IARC MONOGRAPHS ON THE EVALUATION OF CARCINOGENIC ...

AFLATOXINS 203
rose to 5.5 (95% CI, 1.2–25). There was also an extremely high risk associated with positive
HBsAg status (OR = 129; 95% CI, 25–659). The authors surmised that peanut contamination
was a major source of aflatoxin in this population.
Wang et al. (1996a) carried out a cohort study in seven townships of Taiwan, China,
including three on the Penghu Islets and four on Taiwan Island. Of the total population
of 89 342 eligible subjects selected from local housing offices and mailed an invitation
in a cancer screening project, 25 618 (29%) volunteered to participate. Among participants,
47% were men and enrolment occurred from July 1990 to June 1992. Participants
were interviewed to elicit information on sociodemographic characteristics,
alcohol and smoking habits, and medical history. Fasting blood and spot urine specimens
were collected and stored frozen. Serum samples were assayed for HBsAg and α-fetoprotein,
anti-HCV and various markers of liver function. Abdominal ultrasonography
was carried out among a subgroup of high-risk persons from two Penghu Islets. All participants
were recontacted by invitation to local research centres or by telephone interviews
between 1992–94. Periodic searches for death certificates from local housing
offices and in June 1995 through linkage with the national death and cancer registries
were carried out. The overall follow-up rate was > 98%. Between February 1991 and
June 1995, 56 HCC cases were identified in the cohort, of which 22 were histologically/cytologically
confirmed. For each case, four controls were selected among cohort
members who were free of liver cancer or cirrhosis at the time of case identification, and
who were matched for age, sex, township and recruitment date. Altogether there were 56
HCC cases and 220 controls. Serum and urine specimens were available for analysis on
subsets: serum for 52 cases and 168 controls, and urine for 38 cases and 137 controls.
Urinary aflatoxin metabolites were determined using a monoclonal antibody with high
affinity to aflatoxin B 1 and significant cross-reactivity to aflatoxins B 2, M 1, G 1 and P 1.
Serum aflatoxin–albumin adducts were measured. Using conditional logistic regression,
the OR for liver cancer corresponding to detectable levels of aflatoxin–albumin adducts
was 4.6 (95% CI, 2.0–10) before adjustment for HBsAg and 1.6 (95% CI, 0.4–5.5) after
adjustment. [The Working Group noted inconsistencies in numbers of available controls
for serum aflatoxin–albumin adducts.] For high levels of urinary aflatoxin metabolites,
the OR was 3.3 (95% CI, 1.4–7.7) before adjustment for HBsAg and 3.8
(95% CI, 1.1–13) after adjustment. While there was little or no effect of aflatoxin biomarkers
on HCC among HBsAg-negative subjects, there were quite strong effects
among HBsAg-positive subjects, especially in the analysis using aflatoxin metabolites as
the exposure. [It seems that the present cohort and the cases identified in it include the
cases studied by Chen et al. (1996) in the Penghu Islets. The low participation rate would
not have affected the validity of the results unless individuals with preclinical liver
cancer symptoms and high aflatoxin exposure were more likely to volunteer for participation
in the study than others in the same population.]
Sun et al. (2001) reported the results from a nested case–control study of an extended
follow-up of the cohort described by Wang et al. (1996a). Seventy-nine HBsAg-positive
cases of HCC were identified between 1991 and 1997, and matched for age, gender,