Background
Gastric cancer is one of the most common malignant tumors in east Asia. In China, the 5-year survival rate of this disease is relatively lower because a large proportion of gastric cancer patients were already at advanced stage when diagnosis [
1], and about 500,000 people died from it during the year 2015 [
2]. At present, the major treatment option for advanced unresectable gastric cancer is chemotherapy. Unfortunately, doctors have to face that many patients give up further treatments just because of the economic reasons in the less developed regions. Previous studies have demonstrated that some cancer endogenous factors, such as differentiation status [
3,
4] or certain genes expression [
4‐
6], may influence the efficiency of chemotherapy. However, we still know very little about whether there are other affordable pre-treatment factors can better predict progression free survival in advanced unresectable gastric cancer patients, which may help to optimize the treatment strategies.
It was reported that systemic inflammatory response plays important roles in the progression of various cancers [
7‐
9]. Circulating inflammatory cells can release many biological active factors and thus lead to tumor growth, progression, and metastasis [
10‐
12]. Previously, we have demonstrated a close cooperation between neutrophils, platelets and lymphocytes in basic experiments [
13,
14]. In clinical studies, neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) were adopted for prognosis evaluation in many cancers [
15‐
23]. Recently, a study showed that preoperative combined NLR and PLR Score (cNPS) better predicted the postoperative survival in early stage gastric cancer than NLR or PLR alone [
21]. However, it remains unclear whether cNPS has better prognostic prediction value for the first-line chemotherapy in advanced unresectable gastric cancer patients. Moreover, cNPS only implied the systemic inflammatory status. We suppose that taking the endogenous factor (e.g. differentiation status) into account together may further improve the prediction efficiency.
Therefore, we introduce a novel combined Neutrophil/platelet/lymphocyte/differentiation Score (cNPLDS), together with cNPS and other clinical indices, to find out the optimum parameters to predict survival and clinical responses of first-line chemotherapy for advanced gastric cancer.
Methods
Sample size estimation
Sample was estimated by statistical methods (
http://powerandsamplesize.com/). Using the Cox PH, 2-Sided Equality Model online, the inputted parameters of estimated Hazard Ratio was 1.75 to 2.25 according to pervious similar study on advanced GC [
16], the overall probability of event was 0.8, proportion of samples in groups was 1:1. As a result, the estimated sample size ranged from 60 to 126 by the power of 0.8.
Patients and eligibility criteria
A retrospective analysis was performed based on the medical records at Nanfang Hospital, Guangzhou, China. From February, 2011 to August, 2017 at this center, there were 589 advanced gastric cancer patients who lost the radical resection opportunity. Among them, we only focused on the patients who initially treated with standardized palliative first-line chemotherapy with regular follow-up. Those patients who initially received second or more advanced line chemotherapy at our hospital were excluded from the study.
All the patients should have a Zubrod-ECOG (WHO) score ≤ 2. The histopathology diagnosis of gastric cancer was confirmed by at least two experienced pathologists in our hospital. According to WHO classification of stomach cancer histologic types, well differentiated is defined as being composed of well-differentiated adenoid structures, and intestinal metaplasia sometimes could be seen. Poorly differentiated is defined as being composed of irregular-differentiated glands and sometimes even hard to identify the adenoid structures. The microscopy appearance of moderately differentiated is between those of well differentiated and poorly differentiated. Patients had no history of hematological diseases. Antineoplastic therapies (chemotherapy, radiotherapy or immunotherapy, etc.) were not received within two months before the start of first-line chemotherapy, and other treatments that could affect bone marrow hematopoiesis were also not received before the start of first-line chemotherapy. The chemotherapy regimens were limited to the protocol of advanced gastric cancer recommended by guidelines of the Chinese Society of Clinical Oncology (CSCO), the National Comprehensive Cancer Network (US) (NCCN) and the European Society of Medical Oncology (ESMO). The patients changing the chemotherapy regimen before meeting the PD standard, or simultaneously receiving other antineoplastic therapy (including radiotherapy or immunotherapy, etc.) during the treatment, were excluded from the study. As a result, a total of 136 patients met the eligibility criteria and were included in the studies.
Follow-up and assessment
The response to treatment was assessed about every two cycles of three-week regimens or every four cycles of two-week regimens (i.e., every 6 to 8 weeks) on the basis of the rules established by the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 [
24]. Baseline assessment was performed within 2 weeks before the first-line chemotherapy. The chemotherapy response status was defined as complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD). Progression free survival (PFS) was calculated from the beginning of the first-line treatment until PD. Platelet counts, neutrophil, and lymphocyte values were collected before the initiation of first-line treatment.
Statistical analysis
Student’s t test or Analysis of variance (ANOVA) were used to compare PLR, NLR and the clinical pathologic characteristics. Correlation between NLR and PLR, and the disease control rate (DCR; the rate of CR + PR + SD) between groups were assessed by Pearson’s correlation analysis. Association between differentiation score and other parameters were assessed by Chi-square test. Survival analysis was carried out using Kaplan-Meier methods and compared by the Log-rank test. The receiver operating characteristic (ROC) curve was applied to determine the optimal continuous variables cut-off points based on the largest Youden’s index. The prognostic prediction priority of different clinical parameters was compared by areas under the ROC curve (AUC). Univariate and multivariate Cox-regression analyses were performed to identify the independent predictors for PFS. All analyses were performed using the SPSS 23.0 statistical software program (IBM, USA). All statistical tests used in this study were two-sided and P < 0.05 was considered statistically significant.
Discussion
Chemotherapy is the main treatment for advanced unresectable gastric cancer, and the sensitivity to first-line chemotherapy has a direct impact on overall survival prognosis [
25]. In this study, we introduce a novel prognostic parameter cNPLDS, integrating systemic NLR and PLR and local tumor differentiation status to predict the PFS of the first-line chemotherapy in advanced gastric cancer. We proved that the AUC of cNPLDS is larger than NLR, PLR, and cNPS, showing its priority in predicting prognosis. In addition, as an adverse prognostic factor, cNPLDS covers more high-risk population, and further refines the hazard classification. Moreover, cNPLDS is an economical clinical tool, which just depends on the basic diagnostic indices and dispenses with additional examinations. Hence, it is easy to apply in the less developed areas, where gastric cancer is generally of high prevalence and high mortality.
Previously, many studies used peripheral blood routine examination indices, e.g. NLR or PLR, and their derived cNPS, as predictors for curative resected gastric cancer recurrence or overall survival [
17‐
19,
23]. However, studies about these parameters for chemotherapeutic response prediction in advanced unresectable gastric cancer are not too much [
26‐
28]. Recently, a study based on the REAL-2 trail advanced oesophago-gastric cancer population reported that elevated NLR status demonstrate a significant negative prognostic effect on the first-line chemotherapy response [
26]. It was also demonstrated that F-NLR score, a combining score of NLR and fibrinogen, predicted the therapeutic response of advanced gastric cancer to chemotherapy or chemoradiotherapy [
28]. Interestingly, another study of advanced gastric cancer showed an association between the mismatch repair system status and NLR, which both influence the PFS and overall survival (OS) [
27]. This suggested that the great histopathological heterogeneity should not be overlooked. Therein, tumor differentiation is also one of the most basic histopathological index. Given these, the novel cNPLDS was introduced, which reflects both systemic immunological status and local pathological characteristics. In this study, such combination was showed to improve the prediction efficiency. Moreover, since cNPLDS is a complicated prognostic parameter, we believe it would be more stable than single index. No statistical correlation was found between tumor differentiation and NLR or PLR, which suggested they were two relatively independent sets of mechanism that both influence therapeutic response. This is probably another reason why cNPLDS is superior to differential status, NLR, PLR or integrated cNPS.
It is known that inflammation accompanies along with cancer progression [
10]. In addition to local inflammatory response, systemic inflammatory response is also an important prognostic factor in cancer. Many experimental studies documented that neutrophils, platelets and lymphocytes participate in the cancer generation or malignancy progression [
29‐
31]. It was reported that local cancer associated neutrophils of N2 phenotypes promote tumor growth and invasion by altering the extracellular matrix and inhibiting the function of other immune cells [
31]. Activated platelets, which release various growth factors and cytokines, promote tumor metastasis [
32,
33], angiogenesis [
32,
34], and chemoresistance [
32,
35]. Platelets can help the circulating tumor cells forming cancer embolus, and protect them from the immunocytes [
32,
36]. Recently studies showed that platelet activation triggered platelet-neutrophil interaction and alters the immunocyte subpopulations [
37]. In this study, we found a close correlation between NLR and PLR, suggesting that peripheral circulating neutrophils and platelets may play a synergetic effect on primary chemoresistance. Conversely, lymphocytes possess anti-tumor effect [
38]. We reported that platelets could enhance the differentiation and cytokine production of CD4
+ T lymphocytes [
39], which could be regarded as a physiological balance. However, when the balance is broken, like the increase of the neutrophils or platelets but decrease of lymphocytes, the systemic immunity will turn abnormal. This probably explains why high NLR, PLR or cNPS patients has shorter PFS.
In the cNPLDS, tumor differentiation status is a relative steady composition, while other elements of neutrophils, platelets and lymphocytes were dynamic. This made us curious whether intervention of the latter might cover the deficiency of poor differentiation status on prognosis. Recently, increasing clinical data showed that tartigeting-PD-1 or PD-L1 treatment plus chemotherapy received more favorable efficiency [
40], suggesting activating lymphocyte may be a promising approach for chemotherapy optimizing. Besides lymphocytes activation, we also wonder whether inhibiting neutrophils or platelets designedly to certain level or suppressing their activities may improve the therapeutic effect. Few studies explored pharmacological inhibition on neutrophils, which is probably because of catastrophic side effects, such as agranulocytosis or consequent sever infection. However, recent in vitro and in vivo study showed that some antiplatelet activation drugs, such as aspirin [
41], P2Y12 inhibitor [
42], and Integrin β3 inhibitors [
43], may play tumor suppressive roles. Although some clinical trials reported the value of aspirin in cancer prevention and improving forward prognosis [
44‐
46], there were also some negative results documented [
47,
48]. On the other hand, there were no good indicators to determine which subgroups of patients might benefit from platelet intervention. This probably resulted in the trail failure. In this study, cNPLDS is introduced as a prognostic factor integrating platelets, neutrophils, lymphocytes and tumor differentiation, and we suppose that those high cNPLDS patients may probably benefit from the pharmacological interventions. However, this need further studies and larger data to support.