Patterns of immune-related adverse events in patients treated with immune checkpoint inhibitors: a Brazilian real-world analysis

Immune checkpoint inhibitors (ICIs), such as anti-cytotoxic T-lymphocyte antigen-4 and anti-programmed cell death 1 (anti-PD-1) and its ligand, have revolutionized the treatment of several malignant neoplasms.1 2 Initially, these drugs were approved as single agents; however, their use in combination with distinct ICIs, targeted agents, and chemotherapy has emerged as a successful strategy.3

Given the importance of checkpoint molecules in the modulation of the immune response, the clinical use of ICIs has resulted in a new toxicity profile, the so-called immune-related adverse events (irAEs).4 Such events are mediated by the hyperactivation of the immune system, leading to autoimmunity-like reactions.5 6

Data generated during clinical trials cannot necessarily be generalized to real- world situations, especially because the study population is often underrepresented by particular subgroups, such as elderly people, those with borderline functionality (ECOG- PS≥2), or with brain metastasis, owing to the study's strict inclusion criteria.2 7 8

Therefore, this study aimed to describe the occurrence of irAEs among unselected patients with advanced solid tumors treated with ICIs in two Brazilian cancer centers and their association with clinical characteristics and disease outcomes.

This real-life retrospective study describes the use of ICIs in a Brazilian tertiary center outside of a clinical trial. In this cohort, most patients presented with NSCLC and melanoma and 40% of the patients have received ICIs in the first line. Our study revealed that 49.7% of patients presented with at least one irAE, and 8.6% developed a grade 3 or 4 event. The permanent discontinuation and hospitalization rates due to irAEs were 2.7% and 7.0%, respectively. A total of 15.1% of patients received supraphysiological dose of steroids for the management of these events, and two patients required biological treatments. Median PFS and OS were 8.26 and 29.6 months, respectively, and the occurrence of irAEs was associated with longer survival outcomes in the overall population.

Despite the heterogeneity regarding the type of tumor and treatment used, patient characteristics were similar to those of other real-world cohorts.8 9 Moreover, our study demonstrates that uninvestigated subgroups, such as PS=2 (9.7%), older age (≥75 years, 25.9%), and BM (18.9%), represent an important proportion of real-life patients treated with ICIs, which is in accordance with retrospective studies.9

Similar to our study, a prospective clinical trial that evaluated pembrolizumab in NSCLC showed a frequency of any adverse events of 71% but a slightly higher frequency of grade ≥3 events (13%).2 Notably, in that study, the frequency of irAEs was lower than that in our study (17% vs. 49.7%).2 This divergence might have occurred because irAEs vary according to clinical experience and the criteria used. Other retrospective studies involving four types of cancer demonstrated similar frequencies of irAEs (50.9% and 46.2%) and similar types of the most frequent events (hepatitis, rash, colitis, pneumonitis, and hypothyroidism) as in our study.10 11 Another Brazilian retrospective cohort study also showed that endocrine and cutaneous AEs were predominant.12 A retrospective real-world study was conducted on 1.905 patients with advanced lung cancer in China which showed a different spectrum of irAEs, with higher incidence of pneumonitis and lower incidence of gastrointestinal toxicities and arthritis.13 With regard to irAEs ≥ grade 3/4, real-world data demonstrated a similar incidence (5.8-13.1%).7 9 11 13 Of note, Shi et al. (2022)13 and Skribek et al. (2021)7 described pneumonitis as the most common irAEs ≥ grade 3/4 and Nice et al. (2021),11 colitis followed by pneumonitis.

Our study showed that only a minor proportion of patients required supraphysiological dose of steroid administration for managing irAEs, which agrees with the results of Skribek et al. (2021)7 (15.8%). The need for biological treatments was even less frequent, as previously observed.7 12 The hospitalization incidence due to irAEs was similar to that reported by Nice et al. (2021)11 (7%); the most common causes also coincided (colitis and pneumonitis). Ahern et al. (2021)14 reported a higher hospitalization rate than that observed in our study (11%). Finally, it is important to note that adverse events were not an important cause of treatment discontinuation, possibly because of the predominance of monotherapy. However, the discontinuation rate was divergent from that reported by Bjørnhart et al. (2019)8 (24% of patients terminated ICI monotherapy due to irAEs); but is in accordance with other studies.15 We did not observe any deaths in our cohort due to adverse events, similar to the larger cohorts.16 Nevertheless, Shi et al. (2022)13 revealed 11 deaths (11/1905, 0.6%) due to irAEs (pneumonitis, liver failure and myocarditis).

The median follow-up period after the first ICI was 12.4 months. The efficacy data were in accordance with those reported in some clinical trials. The KEYNOTE-001 study involving patients with NSCLC and melanoma, the most frequent types of cancer in our cohort, showed a median OS of 22.3 months (95%CI: 17.1 to 32.3 months) in NSCLC treatment-naive patients and 10.5 months (95%CI: 8.6 to 13.2 months) in previously treated NSCLC patients, which are less than those demonstrated in our group.2 In a melanoma study, the patients exhibited a shorter OS, 23.8 months (95%CI: 20.2-30.4), but a similar PFS of 8.3 months (95%CI: 5.8-11.1).16 However, real world data have demonstrated shorter median OS and PFS rates. An Italian cohort of patients with NSCLC had a median OS of 11.3 months and a median PFS of 3 months for the pre-treated patients.17 Another retrospective study of Danish NSCLC patients showed a median OS of 16.1 months and a median PFS of 6.4 months in the general cohort.8

In our study, the exploratory analysis suggested that the occurrence of irAEs was associated with longer survival. This finding agrees with those of several retrospective and prospective studies.13 15 18 19 One possible explanation for this finding is the hypothesis that a more pronounced state of basal immune activation would make patients more likely to present with autoimmune events but would also enhance the antitumor response.10 Studies that used landmarks or considered time-varying covariates also showed this correlation.18 20 21 Multiple studies have shown that vitiligo is associated with a lower risk of disease progression and death in patients with melanoma treated with ICIs.22 23 Maher et al. (2019)24 found that responders and non-responders to ICIs for urothelial cancer had a differential proclivity for developing irAEs, given the same duration of exposure. Shimozaki et al. (2020)10 suggested that irAES would be a predictive factor and that the development of multiple irAEs (more than one in the same patient) was correlated with longer OS compared to single irAEs. In addition to irAE development, ECOG PS and line of treatment were associated with outcomes in the multivariate analyses. However, in contrast to previous studies, patients aged ≥75 years had worse OS in our multivariate analyses.9

The limitations of our study are its retrospective nature, the heterogeneity and small size of the study population, and the fact that it was based on reference center data. Moreover, patients whose disease progresses or who die have a shorter follow-up period and less treatment exposure, which can impact the association between irAE development and outcomes. Given the retrospective nature of the study, it is possible that grade 1-2 adverse events may have been underestimated, although more severe AEs are likely reported in the patients' charts. However, there are few studies on ICI administration in Brazilian services, and our study illustrates this scenario properly.

This study shows that irAEs are frequent in real-world cohorts; most cases are easily manageable and do not require biological treatments. Moreover, they were responsible for a low percentage of hospitalizations and treatment interruptions. Notably, we found an association between irAEs and improved outcomes, which increases the importance of timely diagnosis and a correct management of these adverse events.