Letter re Marsh et al

Dear Editor,

Marsh and colleagues have recently published a paper (Marsh et al., 2011) in which they conclude that ‘elevated lung levels of tremolite in the lungs of brake repair workers with elevated levels of amphiboles arose from concurrent exposures to commercial amphibole and chrysotile asbestos in occupational settings other than brake repair work.’ As described in their Methods section, the authors ‘abstracted from Butnor et al. (Butnor et al., 2003) data on the levels of commercial amphibole fiber and tremolite fiber found in the lungs of 10 mesothelioma cases used in the Finkelstein (Finkelstein, 2008; Finkelstein, 2009) analyses.’ Marsh and colleagues then proceeded to perform regression analyses of the levels of tremolite fibers against the levels of commercial amphibole fibers. Unfortunately these analyses lack validity because the data values analyzed by the authors do not, in many cases, correspond to the fiber counts recorded by Dr Roggli on his laboratory counting sheets.

In January 2011, I was provided with a copy of Dr Roggli’s case files pertaining to the series of brake mechanics described by Butnor et al. (2003). The documents were obtained as a part of the discovery process in the American legal system and were sent to me for review by a plaintiff’s attorney. The package included the counting sheets upon which the electron microscopist recorded the results of his examination of the tissue samples. Having compared these counting sheets with the data presented in Table 1 of Marsh et al., I observed discrepancies in 7 of the 20 (35%) data points among the Butnor cases. Details are presented in the Table and discussion below.

Table 1.  Comparison of fiber counts (number of fibers) recorded on Dr Roggli’s laboratory count sheets with tissue levels of commercial amphiboles and tremolite (fibers per gram of wet lung tissue for fibers 5 microns or greater in length) reported in Table 1 of Marsh et al. Discrepancies between the Laboratory count sheets and the published Table are indicated with *.

Case	Source	Uncoated fibers	Amosite	Crocidolite	Total commercial amphiboles	Tremolite
1	Lab sheet	5	0	0	0*	1
	Marsh table 1				3270*	2180
2	Lab sheet	12	6	0	6	0*
	Marsh table 1				3970	440*
3	Lab sheet	14	1	0	1	0*
	Marsh table 1				970	4110*
4	Lab sheet	5	0	0	0	0*
	Marsh table 1				<580	1160*
5	Lab sheet missing. I retrieved data from summary sheet	8			0	2
	Marsh table 1				<580	1160
6	Lab sheet	3	0	0	0*	1
	Marsh table 1				490*	490
7	Lab sheet	2	1	0	1	0
	Marsh table 1				340	<340
8	Lab sheet	49	1	0	1	1
	Marsh table 1				120**	240**
9a	Lab sheet	9	0	1	1	1
	Marsh table 1				6000**	3280**
9b	Lab sheet	26	0	4	4	4
	Marsh table 1				6000**	3280**
10	Lab sheet	16	2	0	2	3
	Marsh table 1				1440	2170

**Number of amphibole and tremolite fibers identical, but concentrations reported to be different.

• Case 1: No amosite or crocidolite fibers were detected in Case 1. Table 1 reports 3270 fibers per gram of wet lung tissue for fibers 5 µ or greater in length.

• Case 2: No tremolite fibers were detected in Case 2. Table 1 reports 440 fibers per gram of wet lung tissue for fibers 5 µ or greater in length.

• Case 3: No tremolite fibers were detected in Case 3. Table 1 reports 4110 fibers per gram of wet lung tissue for fibers 5 µ or greater in length.

• Case 4: No tremolite fibers were detected in Case 4. Table 1 reports 1160 fibers per gram of wet lung tissue for fibers 5 µ or greater in length.

• Case 6: No amosite or crocidolite fibers were detected in Case 6. Table 1 reports 490 fibers per gram of wet lung tissue for fibers 5 µ or greater in length.

• Case 8: One amosite fiber and 1 tremolite fiber were detected in Case 8. Table 1 reports 120 fibers per gram of wet lung tissue for fibers 5 µ or greater in length for commercial amphiboles and 240 fibers per gram of wet lung tissue for fibers 5 microns or greater in length for tremolite. Since the number of fibers counted was the same, these differing concentrations cannot both be correct.

• Case 9: The microscopist actually analyzed two separate samples for Case 9. Marsh et al. report the results for only one sample (or do not describe how the samples were combined). In both samples the numbers of crocidolite and tremolite fibers counted were the same. Table 1 reports different values for the concentrations.

• Case 9 is of particular interest because this subject worked as a supervisor in a commercial automotive repair shop where the brand of brake shoes that were used contained crocidolite asbestos fibers. The presence of both crocidolite and tremolite in this man’s lungs is consistent with the deposition and retention of asbestos fibers from friction products in the lungs of individuals working in the repair environment.

It can thus be seen that there are discrepancies between the counts of both commercial amphiboles and tremolite recorded in the laboratory counting sheets and the concentrations of fibers reported and analyzed by Marsh and colleagues. I have recently learned from Dr Roggli that the reported concentrations of commercial amphiboles may be the sum of fibers 5 µ or greater in length and asbestos bodies. This is an unusual procedure and is nowhere described in the published paper. If true, the footer number (c) to Table 1 is misleading and should state “asbestos bodies plus fibers per gram of wet lung tissue for fibers 5 µ or greater in length”.

In summary, the data Marsh and colleagues used in their regression analyses do not, in many cases, correspond with the fiber count data recorded by the electron microscopist. It is my suggestion that the paper of Marsh and colleagues be withdrawn, that the data be corrected and re-analyzed, and that the paper be resubmitted. It is only when we can be confident that the analysis is based upon the true data from Dr Roggli’s laboratory that the substantive issues raised by this work can be addressed.

Declaration of interest

Dr Finkelstein has served as a consultant to plaintiff’s attorneys in asbestos litigation.