Predictive Value of Albumin and Neutrophil Combined Prognostic Grade in Advanced Gastric Cancer Patients Treated With anti-PD-1 Therapy

Background The application of immunotherapy is gradually increasing in advanced gastric cancer (AGC), but only some patients could benet from it. Validated biomarkers can screen out the beneciaries. The objective of this research is to explore the predictive value of albumin and neutrophil combined prognostic grade ( ANPG) in AGC patients receiving immunotherapy. Methods A total number of 268 AGC patients were included. The cut-off value of albumin was 38 g/L obtained by the median value, and neutrophil was 4.16 g/L estimated by the average value. The high levels of albumin ( ≥ 38 g/L) and neutrophil ( ≥ 4.16 g/L) were considered to be two risk factors for ANPG. Based on these two risk factors, patients were categorized into 3 groups(cid:0)the risk factor number for the "good" group was 0, for the "intermediate" group was 1, and for the "poor" group was 2. Results Patients with the good ANPG was related to longer progression free survival (PFS) and overall survival (OS), compared to those with the intermediate and the poor ANPG (5.6 months vs 5.3 months vs 3.4 months, 17.8 months vs 11.8 months vs 8.2 months). The poor group was independently correlated with an over 1.9 times risk of disease progression (HR=1.901; 95% CI, 1.314-2.750; P=0.001) and an over 2 times risk of death (HR=2.003; 95% CI, 1.306-3.072; P=0.001) than the good group. The intermediate group was independently correlated with an over 1.3 times risk of disease progression (HR=1.385; 95% CI, 1.004-1.911; P=0.048) and an over 1.4 times risk of death (HR=1.484; 95% CI, 1.046-2.106; P=0.027) than the good group. patients receiving immunotherapy. Patients with the good ANPG could benet from immunotherapy. microsatellite steady-state (MSI);lung immune prognostic index (LIPI); neutrophil-lymphocyte ratio (NLR); platelet-lymphocyte ratio (PLR); hemoglobin (Hb); progression free survival (PFS); albumin and neutrophil combined prognostic grade (ANPG); non-small cell lung cancer (NSCLC); eastern cooperative oncology group performance status scores (ECOG PS); disease control rate (cid:0)DCR); overall response rate (ORR) ; complete response (CR); partial response (PR) ; stable disease (SD) ; progressive disease (PD); c-reactive protein (CRP);glasgow prognostic score (GPS); carcinoembryonic antigen (CEA)

evaluation of nivolumab was carried out 2-4 weeks after the 3rd intravenous injection, but, the evaluation of toripalimab, sintilimab, and pembrolizumab were carried out 3-5 weeks after the 2rd intravenous injection. The course of trastuzumab was 3 weeks. For the rst course, the dose was 8 mg/kg, applied by intravenous injection for 90 minutes. Starting from the 2rd course, the dose was lowered to 6 mg/kg. And for the infusion time, if the patients tolerate trastuzumab well in the rst course. Anti-angiogenic drugs involve apatinib (850 mg, orally administrated 30 min after a meal, once a day), and bevacizumab (5 mg/kg body weight, once every two weeks; or 7.5 mg/kg body weight, once every 3 weeks). The chemotherapy regimens include 1) XELOX regimen: capecitabine (1000mg/m²) was used 2 times a day orally after breakfast and dinner for 14 consecutive days with 7 days of rest as a treatment cycle. oxaliplatin (130mg/m²) was added at the rst day of each cycle by intravenous injection; 2) SOX regimen: tiggio: (40-60mg) was used 2 times a day orally after breakfast and dinner for 14 consecutive days with 7 days of rest as a treatment cycle. oxaliplatin (130mg/m²) was added at the rst day of each cycle by intravenous injection; 3) DCF regimen: docetaxel (75mg/m²), cisplatin (75mg/m²), and uorouracil (750mg/m²) were applied by intravenous injection. On the rst day of every course, each course lasted 21 days. 4) The combined regimen of irinotecan and oxaliplatin: irinotecan (180 mg/m²) and oxaliplatin (130mg/m²) were applied by intravenous injection. On the rst day of every course, each course lasted 14 days. 5) The combined regimen of irinotecan and raltitrexed: irinotecan (180 mg/m²) and raltitrexed (3 mg/m²) were applied by intravenous injection. On the rst day of every course, each course lasted 14 days. 6) others. The choice of the above regimens was based on the patient's pathological stage, general health conditions. 2.3 Assessment For e cacy evaluation, the disease control rate(DCR)and the overall response rate (ORR) is termed as the percentage of patients with complete response (CR), partial response (PR) and stable disease (SD) and the percentage of patients with CR and PR, respectively. For prognosis analysis, OS and progression free survival (PFS) are the time from the begining of immunotherapy to death and the time between the onset of ICIs and the progression or death of the tumor, respectively.
2.4 Neutrophil, albumin, and ANPG We analyzed the value of neutrophil, and albumin 7 days before implementing immunotherapy. The cut-off value of albumin was estimated by the median value and neutrophil was obtained by the average value. The high levels of albumin (≥ the median value ) and neutrophil (≥ the average value) were considered to be two risk factors for ANPG. Based on these two risk factors, patients were categorized into 3 groups: the risk factor number for the "good" group was 0, for the "intermediate" group was 1, and for the "poor" group was 2.
2.5 Statistical analysis SPSS 26.0 software was used to perform all statistical analysis. Datas were summarized as the minimum-maximum range and median for non-normally distributed continuous variables. Datas were reported as percentages and counts for categorical variables. χ2 or Fisher exact test was carried out to evaluate the relationship between clinical response and ANPG groups of AGC patients. The survival curve was depicted by Kaplan-Meier analysis. Logistic regression models and Cox proportional hazard were applied to assess the prognostic values of ANPG groups for DCR, and survival, respectively. Values of P less than 0.05 (p<0.05) were considered statistically signi cant.

Results
3.1 Baseline characteristics A total of 268 AGC patients receiving ICIs were retrospectively reviewed in the study. The clinical feature of patients were provided below ( Table 1). The cut-off value of albumin and neutrophil were 38 g/L and 4.16 g/L, respectilvely. After grouping, the numbers of patients with the good ANPG, with the intermediate ANPG, and with the poor ANPG were ninety ve, one hundred and eighteen, and fty-ve. The median age of patients with the good ANPG, the poor ANPG,and the poor ANPG were 56-year-old, 60-year-old, and 59-year-old, respectively. Patients were predominantly male (74.3%), no history of smoking (62.3%), short history of smoking exposure (smoking exposure history is less than or equal to 30 years in 82.1% of cases), ECOG PS of 0-1 ((94%), the tumor location of Body/Fundus (41%), no hepatic metastasis (55.6%), negative expression of HER-2 (66%), no pleural uid (92.9%), no ascites (75.4%) and had fewer organ metastasis (the number of metastatic sites of 0-2 in 75.4% of cases)  (Figure 1a and Table 3). Univariate and multivariate analyses of factors associated with PFS were shown in Table 3. In univariate analysis, patients with a good PS (ECOG PS of 0-1), with no ascites, or with no pleural uid showed improved OS. Patients did not reach PD before immunotherapy, and those who were treated with ICIs combined with chemotherapy, who were treated with the 1st line ICIs, and who were treated with pembrolizumab were also associated with improved PFS. Multivariate analysis revealed that the poor ANPG and the intermediate ANPG were independently correlated with an over 1.9 times risk of disease progression (HR=1.901; 95% CI, 1.314-2.750; P=0.001) and an over 1.3 times risk of disease progression (HR=1.385; 95% CI, 1.004-1.911; P=0.048) than the good ANPG.
Moreover, patients who were treated with ICIs combined with chemotherapy, who were treated with the 1st line ICIs, and who were treated with pembrolizumab were also independently associated with improved PFS. Patients treated with ICIs after 1st line were independently correlated with an over 1.6 times risk of disease progression (HR=1.699; 95% CI,1.244-2.320; P=0.001) than those treated with the 1st line ICIs. Patients treated with nivolumab were independently correlated with an over 1.9 times risk of disease progression (HR=1.972; 95% CI,1.259-3.089; P=0.003) than those treated with pembrolizumab. Patients treated with ICIs combined without chemotherapy were independently correlated with an over 1.  (Figure 1b and Table 3). Univariate and multivariate analyses of factors associated with OS were shown in Table 3. In univariate analysis, In univariate analysis, patients with fewer organ metastases (<3), with a good PS (ECOG PS of 0-1), with no ascites, or with no pleural uid showed improved OS. Patients did not reach PD before immunotherapy, and those who were treated with ICIs combined with chemotherapy, who were treated with the 1st line ICIs, and who were treated with pembrolizumab were also associated with improved OS. Multivariate analysis revealed that the poor ANPG and the intermediate ANPG were independently correlated with an over 2 times risk of death (HR=2.003; 95% CI, 1.306-3.072; P=0.001) and an over 1.4 times risk of death (HR=1.484; 95% CI, 1.046-2.106; P=0.027) than the good ANPG. Moreover, patients with no pleural uid and with a good PS (ECOG PS of 0-1) were independently associated with improved OS. Patients who were treated with the 1st line ICIs and who were treated with pembrolizumab were also independently associated with improved OS.
Patients with pleural uid were independently correlated with an over 2.2 times risk of death (HR=2.252; 95% CI, 1.333-3.804; P =0.002) than those without pleural uid. Patients with a good PS (ECOG PS of 0-1) were independently correlated with an over 1.9 times risk of death (HR=1.996; 95% CI, 1.122-3.552; P =0.019) than those with a poor PS (ECOG PS of ≥2). Patients treated with ICIs after 1st line were independently correlated with an over 1.8 times risk of death (HR=1.817; 95% CI,1.315-2.512; P<0.001) than those treated with the 1st line ICIs. Patients treated with nivolumab were independently correlated with an over 1.5 times risk of death(HR=1.577; 95% CI,1.005-2.476; P=0.048) than those treated with pembrolizumab.
3.6 Association of the ANPG with outcomes in lines of immunotherapy of 1 or a large number of lines of immunotherapy (≥2): Subgroup Analysis Multivariate analysis revealed that patients treated with the 1st line ICIs were independently correlated with improved OS and PFS. Our study then conducted subgroup analysis based on different lines of immunotherapy. Univariate analyses of association of the ANPG with outcomes in lines of immunotherapy of 1 was shown in Table 4 (Figure 2b). Univariate analyses of association of the ANPG with outcomes in lines of a large number of lines of immunotherapy (≥2) was shown in Table 5 3.7 Association of the ANPG with outcomes in ICIs combined with chemotherapy or combined without chemotherapy: Subgroup Analysis Multivariate analysis revealed that patients treated with the ICIs combined with chemotherapy were independently correlated with improved OS and PFS. Our study then conducted subgroup analysis based on different schemes of immunotherapy. Univariate analyses of association of the ANPG with outcomes in ICIs combined with chemotherapy was shown in Table 6.

Discussion
The status of immunotherapy in the eld of GC treatment has an increasing trend worldwide annually [7,18] . Judging from the current clinical trial results, the effect of immune checkpoint inhibitors is very obvious [19] .Whereas, immunotherapy drugs are expensive and prone to drug resistance and even super-progress [20,21] . Therefore, looking for predictive indicators is an important problem to be solved urgently that can accurately identify the population with the advantage of immunotherapy to achieve precise immunotherapy as much as possible. However, the current evaluation of biomarkers for immunotherapy is relatively limited [15] . Peripheral blood in ammatory complex index such as LIPI, NLR, PLR, Hb levels had demonstrated a convenient and promising prognostic biomarker for GC [12][13][14][15][16] . The highly heterogeneous characteristics of GC may limit the accuracy of a single biomarker for screening immunotherapy bene ting populations [22] . In contrast, the combination of multiple indicators can provide more targeted information for the detection of potential immune bene t subgroups. Sun H et al. found that patients with the good ANPG was independently associated with better prognosis for NSCLC patients [17] . ANPG was a composite biomarker of neutrophil and albumin [17] . The high levels of albumin (≥the cut-off value) and neutrophil (≥the cut-off value) were considered to be two risk factors for ANPG. Based on these two risk factors, patients were categorized into 3 groups: the risk factor number for the "good" group was 0, for the "intermediate" group was 1, and for the "poor" group was 2 [17] . However, there is no research on ANPG in immunotherapy of patients with tumor. Thus, we carried out the present study to explore the clinical value of ANPG in predicting prognostic outcomes of AGC patients following ICIs treatment. In our study, the cut-off value of albumin was 38 g/L obtained by the median value, and neutrophil was 4.16 g/L estimated by the average value. We found that patients in good ANPG group were independently related to longer PFS and OS, compared to those in the intermediate and poor ANPG group. The prognostic value of ANPG in AGC patients treated with ICIs was consistent with in NSCLC patients of the study of Sun H et al [17] . Moreover, subgroup analysis of our study, based on different lines of immunotherapy, found that ANPG was correlated with PFS but not OS of AGC patients treated with the 1st line ICIs and was correlated with PFS and OS of AGC patients treated with ICIs in subsequent lines. Subgroup analysis of our study, different schemes of immunotherapy, found that ANPG was correlated with PFS but not OS of AGC patients treated with ICIs combined with chemotherapy and was correlated with PFS and OS of AGC patients treated with ICIs combined without chemotherapy. Baicun Hou et al. noticed that patients with a good PS (ECOG PS of 0-1) were also independently associated with PFS and OS for AGC patients treated with ICIs [16] . But, in our study, patients had a good PS (ECOG PS of 0-1) were independently associated with improved OS, but without improved PFS. Baicun Hou et al. noticed that patients treated with combination of immunotherapy and other therapies were associated with improved OS and PFS [16] . Patients receiving ICIs combined without chemotherapy were associated with shorter PFS but not OS in our study. We found patients treated with the 1st line ICIs were independently associated with improved PFS and OS. However, no clear difference of OS and PFS were observed between patients treated with the 1st line ICIs and treated with ICIs after the 1st line in study of Baicun Hou et al. [16] . Baicun Hou et al. found that patients had fewer organ metastases (< 2) and treated with the 1st line ICIs were not independently associated with improved PFS and OS than those had more organ metastases (≥2) [16] . We also found patients had fewer organ metastases (< 3) were not independently associated with improved PFS and OS. Baicun Hou et al. found that the type of ICIs was not an independent factor affecting survival of AGC patients receiving immunotherapy [16] . However, our study found that patients treated with pembrolizumab were independently correlated with improved OS and PFS than those treated with nivolumab. In addition, our study rstly found that patients without pleural uid were independently correlated with improved OS than those with pleural uid, but without improved PFS.
However, the mechanism of the correlation between these peripheral blood in ammatory complex index and the tumor prognosis is relatively complicated, and it still needs to be further explored through basic experiments and clinical trials. Some studies have found that this probably due to the tumor-immune microenvironment of patients [23,24] . In addition to direct immune killing effects on tumor cells, these biomarkers are also related to tumor immunostimulatory signals and the activation of effector cells.
Neutrophils, one type of the most vital and abundant leukocytes in circulating blood, are the rst-line defense to protect host from tissue damage and infections [25] . Neutrophils are derived from bone marrow hematopoietic stem cells and have chemotaxis, phagocytosis and bactericidal effect [26] . The reactive oxygen species released by neutrophils can damage DNA, which is related to the occurrence and development of tumors [27] . Increased number of neutrophils in peripheral blood can promote tumor metastasis and growth by releasing in ammatory mediators [28] . In addition, it can not only enhance the growth of tumor cells under the effect of tumor, its microenvironment reproduction and invasion can promote angiogenesis and mediate tumor immunosuppression [29] . Lymphocyte is an important component of cell for body's immune response function [30] . Elevated neutrophils can inhibit the immune attack ability of lymphocytes [31] . Albumin one type of the most vital biochemical Indicators, is related to nutritional status of patients with cancer [32] . Some studies found that hypoalbuminemia are associated with worse prognosis of patients with GC [33,34] . Some researchers thought if it was associated with other powerful clinical indicators, such as c-reactive protein (CRP), the albumin level as a prognostic indicator would be more convincing [35,36] . Nozoe T et al. found glasgow prognostic score (GPS) which was a composite biomarker of albumin and CRP could be used as an positive prognostic indicator of patients with GC [36] . Zhang J et al. found that a composite biomarker of serum carcinoembryonic antigen (CEA) and brinogen/ albumin ratio also could be used as an positive prognostic indicator of patients with GC [37] . Therefore, the higher level of neutrophils and the lower level of tumor patients get, the worse their prognosis will be. ANPG was a composite biomarker of neutrophil and albumin [17] . In other words, patients with good ANPG will have a better prognosis.
This study had some limitations, including a relatively small sample size with a mixed population of GC of cardia, GC of body/fundus, and GC of pylorus, as well as a lack of comparison of the three ANPG groups among the three cancers.

Conclusion
This study demonstrated that a composite biomarker of ANPG is independently correlated with the survival of AGC patients implementing immunotherapy. However, the possibility of using the complex index as an effective and economic prognostic biomarker to selected patients, those who are best suited to receiving ICIs, needs further investigation in a larger prospective study.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
This study was approved by the Ethics Committee of Chinese PLA General Hospital and was conducted according to the principles of the Declaration of Helsinki.

AUTHOR CONTRIBUTIONS
YP were in charge of writing and analysis. GD and ZK provided the guide and idea. All authors contributed to the article and approved the submitted version.
3 . Nozoe T, Iguchi T, Egashira A, Adachi E, Matsukuma A, Ezaki T. Signi cance of modi ed Glasgow prognostic score as a useful indicator for prognosis of patients with gastric carcinoma. Am J Surg.