Aflatoxin levels and prevalence of TP53 aflatoxin-mutations in hepatocellular carcinomas in Mexico

Abstract

Objective:

To determine the exposure to aflatoxin B₁ (AFB₁) in southern Mexico and the presence of the aflatoxin signature mutation in hepatocellular carcinoma (HCC) tissue from patients from a cancer referral center.

Methods.

We estimated the prevalence and distribution of AFB₁ in a representative sample of 100 women and men from Chiapas. We also examined the presence of the aflatoxin signature mutation in codon 249 (R249S), and other relevant mutations of the TP53 gene in HCC tissue blocks from 24 women and 26 men treated in a national cancer referral center.

Results.

The prevalence of AFB₁ in serum samples was 85.5% (95%CI 72.1-93.1) and the median AFB₁ was 0.117 pg/μL (IQR, 0.050–0.350). We detected TP53 R249S in three of the 50 HCCs (6.0%) and observed four other G>T transversions potentially induced by AFB₁.

Conclusion.

Our analysis provides evidence that AFB₁ may have a relevant role on HCC etiology in Mexico.

Introduction

In Mexico, hepatocellular carcinoma (HCC), follows a unique epidemiologic pattern. Male-to-female mortality ratios are close to one.(1) And Hepatitis B virus (HBV)(2) and hepatitis C virus (HCV)(3) seroprevalence are low. However, contamination of maize tortilla, a staple food in Mexico, with aflatoxin B₁ (AFB₁), a potent liver carcinogen, has been shown to be as high as 95% in store samples in Veracruz.(4) And Chiapas and Guatemala, together they represent the region with the highest HCC burden in the Americas.(1) We aimed to estimate AFB₁ exposure in a representative sample of adults from Chiapas. To further characterize the relevance of aflatoxin in Mexico, we determined the presence of the aflatoxin signature mutation in codon 249 (R249S) of the TP53 gene(5) in HCC tissue from patients from a cancer referral center.

Material and Methods

National Health and Nutrition Survey, 2018-2019 (ENSANUT-18/19)

This nationally representative probabilistic multistage stratified cluster sampling survey (representative at the regional, state, urban/rural, and socioeconomic strata-level) obtained information from 44,069 households.(6) A sample of 899 households from the state of Chiapas was probabilistically selected. Among 869 respondents (96.6%), we identified 308 individuals between 40 and 59 years of age (considered to be at highest risk for exposure) and randomly selected 100 serum samples from the 202 (65.5%) persons who donated a blood sample. Age, sex, education and rural residence distribution of selected participants was comparable to survey participants.

Measurement of AFB₁-albumin (AFB₁-lys) adducts

We assessed AFB₁ by taking 250μL of samples which were stored at −70°C (serum or calibration standard). We then added 250μL of a Pronase-PBS (13mg mL-1) solution, and incubated in water-bath at 37-40°C for 4.5h. Solid-phase extraction (SPE) was conducted on a Waters Oasis MAX 1cc 30mg extraction cartridge with a Phenomenex 24-port vacuum manifold. We conducted analyses using UPLC-TOF-MS/MS Waters, model Synapt G2 SI, equipped with a BEH C18 column (2.1x50μm, 1.7μm), with electrospray ionization in positive ion mode as previously described.(7)

Hepatocellular carcinoma tissue samples

We identified 61 women and 85 men from mostly Mexico City and surrounding areas treated in 2005-2015 for HCC at the Instituto Nacional de Cancerología for whom formalin-fixed paraffin-embedded HCC tissue blocks were available. We extracted date and place of birth, residence, HBV/HCV serostatus, regular alcohol use, ever smoking, diabetes diagnosis, and anthopometry from medical records. We confirmed the HCC diagnosis by central review. We included all women (n=26) and a random sample of 26 men (out of 40) with HCC risk factor information. We excluded two women born outside of Mexico.

DNA extraction and sequencing

Chilled blocks were sectioned until the entire tissue was accessed. Duplicate samples of two 10uM sections were collected, and the blade and microtome cleaned before the next sample was processed. FFPE sections were incubated with lysis buffer for 15 min at 80°C. Proteinase K was added, and samples incubated at 70°C overnight. The clear solution under the wax was removed and the DNA was extracted using Mag-Bind FFPE DNA kit (Omega Biotek, Norcross, GA) and eluted in 40 μl of ddH2O. DNA was quantitated using the Qubit Fluorometer dsDNA high sensitivity assay kit (ThermoFisher Scientific).

Targeted capture was performed on all exons of 245 known cancer-related genes using a NimbleGen capture array. Libraries were prepared with the Kapa HyperPlus kit, quantified using the PicoGreen dsDNA Reagent, normalized, and pooled. The pooled samples were captured with the custom NimbleGen Roche SeqCap EZ Choice custom panel, and 2x150bp sequencing was performed on either an Illumina HiSeq4000, or NovaSeq. Sequences were aligned, and TP53 mutations were identified. Variants passing quality control and filtering were visually confirmed using the Integrative Genomics Viewer (IGV). Variants observed 1 or 2 times, and likely to be enriched in somatic mutations were analyzed for mutational signatures using Mutagene (https://www.ncbi.nlm.nih.gov/research/mutagene/identify).

Statistical analysis

AFB₁-lys were log10-transformed before analysis. We estimated AFB₁-lys exposure prevalence and geometric means and 95% confidence intervals (95%CI) using non-response-adjusted sampling weights based on probabilities of selection of households, individuals, and blood sample collection participants. We used linear regression to estimate the ratio of geometric means and 95% CIs across age, sex, habitual residence (urban/rural), ethnicity, and education and Poisson regression to estimate prevalence ratios (and 95%CIs) using STATA (StataCorp. 2015. Release 14. College Station, TX). The study was approved by institutional review boards.

Results

In a representative sample of adults aged 40-59 from Chiapas (mean age, 48.3 years; SD ± 6.0;) nearly 50% were from rural areas and approximately one third indigenous. The overall prevalence of detectable AFB₁-lys was 85.5% (95%CI 72.1, 93.1) representing 970,702 urban and rural individuals in Chiapas. Prevalence appeared to be higher in younger, indigenous adults who live in rural areas. (Table I) Median AFB1-lys was 0.117 pg/μL (IQR, 0.050–0.350) and the geometric mean 2.03 pg AFB1-lys/μL (95%CI 1.11, 3.72; Table II). Adduct levels were three-fold higher in men relative to women (3.68 vs.1.20 pg AFB₁-lys/μL) and in participants living in rural as compared to those in urban areas (3.67 vs.1.22 pg AFB₁-lys/μL).

Table I.

Unadjusted and adjusted prevalence and prevalence ratios of detected serum AFB₁-lys according to participant characteristics, Chiapas, Mexico 2018-2019

	Unadjusted		Adjusted*
	Prevalence (95% CI)	Prevalence ratio (95% CI)	Prevalence (95% CI)	Prevalence ratio (95% CI)
Overall	85.5 (72.1, 93.1)	-
Age
40-49	89.7 (75.5, 96.1)	Ref.	89.1 (87.6, 91.6)	Ref.
50-59	80.2 (59.8, 91.7)	0.89 (0.73, 1.08)	80.8 (76.9, 83.4)	0.90 (0.74, 1.10)
Sex
Male	86.5 (69.7, 94.7)	Ref.	87.2 (81.9, 91.2)	Ref.
Female	84.6 (66.2, 93.9)	0.97 (0.81, 1.16)	84.0 (82.3, 86.9)	0.96 (0.80, 1.15)
Residence
Rural	88.4 (75.6, 94.9)	Ref.	89.0 (85.6, 91.0)	Ref.
Urban	83.1 (57.5, 94.7)	0.94 (0.74, 1.18)	82.5 (78.8, 87.4)	0.92 (0.73, 1.17)
Indigenous
No	88.6 (75.1, 95.3)	Ref.	88.5 (86.6, 90.6)	Ref.
Yes	77.4 (46.2, 93.1)	0.87 (0.62, 1.21)	77.7 (73.3, 81.3)	0.87 (0.62, 1.22)
Education
0-6 years	83.5 (66.9, 92.6)	Ref.	84.1 (80.6, 86.4)	Ref.
≥7 years	89.7 (69.8, 97.0)	1.07 (0.89, 1.29)	88.4 (86.8, 92.6)	1.05 (0.86, 1.28)

Weighted n=1,135,324. Indigenous, participants who reported speaking an indigenous language.

^*All variables were mutually adjusted.

Table II.

Geometric means and geometric mean ratios of serum AFB₁-lys according to participant characteristics, Chiapas, Mexico 2018-2019

	Unadjusted			Adjusted
	N=100	Geometric mean (95% CI)*	Ratio of geometric mean (95% CI)	Geometric mean (95% CI)*	Ratio of geometric mean (95% CI)**
Overall		2.03 (1.11, 3.72)	-
Age
40-49	49	2.74 (0.95, 7.86)	Ref.	2.75 (0.99, 7.64)	Ref.
50-59	51	1.38 (1.06, 1.79)	0.50 (0.16, 1.51)	1.37 (0.98, 1.90)	0.49 (0.17, 1.45)
Sex
Male	56	3.50 (1.00, 12.25)	Ref.	3.68 (1.11, 12.10)	Ref.
Female	44	1.25 (1.10, 1.42)	0.35 (0.10, 1.25)	1.20 (0.85, 1.70)	0.32 (0.09, 1.13)
Residence
Rural	54	3.87 (1.03, 14.57)	Ref.	3.67 (1.15, 11.76)	Ref.
Urban	46	1.17 (1.09, 1.25)	0.30 (0.08, 1.13)	1.22 (0.90, 1.66)	0.33 (0.10, 1.05)
Indigenous
No	72	1.67 (1.04, 2.66)	Ref.	1.92 (1.04, 3.57)	Ref.
Yes	28	3.41 (0.50, 23.05)	2.04 (0.27, 15.14)	2.33 (0.46, 11.69)	1.21 (0.20, 7.31)
Education
0-6 years	68	2.53 (1.03, 6.23)	Ref.	2.52 (1.10, 5.77)	Ref.
≥7 years	32	1.28 (1.13, 1.45)	0.50 (0.20, 1.25)	1.29 (0.82, 2.04)	0.51 (0.20, 1.28)

Weighted n=1,135,324. Indigenous, participants who reported speaking an indigenous language.

^*In pg AFB₁-lys/μL.

^**All variables were mutually adjusted.

Characteristics of the 50 HCC patients are shown in Table III. The median age was 63 years (interquartile range, 20). The prevalence of chronic HBV infection (HBsAg+) was 4.0% and the prevalence of HCV infection (anti-HCV+) was 14.0%. Overweight and obesity were common in both women and men. Cirrhosis was present in 25% of the women and 42.3% of the men. We detected the TP53 R249S mutation in 3 patients (all from mostly rural regions) of the 50 HCCs (6.0%). Two of them were from Veracruz and one was born in Oaxaca (but lived in Mexico City). Four patients with other TP53 G>T transversions (two samples with TP53 V157F, one with TP53 V203L, and one with TP53 G245V). In total, seven (14%) of patients had mutations that could be related to AFB₁ exposure. Within this group, three tumors were from women and evidence of chronic HCV infection was present in only one patient. A preliminary analysis of mutational signatures revealed that up to 10% of somatic mutations in these tumors may be due to aflatoxin (Figure I).

Table III.

Characteristics of all HCC patients included in the study (n=50) diagnosed 2005-2015 in Mexico City by gender

	Women n= 24	Men n= 26
Median age, years (IQR)	64.5 (27)	63.0 (6)
Hepatitis B virus +	1 (4.2)	1 (3.8)
Hepatitis C virus +	3 (12.5)	4 (15.4)
Regular alcohol intake	3 (12.5)	17 (65.4)
Ever smoker	5 (20.8)	12 (46.1)
Mean BMI, kg/m2 (± SD)	24.9 (5.4)	25.5 (3.3)
BMI categories
Normal weight, <25	13 (54.2)	15 (57.7)
Overweight, 25-29.9	7 (29.2)	7 (26.9)
Obese, ≥30	4 (16.6)	4 (15.4)
Diabetes	7 (29.2)	9 (34.6)
Missing information	6 (25.0)	3 (11.5)
Cirrhosis	6 (25.0)	11 (42.31)
Fatty liver	7 (29.2)	6 (23.1)
Differentiation
Well	8 (33.4)	11 (42.3)
Moderate	11 (45.8)	8 (30.8)
Poor	5 (20.8)	7 (26.9)
Mutations
TP53 R249S	0 (0)	3 (11.5)
Other TP53 G>T*	3 (12.5)	1 (7.7)

IQR, interquartile range; BMI, body-mass index=weight in kilograms divided by height in meters squared; SD, standard deviation;

Data presented as number (%) unless otherwise specified;

^*includes TP53 V157F, V203L, and G245V

Figure I.

Frequency of mutations* observed 1 or 2 times in 50 men and women with HCC diagnosed 2005-2015 in Mexico City

*mutations were analyzed using Mutagene Identify (https://www.ncbi.nlm.nih.gov/research/mutagene/identify).

3. Discussion

We found a high prevalence of exposure to aflatoxin in urban and rural Chiapas. However, circulating AFB₁ levels were moderate. Also, aflatoxin-associated TP53 R249S mutation prevalence was higher than observations in low aflatoxin exposure regions.

In Chiapas, AFB₁ exposure was >85% in urban and rural adults. While the magnitude of the exposure was important (0.117 pg/μL; 2.78 pg/mg), it was not as high as in Guatemala (in the Southern border of Chiapas) where AFB₁ was detectable in all participants and the median AFB₁-lys was 0.359 pg/μL.(8) Our results are consistent with an initial estimate in adult indigenous women, where the exposure was widespread but at a more moderate level relative to very high HCC burden areas.(9) Nevertheless, these exposure levels confer increased risk of HCC.(10)

AFB₁ undergoes enzymatic conversion in the liver by CYP3A4 to the AFB₁-8,9-epoxide, the active metabolite that reacts with DNA to form an AFB-N7-guanine adduct. These adducts principally cause G:T/C:A mutations on the transcribed strand through the transcription-coupled pathway of nucleotide excision repair.(11) Almost two decades ago a small study in Northern Mexico documented the presence of the AFB₁ codon 249 mutations in HCC tissue (19% prevalence).(12) In another study, this mutation was not present in any of the 69 HCC tissue samples from highly selected patient population in Mexico City most of whom with cirrhosis.(13) The prevalence of this mutation in our study was lower that what was observed then and in Guatemala (24% prevalence) but higher than the 1-3% seen in low aflatoxin regions (the US, Europe, Korea).(5, 14) While R249S is the dominant TP53 aflatoxin-induced mutation, this mutation has been studied almost exclusively in high HBV prevalence regions. We also saw several examples of mutation G T/C:A in V157F. This mutation is another hotspot in TP53, but has not been causally linked to aflatoxin exposure in addition, R249S and V157F have been reported to be associated with a higher stem-cell-like gene expression and poorer survival.(15)

Major strengths of our analyses are generalizability of AFB₁ exposure estimates to adults in Chiapas and detailed sequencing of TP53 mutations. However, AFB₁ analyses were done in a limited number of participants affecting confidence on subgroup analysis. The examination of the TP53 mutations was performed in HCC cases that may not be generalizable to all patients seen in Mexico. Currently, the majority of HCC cases are diagnosed without pathology confirmation. Similar to other countries, only a fraction of HCCs for which diagnosis is unclear were biopsied. Also, the patient population included in our study may not adequately represent the source population of HCC patients in Mexico. In conclusion as the HCC burden remains understudied, our analysis provides evidence that in Mexico AFB₁ may have a potentially important role in the burden of HCC. The five areas in Mexico with the highest HCC burden in order are Chiapas, Veracruz, Yucatán, Tabasco, and Campeche. Future research should focus in these geographic areas and accurately estimate aflatoxin exposure, identify subpopulations at risk and characterize sources of exposures.