Correlation between the Immune Checkpoint Inhibitors Prognostic Index and Outcomes in Nonsmall Cell Lung Cancer: A Multicentre Analysis

Objective To evaluate the prognostic value of the immune checkpoint inhibitor prognostic index (ICPI), based on the albumin (ALB) and derived neutrophil-to-lymphocyte ratio (dNLR), for nonsmall cell lung cancer (NSCLC) patients receiving immune checkpoint inhibitors (ICIs). Methods We conducted a multicentre retrospective study with an ICIs cohort (n = 143) and a chemotherapy control cohort (n = 84). A Cox proportional hazards regression and logistic regression model were used to find the independent risk factor for progression-free survival (PFS) and overall survival (OS) and disease control rate (DCR) in NSCLC patients. The Kaplan–Meier was used to evaluating the PFS and OS. Results The ALB <35 g/L and dNLR >3 were correlated with worse PFS and OS for NSCLC patients receiving ICIs, respectively. The moderately high-risk ICPI had a significantly increased risk of progression (hazard ratio (HR) 1.83, 95% confidence interval (CI) 1.14–2.91; P=0.012) and of death (HR 2.33, 95% CI 1.12–4.87; P=0.024) and of nondisease control (odds ratio (OR) 3.05, 95% CI 1.19–7.83; P=0.021) and was correlated with worse PFS and 1-year survival rates (4.0 months vs. 7.2 months; P=0.001; 44.3% vs. 76.1%; P=0.001) compared with low-risk ICPI when it was characterized two groups. When ICPI was further divided into three groups, the results showed that the high-risk ICPI was correlated with worse PFS and 1-year survival rates. However, there was no difference in the chemotherapy cohort. Conclusion The ICPI was correlated with worse outcomes for NSCLC patients receiving ICIs but not for patients with chemotherapy.


Introduction
e success of immunotherapy has not only revolutionized the pattern but also the landscape of nonsmall cell lung cancer (NSCLC) treatment [1]. e immune checkpoint inhibitors (ICIs), principally represented by cytotoxic T lymphocyte antigen-4 and programmed death 1/ligand 1 (PD-1/PD-L1) inhibitors, have been widely and successfully used in clinical practice [2]. Increasing evidence shows that up to 80% of NSCLC patients do not benefit from ICIs [3], and what is more, some patients even develop severe immunotoxicity and financial toxicity, although biomarkers promise new dawn for patients. e tumor-related biomarkers, such as PD-L1 expression are widely used in clinical applications. A correlation between high PD-L1 expression and good outcomes has been observed in NSCLC patients receiving ICIs. In contrast, some studies showed that nearly 60-70% of patients did not benefit from ICIs even in the PD-L1 positive population [4][5][6]. In some circumstances, some patients show clinical benefits regardless of the expression level of PD-L1 in tumor cells [7]. Besides, PD-L1 has no uniform detection platform and cutoff value [5,8]. Another biomarker is tumor mutation burden (TMB). Numerous studies indicate that patients with high TMB have a higher overall response rate (ORR), progression-free survival (PFS), and overall survival (OS) [9,10]. However, the limitations of TMB are salient, including costly and time-consuming detection, and a lack of a standardized detection platform and uniform cutoff value. e imperfections of these tumor-related biomarkers are becoming increasingly apparent.
An increasing amount of research has confirmed that the systemic inflammatory response (SIR) is inextricably related to the occurrence and development of tumors, and also affects the immune response of cancer, which may be associated with the effect of immunotherapy [11][12][13]. Numerous routine blood parameters have been demonstrated as SIR-related biomarkers such as circulating white blood cells (WBC), absolute neutrophil counts (ANC), platelet counts (PLT), lactate dehydrogenase (LDH), albumin (ALB), and even neutrophil-to-lymphocyte ratio (NLR) [14], which were associated with poor prognosis in several malignant solid tumors, including NSCLC [15]. However, the prognostic and predictive value of SIR-related biomarkers in NSCLC with ICIs has not yet been completely elucidated. In the present study, we sought to explore a novel, convenient, practical, and economical combined prognostic index to predict the outcomes of NSCLC patients receiving ICIs, and help clinicians determine and screen NSCLC patients who are ineligible for ICIs in order to avoid unnecessary immunotoxicity and financial toxicity.

Study Population.
We conducted a multicentre retrospective study of a cohort of patients with NSCLC receiving ICIs from 6 departments at 2 academic centers, the respiratory (n � 22) and oncology (n � 3) departments of Xijing Hospital and the respiratory (n � 15), oncology (n � 43), thoracic surgery (n � 56) and Traditional Chinese medicine (n � 4) departments of the Tangdu Hospital ( Figure 1). e patient collection was based on the following inclusion criteria: (1) adult patients over 18 years old; (2) patients, who were pathologically diagnosed with NSCLC; (3) at least one radiological assessment per Response Evaluation Criteria in Solid Tumors (RECIST) v1.1 [16]; and (4) patients, who received ICIs. Patients, who matched any of the following criteria were excluded: (1) patients, who had ongoing noncancer related inflammation, immune disease, end-stage liver disease, or hematologic disease within 1 week before treatment; (2) patients with EGFR mutation or ALK and ROS1 gene fusion; (3) patients with other previous or concomitant cancers; and (4) patients with allergies or intolerance to ICIs or chemotherapy. A total of 143 patients from the Xijing Hospital (n � 25) and the Tangdu Hospital (n � 118) treated with ICIs between January 2018 and July 2019 were enrolled in the immunotherapy cohort and followed up until July 2021. A control cohort of 84 patients with NSCLC from the Xijing Hospital was exclusively treated with chemotherapy between June 2014 and April 2015.

Parameters and Assessments.
Peripheral blood cell counts and ALB levels at baseline before ICI treatment were extracted from electronic medical records. Demographic, clinical, pathological, and molecular data were also collected. PD-L1 expression was analyzed on tumor cells by immunohistochemistry, according to the standard practice for each center. Expression of at least 1% was considered positive.
Radiological assessments were performed every 6 weeks as per RECIST v1.1 [16] as per the investigator's discretion in the immunotherapy cohort and the chemotherapy cohort. e objective remission rate (ORR) refers to the percentage of complete responses (CR) + partial responses (PR) patients out of the total number of patients, and the disease control rate (DCR) refers to the percentage of CR + PR + stable disease (SD) patients out of the total number of patients. OS was calculated from the date of initial immunotherapy administration until death (event) owing to any cause or the last follow-up (censored). PFS was calculated from the date of initial immunotherapy administration until disease progression or death (event) due to any cause.

Statistical Analysis.
e dNLR was calculated as follows: dNLR � ANC/(WBC−ANC) [15]. e optimal cutoff value for the dNLR was greater than 3 and the ALB level was lower than 35 g/L based on previous largest published studies [15,17]. e chi-square test and Fisher's exact test were used to analyze the distribution of clinical characteristics data. Significant parameters identified in univariate analysis (P < 0.05) were incorporated into multivariate Cox regression analysis to determine the independent factors associated with OS and PFS, and the hazard ratio (HR) was calculated. Variables associated with DCR were identified with logistic regression in the final multivariate model and were selected according to statistical significance in univariate analysis (P < 0.05), and the odds ratio (OR) was calculated. e α level was 5%. e results are presented as HR and OR and with 95% confidence interval (CI). Survival analyses were performed using the Kaplan-Meier diagram and compared by the log-rank method. All P < 0.05 were considered statistically significant. Data analysis was performed using SPSS software (version 22, IBM) and GraphPad Prism 8 software.

Immune Checkpoint Inhibitor Prognostic Index (ICPI).
e ALB and dNLR were vital for the prognoses of NSCLC patients receiving ICIs. However, the predictive ability of individual indicators is relatively weak, a new prognostic indicator ICPI, based on the ALB <35 g/L and dNLR >3, had been constructed as a result. e ICPI was developed to characterize two groups, the low-risk ICPI (0 factor) and moderately high-risk ICPI (1 or 2 factors).

Chemotherapy Control Cohort.
Whether ICPI was divided into two groups or three groups, the moderately highrisk ICPI or the high-risk ICPI was correlated with worse PFS, OS and DCR for NSCLC patients receiving ICIs. erefore, this study further explored the predictive value of ICPI in NSCLC patients receiving chemotherapy. In the chemotherapy cohort, the 84 patients with lung cancer had a median follow-up of 8.7 months (95% CI 8.2-9.2 months). e median PFS and OS were 4.3 months (95% CI 2.6-6.0 months) and 11.1 months (95% CI 7.6-14.6 months). Baseline characteristics are given in Table 5.
When the ICPI was divided into two groups, 48 (57%) patients had a low-risk ICPI and 36 (43%) had a moderately high-risk ICPI. In contrast to the ICIs cohort, no significant differences in PFS and OS were observed among the moderately high-risk ICPI and low-risk ICPI in the chemotherapy cohort (4.0 months vs. 4.3 months, P � 0.740; 1-year survival rates: 60.0% vs. 32.4%, P � 0.257).
e ICPI was further divided into 3 groups, the median PFS was 4.8 months vs.  Figure 7). In terms of DCR, whether ICPI was divided into two or three groups, the DCR was all 100%, so there was no significant difference between different ICPI groups.

Discussion
In our 143 patients treated with ICIs, the median PFS was 6.2 months (95% CI 5.2-7.1 months), which was similar to the PFS of the Impower 131 [18], Impower 130 [19], and KEYNOTE 407 [20]. e median OS did not reach, the reason might be as follows: first, the proportion of ICIs as first-line was high (32.2%); second, some patients received surgical treatment (35.0%) before ICIs treatment; Last, the period some patients assessed was not every 6 weeks as advised. Although the median OS did not reach in the present study, the 1-year survival rate was 66.2%, which was basically consistent with 61.3% in the Krefting study [21].
In the present study, multivariate analysis showed that the ALB <35 g/L was correlated with shorter PFS (3.0 months vs. 6.9 months, P < 0.001) and 1-year survival rates (28.6% vs. 72.8%, P < 0.001) compared with ALB ≥35 g/ L in NSCLC receiving ICIs, which is consistent with previous findings that high ALB levels are associated with poor Journal of Oncology 5 outcomes in various cancers, including melanoma, pancreatic cancer, lung cancer, gastric cancer, and breast cancer [22]. Kazandjian [17] et al. found that ALB <35 g/L was associated with poor OS and PFS in NSCLC receiving ICIs. is may be related to the following factors: first, for the host, the tumor is accompanied by tumor hypoxia and necrosis, and local tissue damage. In response to these changes, the body system releases proinflammatory cytokines and growth factors, and liver cells increase the production of acute phase proteins, such as CRP, and reduce ALB production [23]; second, liver synthesis of ALB is mainly affected by colloid osmotic pressure and inflammatory state but does not change in nutrient intake and malnutrition state [22,24]. erefore, hypoproteinaemia represents a proinflammatory state rather than a nutritional status in cancer patients [22]. A large number of pieces of evidence showed that hypoproteinaemia has also been found to be associated with a poor prognosis of NSCLC [15,17]. In a multivariate analysis of the present study, the dNLR >3 was correlated with worse PFS (4.0 months vs. 6.6 months, P � 0.009) and 1-year survival rates (48.9% vs. 70.9%, P � 0.038) than dNLR ≤3, which is consistent with previous studies in patients with NSCLC treated with ICIs [15]. As an inflammatory response cell, neutrophil inhibits antitumor immune response by inhibiting the cytotoxic activity of immune cells, especially activated T cells [25]. e reduction of lymphocytes weakens the effect of ICIs and mainly releases the inhibitory signal of T cell function [25]. erefore, researchers proposed the NLR, neutrophil-to-lymphocyte ratio.
In the last 15 years, there has been a movement towards the use of combined prognostic scores [33][34][35]. Since ALB <35 g/L and dNLR >3 were closely associated with unfavorable prognosis in NSCLC patients treated with ICIs, we constructed a new prognostic index, ICPI, based on the two risk factors. e results showed that the ICPI was correlated with worse PFS, OS and DCR for NSCLC patients receiving ICIs. e moderately high-risk ICPI had a significantly increased risk of progression, death, and non-DCR (P < 0.05), and had worse PFS and 1-year survival rates (4.0 months vs. 7.2 months, P � 0.001; 44.3% vs. 76.1%, P � 0.001) compared with low-risk ICPI. Similarly, in further analysis, the ICPI was divided into three groups, and the results demonstrated that the high-risk ICPI was correlated with worse PFS and 1-year survival rates compared with middle-risk ICPI and low-risk ICPI (2.0 months vs. 5.0 months vs. 7.2 months, P < 0.001; 20.0% vs. 49.2% vs. 76.1%, P < 0.0011). However, there were only 9 low-risk ICPI patients (6%), which may impact the results, and           017 [7] both describe that regardless of PD-L1 expression level or even negative, NSCLC patients showed clinical benefits from ICIs treatment.
However, there are some limitations in this study. First, the present study was a retrospective evaluation with potential biases due to missing trials or missing laboratory values, such as the LDH level and ECOG PS. Second, the identified cutoff values for the dNLR and ALB need to be validated in external populations.
ird, the information of some patients including concurrent conditions and medications is missing. Comorbidities (such as infections or inflammation) and the use of steroids that may cause changes in peripheral blood cell counts are lacking. Future modeling could incorporate other known prognostic markers such as the performance status, other baseline factors, tumor genomic, transcriptomic, proteomic, and metabolomic markers.

Conclusions
e ALB <35 g/L and dNLR >3 were correlated with worse PFS and OS for NSCLC patients receiving ICIs. e ICPI was correlated with an unfavorable prognosis for NSCLC patients receiving ICIs, but not for patients with chemotherapy, suggesting that baseline ICPI might be useful for identifying patients, who are unlikely to benefit from treatment with ICIs and avoiding unnecessary immunotoxicity and financial toxicity.
Data Availability e datasets generated during and/or analyzed during the current study are available from the corresponding author upon request.

Ethical Approval
e study was conducted in consent to the Declaration of Helsinki and was approved by the Institutional Ethics Committee of the Xijing Hospital of Fourth Military Medical University (Approval Number: KY20202077-C-1).

Consent
Informed consent was obtained from patients.

Disclosure
Ying Zhou, Bin Wu, and Tian Li are the co-first authors.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
YZ and JZ involved in conceptualization, data curation, formal analysis, investigation, methodology, supervision, writing original draft, and review and editing. BW and Tianqi Xu involved in data curation, writing-original draft, and review and editing. BW, TL, YZ, and NC involved in data curation, formal analysis, investigation, and methodology.
Acknowledgments is work was supported by the National Natural Science Foundation of China (81773153).