INWORKS Solid Cancer Mortality Risk Estimate is Much Higher than the Equivalent LSS Risk Estimate

Abstract

Risk estimates for solid cancer mortality are much higher in the INWORKS study than in the LSS study.^1,2 However, some analysts have reached the opposite conclusion by comparing non-equivalent risk estimates in the 2 studies.³

Solid cancer mortality has been studied in the INWORKS worker cohort by Richardson et al. and the Hiroshima/Nagasaki Life Span Study (LSS) cohort by Brenner et al.^1,2 Linear No Threshold (LNT) based risk estimates, namely Excess Relative Risk (ERR), were produced for both cohorts. The Canadian Nuclear Safety Commission (CNSC) web site says: “Results from INWORKS (2015, 2023) are comparable to Life Span Study results”. ³ This conclusion is based upon an INWORKS ERR of 0.47/Gy and an LSS ERR of 0.44/Gy. Brenner’s LSS ERR is just 6% less than Richardson’s, and analysts concluded that this gives credibility to both studies and to LNT. However, the 3 factors discussed below show that these are the wrong ERR estimates to compare.

First, 87% of the INWORKS cohort is male and 97% of the dose was to males. ¹ Male-only ERR estimates were produced for both the LSS and INWORKS cohorts. The LSS male-only ERR estimate was 45% less than INWORKS, instead of just 6%. Second, the INWORKS highest dose bin was only 0.6 Gy whereas the LSS highest dose bin was 4 Gy. Brenner published the LSS ERR estimate for the dose range up to 0.5 Gray which is 70% smaller than the equivalent INWORKS ERR. Third, quadratic fits to male solid cancer mortality in the LSS cohort fit the data better than LNT and they produce risk estimates in the low dose range of interest that are about 0, or even slightly negative.

In conclusion, the Brenner LSS solid cancer mortality risk estimates are much lower than INWORKS’s, even showing hormesis in the dose range of interest for the quadratic fits. The LSS studies are the most highly regarded for cancer risk, suggesting that the INWORKS studies overestimate risk. Furthermore, one would expect that the ERR from Hiroshima/Nagasaki would have higher solid cancer mortality risks than INWORKS since: LSS doses are acute, the LSS cohort contains children, the cohort was suffering from wartime privation, and medical treatments available before 1980 were generally worse in Japan than for the INWORKS cohort. This further suggests that risk is overestimated in the INWORKS cohort. Secondly, the Brenner study shows significant curvature in the low dose range requiring a quadratic fit and invalidating LNT.

Two articles were recently published in Dose-Response. One explains why the Richardson ERR estimates are too high, and the other shows that the INWORKS leukaemia LNT risk estimates are misleading.^4,5