Department of Radiotherapy, University Medical Center Utrecht, Utrecht, The Netherlands.
Department of Medical Oncology, University Medical Center Utrecht, Utrecht, The Netherlands.
Int J Radiat Oncol Biol Phys. 2014 Nov 15;90(4):739-47. doi: 10.1016/j.ijrobp.2014.07.051. Epub 2014 Sep 24.
Patients with bone metastases have a widely varying survival. A reliable estimation of survival is needed for appropriate treatment strategies. Our goal was to assess the value of simple prognostic factors, namely, patient and tumor characteristics, Karnofsky performance status (KPS), and patient-reported scores of pain and quality of life, to predict survival in patients with painful bone metastases.
In the Dutch Bone Metastasis Study, 1157 patients were treated with radiation therapy for painful bone metastases. At randomization, physicians determined the KPS; patients rated general health on a visual analogue scale (VAS-gh), valuation of life on a verbal rating scale (VRS-vl) and pain intensity. To assess the predictive value of the variables, we used multivariate Cox proportional hazard analyses and C-statistics for discriminative value. Of the final model, calibration was assessed. External validation was performed on a dataset of 934 patients who were treated with radiation therapy for vertebral metastases.
Patients had mainly breast (39%), prostate (23%), or lung cancer (25%). After a maximum of 142 weeks' follow-up, 74% of patients had died. The best predictive model included sex, primary tumor, visceral metastases, KPS, VAS-gh, and VRS-vl (C-statistic = 0.72, 95% CI = 0.70-0.74). A reduced model, with only KPS and primary tumor, showed comparable discriminative capacity (C-statistic = 0.71, 95% CI = 0.69-0.72). External validation showed a C-statistic of 0.72 (95% CI = 0.70-0.73). Calibration of the derivation and the validation dataset showed underestimation of survival.
In predicting survival in patients with painful bone metastases, KPS combined with primary tumor was comparable to a more complex model. Considering the amount of variables in complex models and the additional burden on patients, the simple model is preferred for daily use. In addition, a risk table for survival is provided.
患有骨转移的患者的生存时间差异很大。为了制定适当的治疗策略，需要对生存时间进行可靠的评估。我们的目的是评估简单的预后因素，即患者和肿瘤特征、卡诺夫斯基表现状态（KPS）以及患者自评的疼痛和生活质量评分，在预测有骨转移疼痛的患者的生存时间方面的价值。
在荷兰骨转移研究中，1157 例患者接受放射治疗以缓解骨转移疼痛。在随机分组时，医生确定 KPS；患者使用视觉模拟量表（VAS-gh）评估一般健康状况，使用口头评分量表（VRS-vl）评估生命价值和疼痛强度。为了评估变量的预测价值，我们使用多变量 Cox 比例风险分析和 C 统计量进行区分能力评估。对最终模型进行校准评估。在 934 例接受放射治疗治疗椎体转移的患者数据集中进行外部验证。
患者主要患有乳腺癌（39%）、前列腺癌（23%）或肺癌（25%）。在最长 142 周的随访后，74%的患者死亡。最佳预测模型包括性别、原发肿瘤、内脏转移、KPS、VAS-gh 和 VRS-vl（C 统计量=0.72，95%CI=0.70-0.74）。一个包含 KPS 和原发肿瘤的简化模型显示出相似的区分能力（C 统计量=0.71，95%CI=0.69-0.72）。外部验证显示 C 统计量为 0.72（95%CI=0.70-0.73）。推导和验证数据集的校准显示生存时间被低估。
在预测有骨转移疼痛的患者的生存时间方面，KPS 联合原发肿瘤与更复杂的模型相当。考虑到复杂模型中变量的数量以及对患者的额外负担，简单模型更适合日常使用。此外，还提供了生存风险表。
