it is difficult to conclude with confidence that the risk is more strongly associated with any specific exposure in the broad class of phenoxy acids and related compounds. There was evidence of increasing risk with increasing cumulative exposure to several agents, including TCDD, and 2,4-D, a herbicide that does not contain TCDD.
Collins et al. (1993) point out that "All but one of the confirmed STS cases among more than 5000 workers in 12 plants mentioned in the Fingerhut et al. [(1991]) study occurred among the 754 persons in the [Monsanto] study," and, based on a detailed analysis of the exposure histories of the STS cases, argue that the TCDD is unlikely to be responsible and that 4-aminobiphenyl may be.
Several authors have added additional years of follow-up to occupational cohort studies (Bloemen et al., 1993; Lynge, 1993; Asp et al., 1994). Lynge (1993) found the risk of STS similar to that reported in the earlier study of this cohort of Danish herbicide manufacturers (Lynge, 1985). There were five cases of STS observed, versus 2.5 expected (relative risk = 2.0, CI 0.7 to 4.8). When the definition of exposure was restricted to those with at least one year of work in exposed areas, and a ten year interval was applied between the start of exposure and the start of follow-up time considered to be at risk, there were 3 observed cases, compared to 0.5 cases expected (RR = 6.4, CI 1.3 to 18.7). These workers were engaged in the manufacture of 2,4-D and a related herbicide, 4-chloro-2-methylphenoxyacetic acid (MCPA), but not 2,4,5-T.
Despite additional follow-up, the Bloemen et al. (1993) study of 2,4-D workers was still of insufficient size to be useful for an evaluation of STS. This study was an extension of a cohort mortality study of 878 Dow Chemical Company workers exposed to 2,4-D (Bond et al., 1988). There were no deaths observed from this cancer, and considerably less than one case would have been expected.
Similarly, the Asp et al. (1994) study was too small to be useful for detecting an evaluation in risk of STS. This study was an update of a cancer incidence and mortality study of 1,909 herbicide applicators in Finland (Riihimaki et al., 1982, 1983). The authors noted that with 0.8 cases of STS expected, their study had sufficient power to detect only relative risks of 7.0 or greater with 90 percent confidence.
In the United States a PCMR study was performed for farmers in 23 states, using occupational information from death certificates (Blair et al., 1993). Based on 98 deaths from STS in white male farmers, the PCMR was 0.9 (CI 0.8-1.1). The numbers of deaths due to STS were small and nonsignificant in the other racial and gender groups: nonwhite males and white and nonwhite females.
The Bertazzi et al. (1993) study of cancer incidence at Seveso yielded results similar to those reported in earlier publications from this group and summarized in VAO (Bertazzi et al., 1989a,b; Pesatori et al., 1992). In the small, most heavily exposed group (Zone A), there were no cases of STS observed, when the class is defined as those tumors in ICD 171: "malignant neoplasms of connective and other soft-tissues." There were two cases of "soft-tissue sarcomas of parenchymal