Introduction
In recent years, immune checkpoint inhibitors (ICIs) targeting checkpoints such as cytotoxic T-lymphocyte-associated-4 (CTLA-4), programmed cell death 1 (PD-1), and programmed death-ligand 1 (PD-L1) have been found to be essential components in the treatment of a wide range of cancers [
1‐
3]. The binding of PD-1 to its ligand PD-L1 and the binding of CTLA-4 to its ligand CD80/CD86 downregulate T-cell activation, leading to immune escape of tumor cells [
4]. ICIs can boost the activation of T cells by blocking the engagement of the above receptors and ligands [
5]. Although the survival benefit of ICIs is well recognized, they also result in immune-related adverse events (irAEs) [
6], especially in patients with preexisting autoimmune diseases (PADs).
Rheumatologic PAD is a common PAD associated with cancer. Compared with the general population, patients with rheumatologic PADs, such as rheumatoid arthritis (RA) [
7], systemic sclerosis [
8] and Sjogren’s syndrome [
9], have an increased risk of specific cancers. It is worth noting that some patients suffer from both rheumatologic PAD and cancer. A study by Khan et al. included 210,509 lung cancer patients, of which 9.4% also had rheumatologic PAD [
10]. Studies have reported that the PD-1 pathway and CTLA-4 pathway may play potential roles in the occurrence and development of rheumatologic PAD [
11‐
13]. These findings suggested that ICIs may theoretically increase the risk of rheumatologic PAD flares. For this reason, patients with rheumatologic PAD have largely been excluded from clinical trials. At present, available evidence regarding the exact incidence of disease flares, new-onset irAEs or cancer treatment outcomes in this population remains scarce. Considering the increasing dependence on ICIs for the treatment of tumors and the high proportion of cancer patients with rheumatologic PAD, it is especially essential to explore the influence of rheumatologic PAD on ICI treatment outcomes.
Hitherto, there has been increasing interest in the eligibility of receiving ICIs in patients with PAD, and several meta-analyses have assessed the safety and effectiveness of ICIs in patients with cancer and PAD [
14‐
16]. However, these studies focused on a wide variety of autoimmune diseases, and none of them separately reported the incidence of rheumatic disease flares, new-onset irAEs or ICI efficacy in cancer patients with preexisting autoimmune rheumatic diseases. Although multiple studies have reported the efficacy and safety of using ICIs in these patients, a definite conclusion has not been reached according to the results of each single study [
17‐
39]. Based on the above background, our study aimed (i) to summarize the incidence of rheumatologic PAD flares, the incidence of new-onset irAEs, and the rate of discontinuation; (ii) to investigate the objective response rate (ORR) or disease control rate (DCR); and (iii) to explore the impact of baseline anti-rheumatic treatment on the efficacy and safety of ICI treatment in patients with rheumatologic PAD and cancer.
Discussion
PD1/PD-L1 and CTLA-4 are the fundamentals of immune regulation [
42]. Although ICIs, including antibodies against PD-1/PD-L and CTLA-4, have yielded satisfying outcomes in terms of patient survival, they can also disrupt self-tolerance and lead to unique irAEs [
43,
44]. With the widespread application of ICIs in patients with advanced malignant tumors, irAEs caused by ICIs have been adequately discussed in a series of clinical trials [
45,
46]. Translational studies have shown that multiple pathways, such as cytokine, autoreactive T-cell and autoantibody pathways, may affect the development of irAEs [
42]. However, the safety and efficacy of ICIs in patients with rheumatologic PAD are unknown, as these patients have largely been excluded from clinical trials because of the increased risk of flares. Given the negligible number of rheumatologic PAD patients who may benefit from ICI therapy, an accurate determination of the risk-benefit ratio of ICIs in rheumatologic PAD patients is crucial. To our knowledge, our meta-analysis is the first study to investigate this issue. In this meta-analysis, we conducted a meta-analysis of 23 studies to investigate the impact of rheumatologic PAD on TirAEs, flares, new-onset irAEs and treatment efficacy in patients treated with ICIs. Overall, in this meta-analysis, the pooled incidence of TirAEs was 64% (95% CI 55%-72%). In addition, the incidence of TirAEs was 77% in the anti-CTLA-4 therapy group and 58% in the anti-PD-1/PD-L1 therapy group.
Immune cells in the system affected by PAD are generally abnormally activated, and this system is prone to adverse reactions when patients receive drugs to manipulate the immune environment. In contrast, when the immune cells of other systems unrelated to PAD are in a normal status, the frequency of irAEs is similar to that in the normal population. A recent study from Cai et al. [
47] indicated that PAD patients tended to experience irAEs in the same system involved in PAD, whereas the incidence of irAEs concerning other systems that were not affected by PAD was similar to the incidence of irAEs in the non-PAD group. In this meta-analysis, we also reached a similar conclusion: the incidence of flares (41%) was greater than the incidence of new-onset irAEs (33%). The incidences were 7% (95% CI 2%-14%) for Grade 3–4 flares and 12% (95% CI 9%-15%) for Grade 3–4 new-onset irAEs. In addition, 24% of patients discontinued ICIs because of rheumatologic PAD flares or new-onset irAEs. Notably, permanent discontinuation of ICIs may be required because of severe irAEs. Rheumatologic PAD patients require close monitoring for irAEs.
Several previous studies [
48,
49] have shown that irAEs caused by ICIs are associated with improved treatment outcomes in cancer patients. Thus, some researchers speculate that PAD patients may benefit more from ICIs due to their tendency toward immune activation. In this meta analysis, the pooled ORR was 30% and DCR was 44% in rheumatologic PAD patients. Additionally, our results showed that flares of potential rheumatologic PAD did not have an association with the ORR (RR = 1.14, 95% CI 0.62–2.08). It is necessary to interpret the results with caution due to selection bias. Patients with a good prognosis may not be treated with ICIs in order to avoid flares of PAD, while patients with a poor prognosis have no choice but to receive ICIs. Due to the heterogeneity of tumor type and tumor stage, we did not further classify the patients into subgroups based on these factors. More prospective studies focused on specific tumor type are required.
A meta-analysis of Xie et al. involving diverse PAD patients revealed that the pooled incidence of flares was 35% (95% CI 29–41%), and compared with other systemic PAD patients, rheumatologic PAD patients had an increased risk of flares; however, these two groups were not significantly different [
15]. In the present study, the pooled rate of underlying rheumatologic disorder relapse (41%) was greater than that in the previous meta-analysis. A meta-analysis involving 123 patients also revealed that RA flares were the most common [
16]. Previous studies have verified that the expression of PD-1 and PD-L1 is significantly elevated in both early and established RA and that the expression of PD-1 is correlated with the severity of synovial inflammation [
50,
51]. In addition, the expression and function of CTLA-4 have been confirmed to be likely related to the pathogenesis of RA [
52]. Similarly, the PD-1/PD-L1 and CTLA-4 pathways also play roles in the pathophysiology of other rheumatologic PADs, such as myositis, SLE and Sjögren syndrome [
53,
54]. A possible explanation for the greater rate of flares in RA patients was that RA had a stronger connection with the PD-1/PD-L1 and CTLA-4 pathways than other PADs. Another possible explanation was the difference in the use of anti-rheumatic drugs between RA and other rheumatic diseases. Rheumatologists commonly use methotrexate, biologics, and JAK inhibitors for RA, which are typically discontinued when patients are diagnosed with cancers. However, for other rheumatic diseases, glucocorticoids still play a central role. When patients are diagnosed with cancers, glucocorticoids are often continued because they cannot be abruptly discontinued. In this meta-analysis, further exploration of relapses between RA and other subtypes revealed that the rate of flares in RA patients was statistically significant higher than patients with other rheumatologic diseases. Due to the significant heterogeneity in diagnosis of flares identified by different clinicians, the higher rate of flares in the RA population needed to be further confirmed based on individual level data. Interestingly, the rate of new-onset irAEs in RA patients was not greater than that in other rheumatologic PAD patients, which was exactly the opposite of the rate of flares. The contradictory effects of ICIs on flares and new-onset irAEs suggest that there are different mechanisms that lead to the occurrence of flares and new-onset irAEs.
We further conducted subgroup analyses to investigate whether region and ICI type had an impact on the safety of ICIs. We found that regional factors may influence drug safety assessments. The incidence of TirAEs and flares in the Australian population was significantly greater than that in other regions. Patients from different regions may have various ethnic backgrounds, disparate treatment standards, different geographic environments, and CTLA-4 gene polymorphisms [
53]. These factors may lead to differences in clinical outcomes. In addition, after removing the two Australian studies, the pooled results were only slightly lower than before. The small sample size and low weight of the Australian studies could explain this phenomenon. In the subgroup analysis of new-onset irAEs stratified by region, only one study was included in Australia, so no significant results were obtained. With knowledge of this difference, further race-conscious research is urgently needed to confirm the association between geographical region and the safety of ICIs.
Regarding the types of ICIs, flares occurred more often in the anti-CTLA-4 therapy group than in the anti-PD-1/PD-L1 therapy group. We also found a similar trend in the subgroup analysis of the TirAEs. These findings was consistent with the conclusion that the risk of any grade of irAEs was greater in the anti-CTLA-4 therapy group than in the anti-PD-1/PD-L1 therapy group in the general cancer population [
55]. However, another meta-analysis conducted by Abdel-Wahab et al. reached the contradictory conclusion that the rate of flares in patients receiving anti-PD-1 therapy was greater than that in patients receiving anti-CTLA-4 therapy [
16]. This disparate result may be due to the difference in the types of autoimmune diseases between our meta-analysis and previous meta-analysis and the small sample size of the previous study. Another potential confounding factor for these results was the imbalance in terms of the number of patients receiving different ICI regimens in this meta-analysis. There were 11 studies that included 159 patients who received anti-PD-1/PD-L1 therapy, while there were only 3 studies that included 32 patients who received anti-CTLA-4 therapy. The number of patients receiving anti-PD-1/PD-L1 monotherapy was significantly greater than that receiving anti-CTLA-4 monotherapy. The fact that treatment regimens were not balanced in rheumatologic PAD patients may have affected the results.
Horvat et al. reported that the use of corticosteroids for the treatment of irAEs in cancer patients was not associated with the efficacy of ICIs [
56]. In this meta-analysis, we examined the association between the use of baseline anti-rheumatic therapy for rheumatic disease and patient outcomes. The results suggested that the incidence of flares, incidence of new-onset irAEs and ORR in patients with baseline anti-rheumatic therapy at the start of treatment were similar to those in patients without anti-rheumatic therapy, in line with prior studies focused on patients with diverse PADs [
28,
30]. However, Arbor et al. [
57] reported that the use of baseline corticosteroids had a negative influence on the efficacy of ICIs. The underlying mechanism may be that baseline corticosteroid treatment weakened the proliferative burst of CD8-positive T cells required in response to ICIs. The impact of anti-rheumatic therapy on irAEs and outcomes may be different for patients receiving different types of anti-rheumatic therapy for their rheumatologic diseases. However, because the anti-rheumatic therapy regimens used in each study were not uniform and included hydroxychloroquine, methotrexate, prednisone, azathioprine, etanercept, and sulfasalazine, we did not stratify patients according to the type of anti-rheumatic therapy. This phenomenon needs to be further studied in future research. Notably, due to retrospective nature of the most included studies, there was inherent selection bias in evaluating the impact of baseline anti-rheumatic treatment. Patients with higher risk of flares or active rheumatologic PAD were more likely to require anti-rheumatic therapy.
There are also several factors that may affect the safety and effectiveness of ICIs in rheumatologic PAD patients. First, patients who have clinically active disease or more severe disease at the time of ICI therapy may have a greater rate of flares [
18,
28]. However, only two articles have reported whether rheumatologic PAD patients have active symptoms, making comparisons of associations between different activities at baseline and outcomes difficult. Second, the incidence of irAEs may be affected by treatment strategies. A previous meta-analysis suggested that ICIs combined with chemotherapy reduce the incidence of irAEs in advanced NSCLC patients compared with that of ICI monotherapy, possibly due to the immunosuppressive effect of chemotherapy. Bone marrow suppression by chemotherapy may limit immune system overactivation [
58]. There was no detailed explanation of whether systemic chemotherapy was administered prior to ICI therapy or during ICI therapy in the included studies. Therefore, subgroup analysis based on treatment strategies could not be carried out. In addition, due to a lack of data, the impacts of sex, stage, line of ICI treatment, presence of PAD-related autoantibodies, ICIs type, drug dose, underlying disease type, underlying disease activity, comfort of treating physicians, patient preference and timing of cancer diagnosis on outcomes were not assessed. Discussing the association between these potential influencing factors and irAEs will be the research area of future studies.
To date, our study is the first meta-analysis to evaluate the safety and efficacy of ICIs in patients with rheumatologic PAD and fill gaps in knowledge. Unlike previous studies, this study specifically focused on individuals with rheumatologic PAD rather than individuals with various immune diseases. Our study can reduce heterogeneity caused by PADs in different systems. Immune-mediated endocrine dysfunction, such as thyroid dysfunction, is one of the most common irAEs and does not hinder further treatment with ICIs [
59]. If the PAD included the endocrine PAD patient subtypes, there was a confounding effect on the results [
60]. In addition, we further discussed the impact of risk factors such as region, type of ICI, and rheumatologic disease subtype on the safety and effectiveness of ICIs and provided a reference for clinical physicians to identify high-risk populations.
However, there are certain limitations to the present study. First, almost all the included studies were single-arm observational studies, and patient information was obtained from the patients’ medical records. In such studies, incomplete reports may lead to an underestimation of the incidence of irAEs. Second, the largest confounding factor was not addressed in the included studies. There are no unified standards in terms of patients with PAD who were offered ICIs. Compared to patients with active disease requiring immunosuppressants and with better prognosis cancers, clinicians prefer to use ICIs in patients who are inactive and not receiving immunosuppressive therapy and who have a worse prognosis. Most included patients had quiescent disease and inactive symptoms in some studies, while patients with more severe rheumatic disease were under-represented, resulting in selection bias in this meta-analysis. Third, there was another large confounding by indication bias in our meta-analysis. Some authors of the included studies were rheumatologists, and for patients with less severe disease who were not referred for rheumatologic assessment and were managed by the oncology team, such patients may not be included. Only patients with symptoms significant enough to be referred for the assistance of rheumatologists were reported by rheumatologists. Finally, the final sample size was relatively small, although all studies reporting the safety and effectiveness of ICIs for the treatment of rheumatologic PAD were included, which may limit the generalizability of our results. More large-scale prospective studies are necessary to improve the level of evidence.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.