Endogenous aldehyde accumulation generates genotoxicity and exhaled biomarkers in esophageal adenocarcinoma

Abstract

Volatile aldehydes are enriched in esophageal adenocarcinoma (EAC) patients' breath and could improve early diagnosis, however the mechanisms of their production are unknown. Here, we show that weak aldehyde detoxification characterizes EAC, which is sufficient to cause endogenous aldehyde accumulation in vitro. Two aldehyde groups are significantly enriched in EAC biopsies and adjacent tissue: (i) short-chain alkanals, and (ii) medium-chain alkanals, including decanal. The short-chain alkanals form DNA-adducts, which demonstrates genotoxicity and confirms inadequate detoxification. Metformin, a putative aldehyde scavenger, reduces this toxicity. Tissue and breath concentrations of the medium-chain alkanal decanal are correlated, and increased decanal is linked to reduced ALDH3A2 expression, TP53 deletion, and adverse clinical features. Thus, we present a model for increased exhaled aldehydes based on endogenous accumulation from reduced detoxification, which also causes therapeutically actionable genotoxicity. These results support EAC early diagnosis trials using exhaled aldehyde analysis.

Conflict of interest statement

G.H. is named on patents related to breath analysis, and is a shareholder in VODCA, a cancer detection company. R.C.F. is named on patents related to the Cytosponge and associated assays and is a shareholder and consultant for Cyted an early detection company. The other co-authors declare no competing interests.

Figures

Fig. 1. Loss of aldehyde detoxification is conspicuous in esophageal adenocarcinoma.

a Geneset enrichment analysis comparing squamous mucosa (SqN, n = 19) and esophageal adenocarcinoma (EAC, n = 21) samples in GSE26886. Five most significantly enriched Gene Ontology v4.0 genesets in squamous mucosa. P values calculated by the permutation test in the GSEA analysis. b Enrichment plot of “oxidoreductase (OR) acting on aldehydes” geneset from a. c Relative expression of candidate ALDH isoenzymes in EAC endoscopic biopsies compared to paired SqT biopsies (n = 67), quantitative reverse transcriptase polymerase chain reaction (median and 95% confidence interval provided. P values calculated with two-sided Wilcoxon matched-pairs signed rank test; ALDH1A3 P = 0.0026, all others P < 0.0001). d, e Representative immunostaining for ALDH3A1 (d) and ALDH3A2 (e) in an esophageal cross-section featuring an adenocarcinoma (bar indicates 5 mm). f Summary of ALDH3A1 and -3A2 immunostaining in cancer-adjacent squamous tissue (SqT, n = 75), Barrett’s metaplasia (BM, n = 14); Barrett’s dysplasia (BD, n = 12), and EAC (n = 571) drawn from nine UK hospitals. Mean ± SEM provided, Kruskall–Wallis test with Dunn’s correction **P < 0.01, ****P < 0.0001. Source data are provided in the Source Data file.

Fig. 2. Biopsies of esophageal adenocarcinoma and adjacent squamous tissue samples are enriched for genotoxic and medium-chain aldehydes.

a Principal component analysis (PCA) of esophageal tissue aldehyde profiles measured by UPLC-MS/MS. Squamous tissue from healthy patients (SqN, n = 32 independent patients), esophageal adenocarcinoma (EAC)-adjacent squamous tissue (SqT, n = 59), EAC-adjacent Barrett’s (BAR, n = 8), and EAC (n = 59) were compared. b Correlation matrix of aldehydes in EAC tissue samples in a. Scale indicates Pearson’s r. c–l Absolute concentrations of the indicated aldehydes in the tested tissues. Mean ± SEM provided, Kruskall–Wallis test with Dunn’s correction, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are provided in the Source Data file.

Fig. 3. Aldehyde DNA damage is present in esophageal adenocarcinoma, which can be reduced in vitro by metformin.

a Schematic of deoxyadenosine and deoxyguanosine adduction by acetaldehyde (C2:0) and epoxy-hydroxynonanal (EHN), respectively, to form CrodG and 1N6-εdA. b CrodG and 1N6-εdA concentrations in DNA extracts from peripheral leukocytes (n = 9), squamous tissue from healthy patients SqN (n = 5), esophageal adenocarcinoma-adjacent squamous tissue (SqT, n = 20), and esophageal adenocarcinoma (EAC, n = 19). c, d Western blot analysis of the indicated cell lines pre-treated with metformin (MET) and exposed to 150 μM formaldehyde or 500 μM acetaldehyde for 5 h. e Schematic of the ATM and ATR-mediated DNA damage repair f Western blot analysis for DNA damage repair protein phosphorylation in the indicated cell lines pre-treated with metformin and exposed to 500 μM acetaldehyde for 5 h. Mean ± SEM provided for analytical triplicates of biological duplicates or triplicates, two-tailed Mann–Whitney U test or ANOVA, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All blots are representative of three independent experiments with similar results. Source data are provided in the Source Data file.

Fig. 4. Tissue decanal is diagnostic and correlates to exhaled decanal.

a Receiver operating characteristic curve for EAC detection using a random forests multivariable model based on the indicated tissue aldehydes, measured by UPLC-MS/MS. b–g Comparison of tissue and exhaled butanal and decanal. b, e The receiver operating characteristic curve based on tissue butanal (b) or decanal (e) concentrations as a single biomarker (UPLC-MS/MS cohort). c, f Correlation of EAC tissue butanal (c) and decanal (f) concentration to the paired exhaled breath concentration (data from Markar et al.). P values calculated with a two-tailed Spearman test. d, g Gas-phase concentrations of butanal (d) and decanal (g) in the headspace of healthy squamous (SqN, n = 24) and esophageal adenocarcinoma (EAC, n = 25) samples measured by proton transfer reaction time-of-flight mass spectrometry. Mean ± SEM provided with two-tailed Mann–Whitney U test. Source data are provided in the Source Data file.

Fig. 5. ALDH3A2 loss connects decanal to prognosis and TP53.

a, b Relative aldehyde concentrations in tissue samples of esophageal adenocarcinoma (UPLC-MS/MS cohort) divided into ALDH3A1 (a) and ALDH3A2 (b) high or low tumors (n = 24, two equal groups). c, d Correlations of ALDH3A1 (c) and ALDH3A2 (d) expression (CPKG, copies-per-thousand GAPDH values, n = 27) to paired exhaled decanal concentrations from Markar et al.. e Overall survival in the multi-center immunohistochemistry cohort, stratified by ALDH3A2 expression (immunoscore > 1 used as cut-off), log-rank test. Patients surviving <3 months from surgery were excluded (total included n = 329). f Overall survival for patients with esophageal adenocarcinoma in the previously published exhaled breath cohort, dichotomized about the median (n = 47). Patients surviving <3 months were excluded. g Overall survival for patients with esophageal adenocarcinoma in the present UPLC-MS/MS tissue aldehyde dataset, dichotomized about the median (n = 52). Patients surviving <3 months were excluded. h GISTIC analysis of single-nucleotide polymorphism array data from The Cancer Genome Atlas (n = 89). Blue indicates shallow deletion red indicates copy gain. i Representative fluorescent in situ hybridization (FISH) image of EAC, bar indicates 25 μm (false colors used). jALDH3A1/2 FISH signal counts compared to ALDH3A1 and ALDH3A2 expression (cut-off immunoscore >1 used, n = 145, P = 0.0135 for ALDH3A2). k Correlation of TP53 FISH signals to ALDH3A1/2 FISH signals. One dot = one patient (n = 134). Pearson’s test. Mean ± SEM provided, two-tailed Mann–Whitney U-test used unless otherwise stated, *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided in the Source Data file.