IARC MONOGRAPHS ON THE EVALUATION OF CARCINOGENIC ...

in central Sudan (Omer et al., 1998). The authors also noted, however, that residents of
the two regions are ethnically different, so effect modification by unmeasured genetic or
environmental factors cannot be excluded. While GSTM1 genotype was not a risk factor
for HCC, it was a strong effect modifier. The excess risk due to peanut butter consumption
was restricted to subjects with GSTM1-null genotype; the OR in the highest quartile
of peanut butter exposure among GSTM1-null subjects was 17 (95% CI, 2.7–105).
2.4 Limitations of recent studies
AFLATOXINS 209
While recent studies have incorporated many methodological improvements over
studies described in the previous monograph on aflatoxins (IARC, 1993), there
nevertheless remain certain problems that limit our ability to fully understand the role of
aflatoxins in liver carcinogenesis.
Many recent studies have used HBsAg as the marker of exposure to HBV. However,
among liver cancer cases that are negative for HBsAg, HBV DNA can be detected in
33% (serum) and 47% (liver) of the cases, notably those from areas of high viral
prevalence. Similarly, HCV RNA can be found in 7% (serum) and 26% (liver) of anti-
HCV-negative liver cancer cases (Bréchot et al., 1998). Thus studies relying on HBsAg
or anti-HCV measurements may underestimate viral exposure and this may affect an
evaluation of interaction between hepatitis viruses and aflatoxin (Paterlini et al., 1994;
Kew et al., 1997; Kazemi-Shirazi et al., 2000).
Biomarkers of exposure to aflatoxin have been increasingly used to assess aflatoxin
exposure. However, measurable urinary metabolites of aflatoxin or aflatoxin–albumin
adducts in serum reflect only exposures in a recent period (days or weeks), and these may
not be related to exposures during the etiologically relevant period (years earlier).
Moreover, it is unclear whether the presence of liver disease before cancer modifies the
levels of the markers found in serum or urine. In the presence of liver disease, comparisons
of levels of the marker between liver cancer cases and controls may be biased.
Follow-up studies of either general populations in areas of different aflatoxin exposure
or of HBsAg carriers investigated with repeated measurements of aflatoxin biomarkers
have not yet accumulated follow-up periods that are long enough to minimize the
possibility that pre-existing liver disease led to bias in measurements of biomarker
levels.
Dietary questionnaires and food measurement surveys at the population level, often
used to estimate aflatoxin exposure, provide crude measurements and may fail to account
for secular trends in exposure or individual variations in exposure. Similarly, mortality
rates used in ecological studies to characterize regions at variable risk of liver cancer
may suffer from misclassification of the diagnosis and reporting systems in some
countries.