Association between PD‐1 inhibitor‐related adverse events and frailty assessed by frailty index in lung cancer patients

Abstract Background The programmed cell death protein 1 (PD‐1) inhibitor, as one of the immune checkpoint inhibitors (ICIs), is the standard treatment for advanced lung cancer. However, immune‐related adverse events (irAEs) remain poorly understood toxicities. It is unclear whether frailty plays a role in the occurrence of irAEs. Thus, we assess whether irAEs occur more often in frail patients than in non‐frail patients according to the Frailty Index (FI). Methods A retrospective study was conducted. Medical records from lung cancer patients treated with PD‐1 inhibitors (Sintilimab, Camrelizumab, Tislelizumab, and Pembrolizumab) at Peking University First Hospital (May 2018–June 2022). Patients were categorized into non‐frail and frail groups according to a cut‐point of 0.25 by FI. The FI calculation included 28 baseline variables, all of which were health deficits measured by questionnaires and body measurements. Results The statistical analysis included 114 advanced lung cancer patients. The median age was 66 years, and the male/female ratio was 4.7:1 (94/20). Approximately 39 (34%) were classified as frail. PD‐1 inhibitor‐related adverse events occurred in 17.5% of patients, and 6.1% experienced irAEs of grade ≥3. There was no significant difference in the occurrence of irAEs (14.7% vs. 23.1%, p = 0.26), grade ≥ 3 irAEs (5.3% vs. 7.7%, p = 0.93), and treatment discontinuation due to irAEs (12.0% vs. 17.9%, p = 0.39) between non‐frail and frail patients. However, frail patients are more likely to have more than one type of irAEs and are more possibly to have checkpoint inhibitor pneumonitis (CIP) than non‐frail patients when they use PD‐1 inhibitors (p < 0.05). Frail patients had a longer hospital stay (6 vs. 3 days, p = 0.01). Conclusions Frailty is not associated with severe irAEs, but is related to CIP. Meanwhile, it predicts more than one type of irAEs and a longer hospital stay. Frailty screening has added value to the decision‐making process for frail patients eligible for PD‐1 inhibitors.


| INTRODUCTION
The programmed cell death protein 1 (PD-1) inhibitor, as an immune checkpoint inhibitors (ICIs), has become the first-line therapy in advanced stages of different tumor types 1,2 and is increasingly used in early-stage disease settings, including advanced lung cancer. 3 ICIs, especially PD-1 inhibitors, which improve outcomes for advanced lung cancer, 4 are more tolerable than conventional chemotherapy. 5,6 However, PD-1 inhibitor toxicity cannot be ignored. PD-1 inhibitors can lead to an excessive inflammatory response, termed immunerelated adverse events (irAEs), involving any organs. 7,8 IrAEs can be mild, potentially severe, or even fatal. Therefore, these toxicities must be detected early and managed appropriately.
ICIs are considered a tolerable treatment option at an older age due to their favorable safety profile. 9 Current data suggest that the incidence of irAEs does not increase with age. 10 However, in these trials, the analysis is based on the patient's chronological age, and the relationship between the occurrence of irAEs and patients of different ages whose biological age exceeds their chronological age is unclear. Aging accelerates when the biological age exceeds chronological age. It is also associated with high morbidity, mortality, and reduced life expectancy. 11 The potential impact of aging on immunotherapy efficacy and tolerability is unknown. The frailty index (FI) is a widely used measure of biological age. 12 Fan et al. constructed a 28-item FI as a proxy for chronological age to identify accelerated aging populations that might help to prevent premature death and extend life expectancy. 13 The 28-item FI applies to patients of all ages, not just the elderly. Consequently, based on the FI, we intend to establish a 28-item FI including basic diseases, dysfunction, autonomous living ability, physiological indicators, and others (number of long-term medications ≥5) to estimate whether non-frail and frail patients are associated with a high risk of irAEs.
The real-world clinical practice data on frailty patients treated with ICIs are woefully inadequate. To our knowledge, the predictive value of FI for irAEs has never been evaluated in lung cancer patients. Therefore, based on a real-world cohort, this study aims to assess whether irAEs and its sequelae are more common in frail patients than in non-frail patients using FI.

Collection
The electronic medical records of 114 patients were retrospectively reviewed with metastatic or unresectable advanced lung cancer who received PD-1 inhibitors (Pembrolizumab, Sintilimab, Camrelizumab, and Tislelizumab) from May 2018 to June 2022 at Peking University First Hospital. Data from these patients were followed continuously until June 1, 2019. None of the patients had a history of previous ICI treatment and interstitial lung disease prior to treatment with the PD-1 inhibitors. Patient's demographic and clinical characteristics, including age, sex, body mass index (BMI), Eastern Cooperative Oncology Group (ECOG) performance status, tumor type, type of PD-1 inhibitors, combination medication, lymphocyte counts, albumin, neutrophil/ lymphocyte ratio, grade ≥ 3 irAEs, clinically relevant irAEs, time to onset of irAEs, use of corticosteroids, treatment discontinuation due to irAEs, hospitalization due to irAEs, hospitalization duration, Intensive Care Unit (ICU) admission, remission of irAEs, and PD-1 treatment restart, were collected. The proportion of patients who continued treatment with PD-1 inhibitors for ≥90 days was analyzed to verify tolerance in both groups. Patients were followed up for 30, 60, 90, and 120 days after the diagnosis of irAE. This study was approved by the Research Ethics Committee of Peking University First Hospital.
The FI was calculated using 28 baseline variables, including basic illness (hypertension, coronary atherosclerotic cardiopathy, myocardial infarction, diabetes mellitus, stroke/transient ischemic attack, chronic obstructive pulmonary disease, bronchial asthma, tuberculosis, peptic ulcer, chronic cardiac insufficiency, benign prostatic hyperplasia, rheumatoid arthritis, chronic kidney disease, osteoporosis/fractures, and tumor), dysfunction (anxiety, depression, somnipathy, and cognitive dysfunction), ability to live independently (eating, urination, diaphoresis, bathing, and acting ability), physiological indicators (weight loss in one year, body-mass index, albumin, neutrophil-lymphocyte ratio (NLR)), and other aspects (multiple medications). These were health status deficits measured using questionnaires and physical examinations. FI is the proportion of an individual whose value is unhealthy among all health measures. FI = the number of K E Y W O R D S frailty, frailty index, immune-related adverse events, lung cancer, PD-1 inhibitor unhealthy indicators in the health indicators/the number of health indicators. Two levels of frailty status, non-frail (frailty index <0. 25), and frail (FI ≥ 0.25), were defined (Table 1).
IrAEs were defined as (1) pathological evidence of irAEs, (2) multidisciplinary judgment including at least two oncological and respiratory specialists, or (3) clinical improvement after treatment based on clinical and radiographic findings of irAEs. [14][15][16] The severity of irAEs was graded according to the Common Terminology Criteria for Adverse Events (CTCAE) 5.0 criteria. Additionally, the duration of PD-1 inhibitor treatment for a patient in the medical record was defined as the time between the initiation of PD-1 inhibitor treatment and any events leading to treatment discontinuation. Clinical evaluation was performed according to the Response Evaluation Criteria for Solid Tumors, version 1.1 (RECIST 1.1), and computed tomography was performed every 6-8 weeks for response evaluation. 17

| Statistical analysis
Skewed distributions of continuous variables were expressed as median (inter-quartile range [IQR]), and normal distributions were expressed as mean ± standard deviation (SD). Student's t-test was used for continuous variables with normal distribution, while the Mann-Whitney test was used for continuous variables with nonnormal distribution. The Chi-square or Fisher's exact tests were used to compare categorical variables between non-frail and frail patients. p-values < 0.05 were considered statistically significance. All statistical analyses were performed using SPSS (version 23.0). This study's figures were produced using GraphPad Prism 8.0.

| Immune-related adverse events in non-frail and frail patients
When we compared the treatment duration between the two groups, patients in the frailty group was treated with PD-1 inhibitors until irAEs appeared that the median duration was longer (122.5 vs. 80, p = 0.41), although it was not statistically significant. When we focused on irAEs severity and the duration of PD-1 inhibitor treatment, patients with grade < 3 irAEs in the frailty group displayed a tendency to continue treatment for ≥90 days, but there was no statistically significant difference (p > 0.05) between frail and non-frail patients (Figure 1). Frail patients were more likely than non-frail patients to discontinue

| Development of IrAEs and clinical outcomes for patients
Tables 4 and 5 demonstrate the development of irAEs and clinical outcomes for non-frail and frail patients. Frail and non-frail patients were similar; in both cases, two patients were treated with a PD-1 inhibitor, and the remaining patients were treated with a PD-1 inhibitor combined with chemotherapy. Five non-frail patients were with grade 1 irAEs, two with grade 2 irAEs, and four with grade 3 irAEs. The grade ≥ 3 irAEs occurred in four patients treated with PD-1 inhibitor combined with chemotherapy. Four frail patients were with grade 1 irAEs, three with grade 2 irAEs, and three with grade 3 irAEs. The grade ≥ 3 irAEs occurred in one patient treated with a PD-1 inhibitor combined with chemotherapy and two treated with a PD-1 inhibitor. All non-frail patients (including four patients with ≥3 irAEs, one with grade 1 dermatitis, and one with grade 2 CIP) who were reactivated with the same PD-1 inhibitor recovered from irAEs. In contrast, one patient was admitted to the ICU with CIP in frail patients (Table 5. Case 6), and the remaining patients recovered from irAEs. Two patients returned to the same PD-1 inhibitor therapy, including one patient with NSCLC who developed grade 2 dermatitis, retreated with the same PD-1 inhibitor, and developed CIP after 259 days (Table 5. Case 8). Another patient with grade 1 CIP was restarted with the same PD-1 inhibitor with no other adverse events (Table 5. Case 1). Patients were followed up for 30, 60, 90, and 120 days after the diagnosing irAEs. Four patients were lost to follow-up, and none of the others died.

| DISCUSSION
In this retrospective study, we employed 28-item FI to assess the effect of frailty on PD-1 inhibitor-related adverse Several studies have investigated the association between frailty and the occurrence of irAEs assessed by scoring tools other than the 28-item FI. A retrospective study of 197 patients with advanced malignant tumors (including 77 patients with NSCLC) using indirect markers of frailty, such as ECOG performance status, Charlson Comorbidity Index (CCI), and NLR, discovered that frail patients in the elderly and non-elderly groups have no association with irAEs or complications. 18 Kubo et al. 19 retrospectively studied the safety of ICIs in 95 patients ≥75 years with NSCLC and calculated a modified G8. They concluded that impaired, modified G8 was not associated with grade ≥ 2 irAEs. Welaya et al. 20 performed a small retrospective study of 28 patients, 11 of whom had lung cancer, and found no association between Geriatric assessment (GA) impairment and the development of irAEs. We did not identify 28-item studies on the relationship between FI and the occurrence of irAEs.
Our data revealed that the incidence of irAEs was 17.5%, of which the grade ≥ 3 irAEs was 6.1% and the CIP was 11.4%. The results are lower than clinical studies reported in the literature. [21][22][23][24][25] Frail patients developed slightly more irAEs of any grade with PD-1 inhibitors than nonfrail patients, although this difference was not statistically significant. Specifically, the most common adverse effect was CIP, which was more common in frail patients. Other irAEs included myocarditis, neurologic toxicity, dermatitis, thyroid dysfunction, hepatitis, cystitis, and renal insufficiency, but no significant differences were observed. Recently, pulmonary adverse events caused by PD-1 inhibitors, including some severe and even fatal events, have been reported. 26 CIP accounted for 35% of deaths associated with PD-1 and PD-L1 inhibitors. Furthermore, the incidence of CIP caused by PD-1 inhibitors was higher in NSCLC patients than in other cancer patients. 27,28 This will make clinicians more cautious when prescribing PD-1 inhibitors to frail lung cancer patients. In our study, all patients with irAEs recovered with appropriate treatment in non-frail patients, and the same PD-1 inhibitor was restarted in cases with grade < 3 irAEs without other adverse effects. In contrast, although most patients recovered after treatment in the frail group, there was a recurrence of adverse effects in patients with <3 irAEs after the reintroduction of the same PD-1 inhibitor. There are conflicting reports on the safety and efficacy of retreatment after irAEs, and no consensus has been reached. Therefore, restarting PD-1 inhibitors after irAEs may be challenging for frailty patients. [29][30][31] Additionally, we discovered that frail patients were less likely to have ≥3 irAEs with the longer duration of continuous use of PD-1 inhibitors until the irAEs occur, although the development of severe irAEs in frail patients is similar to that of non-frail patients. It indicates that PD-1 inhibitors may F I G U R E 1 The treatment period for patients with irAEs of varying severity. Frail and non-frail patients were divided into ≥3 irAEs and <3 irAEs. Among the frail patients, 33.3% (1/3) patients ≥3 irAEs and 57.1% (4/7) patients <3 irAEs could continue treatment with PD-1 inhibitors for ≥90 days; among the non-frail patients, 50% (2/4) patients ≥3 irAEs and 42.9% (3/7) patients <3 irAEs could continue treatment with PD-1 inhibitors for ≥90 days. There was no statistically significant difference between the two subgroups (p > 0.05).

T A B L E 4 Clinical features of the non-frail patients with irAEs
Case have a certain degree of efficacy and tolerance in lung cancer patients. Surprisingly, our study showed that frail patients had longer hospital stays, and irAEs were presented when they had more than one type of irAEs. This supports the importance of frailty for developing irAEs in lung cancer patients treated with PD-1 inhibitors. Multiple adverse events should be considered when frail patients develop irAEs after receiving PD-1 inhibitors. This study has inherent limitations as a retrospective study. First, the sample size of patients was relatively small. Second, some patients with milder symptoms of grade 1-2 irAEs were excluded from the study because they were frequently observed in the outpatient department rather than hospitalized. Neglecting such patients, especially skin and other adverse events, may lead to underestimating the incidence. Finally, we did not exclude patients with inadequate medical records, for example, who were lost to follow-up after 60 days to document as many irAEs as possible.
In conclusion, although this study does not provide sufficient evidence that frailty is associated with the incidence of total irAEs or grade ≥ 3 irAEs according to FI, frailty is associated with CIP. Moreover, once irAEs occur in frail patients, it is often manifested by more than one adverse event.
Assessing the risk of frailty by FI can help identify those frail patients at risk of CIP and predict the length of hospital stay. Therefore, FI assessment in patients receiving PD-1 inhibitors is feasible and should be considered. Ultimately, a deeper understanding of a patient's frailty can serve as a guide to making personalized treatment decisions.