Nomogram based on SIS predicting the survival of non-surgical thoracic esophageal squamous cell carcinoma treated by IMRT

Background: Several inammatory markers have been reported to be associated with clinical outcomes in patients with esophageal squamous cell carcinoma (ESCC). This study was to evaluate several pre-radiotherapy serum inammatory indicators, including the neutrophil / lymphocytes ratio (NLR), platelet / lymphocyte (PLR), systemic immune-inammatory index (SII), systemic inammation score(SIS), and compare which one has the highest predicted survival value. Finally, combining inammatory markers with traditional prognostic factors, a new Nomogram model was developed to predict overall survival (OS) and progression-free survival (PFS) for ESCC patients receiving radiotherapy (RT) or chemoradiotherapy (CRT). Methods: This study retrospectively reviewed the data of 245 patients with thoracic esophageal squamous cell carcinoma (ESCC) underwent RT or CRT in the Fourth Hospital of Hebei Medical University from January 2013 to December 2015. The survival differences of these indexes were compared by the Kaplan-Meier method, and the univariate and the multivariate analyses were performed to determine these prognostic factors of overall survival (OS) and progression-free survival (PFS). Multivariate Cox proportional hazards regression models were used to create nomogram for OS and PFS. Results: 239 patients met the eligibility criteria. The estimated 1-, 3-, and 5-year OS and PFS rates were 74.6%, 36.8%, 26.5% and 58.4%, 31.3%, 20.5%, respectively, for the whole group. The difference in survival between OS and PFS was signicant when univariate analysis were applied based on these inammation-based measures. Multivariate analysis showed that tumor length, T stage, TNM stage, chemotherapy, SIS were predictive variables for OS and PFS in the multivariate model. The nomogram model established based on multivariate models of training data set had good predictive ability, the unadjusted C-index was 0.701 (95% CI, 0.662– 0.740) and 0.695 (95% CI, 0.656 - 0.734) for OS and PFS. Conclusions: This study show that SIS, as a comprehensive indicator of inammation and nutrition, had the strongest predictive power for evaluating prognosis. Moreover, our nomogram can accurately predict OS and PFS after treatment and may provide guidance regarding adjuvant therapy and surveillance. The same results were obtained in our study, the lower the SIS, the higher the survival. Multivariate analysis showed that SIS was an independent prognostic factor for OS and PFS with ESCC patients.In our study, multivariate analysis showed that SIS closely correlated with clinical characteristics and was an independent prognostic factor for OS and PFS in ESCC patients treated by RT or CRT. According to the area under the ROC curve, SIS was found to had higher predictive value than other indicators. The SIS AUC of OS and PFS higher than other inammatory factors were 0.731 and 0.755, respectively.


Background
Esophageal cancer is a relatively common malignant disease, with 455 800 newly diagnosed patients annually, and 400 200 annual deaths from this disease [1]. Esophageal cancer is histologically classi ed into esophageal squamous cell carcinoma (ESCC), esophageal adenocarcinoma (EAC) and other subtypes. Historically, ESCC has been the dominant histological subtype, accounts for approximately 90% of esophageal cancers in eastern Asia, including China and Japan , and there is a trend of increasing incidence in some Asian countries. In China, esophageal squamous cell carcinoma accounts has been over than 90% [2]. Because most diagnoses were in advanced stages and clinical progress was rapid, 40%~60% of patients lose the opportunity of surgery [3].
Due to lack of extensive prospective studies, it is di cult to accurately predict the prognosis of ESCC patients treated with the RT. Although several studies have reported that few parameters, such as the clinical symptoms and general state of the patients, tumour length, clinical staging, lymphatic/distant metastasis and squamous cell carcinomarelated antigen (SCC) levels, are related to the survival, they were not su ciently powerful for evaluating the outcome. Therefore, a better predictive index for the survival time of patients with ESCC were importance and essential.
Relevant studies have proved that immune in ammation and nutritional indicators are closely related to the prognosis of many malignant tumors( [4,5]. Potential prognostic values have been shown by changes in the levels of neutrophil, lymphocyte, platelet and monocyte. Meanwhile, various novel composite indicators have been calculated based on these peripheral blood parameters, including neutrophil / lymphocyte ratio (NLR), platelet / lymphocyte (PLR), lymphocyte / monocyte (LMR) and the composite index, for example, systemic immune-in ammatory index (SII) , consisting neutrophil multiplied by the platelet divided by the lymphocyte, in some neoplasms [6]. In addition, There are several methods of assessing nutritional status in cancer patient of which serum albumin is one of the most commonly used [7]. However, there is no generally accepted optimal threshold for nutritional and in ammatory markers, and no combined in ammatory and nutritional indicators scoring system to incorporate these indicators in ESCC.
This study aims to evaluate the prognostic value of serum-based in ammatory and nutritional biomarkers for nonoperative esophageal squamous cell carcinoma (ESCC), and to establish a comprehensive and innovative nomogram model for evaluating the prognosis of ESCC.

Patients
We retrospectively collected and analyzed pre-radiotherapy clinical and laboratory test data for 245 consecutive patients with thoracic ESCC who underwent radical RT or CRT at the Fourth Hospital of Hebei Medical University Hebei China between January 2013 and December 2015. More speci cally, the inclusion criteria were as follows: (1)Pathologically con rmed ESCC before RT, (2) Karnofsky Performance Score ≥70 points, (4)patients refused surgery or did not undergo surgery for other reasons, (5)no history of malignant disease, and active double cancer at the time of diagnosis of esophageal cancer, (6)complete clinical and follow-up information was available.
The exclusion criteria were (1) active hemorrhage or severe coagulation dysfunction, (2) severe uncontrolled hypertension, (3) severe cardiopulmonary diseases or abnormal liver and kidney function, (4) death from other diseases during follow-up, (5) incomplete data and (6) lost to follow-up.

Data collection
The limitation of the seventh edition of the 2009 AJCC TNM staging criteria is that it is applicable only to patients with surgical treatment alone, but not for patients with non-surgical. n this study, the tumor staging before radiotherapy was classi ed according to the Clinical Classi cation of Esophageal Carcinoma Treated by Non-surgical Methods, the clinical prognostic value and practicability of which has been veri ed [8,9].
All serum parameters were collected with in 7 days before RT, and including neutrophil, lymphocyte, monocyte, platelet counts, and albumin levels. NLR, LMR,PLR, and SII were graded by calculating the optimal cut-off values based on receiver operating characteristic (ROC) curves. NLR, PLR, LMR, SII, and SIS were calculated as follows NLR= neutrophil count/lymphocyte count, PLR=platelet count/lymphocyte count, LMR=lymphocyte count/ monocyte count, SII= neutrophil count*platelet count/lymphocyte count. SIS decreased serum albumin and decreased LMR (≤40 g/L and ≤4.15, respectively) were assigned SIS 2, patients with either decreased serum albumin or decreased LMR were assigned SIS 1and patients with both elevated serum albumin and elevated LMR(>40 g/L and >4.15, respectively) were assigned SIS 0. All the patients signed informed consent forms, and all protocols in this study were approved by the ethics committee of the Fourth Hospital of Hebei Medical University.

Follow-up
According to our computer database patients were recorded and calculated to the date of death due to any cause. Patients who survived were examined in our survival analyses. Medical history, physical examination, laboratory examination, electrocardiogram, abdominal ultrasound, esophageal barium meal, and chest CT were tested every 3 months in the rst two years and then every 6 months. Those who were lost to follow-up were cut off at the last follow-up time. Adverse events were assessed according to the National Cancer Institute's Common Terminology Criteria for Adverse Events (CTCAE v4.0).

Radiotherapy
Radiotherapy was performed according to standards published by the National Comprehensive Cancer Network (NCCN). In this study, 239 patients with ESCC received IMRT(95% PTV / 50-66Gy / 25-33F, 1.8-2.0Gy / F, 5F / W). Imaging data from the computed tomography (CT) simulation scan were transmitted to the RT treatment planning system (Pinnacle, version 9.2, Philips Radiation Oncology System, USA), for delineation of the primary tumor area, enlargement lymph node and the organs at risk. The gross tumor volume (GTV), clinical tumor volume (CTV), and planned tumor volume (PTV) were outlined. The gross tumor volume (GTV) included esophagus with the thickened wall as shown by CT, esophageal lm, and esophagoscopy and swollen lymph nodes with a short diameter ≥1 cm.
The clinical target volume (CTV) was 0.5 cm to 0.8 cm extended from GTV to the front, back, left, and right, and 2.5 cm to 3.0 cm extended on the top and bottom, plus the corresponding lymphatic drainage area. The planning target volume (PTV) was 0.6 cm extended from CTV on all sides. The treatment plan included the prescribed dose of 50-66Gy / 25-33F, 1.8-2.0Gy / F, 5F / Week, and the median dose of 61.2 Gy. The total lungs V20 ≤ 30%, V30 ≤ 20%, V5 ≤ 60% and the lung mean dose ≤ 16Gy. The heart dose V25 ≤ 50%, V40 ≤ 30% and the maximum Spinal Cord dose <45 Gy. The irradiated elds of esophageal lymph drainage were de ned as CTV1. The scope of CTV1 varies according to the lesion location. The PTV1 was based on CTV1 Expand uniformly from 0.5 to 0.8 cm to the periphery. The required 95% PTV1 prescription dose was 46 to 54 Gy, the 95% PTV and 95% PTV-nd prescription dose were 50 to 66 Gy, 1.80 to 2.06 Gy per fraction, and 5 fractions per week.

Chemotherapy
According to NCCN for the concurrent chemoradiotherapy of Esophageal Cancer, three standard chemotherapy regimens were used in some patients: The rst was cisplatin plus 5-uorouracil (5-FU). Two cycles of cisplatin(75-100 mg/m2) on day 1 and 5-FU (750-1000 mg/m²) on day 1-4, at an interval of 4 weeks were performed. As for maintenance chemotherapy, from approximately 4 weeks after CRT, two cycles of cisplatin (75 mg/m2)on day 1 and 5-FU (750-1000 mg/m²) on day 1-4, One treatment cycle was 21 days, and the patients received 2 chemotherapy treatment cycles.
The second was docetaxel plus carboplatin. Docetaxel (7.5 mg/m2) on day 1 and carboplatin AUC=2 on day 1, One treatment cycle was 7 days, and the patients received 4-6 chemotherapy treatment cycles.
The third was docetaxel plus cisplatin. The dosage and administration schedule of docetaxel (7.5 mg/m2) intravenously on day 1 and continuous infusion of cisplatin (75 mg/m²) on day 1. The chemotherapy treatment cycles were the same as for docetaxel plus carboplatin.

Statistical Analysis
The statistical analysis using Graphpad prism eversion 8.0 and R version 4.0.3 (R Foundation for Statistical Computing). Continuous variables were compared using the unpaired t test or Mann-Whitney nonparametric test. The chi-squared test and Fisher's exact test were used to compare the categorical variables. Survival analysis was conducted using Kaplan-Meier method and log-rank test. Multivariate analysis was performed using the Cox regression model for variables found to be signi cant among the univariate analysis, and the corresponding 95% con dence intervals (CIs) were calculated, a two-tailed P-value < 0.05 was considered statistically signi cant.
Receiver operating characteristic (ROC) curves were calculated to determine optimal cutoff values and to assess the predictive power of these in ammation-based indexes for long-term survival. Nomogram for 1 -, 3 -and 5-year OS and PFS probabilities were constructed based on the multivariate Cox proportional hazards regression models. The discriminatory performance of the OS and PFS nomogram was evaluated by calculating Harrell's concordance index (C-index). In addition, we assessed calibration of the nomogram to compare the nomogram estimated risk to the observed risk. Calibration of the nomogram for 1-, 3-and 5-years OS and PFS were illustrated in calibration plots.

Results
Demographic and Clinical Characteristics 239 patients met the eligibility criteria, 4 patients were excluded due to incomplete peripheral blood indicators, and 2 were excluded due to loss of follow-up. The survival sample numbers of 1, 3 and 5 years were 176, 83 and 62, respectively. The baseline characteristics of the patients were summarized in Table1 CRT : Chemoradiotherapy, IFI : Involved-field irradiation,ENI:Elective nodal irradiation, RT:Radiotherapy, NLR : Neutrophil-to-lymphocyte ratio, LMR: Lymphocyte-to-monocyte ratio, PLR : Platelet-to-lymphocyte, SII : Systemic immune-inflammatory index, SIS: Systemic inflammation score. CRT : Chemoradiotherapy, RT:Radiotherapy.

Predictive value of in ammatory indexes
The ROC curve analysis was performed to estimate predictive powerful of the in ammation based indexes for both OS and PFS in ESCC patients. The ROC curves for OS and PFS prediction were calculated with patient's 5 years of follow-up. More speci cally, results showed the AUROC values of SIS were consistently higher than other in ammatory index with area under curve (AUC) of 0.731for OS and (AUC) of 0.755 for PFS. Subsequently, A comparative study revealed that patients with higher SIS occurred unfavorable outcome more frequently, and they had statistical signi cance among different grades of SIS in 1,3,5 OS and PFS (Fig. 3) (Table 5).  (Figure 4,5). The total scores of these prognostic factors can be used to determine the probabilities of 1-3-and 5-years OS and PFS. The performance of the nomogram were evaluated using the C-index and calibration curves. The C-index for OS prediction was 0.701 (95% CI: 0.662-0.740) and PFS prediction was 0.695 (95% CI:0.656-0.734). The calibration graph showed properly probability consistencies between standard curve and nomogram prediction. The calibration curves also had good prediction consistency (Figures 6).

Discussion
Recent studies showed that several in ammation factors including NLR, PLR, LMR and SII, etc. were associated with long time prognosis survival for esophageal squamous cell carcinoma patients [10,11,12]. Although most studies have demonstrated the prognostic relationship between related in ammatory markers in esophageal squamous cell carcinoma, no comparison of prognostic value has been conducted, nor has a comprehensive prognostic model been developed. Therefore In ammatory markers were assessed and rigorous statistical methods were required. Besides, the nomogram model based on traditional prognostic indicators and SIS considered as a trustworthy method to generate more accurate prediction of prognosis.
In this study, We investigated clinical characteristics and prognosis of 239 ESCC patients treated by RT or CRT with periphery blood markers of systemic in ammation and nutrition parameters. We systematically investigated clinical characteristics and demonstrated the prognostic signi cance of pre-radiotherapy in ammatory parameters, including NLR, PLR, SII and the composite index SIS based on the level of lymphocyte-to-monocyte ratio (LMR) and serum albumin (Alb) for predicting survival status. Studies have found that SIS has a high prognostic value in patients with ESCC and is a excellent supplement to the traditional prognostic indicators.
Our studies have shown that patients with later lesion, such as tumor length >6.2cm, TNM stage III showed a signi cant increase in SIS1-2. On the contrary, the number of SIS0-1 in patients with tumor length ≤ 6.2cm and early TNM stage is more. This indicates that with the progression of the disease, the immune in ammatory response of the patients is gradually aggravated. In addition, albumin levels are reduced in patients with late stage due to malnutrition, anorexia or abnormal liver function. So people with high SIS scores tend to have a poor prognosis. In terms of treatment methods, patients with SIS0-1 preferred combined chemotherapy, ENI and prescription RT dose > 61.2, while SIS2 preferred RT alone, IFI and RT dose ≤ 61.2. This suggests that patients with a lower SIS score who can tolerate more thorough treatment tend to have a better prognosis.
The prognostic value of SIS in predicting malignant tumors has been reported. Y chang, et al. establish the SIS de ned as follows: patients with both decreased serum albumin and decreased LMR (<40 g/L and <4.44, respectively) were assigned SIS 2, patients with either decreased serum albumin or decreased LMR were assigned SIS 1and patients with both elevated serum albumin and elevated LMR(≥40 g/L 1and ≥4.44, respectively) were assigned SIS 0. Their nal results showed that SIS was an independent prognostic factors of OS in renal cell carcinoma. High SIS was associated with shorter OS [13]. Shoichi Inokuchi, et al. Using ROC curves showed SIS is a good predictor of OS and RFS for patients with hepatocellular carcinoma.
The same results were obtained in our study, the lower the SIS, the higher the survival. Multivariate analysis showed that SIS was an independent prognostic factor for OS and PFS with ESCC patients.In our study, multivariate analysis showed that SIS closely correlated with clinical characteristics and was an independent prognostic factor for OS and PFS in ESCC patients treated by RT or CRT. According to the area under the ROC curve, SIS was found to had higher predictive value than other indicators. The SIS AUC of OS and PFS higher than other in ammatory factors were 0.731 and 0.755, respectively.
However, though the prognostic value of SIS has been shown for predicting the survival of patients with ESCC, the mechanism is still unclear. A lower of SIS indicates low level of LMR and/or albumin, which might increased nonspeci c in ammation, and immune system dysfunction, and malnutrition.
For one thing, the probable reason for low level of LMR is that increased monocytes and/or reduced lymphocytes.
Monocytes are known to promote tumorigenesis and angiogenesis through local immune suppression and stimulation of tumor neovasculogenesis [14]. Moreover, tumor-associated macrophages developing from mono nuclear cell lineages have been demonstrated to be able to inhibit cancer progression and spread of tumors [15]. This could explain why an elevated monocyte counts to confer poor clinical outcomes of various types of cancers. Noriyuki,H, et al. reported LMR was associated with cancer-speci c survival (CSS) of esophageal cancer patients after curative esophagectomy. In particular, a low LMR was an independent predictor of poor survival in non-elderly patients [16].
There is increasing evidence that lymphocytes are essential for antitumor immune reactions owing to several mechanisms, including the ability to enhance tumor cell apoptosis, inhibition of tumor cell proliferation, and promotion of metastasis [17]. Davuluri, Rajayogesh et al. demonstrated that Lymphocyte reduction during CRT for EC was associated with poor outcomes, suggesting a role of host immunity in disease control [18]. The same Zhang, Jian. also con rmed that lymphatic invasion, is an independent favorable prognostic factor of DFS in patients with NSCLC who underwent lobectomy and lymph node dissection and adjuvant chemotherapy [19]. These may explain the reason that low level of LMR may serve as an independent prognostic factor for OS and PFS.
For another, serum albumin is generally used to assess the nutritional status, severity of disease, disease progression and as an independent prognosticator of survival in various cancers [20]. Malnutrition is closely related to the imbalance of tumor microenvironment. It weaken patient's defense mechanisms such as humoral immunity, cellular immunity and phagocytosis and resulting in increased risk of infection and poor e cacy of anti-tumor treatment [21]. Because nutrition is an important determinant of immune response, decreased serum albumin re ects both malnutrition and sustained systemic in ammation response [22]. Thus, SIS, which is based on LMR and serum albumin level, may predict prognostic as assessments for cancer. This observation provides a rationale to prospectively test chemotherapeutic and radiation treatment strategies that may have a lower impact on host immunity.
Finally, we endued scores based on the weight of each predictor and established a nomogram model [23,24], aimed to better visualize the risk point and predictive power of SIS. Statistical analysis showed that this model had a good tness of data and accuracy in predicting functional outcome. For example, one patient had tumor length=5cm, TNM stage (II), combined chemotherapy, and SIS 2. For this example, the OS nomogram model total point equals 164, and the 1-, 3-, 5-years OS is approximately 81%, 42% and 29%. The PFS nomogram model point equals 162, and the 1-, 3-, 5-years PFS is 59%, 24% and 18%. This nomogram model show that the SIS can provide additional prognostic information to traditional prognostic factors and will be helpful assist clinicians to better optimized care and stratify patients of health care resources to improve ESCC patients prognosis. Meanwhile, it including the conventional prognostic factors and SIS can further improve the predictive ability, and helps to identify OS and PFS rate early. In other words, SIS had an high predictive value for ESCC patients,independent of conventional risk factors.
The advantage of this study is that SIS is superior to NLR, PLR and SII, and nomogram based on SIS could be a new prognostic scoring system model contributes to the risk assessment for patients with ESCC. What's more, the SIS measurements are easy to obtain and low cost because they are based on standard laboratory tests for serum albumin, lymphocyte, and monocyte counts used routinely in clinical practice.
Nonetheless there are a few limitations in present study. Firstly, this study as a retrospective study, a single-center design with the number of patients enrolled in the model was relatively small and no external validation. Secondly, this study did not further compare the dynamic change of peripheral blood in ammatory markers during and after RT. Therefore, a prospective veri cation trials with a great quantity of cases are needed to con rm our present ndings.

Conclusion
In conclusion, the predictive power of several in ammation-based prognostic values were assessed and compared in patients with ESCC. The SIS were independently associated with OS and PFS and the nomogram model based on SIS had more favorable predictive ability. Patients with SIS 0 have better OS and PFS, and those with SIS 1 and SIS 2 should receive more better monitoring and more rigorous treatment to avoid tumor progression. However, further studies are necessary to validate this ndings and explore the mechanisms, to provide evidence for individualized treatment of ESCC patients.