms of CIP.

The mechanism of CIP

As the understanding of CIP continues to deepen, many speculations about the mechanisms of CIP have come out, but there is still no clear understanding of mechanism, as its low incidence and difficulty in constructing animal model. According to current research, the pathogenesis of CIP is attributed to disorders in T cell activity and proportion, reduction of Bcell number and function, changes of inflammatory cytokine profiles, and alteration of autoimmune antibodies.

Disorders in T cell activity and proportion

For the changes in T cell activity and proportion during the occurrence of irAEs, there are currently two main hypotheses, tissue-associated antigen and neoantigen theory. Tissue-associated antigen theory suggests that antigen sharing between tumor cells and normal tissues leads to irAEs, while neoantigen theory points out that the interaction between T cells targeting neoantigens and wild proteins is related to irAEs.⁵⁴ With the development of single-cell sequencing technology, it has significantly enhanced our comprehension of the function and dynamics of T cells.⁵⁵ T cells play a key role in immunotherapy and are also associated with CIP.⁵⁶ The analysis of TCR library in pneumonitis tissues for the first time shows that T cell clones in tumor and pneumonitis tissues are shared, so they may become the source of T cells in CIP.⁵⁷ An unsupervised cluster analysis study found that Tcms, CD4CD45RA^++-CD62L (expressing CD62L), a subset of CD4⁺ central memory cells, was significantly elevated in bronchoalveolar lavage fluid (BALF) samples from patients with CIP.⁵⁸ In a recent analysis of irAEs in melanoma, CD4⁺ T_M cell enrichment was also found in circulating T cells by single-cell sequencing technology in peripheral blood samples.⁵⁹ The increase of CD4⁺ T cells and the change of subsets may be one of the mechanisms of CIP. With the help of CD4⁺ T cells, CD8⁺ T cells acquire the characteristics of immune memory, and can expand and exert cytotoxicity when re-encountering antigens.⁶⁰ Sumit K. Subudhi et al. found that cloning of circulating CD8⁺ T cells preceded grade 2–3 irAEs.⁶¹ In an analysis of the characteristics of T cells in BALF from patients with CIP, a specific enrichment in the proportion of PD-1TIM-3CD8⁺ and TIGITCD8⁺ T cell subset was found, and the proportion of PD-1⁺PD-L1⁺cells increased most significantly.⁶² Previous studies also found that the CD8⁺ TNF-α^hi, IFN-γ^hi population was significantly increased in the BALF samples of patients with CIP, accompanied by increased expression of TNF-α and IFN-γ,⁶¹ which may be the mechanism of generation of CIP. Recently, some studies on the radioactivity of CD8⁺ T cells have also begun to appear in people’s sight.

Th17 cell acts as a special subgroup of helper T cells that can produce IL-17, which changes may be in parallel with changes in the immune cell landscape of tumors. Previously, Seon Hee Chang et al. found that Th17 cells are the main adaptive cells for lung inflammation in the carcinogenic K-ras mutant mouse model built by them, and the IL-17 secreted by Th17 cells can promote the acceleration of inflammation in the lung cancer model.⁶³ Recently, a prospective observational study further explained the relationship between Th17 cells and CIP. In this original study, Amelie Franken et al. mainly analyzed a kind of cell Th17.1 cells with the characteristics of both Th17 and Th1 helper T cells through the method of single cell transcriptomics, and found that it accumulated in a large amount in the BALF samples of CIP. At the same time, they also showed that pro-inflammatory monocytes aggregate Th17.1 cells and achieve inflammation, which contributes to the occurrence of CIP.⁶⁴

Reduction of B cell number and function

The anti-tumor effect of B cells is constantly being proved. A recent clinical trial found significant clonal expansion of B cells during tumor immunotherapy,⁶⁵ and earlier studies also showed that B cells were activated in various degrees during immune checkpoints therapy.⁶⁶ However, a recent study pointed out for the first time that the characteristics and functional defects of the regulatory B cell (Breg) repertoire may be related to the toxicity during immunotherapy，which may be because the B cells of toxic patients cannot produce anti-inflammatory drugs (IL-10) and pro-inflammatory factors (INF-γ, IL-6). In addition, elevated T_FH cells (usually elevated in autoimmune diseases with autoantibody production) due to the absence of helper neo-B cells may also be involved in the development of CIP.⁶⁷ In conclusion, the loss and lack of function of B cells may lead to changes in downstream immune cells and cytokines, thus increasing the risk of CIP.

Changes of inflammatory cytokine profiles

The increase of three cytokines IL-6、IL-17、TNF-α and IFN-γ has been reported in a large number of literatures. IL‑6 is a prototypical pro‑inflammatory cytokine. In CIP, elevated IL‑6 promotes the recruitment of immune cells to the lung, contributing to inflammatory lung injury. Concurrently, IL‑6 modulates Th17 cell differentiation and function, enhances their cytokine secretion, and thereby amplifies the progression of CIP. A retrospective study pointed out that the increase of IL-6 was related to the severity of the grade of CIP and poor prognosis.⁶⁸ The pro-inflammatory cytokine IL-17 plays a tumor promoting or anti-tumor role by inducing angiogenesis or the generation of cytotoxic T lymphocytes.^69,70 IL-17 contributes to immune-mediated lung injury by recruiting leukocytes via chemokines such as CXCL8. BALF in patients with CIP have shown significant elevations of TNFα and IFNγ. In parallel, the soluble cytokine TWEAK, which is highly expressed, engages its receptor Fn14 and, in concert with TNFα and IFNγ, activates the downstream transcription factor NFκB, thereby driving tissue inflammation and injury.⁷¹ In addition to the four major cytokines, IL-6、IL-17、TNF-α and IFN-γ, more and more cytokines are found to be involved in the occurrence of CIP. A longitudinal analysis of 98 melanoma patients receiving immunotherapy showed that the increased expression level of 11 circulating factors (G-CSF、GM-CSF、FRACTALKINE、FGF-2、IFNα2、IL12p70、IL1a、IL1B、IL1RA、IL2 and IL13) is related to high-level irAEs.⁷² In addition, some case reports also reported the increase of cytokines such as IL-2 and IL-35 in patients with CIP, indicating that it has the potential as a biomarker for the occurrence of CIP.^73,74 Yi Na Wang et al. found that not only the increase of IL-17 but also the increase of IL-35 were observed in the blood and BALF samples of patients with CIP. At the same time, with the clinical recovery or remission, the levels of IL-17 and IL-35 will gradually decline.⁷⁴ Interestingly, the anti-inflammatory factor IL-10 can also be observed in some case reports and small sample studies,^68,75 which may indicate a more intense inflammatory reaction and occurrence of more severe CIP. However, IL-8, another anti-inflammatory factor, showed a downward trend in the occurrence of CIP. This may suggest that the occurrence of CIP is not a simple inflammatory mechanism.⁷⁶ Another study also suggested that elevated IL-8 reduces the benefits of ICIs due to its inhibition of antigen presentation.^77,78 When IL-8 decreased, it represented an enhanced immune response and a higher incidence of CIP.

Alteration of autoimmune antibodies

With the development of research on CIP, people found that the change of autoimmune antibodies may be one of the mechanisms of CIP. A study on the autoantibodies of patients with melanoma in the late stage of 133 who received ipilimumab treatment found that the production of autoimmune antibodies may be related to the occurrence of irAEs.⁷⁹ The study of Yukihiro Toi et al. shows that preexisting autoimmune antibodies (anti-nuclear antibody, rheumatoid factor, etc.) also play a certain role in the progress of irAEs.^80,81 Besides, another recent study found that lower baseline autoantibody levels and greater variation in concentrations were associated with more significant irAEs.⁸² Salahaldin A. Tahiretal. et al. found that the level of anti CD74 autoimmune antibodies in patients with pneumonitis significantly increased after treatment, suggesting that anti CD74 autoimmune antibodies may play a role in the occurrence of CIP.⁸³ Although these accumulating findings suggest an association between autoantibodies and irAEs, research focused on identifying specific autoantibodies directly causative of or highly specific for distinct irAEs like CIP remains limited. The current evidence often points to correlations rather than causative roles, and many studies are constrained by sample size and heterogeneity in patient populations and ICI regimens. Therefore, more extensive and mechanistic explorations, including large-scale prospective studies and in-depth investigations into the functional roles of these autoantibodies, are warranted in the future to validate these associations, elucidate their precise mechanisms, and explore their potential as predictive biomarkers or therapeutic targets.

The latest early diagnostic strategy for CIP

Currently, the early diagnosis of CIP remains challenging due to the absence of a single, gold-standard objective biomarker. Diagnosis primarily relies on a comprehensive clinical assessment that integrates clinical symptoms, radiological features, and exclusion of alternative causes (particularly infections), underscoring its nature as an exclusion diagnosis. This process heavily depends on clinical experience and a high index of suspicion. Fortunately, with advancing research into its mechanisms and pathogenic features, several promising tools and strategies for the early detection and diagnosis of CIP are emerging (Figure 3).

Figure 3.

The latest early diagnostic strategy for CIP.

Application of artificial intelligence in imaging

The accurate differentiation of CIP from other pulmonary pathologies, such as radiation pneumonitis (RP), on computed tomography (CT) images remains a significant diagnostic challenge in oncology. To address this, researchers have proposed the conversion of subjective imaging findings into quantitative, data-driven biomarkers to enhance diagnostic specificity and accuracy.^84–86 Radiomics, an emerging field that employs advanced mathematical algorithms to extract and analyze a high-dimensional set of quantitative features from medical images, has demonstrated considerable potential in improving diagnostic precision and facilitating personalized medicine strategies when integrated with machine learning techniques.^87–89 A particular diagnostic difficulty arises in patients receiving both immunotherapy and radiotherapy, where the imaging manifestations of CIP and radiation pneumonitis can overlap. To address this, integrated dosiomics-radiomics models have been proposed. Dosiomics involves extracting quantitative features from radiation dose distribution data.^90,91 Combining these with radiomic features from CT images can create more powerful predictive models for patients undergoing concurrent radiotherapy and immunotherapy. With the deepening of the exploration between machine learning and CIP, some researchers have begun to propose that risk factors, clinical manifestations and other characteristics can also be added to the establishment of the model and various factors can be calculated to make it more effective usability.^92,93 However, as more factors are added, the establishment and verification of the model become more difficult. In the future, multi-center studies with large samples will be required to train and improve the model.

Utility of single-cell sequencing technologies and diagnostic strategies

Single-cell sequencing technologies are emerging as powerful tools for elucidating the pathogenesis and potentially aiding in the diagnosis of CIP. By enabling high-resolution analysis of diverse patient-derived samples, these technologies facilitate the precise characterization of alterations in immune cell subsets and transcriptional programs within the pulmonary microenvironment. Peripheral blood rarely has a specific role in the diagnosis of inflammation in the inherent impression. However, through single-cell sequencing technology, the characteristics of cell subpopulations in it are clearer. Shoiab Bukhari et al. discovered unique cell subsets and gene changes during the occurrence of irAEs through single-cell RNA sequencing, indicating that different irAEs are related to different circulating T cell subsets.⁹⁴ Pleural effusion is also another nonspecific specimen. Earlier, researchers used TCR second-generation sequencing technology to analyze a pleural effusion specimen from a clear cell carcinoma and pointed out that TCR clonality is important in the diagnosis of CIP.⁹⁵ In the future, with the popularization of single-cell sequencing technology, the role of pleural effusion in guiding diagnosis may be valued. As specimens that need to be obtained under bronchoscopy, BALF and lung tissue are closer to the onset site of CIP and have a more specific diagnostic role. Currently, various single-cell sequencing studies on BALF and lung tissue of patients with CIP are emerging. At the same time, these studies have also revealed the immune infiltration environment of CIP.⁷¹ In addition to the pulmonary immune infiltration environment revealed by single cells, the latest studies also indicate that the pulmonary microbiota and acylcarnitine metabolism disorders in the lungs may be related to the pathogenesis of CIP. Microbiome has begun to be used to reveal the mechanism of CIP.⁹⁶ Although single-cell sequencing technology has excellent diagnostic guidance in the detection of various specimens, the cost of this technology is relatively high and there is a lack of a gold standard. In the future, continuous technological advancement and cost optimization are needed to achieve its popularization.

Detection of traditional markers and novel markers

Traditional markers, including blood routine, blood biochemistry, liver and kidney function, etc., have the function of predicting CIP. In addition to white blood cells, neutrophils, monocytes, etc. that represent inflammatory indicators, a latest study indicates that low levels of HB before treatment are an independent risk factor for CIP. Xiaoyu Liu et al. believe that after HB deficiency, This leads to a decrease in oxygen transport capacity, and then hypoxia leads to defects in lung function, making it more susceptible to CIP.⁹⁷ In addition to direct detection of patients own biomarkers, some researchers have also proposed to add nanoreporters that react with specific markers during the immunotherapy process, including imaging probes to monitor pharmacokinetics during the immunotherapy process and post-treatment changes at the cellular and molecular level, and through a complete imaging detection system, an early and comprehensive view of the immunotherapy process can be achieved.⁹⁸ The latest research indicates that the generation, invasion, and proliferation of blood vessels are important regulatory factors in the process of ICIs. A new study pioneered the use of the nodular vascular system to extract tumor vessel tortuosity from CT images as a predictive factor for ICIs response. Although the predictive function for irAEs such as CIP has not been formally studied, this new biomarker is noninvasive, inexpensive, and comes from routine imaging examinations. There is hope to become a member of the early diagnosis plan.⁹⁹ In addition, proteins or molecules that play a role in immune responses, such as semaphorin 4A and serotonin, have also begun to be applied to predict ICIs reactions.^100,101 In the future, their predictive role in irAEs will also be further explored.

Differential genes and non-coding RNA

Compared with cell biomarkers and cytokines, differential genes and non-coding RNA may have better predictive ability as upstream factors. An original study carried out by Diego Chowell et al. pointed out that HLA-I genotype would affect the survival rate of patients using ICIs and change their clinical results.¹⁰² The study of Vivek Naranbhai et al. also showed similar results. They even suggested that the mutant HLA-A *03 had significantly shortened survival time.¹⁰³ The latest article published in Lancet pointed out that the mutation at position 161 of HLA-A is the key to distinguish HLA-A *03 from other lineages, and further validated HLA-A *03 was a key factor in the poor prognosis of immunotherapy through the study of four large renal cell carcinomas.¹⁰³ Afaf Abed et al. were the first to associate HLA-I genotype with irAEs and pointed out that HLA-A *03 increased the risk of irAEs.¹⁰⁴ Pierpaolo Correale et al. studied the relationship between HLA-I and irAEs in more depth. They showed that HLA-A *02 had a longer progression-free survival period than HLA-A *01 with adverse outcomes. At the same time, the heterozygosity at A site showed a worse overall survival period, while the heterozygosity at DRB1 site showed a longer survival period. At the same time, it was found that the highest incidence of CIP was detected in patients with HLA-B * 35 and DRB1 × 11 alleles.^105,106 In addition, tumor mutational burden (TMB), which expresses changes at the gene level in tumor, has also been studied. A study in multiple cohorts, open label, nonrandom, phase 2 KEYNOTE-158 showed that there was a close relationship between high tissue TMB and irAEs, which is a novel and potential biomarker.¹⁰⁷ MiRNA has been proposed as an excellent biomarker for some time, and some studies have suggested that some miRNAs (miR-23b-3p, miR-204, miR-17-5p) play a certain role in the prediction of non-small cell lung cancer.^108–110 A systematic study also analyzed the ability of 38 commonly used miRNAs to evaluate the therapeutic effect of ICIs.¹¹¹ However, the predictive role of these upstream factors in CIP is not clear, and due to the dilution effect of technology and tumor purity, their role as predictors needs to be proved by large-scale experiments.¹¹²

Summary

CIP is a rare but potentially life-threatening irAE associated with ICI therapy. Despite its low incidence, the risk of severe pneumonitis and mortality is high, especially with the increasing use of immunotherapy. At present, the diagnosis of CIP mainly depends on imaging and clinical features, and the hierarchical management method is adopted to treat CIP. However, the treatment of steroid refractory or drug-resistant pneumonitis still needs to be studied urgently with more drugs and protocols. According to current research, the pathogenesis of CIP is attributed to disorders in T cell activity and proportion, reduction of B cell number and function, changes of inflammatory cytokine profiles, and alteration of autoimmune antibodies. With the deepening of mechanistic research, a number of novel early diagnostic strategies have gradually entered clinical practice, including: Application of Artificial Intelligence in Imaging, single-cell sequencing technologies, single-cell sequencing technologies and differential genes and non-coding RNA. Future research should focus on developing more reliable animal and sample models to better understand the mechanisms of CIP and identify new therapeutic targets. At the same time, we also need more reliable, economical, and convenient strategies for early diagnosis of CIP.

Biography

Ximing Xu, Standing Committee of Tumor Radiation Protection Specialized Committee of Chinese Anti-cancer Association, Standing Committee of Tumor Thermal Infusion Chemotherapy Group of Chinese Anti-cancer Association, Standing Committee of Tumor Specialized Committee of Chinese Physicians Association of Integrative Medicine, Vice Chairman of Precision Medicine and Tumor MDT Specialized Committee of Chinese Society of Research Hospitals, Member of Tumor Precision Radiotherapy Specialized Committee of Wu Jieping Foundation, Chairman of Tumor Specialized Committee of Hubei Microcirculation Society, Deputy Chairman of Tumor Molecular Targeted Therapy Specialized Committee of Hubei Anti-Cancer Association. Vice Chairman of Tumor Molecular Targeted Therapy Committee of Hubei Anti-cancer Association, Standing Committee of Internal Medicine Treatment Committee of Hubei Anti-cancer Association.

He has presided over two National Natural Science Foundation of China, one sub-project of National Key Research and Development Program of China, many projects of Doctoral Fund of Ministry of Education, Natural Science Foundation of Hubei Province, and Wuhan Morning Glory Program. He has published more than 40 SCI papers as the first and the corresponding authors, and dozens of papers in Chinese core journals, and he has edited two monographs and has won one award of Provincial Scientific and Technological Progress Award.

Funding Statement

This work was supported by National Natural Science Foundation of China [No. 31971166].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Ethics statement

This article is a review and does not involve human subjects. Therefore, the ethical statement does not apply to this article.