Background/Objectives: There is a need for prediction models which enable weighing benefits against risks of different treatment strategies for individual Hodgkin lymphoma (HL) patients. Therefore, we aimed to predict absolute risk of progression, first relapse or death (PRD) with and without incorporating HL treatment.Methods: The prognostic and treatment information of 2343 patients treated for classical HL at ages 15–60 years between 2008 and 2018 in the Netherlands was used to predict absolute risk of PRD up to 5 years after diagnosis using Cox proportional hazard models allowing for time-varying coefficients. Models were externally validated in 1675 patients treated for classical HL in Denmark between 2000 and 2018.Results:In early stages, gender, leukocyte, and lymphocyte counts were associated with risk of PRD. Additionally, receiving >4 cycles of ABVD (doxorubicin, bleomycin, vinblastine, and dacarbazine) or ABVD plus radiotherapy predicted lower risk of relapse compared with receiving ≤4 cycles of ABVD. In advanced stages, age, albumin and leukocyte counts predicted PRD risk. Receiving (escalated) BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone) predicted lower PRD risk compared to ABVD. In Danish patients treated between 2008 and 2018, adding treatment information improved 5-year Inverse Probability of Censoring Weighted (IPCW) Area Under the Curve (AUC) values from 0.63 (95% Confidence Interval (CI): 0.55–0.72) to 0.71 (95% CI: 0.63–0.79) in early stages (p-value = 0.04) and from 0.59 (95% CI: 0.52–0.65) to 0.62 (95% CI: 0.55–0.68) in advanced stages (p-value = 0.33).Conclusions: We developed well calibrated models with reasonable discrimination, not only incorporating pre-treatment prognostic factors but also treatment effect enabling the prediction of absolute risk of first relapse/progression.
背景/目的：有必要建立预测模型，以权衡霍奇金淋巴瘤（HL）患者个体不同治疗策略的获益与风险。因此，本研究旨在预测纳入与未纳入HL治疗方案的疾病进展、首次复发或死亡（PRD）绝对风险。方法：利用2008年至2018年间荷兰2343例15-60岁经典型HL患者的预后及治疗信息，采用允许时变系数的Cox比例风险模型预测诊断后5年内PRD绝对风险。模型在2000年至2018年间丹麦1675例经典型HL患者中进行了外部验证。结果：在早期患者中，性别、白细胞及淋巴细胞计数与PRD风险相关。此外，与接受≤4周期ABVD方案（多柔比星、博来霉素、长春碱、达卡巴嗪）相比，接受>4周期ABVD或ABVD联合放疗可预测更低的复发风险。在晚期患者中，年龄、白蛋白及白细胞计数可预测PRD风险。与ABVD方案相比，接受（强化）BEACOPP方案（博来霉素、依托泊苷、多柔比星、环磷酰胺、长春新碱、丙卡巴肼、泼尼松）可预测更低的PRD风险。在2008-2018年治疗的丹麦患者中，纳入治疗信息使早期患者5年逆概率删失加权曲线下面积从0.63（95%置信区间：0.55-0.72）提升至0.71（95%置信区间：0.63-0.79）（p值=0.04），晚期患者从0.59（95%置信区间：0.52-0.65）提升至0.62（95%置信区间：0.55-0.68）（p值=0.33）。结论：我们建立了校准良好且具有合理区分度的预测模型，不仅纳入治疗前预后因素，同时整合了治疗效果，能够有效预测首次复发/进展的绝对风险。
肿瘤(癌症)患者之家