Association of Aflatoxin With Gallbladder Cancer in Chile

In Chile, gallbladder cancer is a leading cause of cancer death in women. Other than gallstones, gallbladder cancer etiology remains largely unclear. Exposure to aflatoxin, a liver carcinogen, is associated with bile duct epithelium proliferation in both animals and humans,¹ and with gallbladder cancer in primates.¹ Aflatoxin contamination has been identified in Chile, including in ají rojo (red chili peppers). Ají rojo is associated with gallbladder cancer²; however, the association of aflatoxin with gallbladder cancer in humans has not been directly evaluated.

Methods ∣

We evaluated plasma aflatoxin-albumin adducts and gallbladder cancer in a pilot study conducted from April 2012 through August 2013. We recruited incident gallbladder cancer cases identified through rapid ascertainment at cancer referral hospitals in Santiago, Concepción, and Temuco, Chile. We initially recruited all cases consecutively and later recruited only surgical cases to provide tissue samples for future studies. We aimed at 1:1:1 matching by age and sex, as well as hospital for controls with gallstones (to ensure associations with gallbladder cancer were not solely due to gallstones) or study site for community controls. Pairing of controls with gallstones was limited by the small number of patients older than 50 years who underwent gallbladder surgery.

Community controls were selected from a random listing of the population enrolled in the same health center registry as the cases or through neighborhood sampling. Participants had to be previously cancer-free and covered by public health insurance (>90% of population). Chile and US institutional review board-approved written consent was obtained for data and blood collection at enrollment.

Aflatoxin forms adducts with albumin in peripheral blood that accumulate up to 30-fold higher with chronic vs single exposure.³ Using isotope dilution mass spectrometry,³ we assessed aflatoxin B₁-lysine adduct (AFB₁ adduct) detection (≥0.5 pg/mg of albumin) and level. Conditional and unconditional logistic regression models produced similar results. Therefore, we used polytomous logistic regression as the most powerful analytic approach. We evaluated variables in Table 1 as potential confounders. We retained questionnaire-derived ají rojo consumption (at least weekly), which changed the magnitude of the odds ratio (OR) for AFB₁ adduct detection by greater than 10%. We assessed statistical significance at P < .05 using 2-sided tests. Analyses were conducted in SAS version 9.3 (SAS Institute Inc).

Table 1.

Characteristics of Patients With Gallbladder Cancer Compared With 2 Control Groups

		Controls
	Gallbladder Cancer (n = 36)	With Gallstones (n = 29)	Community (n = 47)	P Valuea
Age, median (range), y	61 (41-77)	57 (44-76)	65 (3779)	.60
Female sex, No. (%)b	32 (88.9)	21 (77.8)	39 (83.0)	.50
Location in Chile, No. (%)
Santiago	17 (47.2)	14 (48.3)	27 (57.4)
Concepción	10 (27.8)	8 (27.6)	10 (21.3)	.90
Temuco	9 (25.0)	7 (24.1)	10 (21.3)
History or presence of gallstones, No. (%)b,c	34 (94.4)	29 (100.0)	8 (17.8)	<.001
Education, median (range), y	6 (2-16)	8 (2-15)	6 (0-25)	>.99
Ever smoking, No. (%)b	12 (33.3)	14 (50.0)	21 (44.7)	.40
Typical weight as adult, No. (%)b
Normal weight	13 (41.9)	7 (29.2)	13 (30.2)
Overweight	10 (32.3)	11 (45.8)	19 (44.2)	.90
Class I obesity	6 (19.4)	4 (16.7)	6 (14.0)
Class II/III obesity	2 (6.5)	2 (8.3)	5 (11.6)
Self-reported diabetes, No. (%)b	9 (25.0)	5 (17.9)	8 (17.4)	.70
Family history of gallbladder cancer, No. (%)b	2 (7.4)	1 (4.8)	1 (2.9)	.80
Consumption at least weekly, No. (%)b,d
Ají rojo paste	17 (47.2)	5 (17.9)	9 (19.1)	.009
Fresh ají verde	20 (55.6)	12 (42.9)	18 (38.3)	.30
Aflatoxin B1-lysine adduct
No. detectable	23	7	9
Level, median (IQR), pg/mg of albumine	7.6 (11.1)	3.5 (3.5)	2.4 (2.5)	<.001

Abbreviation: IQR, interquartile range.

^aCalculated using the Kruskal-Wallis test for difference in medians for continuous variables and the Fisher exact χ² test for categorical variables.

^bPercentages exclude individuals with missing data.

^cGallstone status assessed through questionnaire, medical record abstraction, ultrasound, specimen collection or histology, or a combination of these.

^dParticipants were asked, “Before the last 3 years, how many times per day, week, or month did you eat….”

^eAmong those with detectable levels.

Results ∣

Participation rates were similar for cases with gallbladder cancer (85%, 52/61) and controls with gallstones (86%, 37/43) but lower for community controls (57%, 50/88); male and younger potential community controls refused more often. However, age did not differ among community controls by aflatoxin status (median age, 66 vs 65 years) and none of the 8 males had detectable levels. We included all participants with available plasma: 36 cases (69%), 29 controls with gallstones (78%), and 47 community controls (94%). Cases and controls had similar characteristics except for ají rojo consumption (Table 1).

The AFB₁-adducts were detected in 23 cases (64%), 7 controls with gallstones (18%), and 9 community controls (23%). Levels were highest in cases (median, 7.6 pg/mg; Table 1), who were more likely to have detectable AFB₁ adducts than controls with gallstones (OR, 9.4; 95% CI, 2.8-37.2) or community controls (OR, 13.2; 95% CI, 4.3-47.9) (Table 2). Restricted to participants with AFB₁ adducts, cases with gallbladder cancer had higher levels per change of 10 pg/mg of albumin than controls with gallstones (OR, 4.0; 95% CI, 1.0-78.0) or community controls (OR, 2.5; 95% CI, 1.0-16.7).

Table 2.

Association of Aflatoxin B₁-Lysine Adducts in Patients With Gallbladder Cancer Compared With 2 Control Groups^a

	Aflatoxin B1-Lysine Adduct
	Detection	Change in Level per 10 pg/mg of Albuminb
Controls with gallstones vs community controls	1.4 (0.4-4.4)	0.6 (0.03-5.7)
Patients with gallbladder cancer vs controls with gallstones	9.4 (2.8-37.2)	4.0 (1.0-78.0)
Patients with gallbladder cancer vs community controls	13.2 (4.3-47.9)	2.5 (1.0-16.7)

^aValues are expressed as odds ratios and 95% confidence intervals from polytomous logistic regression models adjusted for at least weekly consumption of ají rojo paste. Estimates obtained using community controls as the referent group in one model and controls with gallstones as the referent group in the other model.

^bRestricted to individuals with detectable levels (≥0.5 pg/mg of albumin).

Discussion ∣

Several lines of evidence support the biological plausibility of the association of gallbladder cancer with aflatoxin, including experimental, animal, and occupational data⁴; low hepatitis B virus prevalence in Chile^2,5; and genetic variation that may affect xenobiotic excretion.⁴ In addition, AFB₁ adduct plasma levels are similar to those associated with increased risk of hepatocellular carcinoma.⁶

Despite the small number of participants, the associations between aflatoxin exposure and gallbladder cancer were statistically significant. Recall bias may affect self-reported variables, but not exposure measurement. We cannot rule out reverse causation (ie, cancer may affect AFB₁ adduct detection) using cross-sectional data. Larger and longitudinal efforts are needed to substantiate these preliminary findings (eg, by identifying aflatoxin-related TP53 mutations), obtain more precise effect estimates, and identify sources of aflatoxin. These findings, if confirmed, may have implications for cancer prevention.