Adapted systemic inflammation score as a novel prognostic marker for esophageal squamous cell carcinoma patients

Abstract Background The adapted systemic inflammation score (aSIS), calculated from serum albumin and the lymphocyte‐to‐monocyte ratio, has been reported to be a novel prognostic marker for some types of cancers. However, the prognostic impact of aSIS in patients with esophageal squamous cell carcinoma (ESCC) remains controversial. This study aimed to examine the prognostic effects of aSIS in a large cohort of 509 ESCC patients. Methods Preoperative aSIS was retrospectively calculated for 509 ESCC patients who underwent curative resection. Time‐dependent receiver operating characteristics (t‐ROC) curves were used for comparing the prognostic impact. Results Patients with high aSIS showed significantly poorer overall survival (OS) than patients with low aSIS (log rank P < .001). The multivariate analysis revealed that aSIS was an independent prognostic factor for overall survival (multivariate hazard ratio 1.76; 95% confidence interval 1.13–2.75; P = .013). The t‐ROC analysis showed that aSIS was more sensitive than other nutritional prognostic factors (controlling for nutritional status, systemic inflammation score, and the neutrophil‐to‐lymphocyte ratio). Conclusion Preoperative aSIS may be a useful prognostic biomarker in ESCC patients who underwent curative resection.

The systemic inflammation score (SIS) was developed as a new scoring system based on serum albumin (Alb) and the lymphocyte-tomonocyte ratio (LMR). SIS has been reported as a powerful prognostic marker for clear-cell renal cell carcinoma and colorectal cancer. 2,3 The cutoff value of LMR in SIS was the median value for patients with renal cell carcinoma. Therefore, Lin et al developed a modified SIS (mSIS) with optimized LMR cutoff values for gastric cancer patients using the software X-tile (Yale University, New Haven, CT). 4 The mSIS has been shown to be superior to SIS as a predictive indicator for gastric cancer. 5 Several studies have examined the prognostic role of SIS and mSIS in esophageal squamous cell carcinoma (ESCC). [6][7][8] However, previous data on SIS, mSIS, and clinical outcomes for ESCC have been inconclusive because previous studies were limited by small sample sizes (n < 443). Considering the importance of a potential prognostic marker for ESCC, the assessment of SIS and clinical outcome using a large number of ESCC cases is needed.
In this study we reoptimized the LMR cutoff values as 3.

| The aSIS and other scoring systems
Serum samples were obtained and analyzed within 7 d before esophagectomy. The aSIS was calculated from serum albumin and peripheral LMR (Figure 1). The LMR cutoff value was calculated using the software X-tile, 4 which determined the LMR cutoff value as 3.3 ( Figure S1).

F I G U R E 1
Flow chart of the analyzed cases and the definition of adapted systemic inflammation score NLR and CONUT scores were calculated as described previously. [11][12][13]

| Statistical analyses
All statistical analyses were performed using JMP v. 10

| Correlations between the aSIS and short-term outcomes of surgery
The short-term outcomes after surgery according to the aSIS are presented in Table 2. High aSIS was significantly associated with a higher incidence of pulmonary morbidity (P < .01). Other complications and complication severity were not associated with aSIS.
We checked the correlation between aSIS and patients' characteristics affecting pulmonary morbidity such as a respiratory function or surgical procedure. Regarding a respiratory function, we ana-

| Adapted systemic inflammation score and patient survival
The impact of aSIS on clinical outcomes was assessed in ESCC patients. There were 168 deaths among the 509 patients, including 96 esophageal cancer-specific deaths. The median follow-up time for censored patients was 4.0 y. The Kaplan-Meier analysis showed that there was a significant difference in the OS and CSS among the three groups (both log rank P < .01). The 3-y OS rates in groups 0, 1, and 2 were 84.1%, 74.6%, and 57.3%, respectively ( Figure 2). The 3-y CSS rates in groups 0, 1, and 2 were 87.5%, 81.3%, and 72.0%, respectively ( Figure 2).

| Survival analysis of interactions between aSIS and other factors
We next determined whether the influence of aSIS on the OS was affected by any clinicopathological factors. The effect of aSIS was not significantly modified by comorbidity, BMI, Brinkman Index, sex, or age (P > .11 for all interactions) ( Figure 3A). Interestingly, a potential modifying effect of preoperative treatment on the relationship between the aSIS and OS was observed (P for interaction = .055). Among patients who did not receive preoperative treatment, the high mSIS had a significantly shorter OS (log rank P < .001). In contrast, among patients who received preoperative treatment, there was no significant difference in the OS for the high aSIS and low aSIS (log rank P = .676) ( Figure 3B). Similarly, a potential modifying effect of cStage on the relationship between the mSIS and OS was also observed (P for interaction = .030). Among patients with cStage I, II, the high mSIS had a significantly shorter OS (log rank P < .001) ( Figure S3).

TA B L E 2 Short-term outcome of surgery
Furthermore, we divided the patients into those who received preoperative treatment and those who did not, and further analyzed the patients' characteristics and prognostic factors.
In patients without preoperative treatment, the relationships between aSIS and clinicopathological factors are presented in Table S1.
Older age, poorer PS, lower BMI, and higher clinical stage were significantly associated with higher aSIS (each P < .05). High aSIS was significantly associated with a higher incidence of pulmonary morbidity (P < .01). In the multivariate Cox regression analysis for OS, aSIS was an independent prognostic factor for the OS (multivariate HR 2.27, 95% CI 1.18-4.19, P = .02; Table S3). These results for patients who did not receive preoperative treatment were in good agreement with the analysis results of all 509 patients.
On the other hand, in patients with preoperative treatment, the relationships between aSIS and clinicopathological factors are presented in Table S2. Older age, and smoker were significantly associated with higher aSIS (each P < .05). Unlike patients without preoperative treatment, high aSIS was not significantly associated with a higher incidence of pulmonary morbidity (P = .62). Furthermore, in the univariate and multivariate Cox regression analysis for OS, aSIS was not an independent prognostic factor for the OS (univariate HR 1.23, 95% CI 0.68-2.26, P = .50; Table S4).

| Comparison of aSIS with other prognostic scoring systems
Time-dependent area under the curve (AUC)-of-ROC curves of aSIS and other prognostic scoring systems (CONUT, SIS, and NLR) was constructed for the prediction of OS (Figure 4). When the predictive performance for the OS was compared after 1-5 y of follow-up, the AUC values of the aSIS were similar to those of the SIS and CONUT, and were significantly higher than those of the NLR.

| D ISCUSS I ON
In this study we examined the prognostic effects of aSIS in a large cohort of 509 patients with ESCC. We found that high aSIS was associated with poor prognosis, suggesting that aSIS status could be used as a marker to identify patients who are likely to have unfavora-   Two previous studies on ESCC showed that a high SIS was significantly associated with poorer OS. 6,7 There was only one study focusing on the prognostic role of mSIS in ESCC patients. 8  Peripheral monocyte count appeared to be useful prognostic marker in a large cohort of patients with ESCC. 18 However, the precise molecular mechanism by which monocytes might predict clinical survival is not understood. One hypothesized pathogenic mechanism is that monocytes are recruited by some cytokines and chemokines around the tumor and differentiate into tumor-associated macrophages (TAMs), which facilitate multiple tumorigenic mechanisms.
Type 2 macrophages, in particular, have a tumorigenic effect that includes tumor initiation, invasion, angiogenesis, and metastasis. 19 We have previously reported that the presence of a high-density of This study has several limitations. First, this was a retrospective study. Second, the patients were included from just our institution.
The significance of aSIS needs to be validated using other cohorts.
In conclusion, our large cohort study with more than 500 cases revealed that the preoperative aSIS was associated with the clinical outcome in ESCC patients who underwent curative resection.
Furthermore, aSIS was more sensitive than other nutritional prognostic factors. Considering the relationship between nutritional status and local tumor immunity, our findings may have considerable

ACK N OWLED G M ENT
We thank Editage (www.edita ge.com) for English language editing.

CO N FLI C T O F I NTE R E S T
No conflicts of interest exist.

E TH I C A L S TATEM ENT
The study was approved by the Institutional Review Board of Kumamoto University (#1909).