Supplementary Material for: Key prognostic factors create a composite risk score to stratify patients into high- and low treatment benefit groups: A multicenter, retrospective data analysis of 84 mCRC patients treated with regorafenib as part of the CORRECT and CONSIGN trials

Introduction: In further-line mCRC treatment, median progression-free survival (PFS) is rather short, and many patients do not benefit from any anti-tumor treatment and should therefore treated according to best-supportive care (BSC). A risk score based on standard laboratory values using markers of tumor inflammation aims to define a patient cohort with high treatment benefit and might offer insights into tumor biology. As regorafenib has been dropped off the German market due to an unfavorable risk-benefit ratio, patient selection is key for any further-line treatment option. Methods: We used cox regression analysis to determine lab markers that are independent prognostic factors of OS and PFS outcome. The influence of these variables was weighted using an estimator, which was calculated using cox regression analysis. The estimators were implemented as multiplication factors, resulting in a risk score. A cut-off value for the resulting risk values was then determined via cox regression analysis resulting in a low- and high-risk subgroup. Results: Using data of 82 patients, a risk score identifying long-term survival in patients with last line mCRC treatment could be calculated. The following parameters were associated with significantly longer survival in multivariate analysis: NLR ≤ 5 (p = <0.001), AP ≤200 U/L (p = 0.001), CRP ≤3.2mg/dl (p = <0.001). The following estimator values were used to calculate a risk score: NLR: 0.132 (p=0.046), AP: 0.004 (p=0.014), and CRP: 0.032 (p=0.039). Implementing the estimators as multiplication factors yielded the following risk score: 0.132*NLR + 0.004*AP + 0.032*CRP = Risk value. Cox regression resulted in low- and high-risk subgroups with risk values below and above 1.4, respectively. In the group with a low-risk score (<1.4), patients had a median OS of 10.5 months after initiating regorafenib. Patients with a high-risk score (>1.4) survived only 3.3 months after starting therapy with regorafenib (n=43, p<0.001, HR=3.76). Discussion/Conclusions: The presented composite risk score stratifies patients into two prognostic subgroups characterized by standard laboratory values. Patients with signs of systemic characterized by elevated NLR, AP, and CRP have a high composite risk score and a significant shorter overall survival. Although this score needs to be prospectively validated in larger cohorts, it may be used to stratify patients suitable for further-line treatment studies.

引言：在后线转移性结直肠癌（metastatic colorectal cancer, mCRC）治疗中，中位无进展生存期（progression-free survival, PFS）普遍较短，且诸多患者无法从任何抗肿瘤治疗中获益，因此应按照最佳支持治疗（best-supportive care, BSC）方案进行管理。基于肿瘤炎症标志物的常规实验室指标构建的风险评分，旨在筛选出可从治疗中获益显著的患者队列，同时可为肿瘤生物学研究提供参考。鉴于瑞戈非尼（regorafenib）因不利的风险获益比已退出德国市场，患者筛选对于各类后线治疗方案而言至关重要。

方法：本研究采用Cox回归分析，筛选出与总生存期（Overall Survival, OS）和PFS结局独立相关的实验室标志物。通过Cox回归分析计算得到的估计量对上述变量的影响进行加权，将估计量作为相乘因子构建风险评分。随后通过Cox回归分析确定风险值的截断值，将患者划分为低风险亚组与高风险亚组。

结果：基于82例患者的数据，本研究成功构建了可识别后线mCRC治疗患者长期生存状态的风险评分。多因素分析显示，以下指标与更长的生存期显著相关：中性粒细胞与淋巴细胞比值（Neutrophil-to-Lymphocyte Ratio, NLR）≤5（p<0.001）、碱性磷酸酶（Alkaline Phosphatase, AP）≤200 U/L（p=0.001）、C反应蛋白（C-reactive Protein, CRP）≤3.2mg/dl（p<0.001）。用于构建风险评分的估计量取值如下：NLR：0.132（p=0.046），AP：0.004（p=0.014），CRP：0.032（p=0.039）。将估计量作为相乘因子后，风险评分计算公式为：0.132*NLR + 0.004*AP + 0.032*CRP = 风险值。经Cox回归分析，截断值为1.4，风险值低于1.4者归入低风险亚组，高于1.4者归入高风险亚组。在低风险评分组（<1.4）中，患者接受瑞戈非尼治疗后的中位OS为10.5个月；而高风险评分组（>1.4）患者在启动瑞戈非尼治疗后仅存活3.3个月（n=43，p<0.001，风险比Hazard Ratio, HR=3.76）。

讨论与结论：本研究提出的复合风险评分可基于常规实验室指标将患者划分为两个预后亚组。伴有NLR、AP及CRP升高的全身炎症表现的患者，其复合风险评分更高，总生存期显著更短。尽管该评分尚需在更大规模队列中进行前瞻性验证，但可用于筛选适合开展后线治疗研究的患者。