Background

For men on active surveillance for prostate cancer, utility of prostate-specific antigen (PSA) kinetics (PSAk) in predicting pathologic reclassification remains controversial.

Objective

To develop prediction methods for utilizing serial PSA and evaluate frequency of collection.

Design, setting, and participants

Data were collected from men enrolled in the multicenter Canary Prostate Active Surveillance Study, for whom PSA data were measured and biopsies performed on prespecified schedules. We developed a PSAk parameter based on a linear mixed-effect model (LMEM) that accounted for serial PSA levels.

Outcome measurements and statistical analysis

The association of diagnostic PSA and/or PSAk with time to reclassification (increase in cancer grade and/or volume) was evaluated using multivariable Cox proportional hazards models.

Results and limitations

A total of 851 men met the study criteria; 255 (30%) had a reclassification event within 5 yr. Median follow-up was 3.7 yr. After adjusting for prostate size, time since diagnosis, biopsy parameters, and diagnostic PSA, PSAk was a significant predictor of reclassification (hazard ratio for each 0.10 increase in PSAk = 1.6 [95% confidence interval 1.2–2.1, p < 0.001]). The PSAk model improved stratification of risk prediction for the top and bottom deciles of risk over a model without PSAk. Model performance was essentially identical using PSA data measured every 6 mo to those measured every 3 mo. The major limitation is the reliability of reclassification as an end point, although it drives most treatment decisions.

Conclusions

PSAk calculated using an LMEM statistically significantly predicts biopsy reclassification. Models that use repeat PSA measurements outperform a model incorporating only diagnostic PSA. Model performance is similar using PSA assessed every 3 or 6 mo. If validated, these results should inform optimal incorporation of PSA trends into active surveillance protocols and risk calculators.

Patient summary

In this report, we looked at whether repeat prostate-specific antigen (PSA) measurements, or PSA kinetics, improve prediction of biopsy outcomes in men using active surveillance to manage localized prostate cancer. We found that in a large multicenter active surveillance cohort, PSA kinetics improves the prediction of surveillance biopsy outcome.

中文翻译：

---

前列腺特异性抗原动力学的精细分析以预测前列腺癌主动监测结果

背景

对于积极监测前列腺癌的男性，前列腺特异性抗原 (PSA) 动力学 (PSAk) 在预测病理重新分类方面的效用仍然存在争议。

客观的

开发利用系列 PSA 的预测方法并评估收集频率。

设计、设置和参与者

从参加多中心金丝雀前列腺主动监测研究的男性中收集数据，测量他们的 PSA 数据并按预先指定的时间表进行活检。我们开发了一个基于线性混合效应模型 (LMEM) 的 PSAk 参数，该模型解释了系列 PSA 水平。

结果测量和统计分析

使用多变量 Cox 比例风险模型评估诊断 PSA 和/或 PSAk 与重新分类时间（癌症等级和/或体积增加）的关联。

结果和局限性

共有 851 名男性符合研究标准；255 (30%) 人在 5 年内发生了重新分类事件。中位随访时间为 3.7 年。在调整前列腺大小、诊断后时间、活检参数和诊断 PSA 后，PSAk 是重新分类的重要预测因子（PSAk 每增加 0.10 的风险比 = 1.6 [95% 置信区间 1.2-2.1，p  < 0.001]）。与没有 PSAk 的模型相比，PSAk 模型改进了风险最高和最低十分位数的风险预测分层。使用每 6 个月测量一次的 PSA 数据与每 3 个月测量一次的模型性能基本相同。主要限制是重新分类作为终点的可靠性，尽管它推动了大多数治疗决策。

结论

使用 LMEM 计算的 PSAk 在统计学上显着预测活检重新分类。使用重复 PSA 测量的模型优于仅包含诊断 PSA 的模型。使用每 3 或 6 个月评估一次的 PSA 时，模型性能相似。如果得到验证，这些结果应有助于将 PSA 趋势最佳地纳入主动监测协议和风险计算器。

患者总结

在本报告中，我们研究了重复前列腺特异性抗原 (PSA) 测量或 PSA 动力学是否可以改善使用主动监测管理局部前列腺癌的男性活检结果的预测。我们发现，在大型多中心主动监测队列中，PSA 动力学提高了对监测活检结果的预测。