Corrigendum

This article refers to:

Are airborne refractory ceramic fibers similar to asbestos in their carcinogenicity?

Are airborne refractory ceramic fibers similar to asbestos in their carcinogenicity? Inhalation Toxicology, 2012; 24(7): 416–424

Alexander M. Walker, L. Daniel Maxim, and Mark J. Utell

Earlier, we (Walker et al., 2012) analyzed data from an ongoing mortality study of workers occupationally exposed to refractory ceramic fiber (RCF). Our paper compared lung cancer and mesothelioma mortality of these workers to what would have been expected if the cohort had been exposed to various types of asbestos at the same fiber concentrations and for the same amount of time. We concluded that the mortality from both lung cancer and mesothelioma would have been significantly greater if the cohort had been similarly exposed to crocidolite asbestos. The data did not permit a similar conclusion to be drawn if the cohort had been exposed to chrysotile or amosite asbestos.

After publication, an external reviewer discovered an error in the lung cancer calculations. We have corrected the calculations and find that it is not possible to distinguish between the observed lung cancer mortality of the RCF-exposed cohort and what would have been expected if the cohort had been exposed to crocidolite asbestos at the same concentrations and for the same time. There was no error in a parallel set of calculations for mesothelioma, so the conclusion that the RCF cohort experienced less mesothelioma mortality (zero in this case) than would be expected if the cohort had been exposed to crocidolite asbestos remains valid.

Details

Lung cancer calculations

The lung cancer mortality calculations were presented in Table 3 of the Walker et al. (2012) paper. The error in the lung cancer computations resulted from failing to convert the cumulative exposure of the cohort from fiber-months per cubic centimeter to fiber-years per cubic centimeter, which are the appropriate units for equation (1) of our original paper. The corrected version of Table 3 is shown below.

Table 3.  Observed lung cancer deaths by RCF exposure category and deaths anticipated under scenarios of asbestos-like effect.

f-m/cc category	Average f-m/cc	Average f-y/cc	Observed lung cancer deaths	Anticipated lung cancer deaths, if RCF were like asbestos
				No effect	Like chrysotile	Like amphibole
				R = 0	R = 0.5	R = 5
0–15	9	0.75	4	3.250	3.262	3.372
>15–45	26	2.17	2	2.490	2.517	2.760
>45–135	78	6.5	2	3.300	3.407	4.373
>135	234	19.50	4	2.760	3.029	5.451
			12	11.800	12.216	15.955
			p value	0.599	0.551	0.196

Referring to the corrected calculations in Revised Table 3, note that the expected lung cancer deaths if the cohort had been exposed to crocidolite asbestos, using the value for R_L = 5 for crocidolite asbestos estimated in a paper by Hodgson and Darnton (2000) [hereinafter H&D], would be ~16 compared to the observed value of 12 in the RCF-exposed cohort. Although the anticipated number of lung cancer deaths assuming exposure to crocidolite asbestos is greater than either that observed (12) or expected (11.8) if there were no effect of fiber exposure, it is not sufficiently greater to assert that the data are incompatible with the hypothesis that RCF has the same lung cancer potency as crocidolite asbestos (corrected p = 0.196 compared to p < 0.001 in our original paper).

Back calculating, the value of R_L applicable to crocidolite asbestos would have to be approximately 9.2 in order to support the original conclusion of our analysis. A value this high (R_L = 10) was reported by H&D (see Table 2 of that paper) for a cohort of Massachusetts cigarette filter workers, but H&D ultimately selected 5 as the best estimate for R_L. Since publication of the H&D study in 2000, various authors (see Walker et al., 2012) have restudied the available epidemiological data and some (e.g. Lenters et al. [2011] and Burdorf & Heederik [2011]) have concluded that applying more rigorous quality criteria to the selection of cohorts would result in higher potency values for asbestos. Were the assumed value of R_L for crocidolite asbestos to increase, the estimated lung cancer deaths shown in Revised Table 3 would likewise increase and the p value would decrease. A. Darnton of the UK Health and Safety Executive (HSE) reviewed these papers (see http://www.iacoc.org.uk/papers/documents/CC201119AnnexDViewofHSEonNetherlandsRiskAssessment.pdf) and ultimately concluded “while [these] analyses do demonstrate a relationship between increasing study quality and increasing lung cancer risk, restricting attention to a small number of studies that meet quality criteria does not necessarily provide a better view of the risk per unit exposure for the purposes of general risk estimation than earlier meta-analyses.” Thus, HSE continues to support the R_L estimates in the H&D meta-analysis.

The corrected values for the expected lung cancer deaths and resulting value of p assuming that the cohort was exposed to chrysotile asbestos shown in Revised Table 3 differ from those in our original paper, but these differences do not affect our original conclusion.

Mesothelioma calculations

There was no error in the calculations related to mesothelioma in our paper. Thus, the actual number of mesothelioma deaths in the RCF-exposed cohort (zero) would have been significantly greater had the cohort been exposed to crocidolite asbestos. Based on the available data, this conclusion remains valid.

Way forward

The RCF mortality study is ongoing. As time progresses the age, exposure duration, and cumulative exposure of this cohort will increase, which will increase the statistical power of the mortality study. The next scheduled update to the mortality study is 2015 at which time new results will be reported.

We apologize for the computational error.

Acknowledgements

We are indebted to Prof. Dr. Thomas Gebel at the Bundesanstalt für Arbeitsschutz und Arbeitsmedizin, who checked our computations and advised us of this error.

Declaration of interest

This work was funded by the HTIW Coalition, which represents producers of RCF and other high temperature insulating fibers in North and South America, Europe, and Asia. The findings and conclusions of this paper are those of the authors alone.