Risk factors indicating immune-related adverse events with combination chemotherapy with immune checkpoint inhibitors and platinum agents in patients with non-small cell lung cancer: a multicenter retrospective study

Immune checkpoint inhibitors (ICI) ushered in a new era for the treatment of non-small cell lung cancer (NSCLC). However, they carry the risk of immune-related adverse events (irAEs). Recently, various studies have been conducted on the predictive factors for irAEs, but there are no reports focusing only on ICI plus platinum agents. The present study aimed to identify the risk factors for irAEs due to ICI combined with platinum-based induction immunochemotherapy in NSCLC patients, focusing only on the period of combined therapy and excluding the period of ICI maintenance therapy. This retrospective study included 315 NSCLC patients who started ICI combined with platinum-based chemotherapy treatment at 14 hospitals between December 2018 and March 2021. A logistic regression analysis was used to explore the predictive factors. Fifty patients (15.9%) experienced irAEs. A multivariate analysis revealed that squamous cell carcinoma (P = 0.021; odds ratio [OR]: 2.30; 95% confidence interval [Cl]: 1.14–4.65), anti-programmed death 1 antibody (anti-PD-1) plus anti-cytotoxic T-lymphocyte antigen-4 antibody (anti-CTLA-4) regimens (P < 0.01; OR: 22.10; 95% Cl: 5.60–87.20), and neutrophil-to-lymphocyte rate (NLR) < 3 (P < 0.01; OR: 2.91; 95% Cl: 1.35–6.27) were independent predictive factors for irAEs occurrence. Squamous cell carcinoma, anti-PD-1 plus anti-CTLA-4 regimens, and NLR < 3 may be predictive factors for the occurrence of irAEs due to induction immunochemotherapy in patients with NSCLC. By focusing on the potential risk of irAEs in patients with these factors, irAEs can be appropriately managed from an early stage.


Introduction
Immune checkpoint inhibitors (ICI) have been developed and approved as an effective option for patients with various cancers and have revolutionized the treatment of advanced or recurrent non-small cell lung cancer (NSCLC). Although ICI occupy an important place in cancer chemotherapy, they carry the risk of immune-related adverse events (irAEs) due to the disruption of immune function homeostasis [1]. IrAEs pose a risk of poor prognosis in severe cases, and ICI-related mortality rates in the range of 0.36%-1.23% have been reported [2]. Furthermore, large case series studies of NSCLC and other tumors have shown that irAE development is associated with improved survival outcomes [3]. Therefore, numerous studies on risk factors for the development of irAEs have been conducted, and several risk factors have been reported, such as a low neutrophil-to-lymphocyte ratio (NLR), high body mass index (BMI), high serum albumin (ALB) level, and low Eastern Cooperative Oncology Group (ECOG) performance status (PS) [4][5][6][7][8]. Recently, significant survival prolongation was observed [9,10], and combination therapy with ICI and platinum has been recommended as the standard chemotherapy for NSCLC [11,12]. However, despite the promising activity of combination therapy with ICI and platinum for NSCLC, the risk factors associated with the development of irAEs are still unknown. Although cancer chemotherapy has historically been considered immune suppressive, it is now accepted that certain chemotherapies promote tumor immunity by inducing immunogenic cell death and by disrupting strategies that tumors use to evade immune recognition [13]. This suggests that the combined use of ICI and chemotherapy has a synergistic effect due to a complex mechanism beyond the additive effect of ICI plus the effect of cytocidal chemotherapy. Combination chemotherapy with platinum drugs results in additional serious side effects, such as hematologic toxicity, anorexia, and diarrhea. Therefore, the risk factors for the development of irAEs may differ from those previously reported.
This study aimed to investigate the risk factors for the occurrence of irAEs with induction immunochemotherapy using ICI and platinum agents in NSCLC patients; the period of ICI maintenance therapy was not evaluated.

Study design
This multicenter, retrospective, and observational study was conducted at 14 medical centers in the National Hospital Organization in Japan.
Patients whose peripheral blood neutrophil and lymphocyte counts were not measured within the week prior to the start of treatment were excluded. We also excluded patients who had a history of prior treatment with ICI.

Data collection
All data were retrospectively collected from the electronic health records of each medical institution. We collected information regarding patient sex, age, BMI, ECOG-PS, cancer stage, history of chemotherapy/radiotherapy, smoking history, comorbidities (endocrine disease and interstitial lung disease), use of immunosuppressive drugs, and peripheral blood data (within 1 week prior to the initiation of ICI plus platinum agents) at baseline. The histological subtype, driver mutation, and tumor proportion score (TPS) were collected as tumor data. The irAE period was up to 3 weeks after the final administration of ICI plus platinum agents. IrAEs were diagnosed based on the judgment of the attending physician and the description of immune-mediated adverse events by PACIFIC Investigators [14]. Cases in which it was not possible to determine whether the adverse event was due to an ICI or a cytotoxic anticancer drug were excluded. The Common Terminology Criteria for Adverse Events version 5.0 were used to assess irAEs.
Serum creatinine (SCr) levels were determined by an enzymatic method. Creatinine clearance (CCr) for men was calculated using the Cockcroft-Gault equation [15]: CCr for women was then calculated using the following equation: The prognostic nutritional index (PNI) was used as a multi-parameter index of nutritional status [16]: NLR and the platelet-to-lymphocyte ratio (PLR) were calculated as immune system parameters of systemic inflammation as follows:

Statistical analysis
Differences in continuous data were compared using the Mann-Whitney U test, and differences in categorical data were compared using the Fisher's exact test. A multiple logistic regression analysis was used to analyze the factors associated with irAEs. To identify the factors associated with irAEs, an univariate analysis was performed using sex, age (< 65 years or over), BMI (< 25 kg/m 2 or over), ECOG-PS (0, 1, or over), smoking status (ever or never), histology (squamous or non-squamous), epidermal growth factor receptor (EGFR) mutation, TPS, ICI (anti-PD-1 antibody + anti-CTLA-4 antibody or anti-PD-1 antibody/anti-PD-L1 antibody), chemotherapy history, thoracic radiotherapy history, tyrosine kinase inhibitor history, corticosteroid use at baseline, ALB(< 3.6 g/dL or over), total bilirubin (< 1.5 mg/dL or over), aspartate aminotransferase (< 30 U/L or over), alanine aminotransferase (< 42 U/L or over), γ-glutamyl transpeptidase (< 64 U/L or over), blood urea nitrogen (< 20 mg/dL or over), CCr (< 60 mL/min or over), PNI (< 40 or over), NLR (< 3 or over), and PLR (< 180 or over) as independent variables. The cut-off values for ALB [8], BMI [7], ECOG-PS [7,8], CCr [17], PNI [16], NLR [5,6], and PLR [18] were determined based on previous reports, and other factors of peripheral blood data were determined based on the normal range in the national hospital organization Beppu medical center. Differences between each factor were tested for statistical significance (P < 0.15) using the Fisher's exact test. Variables with a P < 0.15 in the initial analysis and factors that have been reported as predictors of irAEs were included in the multivariate modeling, and stepwise variable selection with Akaike's information criterion was used. The magnitude of the association was expressed as an odds ratio (OR) and a 95% confidence interval (95% CI). All statistical analyses were performed using commercial software (EZR, Saitama Medical Center, Jichi Medical University, http:// www. jichi. ac. jp/ saita ma-sct/ Saita maHP. files/ statm edEN. html [19]) and a graphical user interface for R (The R Foundation for Statistical Computing, version 4.20). The Wilcoxon signed-rank test was performed to compare NLR at baseline and at the time of irAE occurrence, and the Kruskal-Wallis test with the Bonferroni method was used for multiple group comparisons. The NLR cutoff value before the start of treatment was calculated using receiver operating characteristic curve analysis. Statistical significance was set at P < 0.05.

Patient characteristics
This study included 315 patients; 265 patients (84.1%) were assigned to the non-irAE group and remaining 50 patients (15.9%) to the irAE group (Table 1). At treatment initiation, one patient in each group had comorbid thyroid-and adrenal cortex-related diseases, and three patients had a history of interstitial lung disease, but none had aggravation or relapse. There were no significant differences in the peripheral blood data between the two groups (P value not shown). Table 2 shows the occurrence and severity of irAEs in the irAE group. In the irAE group, five patients experienced multiple irAEs. The most frequent irAEs were skin-related events, followed by thyroid-related events, and interstitial lung disease. Only one case requiring discontinuation of treatment had an endocrine disorder such as thyroiditis, hypothyroidism, and hypoadrenocorticism. The highest incidence of grade ≥ 3 severity was in hepatopathy, followed by interstitial lung disease, with one of them being grade 5. Interstitial lung disease was also the most common cause of treatment discontinuation due to irAEs, followed by hepatopathy and myositis.

Association between NLR and irAEs
The cutoff value of pretreatment NLR for the occurrence of irAEs was 2.79 (area under curve, 0.60; 95% CI: 0.51-0.69; sensitivity, 0.71; specificity, 0.52). In the 50 patients with irAEs, there was no significant change in NLR trends from baseline to the time of irAE occurrence (P = 0.28; Fig. 1a). Baseline NLR values did not correlate with those calculated for irAE time of onset and severity (P = 0.31 and P = 0.46, Fig. 1b and 1c, respectively).

Discussion
In this study, irAEs were observed in 15.9% of patients, despite a short follow-up period that focused solely on the platinum agent concomitant period. As shown in Table 2, skin-related events, endocrine disease, and interstitial lung disease occurred at a high frequency of ≥ 20%. Furthermore, interstitial lung disease, in particular, was more likely to become severe when it occurred, and the treatment discontinuation rate was high. Focusing on risk factors at the early stage of treatment, the period of the platinum agent combination preparations and early detection and response to the onset of irAEs leads to decreased severity of irAEs resulting in continuation of treatment.
In this study, independent factors such as squamous cell carcinoma, the anti-PD-1 antibody plus anti-CTLA-4 antibody regimens, and NLR < 3 were identified as risk factors for irAEs. Recent studies demonstrated that patients with non-squamous cell subtypes have a lower risk of developing checkpoint inhibitor pneumonitis (CIP) [20]. Smoking is strongly associated with lung cancer, with a relative risk of squamous cell carcinoma of 11.7 and 11.3 for men and women, respectively [21]. Moreover, in a large retrospective cohort comprising multiple tumor types, smoking was associated with an increased CIP risk [22]. Interestingly, all eight squamous cell carcinoma patients who developed interstitial lung disease in our study had a history of smoking. Interstitial lung disease as an irAE is included in CIP. This study suggests that, even under treatment with ICI plus platinum-based chemotherapy, not only are smoking and squamous cell carcinoma closely related, but there are also relationships between smoking and CIP as well as between squamous cell carcinoma and CIP. Smoking status was not identified as a risk factor because this study analyzed all types of irAEs together. In order to identify risk factors for interstitial pneumonia, it would be necessary to analyze a higher number of cases. The high incidence of irAEs in the nivolumab plus ipilimumab regimen was consistent with previous reports [23,24]. However, our study is the first to focus on a platinum-based combination therapy. Although the toxicity and efficacy of each regimen were not directly compared, the results suggest that particular attention should be paid in monitoring irAEs in platinum-based regimens combined with anti-CTLA-4 and anti-PD-1 drugs, as in the previously reported combination regimens of anti-CTLA-4 and anti-PD-1 drugs.
NLR is an index that reflects systemic inflammation and can serve as a prognostic factor for solid tumors [25][26][27][28] and a predictor of the clinical benefits of ICI [29][30][31]. The action of neutrophils on the inflammatory response is diverse and complex, and some neutrophils inhibit the immune system, suppressing the function of natural killer cells and CD8 T cells [32]. Lymphocytes, such as T cells, play a central role in the immune response, and lymphocytes are involved in the antitumor immune response and suppress tumor cell growth [33]. A low NLR reflects a low peripheral neutrophil count and high lymphocyte count, indicating maintenance of the immune system. We have previously reported that a low pretreatment NLR may be a predictive factor for the occurrence of irAEs with ICI monotherapy [4]. Despite the effect of cytotoxic chemotherapy on neutrophil and lymphocyte counts, this study showed that a low pretreatment NLR is a useful predictive factor for irAE development in ICI plus platinum agent therapy, similar to ICI monotherapy.
In this study, the cutoff value of pretreatment NLR for the occurrence of irAEs was 2.79. This value is close to the previously reported cutoff value of 3 [5,6]. In addition, we previously explored for irAE risk factors in ICI monotherapy and derived the NLR cutoff value of 2.86 [4]. This value was closer to the cutoff value 2.79 observed this study. Based on these results, it is considered that there is little difference in the pretreatment NLR cutoff value between ICI monotherapy alone and ICI plus chemotherapy.
While there was no difference in the cutoff value of baseline NLR, the transition of NLR after treatment initiation was different from that observed in ICI monotherapy. Although Matsukane et al. report that NLR was elevated during the development of irAEs [34], a similar tendency was not observed in our study. This was probably because lymphocyte and neutrophil counts were affected by the concomitant use of platinum agents in our study. It has been suggested that the relationship between continuous NLR monitoring and irAEs should be considered separately for ICI monotherapy and platinum combination therapy. In this study, pretreatment NLR values did not correlate with those calculated for irAE time of onset or severity. This may be due to the small number of individual cases and the need for further case pooling.
Unlike in previous reports, BMI, ALB, and ECOG-PS were not identified as risk factors in this study. With     [7]. Second, as discussed by Zhang et al. [7], their data include ICI dose setting based on body weight before 2018, which may differ from the currently prescribed fixed dose. To evaluate the relationship between irAEs and BMI, it is essential examine fixed dose and race-specific BMI.
In previous reports, ECOG-PS was identified as a risk factor in univariate analysis but not in multivariate analysis [7,8]. This may be due to the fact that PS assessment varies slightly among physicians, as discussed by Shimozaki et al. [8]. Furthermore, ECOG-PS was not a risk factor according to the univariate analysis in our study. This may be due to differences in baseline ECOG-PS among cases enrolled in each study; ECOG-PS was ≥ 2 in 78.5% of cases in Zhang et al.'s study [7] compared with in 87.0% in our study. We speculate that this is due to the lower baseline ECOG-PS required with combined chemotherapy as compared with ICI monotherapy [12].
In our study, high baseline ALB levels were associated with a higher incidence of irAEs but were not identified as a risk factor. In a previous study investigating ICI monotherapy for which ALB was identified as a risk factor, the observation period was long [8,35]. The median days of onset ranged from 2 to 835 days among the irAEs; in particular, patients with thyroiditis and hypoadrenocorticism showed a relatively higher rate of long-term irAEs [35]. Moreover, a recent study reported that decrease in pretreatment serum ALB and early ALB during ICI treatment was strongly associated with overall survival in patients with advanced NSCLC treated with ICI monotherapy but not in patients treated with chemoimmunotherapy [36]. This indicates that the influence of the ALB level is not limited to the observation period. Cytocidal anticancer drugs with a high binding rate to ALB are considered to increase the proportion of its free form due to hypoalbuminemia, which affects efficacy and side effects. It has been reported that hypoalbuminemia is not only a risk factor for early termination of chemotherapy, but also an independent prognostic factor in the chemotherapy group [37]. For these reasons, baseline high ALB levels may be an indicator of long-term irAEs in particular, and it may need to be evaluated independently for induction immunochemotherapy and ICI monotherapy.
This study had some limitations. First, it was a retrospective observational study and not a randomized or prospective study. Second, the study had a small sample size and included a possible selection bias of treatment, which is inevitable in a retrospective analysis. It is necessary to accumulate a large number of cases to analyze each irAE separately. Third, the observation period was short because this study focused only on the period of concomitant platinum agent use, and the period of ICI maintenance therapy was not evaluated. Therefore, it did not include the period of ICI-only maintenance therapy after completion of platinum-based treatment, or irAEs that developed after completion of ICI therapy.
However, this study is the first report of irAE predictive factors that focused solely on ICI combined with platinum-based induction immunochemotherapy. These regimens play a central role in NSCLC treatment, and it is of great significance to consider and monitor the risk of irAEs. Several studies demonstrate an association between irAEs and response to treatment with ICI [3,[38][39][40][41][42]. In this study, we did not investigate the treatment efficacy. However, early irAEs have also been reported to be associated with improved outcomes after immunotherapy [40], Fig. 1 Association between NLR and irAEs. a NLR trend from baseline through irAE occurrence. There were no significant changes in NLR trends from baseline to the time of irAE occurrence (P = 0.28). b Association between NLR and course number of onset irAEs.
There was no significant correlation between the baseline NLR values and those at the time of irAE onset (P = 0.31). c Association between NLR and irAE severity at onset. There was no significant correlation between the baseline NLR and irAE severity (P = 0.46) and the occurrence of irAEs during the induction immunochemotherapy period may reflect a favorable prognosis. By focusing on the potential risks of irAEs, as clarified in this study, irAEs can be appropriately managed from an early stage. We believe that this will allow practitioners to maximize the benefits of ICI treatment.