Clinical Significance of Serum Biomarkers in Stage IV Non-Small-Cell Lung Cancer Treated with PD-1 Inhibitors: LIPI Score, NLR, dNLR, LMR, and PAB

Background To assess the prognostic value of pretreatment serum biomarkers in stage IV non-small-cell lung cancer (NSCLC) patients treated with PD-1 (programmed cell death protein 1) inhibitors and their value as a predictor of benefit. Methods We performed a retrospective study including patients with stage IV NSCLC who were treated with anti-PD-1 drugs in first or advanced lines of therapy in the Affiliated Tumor Hospital of Nantong University. Serum biomarkers such as NLR, dNLR, LMR, PAB, ALB, and LIPI scores were calculated and analyzed in detail. Results A total of 85 patients with stage IV NSCLC treated with PD-1 inhibitors in the first or advanced lines of therapy were included in this subject. According to the tumor response of PD-1-based treatment, ORR was 42.4% (36/85) and DCR was 68.2% (58/85). The median OS and PFS were 20.0 months and 7.0 months, respectively. The ROC curves showed that the serum biomarkers of NLR, dNLR, LDH, LMR, PAB, and ALB were significantly associated with overall survival and helped to determine the cut-off value. The multivariate Cox proportional hazard analyses for stage IV NSCLC patients treated with PD-1 inhibitors indicated that dNLR (P < 0.001) and ALB (P = 0.033) were independent prognostic indicators of PFS, while liver metastasis (P = 0.01), NLR (P = 0.01), dNLR (P = 0.001), and LMR (P = 0.006) were independent prognostic indicators of OS. Moreover, patients of the good LIPI group showed prolonged PFS and OS than those with intermediate/poor LIPI score (P < 0.001 and P = 0.006, respectively). Conclusions Pretreatment dNLR is an independent prognostic indicator of both PFS and OS in stage IV NSCLC patients treated with PD-1 inhibitors. Pretreatment LIPI, combining dNLR > 3 and LDH>ULN, was correlated with worse outcome for stage IV NSCLC patients treated with ICI. High NLR, high dNLR, low LMR, and low ALB at baseline might be useful as an early predictive biomarker of benefit.


Introduction
Lung cancer is the most frequent malignant cancer and the leading cause of cancer-related death worldwide in recent years [1]. According to the expected number of cancer deaths in 2021 [2], almost one-quarter of all cancer-related deaths are due to lung cancer, among which nearly 82% is directly caused by cigarette smoking. Non-small-cell lung cancer (NSCLC) accounts for 85% of lung cancer cases, and most of the NSCLC patients are diagnosed in advanced stage. For these patients, the poor overall survival (OS) and 5-years survival rate are essential issues.
Over the past decades, cytotoxic systemic chemotherapy remains an important treatment for advanced NSCLC. A majority of patients are still suffering due to drug resistance or side effects of chemotherapy. The primary goal of systemic therapy in metastatic NSCLC is to reduce cancerrelated symptoms and to prolong survival time, with a concurrent goal to improve quality of life [3]. The advances of treatment in NSCLC have been greatly improved recently by further understanding of the pathogenic genomic alterations of NSCLC [4], the development of novel drugs, and biomarker-based evidence to identify patients most probably to respond to immune checkpoint blockade therapy.
The immunotherapy revolution, especially the development of immune checkpoint inhibitors (ICIs), has dramatically changed the landscape of the treatment paradigm of advanced NSCLC [5,6]. The immune checkpoint inhibitors (ICIs) basically include anti-programmed cell death 1 (PD-1) antibodies and anti-programmed cell death 1 ligand 1 (PD-L1) antibodies. These drugs are thought to be functioning by stimulating cell-mediated immunity to recognize and destroy cancer cells and acting or modulating and targeting relevant immune resistance in tumor microenvironment [3,7,8]. Although PD-L1 expression is a potential biomarker of the therapeutic response to ICIs, it remains controversial whether it is an optimal predictor. Based on the FDA approvement, nivolumab and pembrolizumab (anti-PD-1 antibody) and atezolizumab (an anti-PD-L1 antibody) are granted as the second-line treatment of advanced NSCLC on the basis of improvements in OS versus docetaxel [9][10][11]. As a fully human and monoclonal anti-PD1 antibody, nivolumab was the first PD-1 inhibitor which demonstrated meaningful activity in NSCLC. According to the five-year outcome of the phase III trials (CheckMate 017 and CheckMate 057) [12], nivolumab continued to demonstrate the improvement in overall survival (OS) than docetaxel in previously treated advanced NSCLC patients. Besides, five-year OS for advanced NSCLC patients treated with pembrolizumab [13] (KEYNOTE-001) also implicated sufficient antitumor effect and high 5-year OS rates. In the first-line setting, both nivolumab and pembrolizumab have indicated durable antitumor activity and favorable tolerance than platinum-based doublet chemotherapy in patients without EGFR/ALK aberrations and variably PD-L1enriched patient populations. In KEYNOTE 024 study [14], pembrolizumab significantly improved progressionfree survival (PFS) and OS in advanced NSCLC patients with PD-L1 TPS ≥ 50%. On contrast, durvalumab significantly prolonged PFS in unselected patients with stage III NSCLC in the PACIFIC trial [15]. Consequently, immunotherapy biomarkers such as PD-L1 help to enrich clinical benefit but unable to guarantee the benefit or exclude inappropriate patients.
Tumor mutational burden (TMB) is increasingly serving as an alternative predictor of clinical benefit in immunotherapy [16]. TMB is defined as the number of somatic mutations per megabase of an interrogated genomic sequence [17]; it can be assessed by next-generation sequencing (NGS) to quantify the number of nonsynonymous mutations in the entire exome or defined genome [18]. The limitation of expensive cost and time consuming made it difficult to incorporate into clinical practice. Nevertheless, although TMB might be a perfect response biomarker to improve the predictive accuracy for immunotherapy out-comes [16,17,19,20], the prognostic value of TMB still remains uncertain.
The advanced NSCLC consists of IIIB/IIIC stage and IV stage patients, which varies in therapeutic effect, prognosis, and overall survival time. An important unmet need in immunotherapy is to identify predictive factors that may help select patients who are more likely to benefit from ICI, especially the IV stage NSCLC patients. Serum inflammatory biomarkers such as neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-tomonocyte ratio (LMR), and advanced lung cancer inflammation index (ALI) have been explored as predictive or prognostic factors and the treatment monitoring in NSCLC patients treated with chemotherapy or ICIs [21][22][23][24][25]. However, many indicators such as age, ECOG score, treatment diversification, or drug differences could influence the treatment response or prognosis. Hence, it is important to identify biomarkers helping to provide the most benefit from treatment with minimal risk of toxicity.
In this research, we evaluated the predictive and prognostic significance of biomarkers (NLR, dNLR, LMR, PAB, and LIPI scores) in stage IV NSCLC patients treated with PD-1 inhibitors. We also assessed which biomarker was most specific.

Patients.
We conducted a retrospective analysis and enrolled 85 patients who were cytological or histological diagnosed as stage IV NSCLC and treated with anti-PD-1 antibody between June 2018 and Dec 2019 at the Affiliated Tumor Hospital of Nantong University. Inclusion criteria include the following: (1) aged ≥18 years, (2) pathologically confirmed stage IV NSCLC (according to the 8th version of the International Association for the Study of Lung Cancer TNM Staging System), (3) received anti-PD-1 antibody therapy at our hospital from June 2018 to Dec 2019, and (4) complete data collection and follow-up. Patients with a second malignant tumor, severe comorbidities, active systemic inflammatory, autoimmune diseases, and mental disease that could not cooperate with medical treatment were excluded.
The study was performed according to the Declaration of Helsinki and the International Conference on Harmonization Guidelines on Good Clinical Practice. All enrolled patients signed an informed consent form before participating in this study. The Research Ethics Committee of the Affiliated Tumor Hospital of Nantong University approved this retrospective study (2022053).
The following data were collected from the medical records: age, sex, smoking history, Eastern Cooperative Oncology Group Performance Status (ECOG score), clinical   2  Disease Markers   stage, pathology, biopsy site and method, TNM stage, metastatic sites, EGFR mutation status, PD-L1 expression, line of  therapy, treatment regimen and response, and survival outcome (PFS and OS). Blood test results within three weeks prior to the first administration of anti-PD-1 antibody were collected and recorded into the database. The baseline peripheral blood data include total white blood cell concentration (WBC), absolute neutrophil count (ANC), absolute lymphocyte count, total lymphocyte count, platelet count (PLT), monocyte count, hemoglobin concentration (HGB), lactate dehydrogenase (LDH), alanine aminotransferase (ALT), aspartate aminotransferase (AST), serum albumin (ALB) level, serum prealbumin (PAB) level, serum globulin (GLO), carcinoembryonic antigen (CEA), cytokeratin-19 fragments (CYFRA 21-1), neuron-specific enolase (NSE), and squamous cell carcinoma antigen (SCC). 2.6. Statistical Analysis. All statistical analyses were performed using SPSS 25.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 8.0 (GraphPad Inc., San Diego, CA, USA). The receiver operating characteristic (ROC) curves were constructed to calculate all indicators included above.

Evaluation of Treatment
The area under the ROC curve (AUC) and asymptotic 95% confidence interval were calculated and recorded. Then, we determined which of the indicators might be predictive markers and defined the cut-off values with maximum sensitivity and specificity. The optimal cut-off values and prognostic roles of biomarkers were identified according to the ROC curves and Youden's index. A chi-square test was performed to compare baseline clinical characteristics. The Cox regression model was conducted to evaluate the predictive factors for PFS and OS via the univariate and multivariate analysis. PFS and OS were determined by the Kaplan-Meier method using the log-rank analysis. All statistical tests were performed two-sided, and a P value < 0.05 was considered statistically significant.

Patient Characteristics.
The clinical characteristics of the 85 patients with stage IV NSCLC enrolled in this study are listed in Table 1. The median age was 66 years (range 47-80 years), the majority of patients were male (72.9%), current/former smokers (45.9%), and all patients were clinically diagnosed as stage IV NSCLC. The majority of the patients exhibited the ECOG score of 0-1(88.2%), and the pathology showed the LUSC consisted of 43.5% compared with LUAD (56.5%). According to the biopsy site, most of the tissues were obtained from the primary lung tumor (62.4%) and lymph nodes (22.4%), while in certain challenging cases, we could also get the tissue from pleural/pericardial effusion, subcutaneous nodules, and bone and surrounding soft tissue. Based on the biopsy method, bronchoscope (34.1%) and CT/ Doppler ultrasound (43.6%) were the favorite ways physicians; EBUS-TBNA, pleural/pericardial puncture, or lymph node resection can also help to get tissues.

Receiver Operating Characteristic (ROC) Curves of
Indicators. The receiver operating characteristic (ROC) curves were constructed to access all indicators collected above ( Figure S1). The area under the ROC curve (AUC) and asymptotic 95% confidence interval were calculated and recorded separately ( Figure S2). Based on AUC and

Multifactor Model of LIPI Score for Survival Outcome of Stage IV NSCLC Patients Treated with PD-1 Inhibitors.
Based on the following cut-off values, dNLR ≤ 3 and LDH ≤ upper limit of normal (ULN), dNLR > 3 or LDH > ULN, and dNLR > 3 and LDH > ULN, LIPI score was divided into three subsets of scores: good (0), intermediate (1), and poor (2) LIPI. Then, we explored the OS and PFS by the multifactor model of LIPI score. As indicated in Figure 8(a), PFS of patients with good LIPI score were significantly longer than those with intermediate/poor LIPI score (P < 0:001). Similarly, good LIPI score cohort achieved better overall survival time than the intermediate/poor LIPI score cohort (P < 0:006, Figure 8(b)).

Discussions
This study investigated the predictive and prognostic value of serum biomarkers in stage IV NSCLC patients treated with PD-1 inhibitors. The results revealed that low dNLR or good LIPI score predicts better survival outcomes for these patients, no matter PFS or OS. Further, high dNLR and low ALB level were independent prognostic factors of shorter PFS. Meanwhile, liver metastasis, high NLR, high dNLR, and LMR were independent prognostic indicators of shorter OS. Moreover, prognostic biomarkers such as NLR, dNLR, LMR, and ALB might be significant indicators for PFS, while age, liver metastasis, NLR, dNLR, and LMR might be promising biomarkers for the prediction of OS. Furthermore, in the good LIPI group, PFS and OS were significantly longer. In clinical settings, it is important to use accurate and effective markers to guide clinical treatment and predict prognosis. On the one hand, all these involved serum indicators are easy to obtain and record, so monitoring the indicators above provides a simple and convenient method for clinicians. On the other hand, for certain stage IV non-small-cell lung cancer (NSCLC) patients treated with PD-1 inhibitors, the early prediction of prognosis is definitely of great significance.
A major advancement in cancer treatment is the development of immune checkpoint inhibitors (ICI), which have produced long-lasting responses and improved survival rates in a variety of solid malignancies [17]. However, since most patients receiving ICI treatment did not achieve the expected results, it is necessary to identify predictive biomarkers of ICI response to achieve more clinical benefit, as well as to clarify and overcome the mechanism of treatment resistance. Tumor mutational burden is an indicator controversial. Previously, a study reported that tTMB correlates with bTMB and bTMB helps to identify patients who derive clinically significant improvements in PFS from atezolizumab in second-line or posterior treatment in NSCLC [19]. In terms of determining tumor heterogeneity, the outline of simultaneous genomic changes in NSCLC may be more influential than obvious mutations in oncogenic drivers. Based on current clinical research evidence, cooccurring genomic   [6,27]. Such features may serve as biomarkers for drug sensitivity prediction, especially in immunotherapy. Another study [28] investigated the prognostic value of STK11/KEAP1 mutations in an observational real-world lung adenocarcinoma cohort, and the results suggested that STK11/KEAP1 mutations are prognostic, not predictive biomarkers for immunotherapy. Besides, current studies have shown that TP53 and KRAS mutations in LUAD may be a pair of potential predictive factors in guiding anti-PD-1/PD-L1 immunotherapy [29]. All in all, these markers require sufficient tissue or blood samples for testing. Besides, a certain amount of financial support is needed to complete the inspection. Previous studies have investigated the potential utility of routine blood parameters in the treatment of various tumors. However, there is no uniform boundary value for 15 Disease Markers the emerging biomarkers. In a study evaluating the prognostic value of NLR, PLR, and NLR-PLR score in stage IV GC patients [30], NLR-PLR score showed the value of independently predicted survival outcomes. As for the treatment response for early phase SCLC patients treated with immunotherapy [31], NLR at 6 weeks after initial treatment appears to be a biomarker; the cut-off value of NLR is set as 5. In another research exploring the value of lymphocyte-to-monocyte ratio (LMR) in advanced NSCLC patients who received nivolumab monotherapy [25], the result indicated the rapid increase of LMR (increasement ≥ 10%) was significantly associated with treatment response. Tong et al. conducted a retrospective analysis involving 332 newly diagnosed stage III NSCLC patients [32], which demonstrated that SII (cut-off value 660) is an independent prognostic indicator of poor outcomes for patients. Lung immune prognostic index (LIPI) is also an emerging biomarker that deserves attention. In JAMA Oncology [23,33], researches demonstrated that baseline LDH levels and dNLR are important prognostic biomarkers

18
Disease Markers for metastatic NSCLC patients. An Italian study [34] reported that the systemic inflammatory biomarkers such as ALI, LDH, NLR, and LIPI score may help the understanding of survival differences in the clinical management of lung neuroendocrine carcinomas. In one study, enrolling patients with advanced hepatocellular carcinoma underwent immunotherapy [35], pretreatment LIPI (dNLR ≥ 3 and/or LDH ≥ ULN) were associated with poor outcomes. Although the application of serum biomarkers is emerging into clinical practice, there is still controversial on this issue. First of all, some clinicians argued that different treatments or different populations should have their different cut-off values. The difference of the analysis method and the patients included in the study may influence the cut-off value of each study. Secondly, complications including anemia, pneumonia, abnormal hypothyroidism, liver disease, and heart dysfunction may also affect the serum concentrations of the indicators. Given the inevitable interference of these factors in real-world clinical practice, how to avoid interference scientifically and reasonably raises new    21 Disease Markers questions for clinicians. Thirdly, the stage or the metastatic site of the disease may also have a critical effects on the prognosis. For these reasons, we designed this research aimed to evaluate the significance of blood parameters in this specific population of stage IV non-small-cell lung cancer. Our retrospective study revealed that pretreatment dNLR is an independent prognostic indicator of both PFS and OS in stage IV NSCLC patients treated with PD-1 inhibitors. Besides, pretreatment LIPI combining dNLR ≥ 3 and LDH≥ULN was correlated with worse outcome for stage IV NSCLC patients treated with ICI. High NLR, high dNLR, low LMR, and low ALB at baseline might be useful as an early predictive biomarker of benefit.
However, there are still some shortcomings within our investigation. The first one is, as a retrospective analysis of a single center, the sample size is relatively small. In view of the drug approval of checkpoint inhibitors that were first approved by the National Medical Products Administration (NMPA) in China since June 2018, we only included 85 patients with stage IV NSCLC treated with PD-1 inhibitors

22
Disease Markers in detail. In the next step, our research group intends to further expand the sample size and conduct studies in conjunction with multiple clinical centers. In addition, hematology parameters may be affected by some concurrent medications. As far as our study excluded patients with a second malignant tumor, severe comorbidities, active systemic inflammatory, autoimmune diseases, and mental disease, the effect of combination medication is relatively reduced. Finally, this study has not fully elucidated the basic biological mechanisms. Nonetheless, our study provides a simple, economical, and noninvasive method to help clinicians predict the response and prognosis of anti-PD-1 antibodies. Future research will focus on exploring the significance of serum markers in early-stage NSCLC, and the importance of immunotherapy in a distinct populations.

Conclusions
Pretreatment dNLR is an independent prognostic indicator of both PFS and OS in stage IV NSCLC patients treated with PD-1 inhibitors. Pretreatment LIPI, combining dNLR > 3 and LDH>ULN, is correlated with worse outcome for stage IV NSCLC patients treated with ICI. High NLR, high dNLR, low LMR, and low ALB at baseline might be useful as an early predictive biomarker of benefit.

Data Availability
Data is owned and saved by the Affiliated Tumor Hospital of Nantong University and are available on request to the corresponding author. For researchers meeting the criteria for access to confidential data, please contact the following Email address: xiaodongzhang@csco.ac.cn.