Response to Cuttler et al . re: “Low-Dose Childhood Radiation Effects to the Thyroid Follow a Linear Dose–Response Trend and Persist Even 45+ Years After Exposure”

Dear Editor:

We thank Cuttler et al. for their comments (1) regarding a recent Clinical Thyroidology review (2) that summarized the report of nine pooled cohorts that showed a positive association between childhood low-dose radiation exposure and incident thyroid cancer that was consistent with linearity (3). However, they provide no data to support their objections to our interpretation regarding radiation risks. We appreciate the opportunity to address their concerns and further highlight the potential adverse health effects of low-dose radiation exposure.

Cuttler et al. raise three concerns: (i) over-diagnosis of thyroid cancer occurred in both early and recent time periods, which they claim invalidates the demonstration by Lubin et al. of temporal consistency of the linear relationship for thyroid cancer incidence and low-dose radiation; (ii) over-diagnosis of thyroid cancer, as for example from increased screening, nullifies any assessment of linearity; and (iii) analysis of thyroid cancer incidence “is not clinically significant” and mortality-based analyses are preferable.

First, Cuttler et al. provide no scientific evidence supporting their assertion regarding temporal variation of “over-diagnosis” rates. More importantly, they provide no evidence that “over-diagnosis” per se is related to dose, and thus provide no basis for their claim that diagnostic patterns confound the observed linear dose–response relationship. Indeed, evidence indicates the opposite. Lubin et al. describe secondary analyses indicating that there was no effect of either calendar period or screening on the dose–response, which directly contradicts Cuttler et al.'s suggestion of confounding by diagnostic pattern. Second, the reference to a Korean study (4) of changing thyroid cancer diagnostic patterns is irrelevant in relation to radiation dose–response, unless such patterns are associated with radiation dose. The Korean study presented no evidence of such a radiation dose-related association and therefore is uninformative on potential confounding. Third, the letter implies a possible screening bias in the pooled cohort results but again provided no evidence for this assertion. The original publication (3) discusses in detail that although medical screening may have increased the detection of thyroid cancer cases, there was little reason to presuppose that screening was also related to dose and thus confound results. Fourth, Cuttler et al. allude to “over-diagnosis” of thyroid cancer in children exposed to ¹³¹I following the Chornobyl ^* accident. However, studies directly contradict their suggestion. Indeed, studies in Ukraine and in Belarus of thyroid cancer in radiation-exposed children only enrolled screened individuals (5,6). Moreover, both studies reported dose–response relationships that were consistent with linearity at low doses. Finally, regarding concern (iii), in the study of disease etiology, which includes analysis of dose–response, incident events are universally preferred whenever possible, since analysis of mortality outcomes is potentially confounded by non-causal factors associated only with differential survival.

The pooled data from the nine studies represent the best available epidemiologic evidence on the potential adverse effects of external low-dose childhood radiation exposure. Those data suggest that incident thyroid cancer rates are consistent with linearity at low radiation doses, even 45 years later, and thus long-term surveillance would be beneficial toward optimizing health outcomes.