Health risks attributable to radiation

Draft document: Health risks attributable to radiation
Submitted by Richard Wakeford, BNFL
Commenting as an individual

1) One impression I gained when reading this document (which I also felt when reading the low dose extrapolation of radiation-related cancer risk document) is that the consideration of the possibility of a dose threshold was rather disproportionate to the consideration of the possibility that risks are underestimated by the standard LNT model. It is quite right that dose thresholds should be fully discussed, but at least the LNT model would overestimate the risks if a dose threshold exists. In contrast, some supra-linear model (or a DDREF less than 2) would underestimate the radiation-related risk, which would be a problem for radiological protection. A number of potential mechanisms have been advanced for such supra-linear models (e.g. the bystander effect and sensitive sub-groups) and it would seem appropriate to afford equal (if not greater) attention to the possibility of underestimating the risk as to overestimating the risk. I realise that there exist vocal advocates of a threshold or hormetic dose-response, but responding to this pressure should not be at the expense of dealing appropriately with opinions from the other wing of views, an omission which could lead to criticism. 2) I do not feel that the penultimate paragraph of Page 57 provides an appropriate summary of the evidence for an increased risk of cancer following intrauterine exposure to radiation. The OSCC provides about ¾ of the overall statistical information on childhood cancer following antenatal exposure to X-rays during radiographic examinations. The second largest study (Monson & MacMahon, 1984) provides just 10% of the overall information, inevitably leading to greater statistical uncertainty associated with risk estimates derived from this study. For this reason, the relative risk of childhood leukaemia (1.52; 95% CI: 1.18, 1.95) found by Monson & MacMahon, although greater than that for childhood solid tumours (1.27; 95% CI: 0.95, 1.70), is statistically consistent with the solid tumour relative risk, and the implication in the second sentence of this paragraph that these risk estimates are incompatible with each other and with the findings of the OSCC is misleading. Further, the treatment of the cohort studies of intrauterine exposure is also misleading. The only cohort study of antenatal exposure to radiographic X-rays with sufficient power to contradict the findings of the OSCC is that carried out in Edinburgh and London (Court Brown et al., 1960), and the results of this study do indeed seem to contradict the findings of the OSCC. However, one of the authors of this cohort study (Sir Richard Doll) now believes that the results of this cohort study are unreliable because of possible linkage errors (see UNSCEAR 1994 Report). In the absence of this cohort study, the evidence from the remaining cohort studies against the OSCC findings is very weak. The cohort of Japanese atomic bomb survivors irradiated in utero shows no excess of childhood leukaemia (although only 0.2 case would be expected in the absence of radiation exposure), but does show a statistically significant excess of childhood solid tumours (although based on just two cases). The Japanese findings are not incompatible with the findings of the OSCC. I would agree with the proposition that it would appear that the lifetime risk of cancer following exposure in utero is similar to that following exposure in early childhood. The difference between exposure in utero and exposure in early childhood is that the former seems to raise the risk of solid tumours in childhood whereas the latter in general does not (with the clear exception of childhood thyroid cancer – not a typical solid tumour of childhood). This additional component to the overall excess risk is small when weighed against the cancers of adult life, but it should be recognised. The risk coefficient for childhood leukaemia following exposure in utero seems to be similar to that after irradiation in early childhood, and the risk coefficient for childhood solid tumours after exposure in utero seems to be similar to that for childhood leukaemia. The evidence for cancer following intrauterine exposure to radiation has been summarised by Doll & Wakeford (Br J Radiol 1997 70 130-9) and Wakeford & Little (Int J Radiat Biol 2003 79 293-309). (A related small point: at Line 1283, after “power” insert “generally”.) 3) I have not reviewed comprehensively Section 6 (which I found complex, perhaps overly so), but the last paragraph of Page 44 implies that hereditary risk estimates apply only to live births. If my interpretation is correct, should some weight not also be given to hereditary anomalies that result in a miscarriage or stillbirth? It is difficult to see why a late stillbirth should be distinguished from an early neonatal death in such a stark fashion, and at least some justification is required for the current basis for considering live births only. 4) In Section 5, given the potential importance of the risk of non-cancer somatic diseases, it might be argued that ICRP should wait for the deliberations of UNSCEAR to be completed before arriving at a conclusion on the inclusion or otherwise of non-cancer effects in low dose risk estimates. The review of McGale & Darby (Radiat Res 2005 163 247-57) should be referred to. 5) Finally, it is not particularly easy reviewing a document that relies heavily on results using Japanese cancer incidence data that have yet to be published. I trust that this will soon be rectified.