Department of Radiation Oncology, Weifang People's Hospital, Weifang, China.
Clinical School, Weifang Medical University, Weifang, China.
Front Public Health. 2022 Sep 9;10:993443. doi: 10.3389/fpubh.2022.993443. eCollection 2022.
Acute hematologic toxicity (HT) is a common complication during radiotherapy of cervical cancer which may lead to treatment delay or interruption. Despite the use of intensity-modulated radiation therapy (IMRT) with the pelvic bone marrow (PBM) sparing, some patients still suffer from acute HT. We aimed to identify predictors associated with HT and develop a nomogram for predicting grade 2 or higher (G2+) acute HT in cervical cancer following the PBM sparing strategy.
This study retrospectively analyzed 125 patients with cervical cancer who underwent IMRT with the PBM sparing strategy at our institution. Univariate and multivariate logistic regression, best subset regression, and least absolute shrinkage and selection operator (LASSO) regression, respectively, were used for predictor screening, and Akaike information criterion (AIC) was used to determine the best model for developing the nomogram. Finally, we quantified the risk of G2+ acute HT based on this model to establish a risk stratification.
The independent predictors used to develop the nomogram were histological grade, pre-radiotherapy chemotherapy, pre-radiotherapy HT, and radiotherapy [IMRT alone vs. concurrent chemoradiotherapy (CCRT)] which were determined by the univariate and multivariate logistic regression with the minimum AIC of 125.49. Meanwhile, the heat map showed that there is no multicollinearity among the predictors. The nomogram was well-calibrated to reality, with a Brier score of 0.15. The AUC value was 0.82, and the median Brier score and AUC in 1000 five-fold cross-validation were 0.16 and 0.80, respectively. The web version developed together was very easy to use. The risk stratification indicated that high-risk patients (risk point > 195.67) were more likely to develop G2+ acute HT [odds ratio (OR) = 2.17, 95% confidence interval (CI): 1.30-3.05].
This nomogram well-predicted the risk of G2+ acute HT during IMRT in cervical cancer after the PBM sparing strategy, and the constructed risk stratification could assist physicians in screening high-risk patients and provide a useful reference for future prevention and treatment strategies for acute HT.
急性血液学毒性（HT）是宫颈癌放疗过程中的常见并发症，可能导致治疗延迟或中断。尽管采用了骨盆骨髓（PBM）保护的调强放疗（IMRT），但仍有部分患者出现急性 HT。我们旨在确定与 HT 相关的预测因素，并为 PBM 保护策略后宫颈癌的 2 级或更高（G2+）急性 HT 开发预测模型。
本研究回顾性分析了 125 例在我院接受 PBM 保护策略 IMRT 的宫颈癌患者。分别采用单因素和多因素逻辑回归、最佳子集回归和最小绝对收缩和选择算子（LASSO）回归进行预测因素筛选，Akaike 信息准则（AIC）用于确定最佳模型以开发预测模型。最后，我们根据该模型量化 G2+急性 HT 的风险，建立风险分层。
通过单因素和多因素逻辑回归，确定了用于建立预测模型的独立预测因素为组织学分级、放疗前化疗、放疗前 HT 和放疗方式（单纯 IMRT 与同期放化疗（CCRT）），AIC 最小为 125.49。同时，热图显示预测因素之间不存在多重共线性。该预测模型与实际情况具有良好的校准度，Brier 评分为 0.15。AUC 值为 0.82，1000 次五重交叉验证的中位数 Brier 评分和 AUC 分别为 0.16 和 0.80。共同开发的网络版使用非常方便。风险分层表明，高危患者（风险点>195.67）更有可能发生 G2+急性 HT[比值比（OR）=2.17，95%置信区间（CI）：1.30-3.05]。
该预测模型很好地预测了 PBM 保护策略后宫颈癌 IMRT 期间 G2+急性 HT 的风险，构建的风险分层有助于医生筛选高危患者，并为未来急性 HT 的预防和治疗策略提供有用参考。
