Abstract

Purpose

Thyroid cancer of papillary histology (PTC) is the dominant type in radio-epidemiological cohorts established after nuclear accidents or warfare. Studies on post-Chernobyl PTC and on thyroid cancer in the life span study (LSS) of Japanese a-bomb survivors consistently revealed high radiation risk after exposure during childhood and adolescence. For post-Chernobyl risk assessment overexpression of the CLIP2 gene was proposed as molecular biomarker to separate radiogenic from sporadic PTC. Based on such binary marker a biologically-based risk model of PTC carcinogenesis has been developed for observational Chernobyl data. The model featured two independent molecular pathways of disease development, of which one was associated with radiation exposure. To gain credibility the concept for a mechanistic risk model must be based on general biological features which transcend findings in a single cohort. The purpose of the present study is therefore to demonstrate portability of the model concept by application to PTC incidence data in the LSS. By exploiting the molecular two-path concept we improve the determination of the probability of radiation causing cancer (POC).

Materials and methods

The current analysis uses thyroid cancer incidence data of the LSS with thyroid cancer diagnoses and papillary histology (n = 292) from the follow-up period between 1958 and 2005. Risk analysis was performed with both descriptive and biologically-based models.

Results

Judged by goodness-of-fit all applied models described the data almost equally well. They yielded similar risk estimates in cohorts post-Chernobyl and LSS. The preferred mechanistic model was selected by biological plausibility. It reflected important features of an imperfect radiation marker which are not easily addressed by descriptive models. Precise model predictions of marker prevalence in strata of epidemiological covariables can be tested by molecular measurements. Application of the radiation-related molecular pathway from our preferred model in retrospective risk assessment decreases the threshold dose for 50% POC from 0.33 (95% confidence interval (CI) 0.18; 0.64) Gy to 0.04 (95% CI 0.01; 0.19) Gy for females and from 0.43 (95% CI 0.17; 1.84) Gy to 0.19 (95% CI 0.05; 1.00) Gy for males. These improvements are still not sufficient to separate radiation-induced from sporadic PTC cases at very low doses <0.015 Gy typical for the Fukushima accident.

Conclusions

Successful application of our preferred mechanistic model to LSS incidence data confirms and improves the biological two-path concept of radiation-induced PTC. Model predictions suggest further molecular validation studies to consolidate the basis of biologically-based risk estimation.

中文翻译：

---

1958-2005 年日本原子弹幸存者甲状腺乳头状癌的放射风险和生物标志物患病率的生物学建模。

摘要

目的

甲状腺乳头状组织癌 (PTC) 是核事故或战争后建立的放射流行病学队列中的主要类型。在日本原子弹幸存者的寿命研究 (LSS) 中对切尔诺贝利后 PTC 和甲状腺癌的研究一致显示，儿童和青春期暴露后辐射风险很高。对于切尔诺贝利事故后的风险评估，CLIP2 基因的过度表达被提议作为分子生物标志物，以区分放射源性和散发性 PTC。基于这种二元标记，已经为观察切尔诺贝利数据开发了基于生物学的 PTC 致癌风险模型。该模型具有两种独立的疾病发展分子途径，其中一种与辐射暴露有关。为了获得可信度，机械风险模型的概念必须基于超越单个队列研究结果的一般生物学特征。因此，本研究的目的是通过应用于 LSS 中的 PTC 发病率数据来证明模型概念的可移植性。通过利用分子双路径概念，我们改进了辐射致癌 (POC) 概率的确定。

材料和方法

目前的分析使用 LSS 的甲状腺癌发病率数据，包括 1958 年至 2005 年随访期间的甲状腺癌诊断和乳头状组织学 ( n = 292)。使用描述性和基于生物学的模型进行风险分析。

结果

根据拟合优度判断，所有应用的模型对数据的描述几乎一样好。他们在切尔诺贝利和 LSS 后的队列中得出了类似的风险估计。首选的机械模型是通过生物学合理性选择的。它反映了描述性模型不易解决的不完美辐射标记的重要特征。流行病学协变量分层中标记流行的精确模型预测可以通过分子测量进行测试。在回顾性风险评估中应用我们首选模型中的辐射相关分子途径将 50% POC 的阈值剂量从 0.33（95% 置信区间（CI）0.18；0.64）Gy 降低到 0.04（95% CI 0.01；0.19）Gy对于女性，男性为 0.43 (95% CI 0.17; 1.84) Gy 至 0.19 (95% CI 0.05; 1.00) Gy。

结论

我们首选的机械模型成功应用于 LSS 入射数据证实并改进了辐射诱导 PTC 的生物双路径概念。模型预测表明进一步的分子验证研究可以巩固基于生物学的风险估计的基础。