Peripheral blood cell count ratios are predictive biomarkers of clinical response and prognosis for non‐surgical esophageal squamous cell carcinoma patients treated with radiotherapy

Abstract Background Peripheral blood cell count ratios, including the neutrophil‐to‐lymphocyte ratio (NLR), platelet‐to‐lymphocyte ratio (PLR), and lymphocyte‐to‐monocyte ratio (LMR), have been reported to be prognostic factors in many malignancies as markers of inflammation and immune status. The aim of this study was to determine whether NLR, PLR, or LMR can be clinical response and prognostic biomarkers of non‐surgical esophageal squamous cell carcinoma (ESCC) patients treated with radiotherapy. Methods 193 non‐surgical ESCC patients who underwent radiotherapy were retrospectively analyzed. The peripheral blood cell count ratios were obtained before, during (weekly) and at the end of the treatment. Then, we compared the subsequent results with the corresponding pretreatment values and computed the rates of change, which were defined as cNLR, cPLR, and cLMR. Univariate and multivariate Cox regression analyses were used for overall survival (OS). Ordinal logistic regression was used to analyze the clinical response. Results In multivariate analysis, cNLR at week 4(P = .026) and week 5(P = .025) during radiotherapy were significantly associated with OS, along with BMI, tumor stage, tumor length, tumor location, and grade of adverse events. Besides, BMI, tumor stage, tumor length, adverse event grade, cNLR at week 4(P = .044) and week 5(P = .013), and cPLR at week 4(P = .034) and week 5(P = .015) were significantly associated with the clinical response in the multivariate logistic regression analysis. Conclusions The cNLR at weeks 4 and 5 was negatively correlated with the OS and clinical response of non‐surgical ESCC patients treated with radiotherapy. The elevated cPLR at weeks 4 and 5 was only related to poor clinical response.


| INTRODUC TI ON
Esophageal carcinoma (EC) is one of the most common cancers worldwide. The incidence of EC is ranked seventh among all types of malignant tumors, and mortality is ranked sixth, according to Global Cancer Statistics 2018. 1 The 5-year age-standardized net survival of EC is approximately 10%-30% worldwide. 2 Eastern Asia, including China, has a high incidence of EC. Our research group is located in one of the highest incidence areas of EC in China. The major histological type of EC in China is esophageal squamous cell carcinoma (ESCC), which differs from that in Western countries. 3 At present, the diagnosis and prognosis evaluation of ESCC mostly depend on endoscopic procedures and imaging tests, such as gastroscopy, computed tomography (CT), and esophagram. Unfortunately, unlike other digestive system neoplasms, EC lacks blood biomarkers for predicting prognosis and tumor response to treatment and biomarkers for risk stratification.
Evidence has increasingly shown that inflammation plays an important role in tumor development and progression by modifying the tumor microenvironment. [4][5][6] Recently, various studies have shifted their sights to inflammatory biomarkers in peripheral blood. Neutrophil and platelet regularly change with the level of systemic inflammation, while lymphocyte and monocyte can indicate the level of immunity. Therefore, peripheral blood cell count ratios, including the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and lymphocyte-to-monocyte ratio (LMR), can quantify the inflammatory and immune response. They can be used as practical prognostic biomarkers to evaluate patient outcomes in various malignancies. [7][8][9] In EC, many researches have estimated the prognostic significance of pretreatment peripheral blood cell count ratios. 10 Therefore, we performed this retrospective study to evaluate whether these simple and repeatable parameters can be clinical response and prognostic biomarkers of non-surgical ESCC patients treated with radiotherapy.

| Patient eligibility
A total of 755 patients diagnosed with ESCC by pathology underwent radiotherapy at our institute between January 2013 and December 2016. Patients were omitted from this study if they met any of the following exclusion criteria: (a) patients without records of peripheral blood cell counts before or during the study period; (b) patients with any other malignancy before or during the study; (c) patients who had received previous chemotherapy or radiotherapy; (d) patients who had undergone previous cancerrelated surgery; (e) patients who received palliative or supportive treatment only; (f) patients with underlying diseases that might influence peripheral blood cell counts, such as liver cirrhosis or infection; (g) patients who received certain medications within 5 days before blood sample collection, such as granulocyte colony-stimulating factor or thrombopoietin; and (h) patients lost to follow-up. Through the above filters, 193 patients were selected.
In addition to the overall analysis of all patients, we conducted subgroup analyses for patients who received concurrent chemoradiotherapy (CCRT) and RT alone. This study was approved by the Ethics Committee of the Affiliated Huai'an Hospital of Xuzhou Medical University.

| Clinicopathological data
Pretreatment tumor stage was classified based on the Clinical

Classification of Esophageal Carcinoma Treated by Non-surgical
Methods, the accuracy of which has been well confirmed. 12,13 According to this classification system, T stage was assessed based on tumor length measured by barium esophagram, the maximum esophageal diameter of the largest esophageal lesion on CT scans and whether the tumor invaded adjacent organs. N stage was evaluated based on the diameter and location of the lymph nodes on CT scans.
All clinical characteristics were extracted from electronic medical records.

| Treatment protocol
All patients underwent CT simulations before radiotherapy. The radiation range involved the primary tumor along with the prophylactic regional lymph nodes. The radiation treatment was delivered as intensity-modulated radiation treatment (IMRT) using conventional fractionation (CF), simultaneous integrated boost (SIB), or sequential boost (SB

| Peripheral blood cell count ratios
Laboratory data were extracted from the electronic medical records.
Blood samples were obtained before, during(weekly), and at the end of RT and were analyzed at the clinical laboratory of our institute. The NLR was calculated as the absolute neutrophil count divided by the absolute lymphocyte count. The LMR was indicated as the absolute lymphocyte count divided by the absolute monocyte count. The PLR was calculated by dividing the absolute platelet count by the absolute lymphocyte count. Each patient's peripheral blood cell count ratio prior to treatment was defined as the baseline value. Then, we compared the subsequent results with the baseline and computed the rates of change. These rates were defined as cNLR, cPLR, and cLMR for the NLR, PLR, and LMR, respectively. The formulas were as follows:

| Statistical analysis
Univariate and multivariate ordinal logistic regression analyses were performed to assess the association between variables and clinical responses. Independent prognostic factors for OS were determined by univariate and multivariate Cox proportional regression models. To assess the differences between the baseline characteristics of the CCRT and RT alone groups, the chi-squared test or Fisher's exact test was used for categorical variables, and two independent samples t tests were applied to continuous variables. We estimated the OS of patients in each group by the Kaplan-Meier method and applied log-rank tests to compare the survival curves. Variance analysis or the rank sum test was used to analyze the differences in cNLR, cPLR and cLMR between groups, and graphs were drawn to view the trend of these parameters. All statistical analyses were performed using SPSS version 23.0 (SPSS, Inc.), and statistical significance was defined as P < .05.

| Patient characteristics
The baseline patient, tumor, and treatment characteristics are presented in Table 1

| Peripheral blood cell count ratios during radiotherapy
During radiotherapy, the median cNLR and cPLR increased gradually every week, reaching the highest values at week 5 and then declining. In contrast, the median cLMR went down week by week and reached the lowest value at week 5 and then returned up ( Figure 1A). The maximum weekly median cNLR and cPLR were 3.06 (95% CI = 2.76-3.58) and 2.63 (95% CI = 2.33-2.93), respectively.
Most minimum or maximum values appeared in the fourth or fifth week of radiotherapy.

| Association of peripheral blood cell count ratios with clinical response
We analyzed the relationship between clinicopathologic features, including peripheral blood cell count ratios, and clinical response (

| Association of peripheral blood cell count ratios with survival outcomes
By the end of follow-up, the median overall survival was 33.5 months (95% CI = 27.0-39.9). Cox regression analysis was performed for predictors of overall survival (  showed that CCRT remained insignificant (Table S1). The model was statistically significant, and there is no significant collinearity among the parameters (Tables S2 and S3).

| Comparison between CCRT and RT alone patients
The baseline clinical characteristics of patients in the CCRT group and RT alone group are shown in During the treatment, the median cNLR and cPLR in the CCRT group were lower than those in the RT alone group, except at week 4, and the gap narrowed gradually ( Figure 1B,C). At week 4, the median cNLR in the CCRT group became higher than that in the RT alone group, while the median cPLR was equal in the two groups.
The trend of the cLMR was the opposite ( Figure 1D). A detailed comparison is shown in Table 5. In addition to the cLMR at weeks 1 and 2 (P = .045, 0.029), no statistical difference was observed between the two groups.
The median OS of the CCRT group was 37.8 months (95% CI = 28.361-47.239), while that of the RT alone group was 29.5 months (95% CI = 19.209-39.791) ( Figure 2). However, the logrank test showed that there was no statistical difference between the two groups (P = .279). For the past several years, inflammation has been investigated as a hallmark of cancer. 15 Through modulation of the tumor microenvironment, inflammation can induce DNA damage, dysregulate the cell cycle, and lead to angiogenesis. 16,17 There are many biomarkers of the inflammatory response that have been reported to be related to tumor progression. 18,19 Among them, peripheral blood cell count ratios, such as NLR, PLR, and LMR, have received much attention due to their convenience and repeatability. However, though radiotherapy is the main treatment for non-surgical ESCC

F I G U R E 1
The median change rates of neutrophil-to-lymphocyte ratio (cNLR), platelet-to-lymphocyte ratio (cPLR), lymphocyte-tomonocyte ratio (cLMR) trend during radiotherapy (A); The median cNLR (B), cPLR (C), and cLMR (D) in concurrent chemoradiotherapy (CCRT) group and radiotherapy (RT) alone group trend during radiotherapy Circulating neutrophils contribute to tumor progression and invasiveness by secreting cytokines, vascular endothelial growth factor, and chemokines. [19][20][21] As a component of the immune complex, lymphocytes, which can inhibit the proliferation and metastasis of tumors, play a crucial role in antitumor immunity. 22 The NLR represents the balance between inflammatory and immune responses in peripheral blood. Previous studies have shown that a high pretreatment NLR is associated with a poor outcome in many solid tumors, 23-25 while a low post-treatment NLR can indicate a favorable prognosis. 26 Radiotherapy can kill tumor cells directly or indirectly and stimulate inflammatory responses at the same time. Inflammation in radiotherapy is caused by the clearance of dying cells, leading to the modulation of the tumor microenvironment, which is a double-edged sword in cancer treatment. [27][28][29] It can boost immunity within certain limits but also cause tumor recurrence, radiation resistance, and severe side effects within high degree. 27,30,31 Radiotherapy always has an effect on peripheral blood cells. Among leukocytopenias, lymphocytopenia is the most significant due to its high sensitivity to radiation, 32 Platelet count is another index of inflammation. Studies have shown that proinflammatory cytokines can facilitate the proliferation of megakaryocytes and increase platelet production. 34,35 Platelets can promote angiogenesis and tumor growth by producing cytokines, such as VEGF and transforming growth factor β. 4,36,37 The significance of the correlation between PLR and tumor prognosis remains controversial. Some studies have shown that pretreatment PLR is an independent prognostic factor for tumor outcome, 38,39 while others report the opposite findings. 40,41 Decreases in platelet count also appear during radiotherapy but are not as significant as lymphocytopenia. Similar to cNLR, cPLR can also be an indicator to assess variations in the inflammatory response and immunosuppression. In our study, cPLR was upregulated with increased radiation dose and reached its peak at week 5. This was consistent with the uptrend of inflammation during radiotherapy, and the value at the end of the treatment was under the influence of drug therapy and the decreased tumor load.
Elevated cPLR at weeks 4 and 5 was significantly associated with poor clinical outcome. These results reconfirm the close relation- Regarding prognosis, cPLR each week was a statistically significant variable in the univariate analysis but was rejected after the multivariate analysis, which meant that the predictive ability of cPLR was weaker than that of cNLR.
Monocytes also participate in tumor development. Monocytes can differentiate into tumor-associated macrophages (TAMs), which promote tumor progression by secreting growth factors and cytokines. 42,43 Previous studies have suggested that baseline LMR, another indicator of inflammation and immune status, is a prognostic factor in nasopharyngeal carcinoma, 44 gastric cancer, 45 and hepatocellular carcinoma. 46 However, some reports do not support this view. 47 According to the present study, LMR decreased weekly, and the minimum value appeared at week 5. This is in accordance with the dramatic reduction in lymphocytes during RT in ESCC patients. The extent of LMR change showed no connection with either clinical responses or OS. This may be due to the insufficiency of samples or the low absolute value of cLMR, which may affect the effectiveness of statistical test.
Chemotherapy has an impact on peripheral blood cell count and has shown some common patterns. As patients received chemotherapy at week 1, chemotherapeutic drug-induced agranulocytosis began to appear, which caused the patients in the CCRT group to have a lower cNLR, lower cPLR, and higher cLMR than those in the RT alone group.
During the intermittent period of chemotherapy, peripheral blood cells, especially granulocytes, began to recover gradually, so the gap narrowed. The two lines began to cross at week 4. This may be due to several reasons: (a) the removal of tumor cells by chemotherapy leads to a stronger inflammatory response; and (b) chemotherapy has a certain radiosensitization effect, leading to an increased inflammatory response. As to week 5, with the second cycle of chemotherapy, cNLR and cPLR became lower and cLMR became higher again in the CCRT group. This was consistent with the side effects of chemotherapeutic drugs. Although there was no significant difference between the cNLR, cPLR, and cLMR each week in the two groups except the cLMR at weeks 1 and 2, we still found a difference in the median. This may be due to the statistical deviation caused by the large difference in sample size between the two groups.
It is widely accepted that CCRT is one of the standard treat-

TA B L E 3 (Continued)
responses was based on barium radiography and CT but not PET/ CT or biopsy. In addition, although the process of patient selection was rigorous, we could not completely rule out all factors that might potentially impact peripheral blood cell count ratios. Third, we did not conduct a further analysis to confirm the cutoff value of these parameters, and progression-free survival was not taken into consideration because of the lack of accurate data. It is obvious that a large-sample and multicenter perspective study should be conducted in the future to further confirm these results.
In conclusion, we demonstrated that the cNLR at weeks 4 and 5 during RT had a significant negative correlation with the OS and clinical responses of non-surgical ESCC patients treated with radiotherapy, while an elevated cPLR at weeks 4 and 5 was only related to poor clinical response. These findings can be used as a basis for the dynamic evaluation of patient clinical responses and prognosis during radiotherapy through simple and repeatable biomarkers. Abbreviations: BMI, body mass index; CCRT, concurrent chemoradiotherapy; CF, conventional fractionated radiotherapy; RT, radiotherapy; SB, sequential boost; SIB, simultaneous integrated boost *Statistically significant.